Dov Bahat

Avinoam Rabinovitch
Vladimir Frid
Tensile Fracturing in Rocks
Tectonofractographic and Electromagnetic
Radiation Methods
Dov Bahat
Avinoam Rabinovitch
Vladimir Frid

Tensile Fracturing
in Rocks
Tectonofractographic and
Electromagnetic Radiation Methods
With 302 Figures
Authors

Prof. Dov Bahat

Department of Geological and Environmental Sciences
Ben Gurion University of the Negev, POB 653
Beer Sheva, 84105 Israel
Head, the Deichmann Rock Mechanics Laboratory of the Negev
e-mail: bahat@bgumail.bgu.ac.il, phone: +972-8-6472623, fax: +972-8-6472997

Prof. Avinoam Rabinovitch

Physics Department
Ben Gurion University of the Negev, POB 653
Beer Sheva, 84105 Israel
The Deichmann Rock Mechanics Laboratory of the Negev
e-mail: avinoam@bgumail.bgu.ac.il, phone: +972-8-6461172

Dr. Vladimir Frid

Department of Geological and Environmental Sciences
Ben Gurion University of the Negev, POB 653
Beer Sheva, 84105 Israel
The Deichmann Rock Mechanics Laboratory of the Negev
e-mail: vfrid@bgumail.bgu.ac.il, phone: +972-8-6461770

Library of Congress Control Number: 2004108439

ISBN 3-540-21456-9 Springer Berlin Heidelberg New York

This work is subject to copyright. All rights are reserved, whether the whole or part of the material
is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitations,
broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication
of this publication or parts thereof is permitted only under the provisions of the German Copyright
Law of September 9, 1965, in its current version, and permission for use must always be obtained from
Springer. Violations are liable to prosecution under the German Copyright Law.

Springer is a part of Springer Science+Business Media

springeronline.com
© Springer-Verlag Berlin Heidelberg 2005
Printed in Germany

The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant pro-
tective laws and regulations and therefore free for general use.

Cover design: Erich Kirchner, Heidelberg

Typesetting: Büro Stasch (stasch@stasch.com), Bayreuth
Production: Luisa Tonarelli
Printing: Mercedes-Druck, Berlin
Binding: Stein & Lehmann, Berlin

Printed on acid-free paper 32/3141/LT – 5 4 3 2 1 0

Preface

‘Tensile Fracturing in Rocks’ presents field observations on fracturing of sedimen-

tary rocks and granite outcrops from various provinces in three continents. It also
combines results of recent experiments conducted at different laboratories around the
world with current theories on fracturing. In treating faults, this book limits itself to
faults that are associated with joint sets produced by definable causes and occasion-
ally to cases where interaction between the two types of fracture – faults and joints –
is not clear. The book’s subject matter is divided over six chapters, which are briefly
described below.
Chapter 1 summarizes current key concepts in fracture physics. It starts with a pres-
entation of the elastic theory of fracture, and concentrates on the results of linear elas-
tic fracture mechanics. The chapter touches also upon other fracture properties, e.g.,
crack nucleation, dynamic fracturing and slow fracturing processes. Nucleation is ad-
dressed by statistical mechanics methods incorporating modern approaches of ther-
mal and fiber bundle processes. The analyses of dynamic fracturing and slow fractur-
ing focus on the differences, as compared to the linear elastic approach. The contro-
versy in interpreting experimental dynamic results is highlighted, as are the surface
morphology patterns that emerge in fracturing and the non-Griffith crack extension
criterion in very slow fracturing processes.
Chapter 2 concentrates on basic topics in fracture geology. Fractography is applied
to improve the understanding of some intricacies, such as nucleation and fracture
growth in rocks. A review of fourteen lab experiments and a numerical experiment is
followed by a discussion on the properties of mirror planes and fringes on joint sur-
faces. These are exemplified on sedimentary and granitic rocks. Concepts of conjugate
hybrid joints and longitudinal splitting are re-evaluated in the light of recent experi-
ments. Various aspects of joint classification are reviewed. The chapter’s last section (2.4)
relates to mutual relationships of primary and secondary fractures, such as wing cracks,
pinnate joints and branching fractures.
Chapter 3 presents observations on sedimentary rocks in four fracture provinces:
Two in the USA, one in south Britain and one in south Israel. Although not represent-
ing all types and styles of fracturing, the combined data allow a synthesis of joint for-
mation in different tectonic settings.
Chapter 4 presents four fracture provinces in granites (in Germany, the Czech Re-
public and in California) that exhibit a wide spectrum of fracture settings. The
syntectonic, uplift and post-uplift joint features resemble their equivalents in sedimen-
tary rocks. However, granites also display unique fracture features that are associated
with late cooling stages of the pluton.
VI Preface

Chapter 5 deals with a new method of investigating embryonic stages of rock fail-
ure by measuring high-frequency electromagnetic radiation (EMR) from micro-cracks.
We review previous EMR investigations, present our new EMR model and show the
extracted EMR ‘fingerprints’ in rocks and other materials when strained. EMR sig-
nals from laboratory tests as well as from quarry blasting are analyzed. Underground
EMR preceding rock-bursts and gas outburst are studied, and some criteria for their
forecasting based on the obtained parameters are derived. The EMR results described
here are expected to lead to a novel forecasting method of earthquake nucleation.
Chapter 6 addresses some particular problems in fracture geology, including pro-
cedures for the study of joints. A discussion on previous paleostress studies in the
Middle East, particularly in the Israeli Negev, leads to a model on regional jointing in
the broad context of plate tectonics. This chapter contains a new approach in study-
ing joint length and a brief account of the procedure of estimating the fracture
paleostresses on a joint surface by a fractographic method. Also included is a detailed
characterization of the unique fracture geometry of a fault termination zone. Addi-
tional topics are, the description of a variety of fault-joint geometric and genetic rela-
tionships and a characterization of a new joint category, termed ‘surface joints’. We
also take account of hydro-geological research that relies on fracture distribution data
in the Beer Sheva syncline. Finally, we present here major features of a new mechani-
cal model of en echelon segmentation that differs from the more conventional ideas.
This book is a sequel to a previous volume in this series, ‘Tectonofractography’ (1991),
and many premises are not re-introduced here. Some terms are introduced in a special
section on nomenclature.
Acknowledgements

The authors wish to thank all those who contributed to the writing of this book. Sev-
eral persons contributed immensely to the field work and data collection. These in-
clude Peter and Elfriede Bankwitz who collaborated in the study of fracturing in gran-
ites in central Europe, particularly, the pre-uplift fracture during the cooling of the
pluton. Jointing in sandstone (Zion National Park, Utah) and in granites (the Sierra
Nevada Batholith, California) was studied in cooperation with Ken Grossenbacher and
Kenzi Karasaki. Some of the initial observations that were made with them on post-
uplift joints and oblique fractures during my (DB) Sabbatical year at LBNL matured
during the years to concrete data, and they should share the credit for it. Section 4.12
was mostly contributed by Ken Grossenbacher. Important help was also given by Paul
Witherspoon and Sally Benson. This book contains numerous citations from works by
Terry Engelder and his associates that contributed greatly to the understanding of
jointing. Working with Tsafrir Levi on his thesis has been a delight for one of us (DB).
Julia Goldbaum enriched considerably our understanding of EMR phenomena. We also
acknowledge the important contributions by Yoav Avni and Menachem Friedman, and
instructive field trips with Yehuda Eyal and Ram Weinberger. Charles Howard contrib-
uted greatly to the writing of Sect. 6.8.
Special thanks are due to colleagues who critically reviewed chapters in draft: Chap-
ter 2, Ram Weinberger; Chapter 3, Ithamar Perath; Chapter 4, Reginald Shagam; Chap-
ter 6, Meir Abelson and Ronit Nativ.
Figures 2.35e and 3.35c were donated by Ran Kagan and Alon Raviv, respectively, and
Fig. 2.69b and 3.10e were photographed during inspections of quarries with Peter and
Elfriede Bankwitz.
Most of our experiments were conducted in the Ruth and Horst Deichmann Rock
Mechanics Laboratory of the Negev, the Department of Geological and Environmental
Sciences, the Ben Gurion University of the Negev.
For important technical help, we owe thanks to Vyacheslav Palchik, Dina, Isaak
and Michael Frid, Julia and Boris Goldbaum, Avi Sharon, Leonid Baluashvili, Eli
Shimshilashvili, Rivka Eini, Ran Novitsky, Hagit Lev, Ania Arbel, Daniela Vida, Michal
Barima, Ronen Shavit from BGU, and Pascual Benito from LBNL. We also owe thanks
to H. Hertrich from Solnhofer Bodenplatten-Industrie GmbH and Co KG for help in
our experiments.
The help extended by members of the department of the Geological and Environ-
mental Sciences was crucial to the completion of this book. Financial assistance was
received from the Earth Sciences Administration (Ministry of Energy and Infrastruc-
VIII Acknowledgements

ture), and thanks are due to Yosef Bartov for his interest and assistance. Support was
also given by the Israel Science Foundation, the Ezvonot Committee of the Israeli Gov-
ernment Estate Distributions and by the Vatat and NRCN funds.
Many authors and publishers contributed valuable photographs and diagrams upon
our request, enabling us to fundamentally upgrade our book. All readers are invited to
write to us about errors that they find in the book.
The ideas presented in this book were beginning to crystallize when a sudden dete-
rioration in my (DB) wife Hanna’s health overshadowed all literary thinking. Dan and
Merav, Zvi and Jonathan were with us during her last months of suffering. This book is
dedicated to her memory.
Contents

1 Fracture Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 The Griffith Criterion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Stress and Strain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.2 Energy Considerations and the Griffith Criterion . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.3 Fracture Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 Nucleation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.2 Stress Concentration Avenues to Microcrack Nucleation . . . . . . . . . . . . . . . . . 11
1.2.3 Thermal Nucleation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.2.4 Fiber Bundle Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.3 Linear Elastic Fracture Mechanics (LEFM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.3.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.3.2 Equations of Equilibrium and of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
1.3.3 Two-dimensional (2D) Problems and Stress Intensity Factors . . . . . . . . . . . 35
1.4 Dynamic Fracture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
1.4.1 Mott and Mott-like Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
1.4.2 2D Dynamic Crack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
1.4.3 Energy Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
1.4.4 Fracture Pattern Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
1.5 Subcritical Cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
1.5.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
1.5.2 Creep . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
1.5.3 The Dependence of Jointing in Rock Layers Solely on Stress . . . . . . . . . . . . . 63

2 Elements of Fracture Geology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
2.2 Fractography and Tectonofractography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
2.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
2.2.2 The Origin: Crack Nucleation via Fractographic Techniques . . . . . . . . . . . . 83
2.2.3 Fractographic Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
2.2.4 Joint Mirror Planes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
2.2.5 Joint Fringes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
2.2.6 Mirror Boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
2.2.7 Recent Studies on En Echelon Fringes from the Beer Sheva Syncline . . 141
2.2.8 Straight Kinks and Curved Kinks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
2.2.9 Recent Studies from the Appalachian Plateau . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
X Contents

2.2.10 On the En Echelon Fringe Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

2.2.11 Summary of Five Categories of Breakdown Styles . . . . . . . . . . . . . . . . . . . . . . . . 160
2.2.12 Recent Studies from the South Bohemian Pluton . . . . . . . . . . . . . . . . . . . . . . . . . 161
2.2.13 Oblique Extension Joints and Longitudinal Splits (or Splittings) . . . . . . 166
2.3 On Uplift, Post-Uplift, Neotectonic and Surface Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
2.3.1 Neotectonic Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
2.3.2 Exfoliation Joints and Sheet Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
2.3.3 Surface Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
2.4 Primary and Secondary Fractures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
2.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
2.4.2 The Wing Crack Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
2.4.3 Pinnate and Horsetail Fractures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
2.4.4 Fracture Polarization in the Two Extensional Quadrants
along the Fault . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
2.4.5 Termination-Zone Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
2.4.6 Fracture Branching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

3 Four Fracture Provinces in Sedimentary Rocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191

3.1 The Devonian Fracture Province of the Appalachian Plateau, New York . . . . . 191
3.1.1 General Geology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
3.1.2 The Tectonic Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
3.1.3 Sequence of the Dip Joint Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
3.1.4 Progressive Development of the Regional Stress Field . . . . . . . . . . . . . . . . . . . 201
3.1.5 Tectonic History of the Appalachian Plateau Detachment Sheet . . . . . . . 203
3.1.6 Three Main Concepts of Regional Cross-Fold Joint Patterns . . . . . . . . . . . 208
3.2 Fracturing in The Bristol Channel Basin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
3.2.1 General Geology and Previous Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
3.2.2 Fracture Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
3.2.3 Data Collection and Joint Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
3.2.4 Joint Sets in the Blue Lias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
3.2.5 Stresses in the South-Dipping Limestone Beds
during Alpine Inversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
3.2.6 Conclusions by Engelder and Peacock (2001) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
3.2.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
3.3 Post-Uplift Exfoliation Joints and Arch Formation in Sandstone
at Zion National Park, Utah, USA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
3.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
3.3.2 Fracture Characteristics at Zion National Park . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
3.3.3 Fracture Mechanisms at Zion National Park . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
3.3.4 Discrimination between Sheeting and Exfoliation . . . . . . . . . . . . . . . . . . . . . . . 250
3.3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
3.4 Jointing and Faulting in Eocene Chalk Formations
in the Beer Sheva Syncline, Israel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
3.4.1 The Beer Sheva Syncline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
3.4.2 Fractures in the Beer Sheva Syncline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
3.4.3 Fracture Diversity in the Beer Sheva Syncline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
 Contents XI

3.4.4 The Structural Interpretation of En Echelon Fringes

in the Beer Sheva Syncline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
3.4.5 Fold-Crossing Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
3.4.6 Factors Influencing Fracture Diversity in the Beer Sheva Syncline . . . . 273
3.4.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
3.4.8 Open Questions and Future Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
3.5 Comparison of Unrelated Fracture Provinces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
3.5.1 Rock Age, Lithology and Age of Fracturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
3.5.2 Genetic Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
3.5.3 Changes in Stress Fields, Structural Complexity and Overburden . . . . 280

4 Jointing in Granites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283

Part 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
4.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
4.2 The Cloos Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
4.3 Fracture in Granites from the North Bohemian Massif
in the Lusatian Granodiorite Complex and the Erzgebirge . . . . . . . . . . . . . . . . . . . . . 287
4.3.1 The Lusatian Granodiorite Complex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
4.3.2 The Erzgebirge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
4.4 Joints in Granites from the South Bohemian Pluton . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
4.4.1 Geological Setting of the South Bohemian Pluton . . . . . . . . . . . . . . . . . . . . . . . . 294
4.4.2 Fractographic Versatility on Pre-Uplift Joints in the Borsov Quarry . . . 296
4.4.3 “Dynamic Joints” in a Fracture Zone from the Mrákotin Quarry . . . . . . 307
4.4.4 Twist and Tilt Angles on Joints Cutting Granites in Europe . . . . . . . . . . . . . 313
4.4.5 Geological Interpretation of the Joint Fractographies
in the Borsov and Mrákotin Quarries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
4.4.6 Timing, Depth and Temperature of Joint Formation . . . . . . . . . . . . . . . . . . . . . 319
4.5 Joints in Granites from the Sierra Nevada Batholith in California . . . . . . . . . . . . . 322
4.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
4.5.2 Petrographic Study of Two Oblique Joints
from the Knowles Granodiorite in Western Sierra Nevada . . . . . . . . . . . . . . 324
4.5.3 Joint Exfoliation and Arch Formation
at Yosemite National Park, California . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
4.6 Fractographies that Pertain to both Quasi-Static and Dynamic Fractures . . . 348
4.7 Comparative Jointing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
4.7.1 Cooling Joints in Granites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
4.7.2 A Comparison to the Cloos Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
4.7.3 Joint Genetic Grouping in Sedimentary Rocks and in Granites . . . . . . . . 355

Part 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
4.8 Analysis of Fracture Velocity versus Stress Intensity Factor
in the Borsov Joints Based on Fractographic Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
4.8.1 The Velocity versus Stress Intensity Factor Curve . . . . . . . . . . . . . . . . . . . . . . . . 356
4.8.2 Application of Fractography and Fracture Mechanics
to the Study of Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
XII Contents

4.8.3 The Velocity versus Stress Intensity Curve for Granite . . . . . . . . . . . . . . . . . . 357
4.8.4 Methodology for Plotting Joints onto Velocity
versus Stress Intensity Factor Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
4.8.5 Plotting Joints of the Borsov Pluton . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
4.8.6 Limitations of this Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
4.9 Velocity and Stress Intensity Manifestations of Fracture Propagation
in Granites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
4.9.1 Fracture Velocities of Exfoliation Joints
Cutting the El Capitan Granite and Comparable Fractures . . . . . . . . . . . . . 363
4.9.2 Fracture Velocities Relative to Estimated Paleostress Magnitudes . . . . 363
4.9.3 Cycling Fracture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
4.10 New Fracture Area in the En Echelon and Hackle Fringes
on Joint Surfaces from the Mrákotin Quarry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
4.10.1 Fracture Area Increase in the Transition from En Echelon
to Hackle Fringe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
4.10.2 Estimation of the Fracture Areas in the Fringes of Joints A and C . . . . 368
4.10.3 A Comparison between the Fringes of Joint C and Joint A . . . . . . . . . . . . . . 371
4.10.4 How was Energy Released Differently in the Two Fringes? . . . . . . . . . . . . . 371
4.10.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372
4.11 The Index of Hackle Raggedness (IHR) on Joint-Fringes
of the Mrákotin Joint Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372
4.11.1 Criteria of IHR and Application to the Mrákotin Joint Set . . . . . . . . . . . . . . 373
4.11.2 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
4.11.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
4.12 A Method of Drilling in situ Fracture from the Rock . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377

5 Electromagnetic Radiation Induced in Fractured Materials . . . . . . . . . . . . . . . . . . 379

5.1 History of EMR Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380
5.2 Foundations of EMR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
5.3 EMR Model Verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
5.3.1 Experimental Arrangement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
5.3.2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
5.3.3 Amplitude Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
5.3.4 Investigation of the Time up to Envelope Maximum
and its Relation with Frequency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392
5.3.5 Frequency Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
5.3.6 T' / ω Investigation (Eq. 5.8) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
5.3.7 Decay Time Investigation (Rabinovitch et al. 2003a) . . . . . . . . . . . . . . . . . . . . . 400
5.3.8 Summary of the EMR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
5.4 Comparison of the New EMR Model with the Previous Ones . . . . . . . . . . . . . . . . . . 404
5.4.1 Dislocations and Charged Electrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404
5.4.2 Discharge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
5.4.3 Movement of Fracture Tips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406
5.4.4 Movements of Fracture Sides (the “Capacitor” Model) . . . . . . . . . . . . . . . . . . 406
5.4.5 Summary of the New Model and its Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
5.5 EMR Pulses Induced by Rock Fracture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
 Contents XIII

5.5.1 Previous EMR Pulse Investigations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408

5.5.2 Our EMR Pulse Classifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
5.6 EMR and Material Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
5.7 EMR during Percussion Drilling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
5.7.1 The 4th Group and Polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
5.7.2 Fractographic Examination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
5.7.3 Brief Theoretical Consideration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
5.7.4 Examination of ‘b’ Type Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
5.7.5 The Polarization Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425
5.8 Intermediate- and Large-Scale EMR Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
5.8.1 Blasting Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
5.8.2 EMR Prior to EQ and Volcanic Eruptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
5.9 Global Features of EMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
5.9.1 EMR versus Gutenberg-Richter Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
5.9.2 Benioff Strain Release . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
5.10 EMR Induced by Underground Rocks under Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
5.10.1 General Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
5.10.2 EMR in Mines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
5.11 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458

6 Assorted Problems in Fracture Geology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459

6.1 The Importance of Studying Regional Jointing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
6.2 The Procedures of Studying Systematic Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
6.2.1 Qualitative, Measurements, Classification, Plotting
and Division into Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
6.2.2 Statistical Analysis in Support of the Division of Joints into Sets . . . . . . 463
6.2.3 Mapping Joint Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
6.3 Paleostresses in the Israeli Negev . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
6.3.1 Jordan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
6.3.2 The East Mediterranean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
6.3.3 The Sinai-Israel Sub-Plate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
6.3.4 Two Synclines in the Northern Negev . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
6.3.5 The Neqarot Syncline in the Central Negev . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
6.3.6 Two Quasi-Synclines in the South Negev . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478
6.3.7 Summary of Results that Evidence SHmax Clockwise Rotation . . . . . . . . . . 479
6.3.8 Beyond the Mediterranean Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480
6.3.9 Open Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481
6.4 Analysis of the “Fat Tail” of Joint Length Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . 484
6.4.1 Joint Length Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484
6.4.2 The “Fat Tail” Phenomenon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
6.4.3 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
6.4.4 The Fractal Nature of the Tail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
6.4.5 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 490
6.5 Estimation of Paleo Fracture-Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492
6.6 Fracture Variability in a Fault Termination Zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
6.6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
XIV Contents

6.6.2 Description of the Fault Termination Zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494

6.6.3 Fracture Parametrization in the Fault Termination Zone . . . . . . . . . . . . . . . 500
6.6.4 Open Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503
6.7 Fault-Joint Relationships in the Beer Sheva Syncline, Israel . . . . . . . . . . . . . . . . . . . . 504
6.7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504
6.7.2 Eight Types of Fault-Joint Relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505
6.7.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
6.8 The Path from Geological Joints to Soil through a Transitional Layer
of Surface Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
6.8.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
6.8.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514
6.8.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516
6.9 Hydro-Geological Research in the Beer Sheva Syncline . . . . . . . . . . . . . . . . . . . . . . . . . 516
6.9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516
6.9.2 Summary of the Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 518
6.9.3 Open Questions Related to Water Flow
in Heterogeneous Fracture Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524
6.10 A New Fringe Characterization and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525

Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529
Physical Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529
Nomenclature of Main Fractographic Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 530
Nomenclature of Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
Selected Supplementary Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 557
Chapter 1

Fracture Physics

In this chapter we give a brief account of the theory of fracture, restricting ourselves
to brittle fracture. The structure of this chapter is as follows. The first part describes
the Griffith criterion of fracture initiation, since this criterion constituted the turning
point in our understanding of the basics of fracture. We next consider the question of
nucleation both of flaws and of fractures in a solid. This process has recently had a
breakthrough and is related to the origin of fracture initiation. The general elastic
theory of fracture is considered next – linear elastic fracture mechanics (LEFM) as
the only complete theory and the rational extension of the Griffith approach. The analy-
sis of two-dimensional problems, for some of which analytical solutions can be ob-
tained, is emphasized. The last parts deal with two important topics that do not fall
within and can not be dealt with by LEFM, namely, dynamic crack propagation and
creep. In both of these topics, new advances have been made in recent years and both
are therefore a subject matter for contemporary research.
A note of caution is in order. Although the Griffith criterion and the related linear
elastic fracture mechanics methods have helped scientists in dealing with static cracks
and engineers in building safer structures, the basic dynamics of the fracture prob-
lem are as of yet not well understood. This relative ignorance stems from the fact that
on the one hand, fracturing is neither an equilibrium nor a near equilibrium process
and is therefore not amenable to perturbation procedures, and on the other hand, frac-
turing, whose principle part occurs at a very restricted area and is highly nonlinear,
influences the whole extension of the body in question, and it is therefore not easy to
correlate between its different scales.
A full understanding of crack propagation would entail a quantum mechanical
calculation of all atoms in a certain vicinity of the crack. Since each atom contains
several electrons, a full “many-body” calculation (see, e.g., Fabrocini et al. 2002) should
be made. Obviously, even numerically, this procedure is unachievable in the foresee-
able future. A less presumptuous attempt would be to consider each atom as a point
entity and to calculate fracture propagation by assuming a certain interaction among
these entities. Even a classical (rather than a quantum mechanical) calculation would
have to assume an approximation of only two or at most three body forces, and it would
have to assume, moreover, that these two body forces (say of a Lennard Jones type;
see, e.g., Ashcroft and Mermin 1981) can be extended to inter-atomic distances far
greater (or smaller) than equilibrium ones. Even under such approximations, the num-
ber of atoms and the time period necessary for a useful calculation are as of yet be-
yond the scope of available computational means. Thus, the maximum number of at-
oms for which such numerical calculations have been tried is the boasted 109 (Abraham
2 Chapter 1 · Fracture Physics

et al. 2002). However, for a meaningful dynamical calculation a number about a thou-
sand times higher is needed. As for the time periods needed, these should be scaled
by atomic vibrations, which entail at least around 106 time steps even for such a short
time as 1 µs, rendering this task unattainable at the present stage.
There are at present several groups of scientists around the world who try to “con-
nect” the different scales that appear in fracture calculations. The idea is to use a quan-
tum mechanical type of calculation in order to obtain the correct two-body atomic
force even for large inter-atomic distances. These forces are to be used only for a rela-
tively small number of atoms near the crack tip and the so called “process zone” to
obtain an accurate microscopic (classical) calculation. The latter should yield inter alia
the energy needed for crack advance, the stresses and strains near the tip and the ap-
pearance and motion of dislocations for non-brittle materials. The boundary condi-
tions for this calculation would be the connection to the major part of the material,
which is to be treated in a continuous manner (linear or non-linear fracture mechan-
ics). This ambitious plan is still in its infancy but promises to become a real asset in
understanding fracture dynamics in the future.

1.1
The Griffith Criterion
1.1.1
Stress and Strain

Consider the bar in Fig. 1.1 whose initial length and cross section are l0 and A = b × b,
respectively. We apply a force in the x-direction, which causes the bar to elongate by x.
We define σxx = f / A, as the stress operating inside the bar in the x-direction. Hence,
stress has the units of force per area. It is assumed that the bar elongates with the contin-
ued application of the force in such a way as to preserve elasticity, namely that it oper-
ates like a spring, maintaining a linear relationship between the applied stress f / A and
the elongation, x, f = kx (Hook’s law). This relationship together with Newton’s law
f = ma are the basis of all of elasticity theory (see, e.g., Timoshenko and Goodier 1984).
Defining the non-dimensional variable εxx = x / l0 as the strain in the bar (the change
in length divided by the initial length), one can write Hook’s law as

(1.1)

Fig. 1.1.
Schematic stress-strain behav-
ior of a bar (l0 × b × b) under
uniaxial force f
 1.1 · The Griffith Criterion 3

where E is Young modulus, related in the case of the bar, to the Hook’s constant k through

(1.2)

Under the same stress, there also occurs a (small) reduction of size of the bar in
the lateral directions. Thus, if in the y-direction the decrease of width is denoted by
–y (the minus sign comes to denote a reduction of size), the strain in this direction is
εyy = –y / b and for elastic materials is given by

(1.3)

where ν is a non-dimensional constant called the Poisson ratio, which for different
materials usually assumes values between 0.1 and 0.5. A similar reduction occurs in
the z-direction (εzz).
An important issue to be used frequently in the following is the value of the elastic
energy stored in the material. For the bar problem, suppose we apply the force quasi
statically to avoid acceleration and kinetic energy and increase its magnitude from zero
to a final value F. The elastic energy is given by the work needed to create the final state:

(1.4)

where ∆l is the final elongation of the bar, εf = ∆l / l0 is the final strain, σf is the final
stress (= Eεf) and V is the initial bar volume. We have therefore obtained the important
result that the elastic energy per unit volume (W / V) stored in the sample is given by

(1.5)

Although this result was obtained for the specific case of a bar acted upon by a uniaxial
force, this result is quite general and we will use it later, e.g., for the Griffith criterion.
In the general case, i.e., for a body of a general shape acted upon by a general distribu-
tion of forces and displacements, elasticity theory considers an infinitesimal cube inside
the body and the force components per unit area on each side of the cube constitute the
stress tensor σij, where i = 1, 2, 3 and j = 1, 2, 3 denote the x-, y- and z-coordinates, respec-
tively. Thus, for example σ12 = σxy denotes the force (per unit area) component in the y-
direction acting on the cube face, which is perpendicular to the x-direction. This tensor can
in general depend both on the location of the cube within the body, r, and on time. There-
fore, σ is generally written as σij(r, t). It can be shown that σ is a symmetric tensor, namely

(1.6)

The force per unit area acting on a general surface perpendicular to a unit vector
n$ = n1x$ + n2 y$ + n3z$ (x$ , y$ , z$ are unit vectors in the x-, y- and z-directions respectively)
is called the traction t and its components are given by
4 Chapter 1 · Fracture Physics

(1.7)

Next, displacements from equilibrium position are defined such that under the
action of forces, a point within the body initially (with no force) located at the posi-
tion (x, y, z) is displaced to the position (x + u1, y + u2, z + u3) by the action of the forces.
The displacement vector is thus denoted by u = u1x$ + u2 y$ + u3z$ . Newton’s equation of
motion (force = mass × acceleration) is transferred into the infinitesimal cube to yield
(per unit volume)

(1.8)

where the first term on the left denotes the effect of the surface forces (by measuring
their difference in the i-direction between the two sides of the cube, while fi is the
ith component of the so-called “body force” if such a force exists (e.g., gravity when it
is relatively large compared with surface forces and can not be ignored) and ρ is the
density (mass per unit volume) at the position (x, y, z). Equation 1.8 is the basic equa-
tion of motion of an elastic medium and all further linear elastic considerations are
conducted in order to solve it. An important special case of Eq. 1.8 is static equilib-
rium, where all movements are zero. In that case,

(1.9)

or, when no body forces exist,

(1.10)

In order to solve Eq. 1.8, one should have a way to connect σ with u. This is done
via the generalized Hook’s law in the following way. Firstly, a strain tensor εij is de-
fined as

(1.11)

As can be seen, ε is also a symmetric tensor, i.e., εij = εji. The most general Hook’s
law can now be written as a linear relation between σ and ε,

(1.12)

where Cijkl is a constant tensor of the 4th rank replacing the Young modulus of Eq. 1.1.
It is a material property. For isotropic homogeneous materials, Cijkl has only two
components λ and µ (called the Lamé constants) different from zero and the law can
be written as
 1.1 · The Griffith Criterion 5

(1.13)

where δij is the Kronecker δ,

(1.14)

For example, σxz = σ13 equals 2µε13 while σzz = σ33 = 2µε33 + λ(ε11 + ε22 + ε33). The
Lamé constants are connected to the Young modulus and the Poisson ratio ν through

(1.15)

Now, the right hand side of Eq. 1.8 can be differentiated with respect to xi and xj to
relate it through Eq. 1.11 with εij, and the left-hand side of Eq. 1.8 in its turn can be
transformed by Eq. 1.12 or 1.13 to be given in terms of εij (as well). In this way, the
ensuing equation will have similar variables on both sides. There is, however, a prob-
lem arising from the connection between the displacements and the strain (Eq. 1.11).
Although the displacement u (a vector with three components) uniquely defines εij
(a symmetric tensor with six components), the εij does not uniquely define the dis-
placements. There is a redundancy. In order to insure a unique u, the strain compo-
nents must satisfy an additional criterion (called the compatibility equations, see also
Sect. 1.3.3.1)

(1.16)

The mathematical complexity of this set of equations (Eqs. 1.8–1.16) both for the
dynamic but even for the static cases makes exact analytic solutions for elastic prob-
lems almost unobtainable. Exceptions are problems that due to their symmetry can
be reduced to simpler forms. The most common solvable problems are “two-dimen-
sional” (2D) ones, where the influence of the third dimension is ignored. Two types of
such problems are of importance. In the first, we deal with a thin plate, where it is as-
sumed that the thickness is much smaller than the other dimensions and can be ig-
nored (plane stress); and in the second a very thick specimen is considered where the
third dimension is ignored (plane strain) under the condition that no change of the
applied force occurs across the thickness of the plate and the whole system is homo-
geneous and therefore solutions do not change with the coordinate in this direction.

1.1.2
Energy Considerations and the Griffith Criterion

Fractures occur in solids at much lower stresses than those which can be deduced from
simple estimations. For example, if we assume that the solid breaks when the distance
between its atomic planes become elongated by, say, 1 / 10 of the equilibrium distance,
this would mean that ε ~ 1 / 10 and σ = Eε ~ E / 10. Most solids break under loads that
6 Chapter 1 · Fracture Physics

are two or three orders of magnitude lower than this value. This discrepancy was re-
solved by Griffith (1920). In principle his argument is that the stress is not transferred
to the solid homogeneously. Rather, due to the existence of flaws in the body (and al-
most all solids contain flaws, unless grown in a very special method – controlled whis-
ker procedure), the stress becomes concentrated at the tip of such a flaw. Hence, strains
and stresses at this location become much larger than those prevailing in the rest of
the material – leading to fracture there. In fact, it is the longest flaw (favorably ori-
ented), which sets the limit for the critical stress, the stress above which fracture would
occur, as is shown presently.
Consider (Fig. 1.2), a large plate of thickness b and cross section A, which is oper-
ated on by a constant force f or a stress σ0 (= f /A) and in which there exists a flaw (frac-
ture) of length c. The fracture evidently has caused the elastic energy in its vicinity to
be relieved. We can estimate the volume wherefrom the elastic energy was relieved to
be approximately equal to the half cylinder shown.
This volume is

(1.17)

And according to Eq. 1.5, the total relieved energy is

(1.18)

Therefore, should the fracture propagate by an additional increment ∆c, the change
in relieved energy would be

Fig. 1.2.
A through crack in a semi infi-
nite slab and the (schematic)
region in which energy is re-
laxed
 1.1 · The Griffith Criterion 7

(1.19)

By the way, the energy per unit length ∆WE / ∆c is usually denoted by the (not very
fortunate) term “energy release rate” G (in honor of Griffith).
According to Griffith, the energy thus released from the elasticity stored reservoir
should go in its entirety (or almost so) to create the areas of the new addition to the
fracture surface. If we assume that the energy needed for the creation of a new unit of
fracture area, Γ , is a material constant (it is in fact half of the energy needed to break
the chemical bonds between the two newly created surfaces), then the energy required
for an increment ∆c of the fracture is

(1.20)

The factor of two comes because two new surfaces are created above and below the
increment. Energy balance entails that the change in the released increment of elastic
energy be equal to the increment of surface energy, i.e.,

yielding

(1.21)

A similar calculation applies for an internal flaw (crack). Assume that this crack is
circular of radius c. The volume from which the elastic energy was released is

and therefore in this case

where σ0 is the remote stress perpendicular to the flaw. The surface energy needed to
create this flaw is Ws = 2pc2Γ. Griffith’s incremental energy balance,

yields

(1.21')
8 Chapter 1 · Fracture Physics

Although the method used here to obtain Eq. 1.21 or 1.21' is somewhat crude, the
result, except perhaps for the coefficient, is accurate. This is Griffith’s condition. Let
us first calculate the value of σ0 needed to initiate a crack and compare it to the ho-
mogeneous estimate of E / 10 we had before. The decrease with respect to E / 10 of σ0
needed for fracture commencement can be seen as a measure of the enhancement of
stress at the crack tip relative to σ0. From Eq. 1.21,

(1.22)

Therefore, the magnitude of q is a measure of stress increase (relative to σ0) at the

tip of a flaw of length c.
Let us examine soda lime glass as an example. The relevant constants for glass are
E ~ 6.2 × 1010 Newton m–2 and Γ ~ 0.54 Joule m–2, hence for glass q ~ 3.4 × 105 √ ⎯c . This
relation is plotted in Fig. 1.3. It is seen that our first estimate of q, q ~ 10 is obtained
only for flaw lengths of the order of atomic distances, while for usual flaw lengths of
the order of 0.1 mm, q is of the order of several thousands.
According to the Griffith criterion, Eq. 1.21, therefore, the important experimental
variable “responsible” for crack initiation is the product σ02c of σ0 √⎯c and not σ0 itself.
When this variable reaches a certain threshold, which depends on the specific material
and on the geometry of the loading system, fracture should commence. This threshold
according to Eq. 1.21 depends on the Young modulus and the surface energy of the frac-
turing sample. As will be shown next (Sect. 1.3), a more elaborate and accurate treatment
of linear elasticity (called linear elastic fracture mechanics (LEFM) an established branch
of mechanical engineering nowadays) yields the same result as Eq. 1.21, up to a change
of coefficient, for uniaxial constant loading. This LEFM treatment has been used for many
years now to achieve solutions for engineering static safety problems and has proven
extremely useful for that purpose. Basically, it is an extension of the Griffith approach.
As we shall point out, however, this approach, so successful for regular static cracking
problems, is inadequate when dealing with the following two important issues:

1. Fracture development under low stresses in very long time periods (the so-called
“creep” situation).
2. Dynamic crack propagation.

We dedicate a section for each of these problems.

Fig. 1.3.
The amount q by which the
stress σ0, needed to cause frac-
ture, is decreased from E as a
function of the Griffith-frac-
ture length c
 1.2 · Nucleation 9

Fig. 1.4. The three fracture modes

1.1.3
Fracture Modes

Cracks can appear in solids under stress only in three different geometrical situations, called
modes, or in situations that are combinations of these modes. Figure 1.4 shows these three
geometries called mode I, mode II and mode III. Almost all engineering treatments of frac-
ture problems are developed in a mode I geometry, which is the prevalent mode of frac-
ture. The presence of mode II and III in addition to mode I is manifested by changes of crack
directions as in the appearance of lancets, striae or plumes, undulations etc. In geological
cracks, we discern between “joints” that are created in a mode I geometry and “folds” that
are created under the other mode types. In the description of all three modes (Fig. 1.4), the
fracture surface is in the x-z-plane and the fracture propagates in the x-direction. In mode I,
stress (tension or compression) is applied in the ±y-direction and the displacements u also
occur in the same direction. Modes II and III are shear modes. In mode II, the shearing stress
and the displacements are in the same direction (±x) as that of the crack propagation. Mode III
fracture is also called an “antiplane” mode, since stress and displacements occur in the
±z-directions, out of the plane of the fracture and normal to the plane of the specimen.

1.2
Nucleation
1.2.1
General

Let us first define the three different nucleation processes that are used here (see also
Sect. 2.2.2 for additional applications of the term “nucleation”):

1. “Flaw nucleation” is the process or processes by which a seed for the subsequent
creation of a propagating fracture is created. The seed is usually a flaw in the other-
wise “perfect” or homogeneous solid, which facilitates the appearance of a crack in
its continuation (see Sect. 1.1.2).
10 Chapter 1 · Fracture Physics

2. “Microcrack nucleation” is the process or processes by which a flaw (which is itself

not a microcrack) develops small cracks at the positions of highest stress concen-
trations such that
a the combined structure, flaw plus small cracks, can be considered a microcrack
for subsequent developments,
b this combined structure is stable, and
c it has not reached the critical (Griffith) condition for propagation.
3. “Crack nucleation” is the process or processes by which a critical crack (i.e., a crack
fulfilling the Griffith criterion for propagation) is created.

Crack nucleation can be “homogeneous”, i.e., with no flaws present (a rare occur-
rence) or “heterogeneous”, where a flaw distribution already exists in a solid. Note that
in the literature the term “crack nucleation” is used to describe both flaw- and crack-
nucleation processes. The terms homogeneous and heterogeneous are also used to de-
scribe materials. Thus a homogeneous material is an ideal material, which is made up
of the same chemical substance, is devoid of grains and is completely devoid of flaws.
The only existing material that is almost completely homogeneous is a glass wisker, i.e.,
a glass fiber grown under very strict conditions of purity. Since no flaws (or almost no
flaws) are created in the production phase, q of Eq. 1.22 is of the order of ten and this
material fractures only under very high stresses. Its fracturing is also “homogeneous”,
namely all parts of it separate at the same time and the material integrates into dust
under a large blast. Heterogeneous brittle materials can be divided into three groups
according to the degree of severity of the heterogeneity. The least heterogeneous mate-
rial is regular glass, containing flaws of all kinds (see below). Next in line are rocks or
ceramic materials having the same chemical composition. These materials are grainy
materials containing grain boundaries and “triple points” as inherent flaws. This group
further divides into isotropic (i.e., having the same properties in any direction) and
unisotropic. The most heterogeneous (third) group of materials are ceramics or rocks
which consist of substances having different chemical compositions. These are usually
both grainy and unisotropic. Homogeneous (heterogeneous) nucleation is a nucleation
process in a homogeneous (heterogeneous) material.
We use the term “flaw” here as a general term for defect. This is the usual meaning of
the term in the fracture literature, while, e.g., in solid state physics the common term is
“defect”. A flaw can be two-dimensional in the form of a microcrack, or three-dimensional,
such as a pore, an inclusion, a vacancy, a dislocation etc. The term “Griffith flaw” is used in
the literature generally to describe a flaw in the form of a microcrack (the common form
of flaw in glass). A three-dimensional flaw can be transformed to a microcrack by devel-
oping actual microcracks at points where high stress concentrations exist (critical or sin-
gular points; see below). Note that regular glass, besides Griffith flaws, can contain other
flaws (voids etc.) generated during its formation stages or by external environmental deg-
radation. In general, the appearance of flaws in a solid or on its surfaces can be induced by
the activity of two mechanisms, “external agents” operation or internal processes. The first
mechanism includes surface abrasions and erosion, corrosion by chemical substances,
radiation and impact effects, etc. The operation of each agent is subject matter for exten-
sive research. A short review can be found in Lawn (1993, Chap. 9). We shall not consider
here the flaws that develop during the production phase and include pores, vacancies, grain
boundaries, inclusions etc. (see ibid). In this section, we will concentrate on the internal
 1.2 · Nucleation 11

processes and specifically on microcrack and crack nucleations. Note that the use of the
term “homogeneous nucleation” in the literature is somewhat controversial. In the words
of Frost (2001), “roughly it means the appearance of a crack where there was none present
before. But the condition of there being no crack before depends on the precision of our
observation.” In physics and metallurgy, nucleation means an appearance of a new “phase”
in a position “where absolutely nothing of the sort existed before.” In fracture nomencla-
ture, this term usually means “where absolutely nothing could be seen (optically) before.”
We will have more to say on this subject when discussing thermal nucleation below.
A nucleation of even a microcrack in a perfect (homogeneous) solid would entail
stresses of the order of E / 10 (or G / 10; Sect. 1.1.2), or corresponding amounts of en-
ergy. Following the Griffith model, we look for stress concentration mechanisms that
could facilitate this process. In Sect. 1.2.2 we shall review (Frost 2001 and Evans 1974)
the possible stress concentration mechanisms capable of reducing the stress (energy)
needed for microcrack nucleation, and in Sect. 1.2.3 and 1.2.4 we will present two mod-
ern theoretical approaches to crack nucleation. The first one is “thermal nucleation”,
which has recently been verified experimentally and the second is the so-called “fiber
bundle model”, which can apparently be used for general grainy materials.

1.2.2
Stress Concentration Avenues to Microcrack Nucleation

The stress concentration mechanisms (see Frost 2001) are elastic stress concentrations
resulting from

1. Interior flaws.
2. Elastic anisotropy.
3. Thermal expansion mismatch or thermal expansion coefficient anisotropy, and
inelastic mechanisms.
4. Dislocation pile-up.
5. Grain-boundary sliding.

We shall briefly touch upon two of the elastic mechanisms. For the inelastic ones
see Frost (2001) and references therein.

1.2.2.1
Elastic Anistropy

We shall treat only heterogeneous anisotropic materials of the second group

(Sect. 1.2.1). In heterogeneous materials of the third group (which are made up of
grains of different substances), there are additional stress concentration possibilities
resulting from the mismatch of elastic constants between neighboring grains, which
can be treated in a similar way to the one presented here.
Tvergaard and Hutchinson (1988) considered stress concentrations in 2D solids. The
treatment was extended to 3D problems by Ghahremani et al. (1990), who treated
grains in hexagonal tiling where all boundaries meet at 120° – which is the case we
will consider here. Picu and Gupta (1996) extended the treatment to general 2D con-
figurations and Picu (1996, 1997) extended it to several 3D problems.
12 Chapter 1 · Fracture Physics

Fig. 1.5. Geometry used for the calculation of stress singularity due to anisotropic elasticity (Frost 2001)

Fig. 1.6.
A scanning electron micrograph
of a typical triple point void in
sintered Al2O3 (Evans 1978)

Each crystalline grain of an anisotropic solid has a “stiff ” elastic direction and a
“compliant” direction, i.e., the strain in the first direction is smaller than the one in
the second direction for the same applied stress. When adjacent to each other, grains
can have several orientation angles. Consider Fig. 1.5, which shows a rather simple
geometry, in a 2D situation. Even here there are several (five) orientations: α1, α2 (here
120° and 240°, respectively) specifying the relative orientations of the grain bound-
aries and three θ measuring the angles of the stiff directions in each grain. Picu and
Gupta (1996) calculated the stress singularity, which arises in such a geometry, in po-
lar coordinates (r, θ ) around the triple point (Fig. 1.6):

(1.23)
 1.2 · Nucleation 13

where k is a constant, γ is an exponent defining the stress singularity and φij (θ ) is a func-
tion of θ only with no singularities. The value of γ is crucial for nucleation. In order to
achieve stress concentration, γ has to be negative (in which case there exists a real
σ singularity, i.e., σ → ∞ for r → 0), and for crack nucleation to occur, γ has to be less than
–0.3 (Frost and Gupta 1993), since only then a σ singularity that is sufficiently high to over-
come atomic cohesion extends beyond atomic dimensions. Picu and Gupta (1996) consid-
ered random orientations of the crystal axes for the 120–240° triple junctions and for other
symmetric (θ1 = –θ3) situations. Specific problems should be checked there. For example,
for ice they have shown that no crack nucleation can occur from elastic anisotropy.

1.2.2.2
Thermal Expansion Mismatch or Anisotropy (Evans 1978)

Evans (1978) calculated the stress concentration, which occurs in polycrystalline

(grainy) solids due to the thermal expansion anisotropy. As for the elastic anisotropy
case (Sect. 1.2.2.1), thermal expansion singularities were found at grain corners (triple
points; Fig. 1.6) and were shown to decrease rapidly with distance from these corners.
The maximum stress concentration for defects located on grain boundaries was again
found for those defects residing at the triple points. Grain arrays were shown (Fig. 1.7)

Fig. 1.7. The normalized stress intensity factor plotted as a function of relative crack length for regular
hexagonal grains; the single grain solution, as well as the solutions for a grain pair and a grain array are
shown (Evans 1978)
14 Chapter 1 · Fracture Physics

Fig. 1.8. The variation of the predicted critical facet length with the triple point defect size for Al2O3.
Also shown is the average value obtained experimentally (Evans 1978)

to magnify the stress intensity factor (see Sect. 1.3.3) over that of a single grain by up
to ~4. “The incidence of microfracture will thus be strongly dependent on the relative
orientation of neighboring grains” (Evans 1978). Stress intensity factor values are cru-
cially dependent on the ratio between the defect (flaw) size (which Evans transforms
to an equivalent crack size; Evans 1974) to the size of the grain. Results also yield a
critical grain size (facet length) lc, for inducing microfracture (Fig. 1.8).

1.2.3
Thermal Nucleation

1.2.3.1
Thermal Fluctuations and Arrhenius Process

In many areas of physics and chemistry, the probability of crossing an energy barrier
is encountered. The theory is usually named after Arrhenius or also by the name “reac-
tion rate”. For a theory based on the Fokker-Planck equation, see e.g., Gardiner (1990).
A naïve explanation of this approach is as follows. We are dealing with a system (A) that
 1.2 · Nucleation 15

Fig. 1.9.
A schematic “barrier passing”.
The energy of the system is
shown as a function of a para-
meter x (which can represent
different variables, e.g., crack
length, stress, strain etc.); x is
often called the “reaction pa-
rameter”

resides in an equilibrium (minimum energy) situation (Fig. 1.9), at a certain “initial” en-
ergy level E1 and interacts with a big system (such as the atmosphere) called a “thermal
bath”, which keeps the temperature of A constant. Due to the heat exchange with the bath,
the system A constantly changes its energy level around E1 (say by moving along positions
1, 2, 3, 4, etc.). These changes are called “thermal fluctuations”. Now suppose that A has a
second equilibrium energy state E2 (>E1). Note that E2 of Fig. 1.9 is not an equilibrium point,
since it is not an energy minimum. If we now start with a large number (an ensemble) of
identical systems, each having two energy levels E1 and E2, the thermal fluctuations will
establish a steady state whereby the ratio of the number of systems residing in E2 to the
number of systems residing in E1 equals exp{–(E2 – E1) / (kT)}, where k is a universal
(Boltzmann) constant. This situation is called Maxwell’s distribution. On this basis, we
assume (to get the Arrhenius process) that the probability of a single system (such as that
depicted in Fig. 1.9), having only one equilibrium energy level, E1, and being in contact
with a heat bath that keeps it at a constant temperature T, to reach a higher energy level
E2 (>E1), is proportional to exp{–(E2 – E1) / (kT)}. This is evidently an approximation,
since E2 is not an equilibrium state. Thus, according to this theory, the probability of a
system overcoming an energy “barrier” of ∆E = E2 – E1 (Fig. 1.9), per unit time is

(1.24)

where ν0 is a “normalization” factor and ∆E = E2 – E1. In a more generalized case where

pressure or volume changes can occur during the process, ∆E would be replaced by
the enthalpy difference of the system between the two states.

1.2.3.2
Fracture and Microfracture Nucleation

Figure 2.1 depicts the pattern observed on a fractured glass sample. Here we relate only
to the two smallest forms, i.e., ci and ccr. The former is created by the initial flaw (or
actually the microcrack developed from it), while the latter, ccr, is the critical fracture
length wherefrom the crack starts to propagate (above which the Griffith criterion
holds). For a physical understanding consider (Fig. 1.10) the total energy of the system as
a function of crack length, c. See also Sect. 2.2.2.1 for characteristic lengths of ci and ccr.
16 Chapter 1 · Fracture Physics

Fig. 1.10.
A schematic representation of
the difference between “micro-
fracture nucleation” (E1 → E3
via E2) and “fracture nucle-
ation” (E1 → E4 → E5, E6 or
E3 → E4 → E5, E6). A: System
with no flaws (no microfrac-
tures); B: System with flaws

As discussed above (Sect. 1.2.1), the term “nucleation” can be used for “microfracture
nucleation”, i.e., nucleation of a microfracture ci (around a void, an inclusion or even
a small existing microfracture), whose length is smaller than ccr; and it can be used
for “fracture nucleation”, namely the appearance of the critical crack ccr that causes
material failure. While both of these “nucleations” usually proceed in the same man-
ner, i.e., by thermal fluctuations, there is a major difference in their final stability.
Microfractures, once nucleated, can remain stable (and do not heal) because they reach
a stable or quasi-stable equilibrium, i.e., an energy minimum (or a local minimum) in
which the system resides (such as E3, Fig. 1.10). To remove it from this equilibrium
position back to E1 (to heal the flaw), an energy barrier (E2 – E3) has to be overcome.
In order to have nucleated this flaw, from the equilibrium position E1, the thermal fluc-
tuations should obviously have taken the system over the barrier E2 – E1 (and would
eventually transfer the system from E1 to E3). On the other hand, “crack nucleation”
would bring the system from E1 (if no flaws exist) or from E3 (in a system with a
microcrack) to E4. Then the system ceases to be stable. It reaches in fact the Griffith
criterion, which means that the crack starts propagating indefinitely, thus gaining
energy from the field (going to E5, E6, etc).
We shall describe only the second procedure, namely the nucleation of a crack lead-
ing to eventual rupture. The procedure of “microcrack nucleation” is principally similar.
For microcrack nucleation, however, the derivation of the energy barrier (E2 – E1) from
the physical situation is of course different from the one we show here and depends
on the existing flaw (see Sect. 1.2.2).

1.2.3.3
Single versus Multiple Crack Nucleation

Two scenarios are conceived in the literature for the fracture process of brittle mate-
rials. The first consists of a nucleation and growth of a single critical crack that after
reaching the Griffith criterion totally breaks the material. If the rate of stress increase
when loading the material is high enough and a critical crack or an effective critical
flaw already exists there, then the nucleation part becomes very short. The second
scenario (e.g., Spyropoulos et al. 2002), which is assumed to hold in “grainy” materi-
als, i.e., groups 2 and 3 heterogeneous materials (Sect. 1.2.1) and below the critical
stress, comprises three to four stages:
 1.2 · Nucleation 17

1. Nucleation of “many” microcracks under the combined effects of the existing stress
and thermal fluctuations.
2. Initial growth of such microcracks.
3. Coalescence of the microcracks into macrocracks.
4. Rupture of the material. In this section we treat the thermal nucleation process,
which is related to the first scenario and also briefly the “fiber bundle” model for
grainy materials. Other models such as percolation (Hermann and Roux 1990) and
self-organized criticality (P. Bak et al. 1987, 1988) will not be touched upon. The
models chosen here were those which (recently) have been shown to agree with
experiments.

1.2.3.4
Thermal Crack Nucleation

For many years, the time to failure of a material under a constant remote stress, σ0,
below that which would cause immediate fracture (Griffith) was calculated by the
Zhurkov (1965) or the Mogi (1962) assumption

where τ0 and b are constants. Careful measurements (see below) by Ciliberto and his
collaborators have shown that this was not the correct τ (σ0) relation.
We shall follow an adapted version of the model proposed by Pomeau (1992) for
fracture of two- and three-dimensional solids. A similar model was proposed by
Golubovic and Feng (1991), and was extended to fractal dimensions by Bonn et al.
(1998), who carried out an experiment on polymer gels to verify their theory.

Crack nucleation in 2D and 3D (no flaws – homogeneous nucleation). Nucleation

of a critical crack for a material under a constant remote stress, σ0, i.e., the probability
of the appearance of a crack large enough as to fulfill the Griffith criterion, is accom-
plished by thermal fluctuations, via an Arrhenius-type process (Sect. 1.2.3.1). Assum-
ing there are no initial flaws in the material, the rate of appearance of a critical crack is
given by (Pomeau 1992)

(1.25)

where ν0 is a normalizing rate, k is the Boltzmann constant, T is the absolute tempera-

ture and W is the “liberated” enthalpy. Here we assume that volume and pressure are
unchanged in the process and therefore W becomes the energy barrier needed to be
overcome by the process. In a method similar to the one described in Sect. 1.1.2, for
a 2D solid, the elastic energy that would be liberated by a fracture of length c is
(Fig. 1.11):

(1.26)
18 Chapter 1 · Fracture Physics

Fig. 1.11.
A schematic “2D” homogeneous
nucleation of a crack of length c
in an infinite plate under remote
tensile stress σ0

where σ0 is the remote stress field in the medium, E is the Young modulus and b is the
thickness (assumed small with respect to the other dimensions) of the sample. The
energy needed to create the two new surfaces is

(1.27)

where Γ is the surface energy. The total energy is therefore

(1.28)

Figure 1.12a depicts W as a function of c. Note that for small c values, W increases
linearly with c and then the parabolic part becomes dominant leading to a maximum.
This maximum is given by ∂W / ∂c = 0, resulting in

(1.29)

Now, if no flaws exist in the material, the critical barrier needed to be overcome by
the creation of a critical crack is equal to Wm, and hence the rate of nucleation of such
flaws is

(1.30)

where

In three dimensions, the calculation is similar to that appearing in section 1, Eq. 1.23'.
The total energy is given (Fig. 1.12b) by

(1.31)
 1.2 · Nucleation 19

Fig. 1.12.
Total energy (relieved elastic
and surface energies) as a func-
tion of c. a 2D case. b 3D case.
Wm: Maximum energy oc-
curring at crack length cm;
W0: Energy for existing crack of
length c0. Note that there should
be a dip (energy minimum)
at W0 (not shown)

the maximal energy, which appears for cm = 2Γ E / σ 02, is given by

(1.32)

and the nucleation rate becomes

(1.33)

The way to experimentally verify these results is to measure the time to fracture
under a constant stress load that is below the critical stress, i.e., in a creep condition.
Thus, in a sequence of papers (Guarino et al. 1999; Scorretti et al. 2001; Ciliberto et al.
2001; Guarino et al. 2002; Politi et al. 2002) Ciliberto and his group performed such
measurements on several materials. According to Eq. 1.30 and Eq. 1.33, the times to
failure for 2D and 3D materials under a constant σ0 are given by the reciprocals of the
rates, namely,

(1.34)
20 Chapter 1 · Fracture Physics

and

(1.35)

respectively. Here τ02 and τ03 are characteristic times, equal to the (1 / ν0), for 2D and 3D,
respectively. Note that the dependence on σ0, especially for the 3D case, is very strong: A
small change in σ0 can lead to a large change in τ. Therefore, the measurements must be
done very delicately to avoid a situation where the material either breaks immediately or
does not break at all (the time to failure is above that of the experiment). Ciliberto’s group
succeeded in obtaining very accurate measurements and verifying the τ (σ0) behavior. For
example, Fig. 1.13 (adapted from Fig. 4 of Guarino et al. 2002) depicts their results for wood
samples (3D). This result, similar results for Perspex by the same group and 2D (Pauchard
and Meunier 1993, 1998) and fractal sample (Bonn et al. 1998) time to fracture results, all
lead to the same conclusion, namely that the new nucleation theory embodied in Eq. 1.34
and Eq. 1.35 is the correct one. For example, Fig. 1.13 shows the inadequacy of Mogi’s (1962)
or Zhurkov’s (1965) approach and the apparent applicability of Eq. 1.35.
The result of Ciliberto’s group did seem to have problems:

1. The distribution of times to failure was not Poissonian as expected.

2. The measured temperature dependence of τ for a linearly increasing load (see be-
low) seemed to yield a temperature of ~3 000 K, while experiments were carried out
near room temperature. To explain this discrepancy, they resorted to the “fiber bundle”
model (see below), arguing that the spread in fiber strengths could lead to such high
effective temperatures.

Fig. 1.13. Measurements on wood samples (Guarino et al. 2002). The time τ needed to break the samples
under an imposed constant pressure P is here plotted as a function of 1/ P4 in a semilog scale. The dashed
line represents the solution proposed by Mogi (1962) (τ = ae–bP). The continuous line is the solution pro-
posed by Pomeau for microcrystals (τ = τ0 exp[(P1 / P)4], Eq. 1.35 with P instead of σ). In the plot τ0 = 50.5 s
and P1 = 0.63 atm. Every point is the average of ten samples. The error bar is the statistical uncertainty.
Dotted line: Results of our calculation Eq. 1.41 and Eq. 1.42 (with average c0 = 1.25 × 10–6 cm)
 1.2 · Nucleation 21

Fig. 1.14.
The absolute value of the mea-
sured exponent of activation
energy vs. force as a function
of the fractal dimension of the
gels. The point at df = 2 is the
measurement of Pauchard and
Meunier (1993) for a two-di-
mensional crystal. The line is
the prediction from the scaling
arguments, Eq. 1.36 (Bonn
et al. 1998)

Two-dimensional solids were fractured by Pauchard and Meunier (1993, 1998), and
the

relation was shown to hold for the averaged breakage times. As mentioned above, time
to failure of polymer gels, whose fractal dimensions were measured by visible light
scattering, was measured by Bonn et al. (1998), and the exponent of σ0 in the delayed
time to fracture was shown to agree with these fractal dimensions. Thus, the critical
length was predicted to be cm = 2(d – 1)Γ E / dσ 02, where d is the fractal dimension of
the solid. Hence the energy barrier should be proportional to

(1.36)

Figure 1.14 (Fig. 5 of Bonn et al. (1998), adapted) shows the agreement between the
(light scattering) measured dimensions, df and the dimensions d, obtained from the
σ0 dependence of the delayed times to fracture (derived from Eq. 1.36). These authors
also discuss the question of “lattice trapping” (see, e.g., Marder 1996b) and show that
the latter cannot be the reason for the delayed fractures since this phenomenon in-
volves delay times that are shorter (or of much lower energy barriers) than those
measured.

Thermal nucleation in the presence of flaws (microcracks, heterogeneous nucle-

ation). Thermal nucleation was considered here (Sect. 1.2.3.4) as a process for which
the thermal fluctuations were supposed to overcome an energy barrier that started
from zero – thus completely disregarding the existing flaw distribution in the mate-
rial. In this section, we treat nucleation where the stressed sample already contains
internal (or surface) flaws, which reduce the energy barrier (Rabinovitch et al. 2004).
The calculation is carried out for the three-dimensional case. Results are easily ex-
tended to solids of other dimensions.
It is therefore assumed that at the moment of loading, the sample already possesses
a distribution of microcracks each having a different length ci perpendicular to the
22 Chapter 1 · Fracture Physics

direction of the highest principle stress. Note that a flaw can be looked upon as an
“effective microfracture” (Evans 1974) in that the highest stress concentration at a point
on its perimeter can be equated to that arising from a 2D microcrack with a (differ-
ent) effective length. According to Fig. 1.12b, the barrier becomes smaller since the
thermal fluctuations need only to augment an existing flaw from c0, say, to cm, rather
than create a crack of length cm from the start. The energy barrier now is given by
(Fig. 1.12b) W = Wm – W0, where

(1.37)

The energy barrier, W, is larger than zero for c0 < cm. The time to fracture under
such conditions is therefore

(1.38)

where

(1.39)

Clearly, Eq. 1.38 shows that flaws facilitate failure. Thus, denoting by

(the Pomeau result, Eq. 1.33), Fig. 1.15 depicts the decrease in τ with c0 (0 ≤ c0 ≤ cm)
from τ1 to τ0.

Fig. 1.15.
The lowering of the time to
failure for an existing crack of
length c0 in a 3D sample due
to the decrease of the energy
barrier. The value c0 = 0 is for
nucleation where no flaws
exist. Here cm = 1.3 × 10–5 cm,
above which τ = τ0
 1.2 · Nucleation 23

Now, the most “important” crack, the one for which the barrier is smallest and hence
leads to the shortest time to fracture if augmented, is the crack with the longest c.
Denoting the length of this crack by c0, Eq. 1.37 and Eq. 1.38 provide the lowest en-
ergy barrier and the shortest (hence the measured) time to failure, respectively.
According to the statistics of extreme values (e.g., Castillo 1988; Kotz and Nadarajah
2000) the longest cracks can have only one of three distributions, Gumbel, Frechet or
Weibull. Since the “basin of attraction” of the Gumbel distribution includes the exponen-
tial and the Gaussian ones, both of which serve as possible distributions of flaws (Lawn
1993), it is conceivable that the longest flaws here assume the Gumbel distribution. The
probability density function of c0 is therefore assumed (Kotz and Nadarajah 2000) to be

(1.40)

where l and m are parameters.

The following procedure was adopted. For specific chosen values of l and m, a
c0 value is chosen randomly from the p(c0) distribution. The τ for this c0 is calculated
by Eq. 1.38 for a definite remote stress σ0 and temperature T. Note that the choice
mechanism can generate c0 values that are already above the Griffith criterion. For these
cases, τ is chosen to be τ0 (see Fig. 1.15). In this way,
_ a histogram of τ for these values
of σ0 and T is obtained, from which the average, τ , of τ under these conditions is _ cal-
culated. Changing σ0 for the same T, these averages are drawn as a_function of σ0, τ (σ0).
This procedure is repeated for different l and m values until the τ (σ0) agrees with the
experimental results, at which point the values of l and m are considered to be the
correct ones for the flaw distribution. Results are shown in Fig. 1.13 (dotted line). It is
seen that although ln(τ ) of Eq. 1.38 is no longer directly proportional to 1 / σ 04, the
changes are minimal and the agreement with experiment is good.
A straightforward understanding of the situation can be obtained by considering
ln(τ ) as a function of x = 1 / σ 04. By Eq. 1.38, we obtain

(1.41)

where

Hence ln(τ ) has a minimum at

(1.42)

below which it diverges. Therefore, if the stress values used in the experiment are much
smaller than σ%, then ln(τ ) would still follow an almost straight line. _For Fig. 1.13, the
largest experimental σ or (P) is ~0.7 atm, while for our simulations c0 = 1.2 × 10–6 cm
and therefore σ% ≈ 3.5 atm.
24 Chapter 1 · Fracture Physics

A second experiment, analyzed, e.g., in Guarino et al. (2002) is where a load, which
is linearly increasing with time (σ0 = Apt for a constant Ap), is applied to different
samples until they fail. The breaking times are drawn as a function of Ap in a log-log
plot yielding a straight line. In the experiment (Guarino et al. 2002) two temperature
values were used, 300 K and 380 K, with almost no difference in the obtained straight
lines. On the basis of this invariance, the investigators concluded that there was some-
thing amiss and invoked (Politi et al. 2002) fiber bundle theory to explain this appar-
ent difficulty.

Fig. 1.16.
Time to failure for a linearly
increasing stress as a function
of stress rate Ap. a For c0 = 0
simulating Guarino et al. (2002)
results. b Averaged times for c0
chosen according to Eq. 1.40
 1.2 · Nucleation 25

The difficulty is however only a sham difficulty as can be seen from the following
“numerical experiment”. First we follow the method used by these authors (Guarino
et al. 2002) to treat breaking time problems where the stress changes with time. We
regard (Guarino et al. 2002) the variable 1 / τ (either of Eq. 1.35 for “no flaw” or of
Eq. 1.38 for nucleation under existing flaw distribution) as the “density of damage”
per unit time. Breaking is certain to occur after a time τ1 if

(1.43)

We now apply Eq. 1.43 to find τ1 numerically by integrating Eq. 1.35 with σ0 = Apt,
or perform a similar calculation using Eq. 1.38 with an appropriate choice process
(of c0) as described before. We obtain (Fig. 1.16) that in both cases results are almost
independent of temperature. The results of Guarino et al. (2002, Fig. 8) are therefore
completely natural and in line with both the “no flaw” and nucleation under existing
flaw distribution theories.
We turn now to the time distribution function. Experimental results (Guarino et al.
2002, Fig. 6) show a distribution, which is definitely neither Poissonian (which would
have been expected from the Arrhenius approach (e.g., Gardiner 1990)) nor Gaussian.
Using Eq. 1.38 and choosing c0 values randomly according to Eq. 1.39, a series of
τ values is thereby generated. In this way, we calculated τ-distribution functions for
several temperatures. Distribution results are shown in Fig. 1.17 for different values
of temperature. It is seen that

Fig. 1.17. Time distributions for different temperatures. a Experimental results (Guarino et al. 2002)
T ≈ 300 K. b Results of our calculations (Eqs. 1.38, 1.39 and 1.43) T = 300 K. c As in b but for T = 400 K.
d As in b but for T = 500 K
26 Chapter 1 · Fracture Physics

1. The distribution function for 300 K is very similar to the experimental results.
2. The change of distribution function with temperature is interesting.

Thus, a relatively small temperature increase causes the “balance” of the distribu-
tion to move somewhat towards shorter times, while a large increase (in Fig. 1.17 from
300 K, say, to 500 K) even alters the whole pattern of the distribution, moving it away
from a decreasing exponential towards a more “bulky” shape.
It is therefore seen that thermal nucleation is indeed the process behind failure of
polycrystalline samples under creep conditions. An improvement of the theory, by the
use of nucleation where account is taken of the existing flaw distribution inside the
sample, leads

1. To an easier fracture and, more importantly.

2. To an accurate change of lifetime distribution functions with temperature.

1.2.4
Fiber Bundle Model

1.2.4.1
Introduction

A brittle heterogeneous solid of the second or third group (Sect. 1.2.1) under an ap-
plied stress can break via what we have called the “second scenario” (Sect. 1.2.3.3),
namely the appearance of a large number of microcracks. These can either grow or in-
crease in number, leading to their coalescence and eventual failure. In this section, we
discuss fiber bundle models which treat the multiple nucleation of microcracks. Such a
nucleation is supposed to occur in rock fracture, as is evidenced by the appearance of
ever-increasing acoustic emission pulses (see, e.g., Mogi 1962; Hirata et al. 1987; Lockner
1993) and of electromagnetic radiation emissions (see Chap. 5 of this monograph)
during such fracture processes. Although the fiber bundle model was developed for and
seems most appropriate to describe a solid made up of fibers such as wood or “fibers in
matrix” composite materials, it can be shown (Hirata et al. 1987) that results obtained
using it agree with “regular” field damage models having nothing whatsoever to do with
fibers. The fiber bundle model (FBM) was thus actually used to treat general cases (e.g.,
Guarino et al. 2002) of fracture. The exact reason for the success of the FBM to describe
non-fibrous materials is still unknown. A crude argument might be that most experi-
ments either in tension or compression are conducted under a cylindrical symmetry
(uniaxial stress or triaxial stress where the perpendicular axes carry identical loads).
For a grainy material, a string of grains parallel to the symmetry axis would constitute
a “chain of beads” not unlike a fiber. Moreover, a fracture of a single grain either by a
transversal through fracture or a fracture along the grain boundary could be looked
upon as a failure of the chain – the fiber.
We shall discuss the FBM only in 1D, namely under the assumption that the lattice-
like connection of the fibers is ignored. This restriction is done

1. Since the 1D case can often be analytically or at least asymptotically solved.

2. Since it retains the main features and results of the more complex models.
 1.2 · Nucleation 27

More realistic models, usually containing entangled numerical calculations, are

based on a “chain of bundles” system (see Hidalgo et al. 2002; Gucer and Gurland 1962;
Smith and Phoenix 1981; Smith 1982).
Treated here are two cases defined according to how the load is redistributed, fol-
lowing a fiber rupture.

1. The stress is redivided equally among all the remaining fibers (ELS). For this treat-
ment, we will follow mainly the approach of Turcotte et al. (2003).
2. The stress is transferred in decreasing amounts to the “neighbors” of the ruptured
fiber (VLS). For this treatment, we will mainly follow the approach of Hidalgo et al.
(2002).

Our treatment is meant to be explanatory rather than exhaustive, since other ap-
proaches to the unequal redistribution of stress are abundant.

1.2.4.2
Uniform (or Equal) Load Sharing (Turcotte et al. 2003) ELS (Mean-Field Approximation)

Consider (Fig. 1.18) a rod made up of N0 fibers. Denote by N(t) the number of unbro-
ken fibers at time t. The rate of fibers failure is given by

(1.44)

where ν is a parameter (called “hazard rate”), which is dependent on the applied fiber
stress at time t (“no delay” is usually assumed). If the rod is loaded by a constant force,
F0, the fiber stress in it at time t is

(1.45)

where s is the area of a fiber. This is the uniform load sharing assumption namely that
following a failure of a fiber, the load is equally redistributed among the other fibers.

Fig. 1.18.
A schematic illustration of the
failure of a fiber-bundle with
uniform load sharing. a Each
of the fibers carries one-quar-
ter of the load F0. b One fiber
has failed and each remaining
fiber carries one-third of the
load F0. c Two fibers have failed
and each remaining fiber car-
ries one-half of the load F0.
d All four fibers have failed
and no load is carried (from
Turcotte et al. 2003)
28 Chapter 1 · Fracture Physics

In order to complete the presentation of the problem, the functional form of ν (σ )

should be specified. An empirical form for ν (σ ) is

(1.46)

where ν0 is the rate related to the initial stress σ0 (= F0 / (N0s)) and ρ is a constant ex-
ponent. In order to agree with experimental results, it is shown by Newman and Phoe-
nix (2001) that the range of ρ values is between 2 to 5. Solving Eqs. 1.44–1.46 yields

(1.47)

The rod fails when N = 0. Hence, the time to failure is given by

(1.48)

Note that this result does not depend on the number of initial fibers but does de-
pend on σ0. In view of Eq. 1.46, no agreement with Eq. 1.34 or Eq. 1.35 of the thermal
nucleation model (Sect. 1.2.3.4) is to be expected. As noted above, Ciliberto et al. (2001)
try to build a combined model of thermal nucleation with fiber bundles to explain
their experimental results.
A very nice feature of the fiber bundle model is its ability to provide an explana-
tion for the energy emission rate from a fracture either by acoustic emission or by EMR
(see Chap. 5). To calculate the rate of say the acoustic emission (ae) in this model, we
assume it to be a fraction, ηae, of the total strain energy release rate, dW / dt:

(1.49)

while dW / dt is given by

(1.50)

where wf is the stored elastic energy in a fiber at the time of its failure. The latter is
evidently given by

(1.51)

where σff is the stress in the fiber at failure, V is the volume of a fiber and E is its Young
modulus (assuming they behave elastically until failure). Using Eqs. 1.44–1.51 yields

(1.52)
 1.2 · Nucleation 29

Fig. 1.19.
Cumulative acoustic energy
emissions ea(t) at time t nor-
malized by the total acoustic
energy emissions etot at the
time of rupture (tf) as a func-
tion of 1 – t / tf. A constant pres-
sure difference was applied
at t = 0 (Guarino et al. 1999)

A comparison with experiment (e.g., Guarino et al. 2002) of the acoustic energy
release as a function of time from failure of chipboard and fiberglass yields an excel-
lent agreement with Eq. 1.52 for ρ = 3.7, or (ρ + 1) / ρ = 0.27 (Turcotte et al. 2003). Fig-
ure 1.19 shows an example of such an agreement for a cumulative acoustic emission
as a function of (1 – t / tf) (Guarino et al. 1998, 1999).

1.2.4.3
Variable Load Sharing (VLS)

The method of stress sharing following a failure of one fiber that is opposite to the
one discussed above is called local load sharing (LLS). Already, Coleman (1957) con-
sidered the influence of friction among fibers. This friction would tend to concentrate
the load in the vicinity of the broken fiber rather than spread it like in the ESL. Here
we follow the numerical approach of Hidalgo et al. (2002) (see the list of references
there for an accumulation of papers in the area). The model assumes a power law de-
crease of stress away from the failed fiber. Denoting the stress in fiber i at time t by σi(t),
the additional stress there at time t + τ, σi(t + τ) – σi(t) is obtained by contributions
from all failed fibers B(τ ) in the interval τ in the following manner:

(1.53)

where

(1.54)
30 Chapter 1 · Fracture Physics

Fig. 1.20.
Illustration of the model construc-
tion. X indicates a fiber that is go-
ing to break, and
is an intact
fiber in the square lattice (Hidalgo
et al. 2002)

is the normalized contribution fraction from a failed fiber j (at point X in Fig. 1.20) to
the fiber i. The sum on i in the normalizing denominator runs over all unfailed fibers
at time t, and rij is the distance from fiber j to fiber i. Hidalgo et al. (2002) have used
periodic boundary conditions on an L × L grid so that the largest value of r used was
⎯ 2. The number of fibers is N = L2. B(τ ), the set of all fibers that break in the
(L – 1) / √
increment τ is defined as follows. The system is driven by first applying on it a small
value of stress (σ ) which is equally distributed among the L2 fibers. The strengths of
the fibers are assumed to be distributed according to a Weibull distribution.

(1.55)

where ρ is the so-called Weibull index, which controls the spread in threshold value, and
σ0 is a normalization parameter. Each fiber is ascribed a certain threshold value σth cho-
sen at random from this distribution. Now the fiber with the lowest value of (σth – σ / N) is
sought, and this minimum value is added to all intact fibers. Thus, at least one fiber is
going to break and its load is transferred via Eq. 1.53 to the other ones. This may cause
more fibers to fail at this stage, defining ultimately the avalanche B(τ ) of all fibers that
failed at this stage. When the failures stop, σ is increased again by the same method.
The important parameter is γ (see Eq. 1.53 and Eq. 1.54). It defines the effective
range of the influence of one failure. For γ = 0, the range is infinite and the model is
equivalent to the ELS discussed above. For a large γ , this range shrinks and for γ = ∞
only its nearest neighbors are affected by the failure.
Results of numerical simulations (Monte Carlo type) show some interesting results.
Figure 1.21 shows the ultimate strength of the complete bundle as a function of γ for dif-
ferent values of L. Here σc is the stress (per fiber) above which all fibers were broken. Three
ranges of γ are observed. For γ ≤ 2, the behavior is similar to that of γ = 0, namely the
ELS model. In this range, the ultimate strength depends neither on γ nor on L and can
 1.2 · Nucleation 31

Fig. 1.21.
Ultimate strength of the mate-
rial for different system sizes
as a function of the effective
range of interaction γ. A cross-
over from mean-field to short-
range behavior is clearly ob-
served (Hidalgo et al. 2002)

Fig. 1.22.
Avalanche size distributions
for different values of the ex-
ponent of the stress-transfer
function γ (L = 257) (Hidalgo
et al. 2002)

in fact be calculated exactly. For the Weibull distribution, exact calculation yields
σc = (ρ e)–1/ρ. For γ ≥ 2 (local load sharing), σc depends on L and, as was shown by
Hidalgo et al. (2002), approaches zero for L → ∞ as σc(N) ~ 1 / lnN. This behavior is in line
with other short range models (see e.g., Harlow 1985; Kloster et al. 1997). However, for
γ ≤ 7, σc depends on γ , actually decreasing with increasing γ, while for γ ≥ 7 (extreme local
load sharing) σc is independent of γ due probably to the discrete nature of the model.
The avalanche size-distribution is shown in Fig. 1.22 for L = 257 and different val-
ues of γ. For small γ values (mean field approximation or ELS), the distribution can be
fitted by a power law.

(1.56)

A similar result was obtained by exact calculations in this limit (Harlow 1985; Kloster
et al. 1997). For a more localized interaction (γ > 2 say), the causality correlation be-
32 Chapter 1 · Fracture Physics

Fig. 1.23.
Cluster size distributions for
different values of the stress-
transfer function exponent γ.
Clearly, two different groups of
curves can be distinguished
(L = 257) (Hidalgo et al. 2002)

tween failing fibers seems to decrease and no relation like Eq. 1.56 can be derived. For
small γ values, “any given element could be near to its rupture point regardless of its
position in the lattice”, while for large γ values “spatial correlation (is) important and
concentration of stress take(s) place in the fibers located at the perimeter of an al-
ready formed cluster” (Hidalgo et al. 2002) thus reducing the avalanche size.
Figure 1.23 depicts the distribution of cluster sizes, namely the spatially connected
(via nearest-neighbors) failed fibers. Shown are clusters just before total failure; the
transition from ELS to LLS is obvious. The cluster structure for a complete mean field
( γ = 0) “can be mapped to percolation clusters generated with the probability
0 < P(σc) < 1” (ibid) except for the exact critical percolation probability, which depends
on the Weibull index. For more localized interaction, we approach the “first scenario”
(Sect. 1.2.3.3), where a single cluster (crack) is formed leading to complete breakdown.
Thus, for large γ values, the total number of clusters decreases.

1.3
Linear Elastic Fracture Mechanics (LEFM)
1.3.1
General

Although the basic physical model (Newton’s second law, Hook’s law and the bound-
ary conditions which assume that the crack sides are traction free) are simple, the
ensuing differential equations (Eq. 1.8–1.16) become mathematically complicated, es-
pecially those related to 3D problems. Therefore, most of the theoretical approaches
were carried out for 2D problems. Even for 2D, the task of solving these equations is
complicated. However, as will be shown, the solution here can be divided into two parts.
Firstly, due to the sharpness of the crack tip, the stresses in the immediate neighbor-
hood of the tip turn out to be singular with asymptotic forms (at the tip), which de-
pend only on the mode of loading (of which there are only three; Sect. 1.1.3) and on
three related constants (one for each mode). Secondly, the influence of the external
boundary conditions, i.e., the geometry of the problem and the applied stresses or
 1.3 · Linear Elastic Fracture Mechanics (LEFM) 33

strains on the surface of the specimen, is manifested only via these constants (called
stress intensity factors and denoted by KI, KII and KIII). The energy supplied to the
crack tip, which leads to the latter’s ability to propagate, is shown to depend only on
these stress intensity factors. A balance between this energy and the energy needed to
form the new surfaces created by the fracture then yields the extension of the Griffith
criterion. Since the surface energy depends on the specific material the specimen is
made of, a corresponding material based critical parameter, Kc is defined. Energy bal-
ance implies that when KI say, equals the critical parameter KIc, fracture becomes pos-
sible. Since an absolute majority of fractures actually occur in a mode I situation, the
usual result of LEFM is the criterion

(1.57)

as a necessary condition of cracking. A second condition, which (together with Eq. 1.57)
constitutes a sufficient requirement, is that the equilibrium implied by Eq. 1.21 would
not be a stable one (see below). The whole effort of static LEFM is therefore concen-
trated in calculating KI for specific geometries and loadings.

1.3.2
Equations of Equilibrium and of Motion

For the sake of completeness, we present here the general equations of motion and
equilibrium for a homogeneous solid in the elastic approximation, resulting from
Eq. 1.8 in three dimensions. They are written in terms of the displacements u in con-
trast to the presentation in Eqs. 1.8–1.16, which are formulated through stresses (or
strains). For actual problems where boundary conditions are usually given in terms
of stresses or tractions, the second representation (Eqs. 1.8–1.16) is preferable. Thus,
in terms of u, the equations become

(1.58)

or, in terms of the components,

(1.58')

where ρ is the material density and λ, µ are (see Sect. 1.1.1) the Lamé constants, and
it is assumed that body forces are negligible. In the case of equilibrium and static con-
ditions, the left-hand side (or sides) of Eq. 1.58 or 1.58' is zero.
34 Chapter 1 · Fracture Physics

There are several good references in which solutions to Eq. 1.58 are given (e.g.,
Erdogan 1983; Leibowitz 1968, and see also next section). Several works discuss the
steady state solutions while others (see Fineberg and Marder 1999 and references
therein) treat the dynamic version. Let us reemphasize that while the steady state
(Griffith) results are very useful for basic understanding and for engineering applica-
tions, the dynamic solutions are controversial and do not, in general, scrupulously agree
with the experimental results (see Sect. 1.4).
Together with Eq. 1.11 and Eq. 1.13, Eq. 1.58 (1.58') constitute a complete set of
equations for the linear elastic problem.

1.3.2.1
Antiplane Stress Fracture (Mode III)

A very important 2D case, the equations for which are derived from Eqs. 1.58, is the
mode III case. In mathematical terms, this is the simplest case to treat, and can be used
as a basis for understanding all types of fracture.
For mode III (see Fig. 1.4), displacements are only in the z-direction; therefore,
ux = uy = 0 and uz does not depend on z, namely uz = uz(x, y). Hence Eqs. 1.58 become

(1.59)

Defining

Eq. 1.59 becomes the so-called “2D wave equation”

(1.60)

where

By Eq. 1.11 the only non-zero strain components in this case are

leading, by Eq. 1.13, to the only non-zero stress components

 1.3 · Linear Elastic Fracture Mechanics (LEFM) 35

Even in this case, and even for a time-independent solution (where Eq. 1.60 becomes
the Laplace equation, ∇2ux = 0), the actual solution may be complicated due to the pre-
vailing boundary conditions.

1.3.3
Two-dimensional (2D) Problems and Stress Intensity Factors

1.3.3.1
General

In two dimensions (mode III, Sect. 1.3.2.1, plane stress, plane strain, see above), the
equations become much simpler, and it is also easier to understand the problems in-
volved in their solution. We shall treat here only equilibrium problems with no body
forces. Extensions to dynamical problems and to problems including body forces can
be found in books specializing in those cases (see above and Sect. 1.4). Thus, the
2D equations of equilibrium for stresses in the x- and y-directions stemming from
Eq. 1.10 become

(1.61)

For an elementary derivation of these equations, see e.g., Timoshenko and Goodier
(1984) p. 26ff, and for a more mathematical approach see e.g., Landau and Lifshitz
(1963) p. 17ff.
It now becomes obvious why the compatibility equations (Eq. 1.16) are needed, since
Eq. 1.61 are a set of two equations for three unknowns (σxx, σyy and σxy) and as such
are not a complete set. The connections with strains are (Hook’s law, Eqs. 1.13 in 2D):

(1.62)

where G ≡ µ is the so-called shear modulus. But the three components of strain are
obtained from only two displacement components ux and uy , which are the basic physi-
cal variables (this is actually the reason why Eq. 1.61 consists of only two equations).
In 2D, the definition of strains (Eq. 1.11) becomes

(1.63)
36 Chapter 1 · Fracture Physics

(three components of strain obtained from two displacements). However, Eqs. 1.63 are
not independent. For taking the second derivative of the first with respect to y, of the
second with respect to x and y and of the third with respect to x, we arrive at the “com-
patibility equations” in 2D

(1.64)

which can be written (Timoshenko and Goodier 1984) in terms of stresses as

(1.65)

where

The complete set to be solved for σ is now Eq. 1.61 and Eq. 1.65. We discuss here
two types of solutions to this set. One is accomplished by the so-called Airy or stress-
function and the second is known as the Westergaard method.

1.3.3.2
Airy-Function (Stress-Function)

A new function Ψ is defined such that the stress components are given by:

(1.66)

In this manner, Eqs. 1.61 are immediately fulfilled and in order to fulfill Eq. 1.65,
Ψ has to satisfy the following equation:

(1.67)

To get a solution for a 2D elastic problem, one has to solve Eq. 1.67 with the appro-
priate boundary conditions. The latter can be given in terms of displacements (ui),
stresses, or tractions (forces, see Eq. 1.7) on the surface of the medium (including the
surface of the fracture itself).

1.3.3.3
Westergaard Method (Westergaard 1939)

This method is based on the properties of analytic functions of a complex variable. A

variable z = x + iy is defined where x and y are the coordinates of the 2D problem and
i = √⎯–⎯1. Analytical functions Z(z) are defined as “complex functions of a complex vari-
 1.3 · Linear Elastic Fracture Mechanics (LEFM) 37

able which possess a derivative (Z') wherever the function is defined” (Ahlfors 1979).
A well-known property of such functions is that their real and imaginary parts are
harmonic, i.e., have the properties

(1.68)

and fulfill the Laplace equation

(1.69)

where ReZ and ImZ are the real and imaginary parts of Z, which can be written as

(1.70)
_ _
Z', Z and Z are defined as the first derivative and the first and second integrals
of Z, respectively so that

(1.71)

Westergaard then shows that for a restricted group of cases the stress components σij
can be derived from Z in the form

(1.72)

Using Eq. 1.68, it is immediately seen that the stress components defined by Eq. 1.72
fulfill the equilibrium equations (Eq. 1.61) for any analytical function Z(z). The restric-
tion mentioned above is seen from Eqs. 1.72 and amounts to the condition that at y = 0,
σxx = σyy and σxy = 0. The displacements ux and uy are given by

(1.73)

where ν is the Poisson ratio, since it can be shown that ux and uy of Eq. 1.73 can gen-
erate the stress components via Hook’s law.
The Westergaard method can be correlated with the Airy or stress function Ψ de-
fined above, for it is easily seen that Ψ can be given by

(1.74)
38 Chapter 1 · Fracture Physics

1.3.3.4
Application to the Internal Crack under Tension

According to Westergaard (1939), if we use for Z the function

(1.75)

then Eq. 1.72 and Eq. 1.73 would describe the situation for a crack (of length 2c) stretch-
ing between x = –c and x = c at y = 0 in an infinite plane under a remote stress field σ0
in the y-direction (Fig. 1.24). For y = 0 by Eq. 1.72 outside of the crack, the tension is

(1.76)

and inside the length of the crack, by Eq. 1.73, the y displacement is

(1.77)

Therefore, the shape of the crack in the vicinity of the tip is parabolic (the com-
plete crack shape is elliptic). At the tips x = ±c, there exists a singularity of σyy, while uy
is regular.
Irwin (1957) used the solution Eq. 1.75 to calculate the (singular) stress fields near
the tip of the crack and to show that these fields can be connected via the energy re-
lease rate to the Griffith criterion. Substituting for z near the tip,

Fig. 1.24. Internal crack (according to Westergaard’s paper), which has opened from z = –c to z = +c
under the influence of a remote tension σ0 and the resulting σyy (at y = 0) as a function of x
 1.3 · Linear Elastic Fracture Mechanics (LEFM) 39

(1.78)

where r is the small distance (r << c) from the tip as an origin

(1.79)

and θ is the polar angle, again with the tip as an origin

(1.80)

the solution of Eq. 1.75 can be developed in powers of (r / c)1/2. The first and domi-
nant term for small r becomes (Irwin 1957)

mode I (1.81)

and (see e.g., Lawn 1993)

where KI (= σ0 √
⎯π⎯ c) is independent of r and θ and is discussed presently. For mode II,
similar calculations lead to a similar result:

mode II (1.82)

and for completeness, the mode III result is obtained from Eq. 1.59 above, with the
left side equaling zero:

mode III (1.83)

40 Chapter 1 · Fracture Physics

A more elaborate approach for the stress field near the tip of the crack was under-
taken by Williams (1957). In his method, the stresses are directly expanded in a series.
The leading term agrees with the Westergaard-Irwin results (Eq. 1.81). The other terms
in σ come in powers of r1/2; thus, the next two terms are a constant (r 0) and a term propor-
tional to r1/2. These terms and especially the sign of the constant term are sometimes used
for a more elaborate criterion for crack propagation (e.g., Cotterell and Rice 1980).

1.3.3.5
Connection between the Stress Intensity Factors KI (KII, KIII) and the Energy Release Rate

Figure 1.25 shows the (elliptic) crack shape (Eq. 1.77) at its tip (to where the origin of
the x-y-coordinate system has been transferred). Irwin (1957) then calculates the en-
ergy needed to drive the crack forward, assuming that it can be calculated from the
crack vicinity. More comprehensive energy calculations where the energy of the whole
medium is accounted for can be found, e.g., in Freund (1972a,b, 1973, 1974). For the
static case, however, the outcome of these calculations yields essentially the same re-
sults as Irwin’s.
Thus, according to Irwin (1957), suppose the crack had moved by a small incre-
ment α (see Fig. 1.25). The energy needed to perform this extension is equal to the
work needed to close it down, since the theory is linear. Now, the force (per unit area)
on the “bulge” for θ ≅ 0 and 0 ≤ x ≤ α, α << c is, by Eq. 1.81, approximately

(1.84)

where β measures the “closure fraction” changing from 0 at the beginning of closure
to 1 (end of closure). The y extension uy for 0 ≤ x ≤ α << c is given by Eq. 1.77 approxi-
mately as

(1.85)

Fig. 1.25.
Irwin’s (1957) linear-elastic
theory crack opening and
stresses near the tip of the
crack
 1.3 · Linear Elastic Fracture Mechanics (LEFM) 41

The work of closure is therefore

(1.86)

and per unit crack extension, namely the energy release rate is

(1.87)

As stated above, this is a simple mathematical method to obtain the energy deriva-
tive with respect to crack length (G).
These results (the set of Eqs. 1.75, 1.77, 1.81–1.83 and 1.87) constitute the basis of
the LEFM for the treatment of 2D equilibrium problems. Thus, by a connection with
Griffith’s criterion, the necessary condition for a flaw to increase in length or for a crack
to develop is G > Gc or KI > KIc. Here, however, we have also obtained a method of
calculating G (or K). For a crack of length 2c in an infinite medium (Fig. 1.26a), the
above results yield (as in Griffith 1920)

(1.88)

For different geometries, in order to calculate KI say, the elastic equations should be
solved only very near the crack’s tip. The value of KI is then obtained by taking the limit
of σyy √⎯⎯2π
⎯ r say as r → 0, for θ = 0. Several methods, both analytical and numerical, were
developed for the calculation of K (see e.g., Peterson 1974; Pook 2000; Murakami 1987,
1992, 2001). We present here only a few characteristic results of KI calculations:

1. Many results are written in the form

(1.89)

where q is a geometric factor and c is a characteristic crack (or flaw) dimension

(explained below). For example, for an edge crack (Fig. 1.26b), q = 1.12 and

(1.90)

2. For an elliptical crack in an infinite plate (Fig. 1.26c) under tensile stress perpen-
dicular to the plane of the crack,

(1.91)

where Q is the elliptic integral of the second kind:

(1.92)
42 Chapter 1 · Fracture Physics

Fig. 1.26. Several cases of KI values (after Murakami): a Internal crack in an infinite plate. b Edge
crack in an infinite plate. c Elliptic crack in the x-z-plane in an infinite body under uniaxial tension σ
in the y-direction. d Semi elliptical surface crack in a semi infinite body under uniaxial tension σ.
e, f Edge crack in an infinite plate under uniaxial tension: e Sharp notch; f blunted notch. g Two point
loads in middle of crack in an infinite plane (Pook 2000)

where a and c are the large and small elliptic radii, and (Rooke and Cartwright 1976)

M = [cos2β + (c / a)2 sin2β]1/4 (1.93)

describes the change of KI along the ellipse. Note, e.g., that M = 1 for β = 0 (in the direc-
tion of c), while M = √
⎯c⎯⎯/a (<1) in the direction of a. Here q of Eq. 1.89 is equal to M / Q.
3. For a penny-shaped crack loaded as in No. 2, a = c and

(1.94)

(q = 2 / π in this case).
4. For a semi-elliptic crack at the edge of a semi-infinite plate (Fig. 1.26d) under re-
mote tensile stress,

(1.95)
 1.4 · Dynamic Fracture 43

and for a semi penny-shaped crack, etc.

(1.96)

5. The value of c depends on the sharpness of the flaw. For instance, for a sharp groove
followed by a crack (Fig. 1.26e), c is the sum of lengths of the groove and the crack,
while for a blunt groove (Fig. 1.26f), only the crack length is taken as c.
6. An important case is a crack in an infinite plate operated on by two point loads as
in Fig. 1.26g. This case can be used as a basis for KI calculations using it within a
Green’s function method, namely considering any load as a superposition of point
loads. In this case,

(1.97)

where P is the value of the point load and b is the thickness of the plate. Note that
here, uncharacteristically, c appears to the power of –½ instead of the usual ½.

1.4
Dynamic Fracture
1.4.1
Mott and Mott-like Approaches

The first attempt to address dynamic fracture was made by Mott (1948). His approach
was based only on an energy balance and dimensional considerations. Thus, accord-
ing to Mott (1948), there are three terms which combine to form the total energy of
the system: The kinetic energy T, the liberated elastic (potential) energy U and the dis-
sipative energy S. The only changing variables are the crack length c and its velocity

Therefore by a purely dimensional analysis, the respective energies (for a crack in

a slab of constant thickness b under a constant load) should be given by

(1.98)

(1.99)

(1.100)

where α, β and γ are constants. Note that both T and U are proportional to the volume c2b
of the material in which elastic energy has been relaxed and which was set in motion
by the moving crack with different velocities across it, which are assumed to be pro-
portional to the crack velocity v. The dissipative energy, S, as before, is assumed to be
proportional to the newly created area (cb) via a surface energy parameter γ.
44 Chapter 1 · Fracture Physics

The total energy of the system is

(1.101)

where P is the total non-kinetic energy.

Mott then went on to assume that the crack develops under the condition ∂Etot / ∂c = 0,
while assuming that ∂v / ∂c = 0. This procedure yields

(1.102)

where the final velocity V = (β / α )1/2, and c0 = γ / 2β. Since α and β are unknown from
first principles, V and c0 cannot be obtained in this manner from known properties
(e.g., sound velocities). But, since for c = c0, v = 0, it is assumed that c0 is the critical
microcrack (or effective flaw) length (ccr) (see Sect. 1.2.3.2 and 1.2.3.4). However, a dif-
ferent interpretation of c0 is presented below in this section.
Mott’s assumption that ∂v/∂c = 0 (which is obviously valid only for large values of v
approaching the final velocity) has incurred some heavy criticism. Thus, foremost
among others, Berry (1960) and Dulaney and Brace (1960) tried to improve on Mott’s
derivation. To understand their calculations, consider the non-kinetic energy P as a
function of c as obtained by Dulaney and Brace (1960) (Fig. 1.27). It is immediately
seen that the (static) Griffith criterion for the initiation of a crack, namely ∂P / ∂c = 0,
occurs at c0 and that for c > c0 there is an ever-increasing gain of non-kinetic energy
for crack elongation. Assuming now that the (dynamic) condition for crack propaga-
tion is Etot = P0, Dulaney and Brace (1960), obtained for the crack velocity

(1.103)

This relation for the crack velocity was deemed to be preferable over Eq. 1.102 (for
example Fineberg, Gross, Marder and Swinney (Fineberg et al. 1991) and Marder (1993))

Fig. 1.27.
Sum of potential and dissipa-
tive energies, P as a function of
crack length c; c0 is the Griffith
criterion; c1 is the beginning of
dynamic fracture (Rabinovitch
1994)
 1.4 · Dynamic Fracture 45

although it suffers from two severe drawbacks. Firstly, for velocities that are close to V,
even according to Dulaney and Brace (1960), Mott’s approximation (Eq. 1.102) should
be valid. Secondly, regarding Eq. 1.103 as a differential equation, for v = dc / dt, it can-
not be integrated to obtain c(t) (the improper integral diverges).
We have assumed (Rabinovitch 1994) that the dynamic crack process does not start
when the crack length is at c0. Rather, a delay of fast propagation occurs and the crack’s
fast movement is initiated only at c1 > c0. That means that we identify c0 with ci (the
initial microcrack or effective flaw) and c1 with ccr (see Sect. 1.2.3.2, 1.2.3.4 and Fig. 2.1).
Note however that this assumption entails that for KI > KIc (KI = KIc occurs at c0 = ci)
the crack propagates with a very low velocity until its length reaches c1 = ccr. While
the origin of this delay might be in dispute (for example Schoeck and Pichl 1990;
Bernstein and Ness 2003), for our purpose it is the fact that between c0 and c1, energy
is dissipated into avenues other than the kinetic or potential energies discussed above,
which is important. It is therefore assumed that the actual critical length occurs at
c1 (≥c0), and only for c > c1 can the energy balance à la Mott be performed. Thus, for
c ≥ c1, the condition Etot = constant with the provision that v ≅ 0 at c = c1 yields

(1.104)

where V = (β / α )1/2, c0 = γ / 2β as before and

(1.105)

Several points should be noted.

1. Equation 1.104 transfers over to the Mott relation (Eq. 1.102) for large values of c.
2. Very close to c1, say for c = c1 + ε (where ε is a small quantity), the velocity behaves
as ε1/2.
3. Equation 1.104 is easily integrated (Gradshteyn and Ryzhik 1980) to yield

(1.106)

where g(c) = (c2 – 2c0c + b)1/2 and t0 is the time where c = c1. Note that g(c1) = 0 and
that, for t >> t0, c ≈ Vt.
4. Equation 1.103 is a (non-generic) special case of Eq. 1.104 for c1 → c0. In this case,
unlike the divergent integral obtained in trying to integrate Eq. 1.103 directly, the
limit c1 → c0 in Eq. 1.106 is well defined.
5. Since Eq. 1.104 contains three parameters V, c0 and c1, which can be adjusted to ex-
perimental results, it is not remarkable that such experiments can be approximated
quite well by this equation. For example, the results depicted in Fig. 1.31 (see below)
by Kobayashi et al. (1974), which were shown by Marder (1993) not to agree with the
constant G(t) approach (see below, Sect. 1.4.3), are in a very good agreement with
Eq. 1.106 with V ≈ 575 m s–1, c0 ≈ 1.33 cm and c1 ≈ 3.10 cm (not shown in Fig. 1.31).

In Sect. 1.4.2 and 1.4.3, we present the linear elastic approach to dynamic fracture.
46 Chapter 1 · Fracture Physics

1.4.2
2D Dynamic Crack

We first consider here the (asymptotic) constant velocity solution. Therefore, starting
from Eq. 1.58 we look for solutions that describe crack propagation at a constant ve-
locity of magnitude (v) in the x-direction (Fineberg and Marder 1999, Sect. 2.4). We
therefore consider displacements u, which have a “wavelike” behavior or, alternatively,
where the origin stays at the crack tip, x = x' – vt, thus,

and Eq. 1.58 becomes

(1.107)

General solutions can be obtained by defining two auxiliary functions Φ (zα)

and ψ (zβ), where now zα and zβ depend on the longitudinal and transversal acoustic
wave velocities (cl and ct respectively, defined in Eq. 1.109)

and (1.108)

(1.109)

The displacements are given (as can be checked by inserting Eqs. 1.110 in Eq. 1.107) by

(1.110)

The stress fields are given by

(1.111)

where
 1.4 · Dynamic Fracture 47

Note that our Φ and ψ are twice the Φ and ψ of Fineberg and Marder (1999).
In order to obtain a solution for a specific problem, “all” that is needed is to find the
appropriate Φ and ψ functions. An important example (Fineberg and Marder 1999) is the
near-tip behavior of a propagating crack under a mode I loading. Using the static solution
(Sect. 1.3.3.4) as a clue and the symmetry properties of the problem, i.e., ux(–y) = ux(y)
and uy(–y) = –uy(y), the appropriate solution is

(1.112)

(where the cuts of the ⎯√ are along the negative x-axis). The boundary condition on the
crack surface is that σxy and σyy are zero there (relaxed stress). Using these conditions yields

(1.113)

As was done for the static case (Sect. 1.3.3.4), near the tip, we can develop the re-
sults in powers of r1/2, where r is the small distance from the crack tip. We denote

(1.114)

and the leading terms of the stress are again of the order of r–1/2. These are (Freund 1990)

(1.115)

where the dynamical stress intensity factor (mode I) is

and (1.116)

(1.117)
48 Chapter 1 · Fracture Physics

As is shown presently (Sect. 1.4.3), K(t) is related to the energy flux into the crack
tip. Eqations 1.115 can be used to calculate the direction of crack propagation (see e.g.,
Fineberg and Marder 1999). Alternatively, it can be used (Rabinovitch et al. 1993, see
Fig. 2.29) to calculate the direction of secondary cracks that develop ahead of the propa-
gating primary crack. The method used there is as follows. A certain radius r0 > 0 is de-
fined. Eqations 1.115 specify the stress field as a function of θ = arctg(y / x). At each θ,
the two principal directions and principal stresses (the stress tensor can be digonalized)
are calculated. θ1 is calculated for which the major principal stress σ1 is maximal and
the axis of this principal stress is evaluated. The secondary crack is assumed to origi-
nate at (r0, θ1) and its direction (α1) to be perpendicular to the axis thus found. Results
(Fig. 1.28) show that both θ1 and α1 depend on crack velocity. For low velocities (v << Cl),
θ1 is ~60° and secondary cracks emerge almost parallel to the primary crack, while for
higher crack velocities, both θ1 and α1 increase in value. The possibility of θ1 > 90° is
unclear. Experimental results (Rabinovitch et al. 2000a) qualitatively confirm these pre-
dictions for low crack velocities.
In addition to Eq. 1.115, the leading terms of the velocity components of the move-
ments near the crack tip are obtained as

(1.118)

Fig. 1.28.
The angle θ1 at which σ1 at-
tains its maximum and the sec-
ondary crack direction α1, vs.
normalized crack speed v / Cl
(Cl: Longitudinal-wave speed)
(Rabinovitch et al. 1993)
 1.4 · Dynamic Fracture 49

Note that unlike the situation in the static case, here r–1/2 singularities exist for u· x
and u· y besides those for σij. Eqations 1.115 and 1.118 demonstrate that the near-tip
LEFM dynamic solutions, like the static case ones, can be divided into two parts: The
(r, θ ) part that does not and the K(t) part that does depend on the geometry of the
sample, the crack length and the remote stresses. It is thus the value of K(t) that should
dominate crack extension even for dynamic fracturing.

1.4.3
Energy Calculations

The 3D elastic (potential) energy stored in the sample per unit volume is given by
(Eq. 1.5, generalized)

(1.119)

In two dimensions, the indices of both ε and σ change only between 1 and 2
(x and y). We denote the indices in the 2D problem by α and β, each of which takes
the values 1 and 2. The potential energy (per unit area) in 2D is therefore

(1.120)

Using the definition Eq. 1.11 of

and the symmetry of σ (σαβ = σβα), e can be written as

(1.121)

The kinetic energy (per unit area in 2D) is given by

(1.122)

where duα / dt is the αth component of the velocity and ρ is the material density.
Consider now (e.g., Kanninen and Popelar 1985) the cracked body depicted in
Fig. 1.29, where S is its outer boundary on which forces (tractions) t are applied, c(t) is
the crack tip position which changes with time with a velocity v, and Γ * is a small con-
tour through which the energy flux is delivered to the tip; this Γ * is eventually reduced
to zero, and R is the volume (area in 2D) of the sample.
50 Chapter 1 · Fracture Physics

Fig. 1.29.
Dynamic crack propagation
in 2D (after Kanninen and
Popelar 1985)

·
Energy balance is enacted as follows. The rate of work, P, performed by the outer
forces (tractions) on S is equal to the rate of the total potential U and kinetic T energy
increase in R plus the energy flux Fl into the tip

(1.123)

where
(1.124)

tα is the αth component of the traction t; the total kinetic energy is

(1.125)

and the total potential energy is

(1.126)

As always, a dot above a variable means a derivative with respect to time of this
· · ·
variable. We are interested in the flux into the tip, Fl = P – U – T . The calculation of
the separate terms is quite involved, and only a sketch is given here. A specific compli-
cation arises in the calculations due to the fact that the contour Γ * propagates in time
·
with the crack tip, making R time-dependent. Therefore, for example, U includes two
terms, one which comes from the rate of change of uα with time and the other is the
contribution from the flux of the density of U through Γ *.
Consider (Fig. 1.30) a small surface (line in 2D), dS = AB which moves with a
velocity v. All through space (area) there exists a density (an amount per unit volume)
of some variable, e.g., energy. In the case of the potential energy, the density is
 1.4 · Dynamic Fracture 51

Fig. 1.30.
Schematic representation of
an area dS (here the line AB)
moving through space filled
with a density of a specific
physical quantity

and in the case of the kinetic energy it is

During the time dt, AB has moved by an amount vdt to the position A'B'. The area
(volume in 3D) traversed by dS is that of the parallelogram ABB'A', namely

where n$ is a unit vector normal to dS. Therefore the amount of energy which passed
through dS in the time dt is e dS v · n$ dt and the contribution of the movement of dS to
·
the time rate of change of energy, ∆U, is

Hence, the contribution from the movement of the whole Γ * is

52 Chapter 1 · Fracture Physics

Therefore,

(1.127)

where vn = vcos θ is the component of v normal to Γ *, θ being the angle between the
·
normal to Γ * and the v-direction. A similar term appears for T . Hence, as can easily
be shown,

(1.128)

The next step (e.g., Kanninen and Popelar 1985) is to use the Gauss theorem after
inserting the relation

(1.129)

After cancellation, this leads to a simpler equation

(1.130)

Three remarks are in order.

1. Equation 1.130 means that the flux of energy into the crack tip can be evaluated by
using the values of the stress and displacement fields only in the immediate vicin-
ity of the tip.
2. The flux of energy into the tip is given by the sum of fluxes of the kinetic and
potential energies into Γ *, as the tip is propagating, and the rate of work done
on Γ * by the tractions from the exterior of Γ *.
3. The flux is only obtained in the limit where Γ * → 0. A non-zero different contour
around the tip would yield a different value of the right-hand side of Eq. 1.130. An
exception to this general state of affairs is the case of a crack moving at a constant
velocity, v = v0 x$ . Here

and Eq. 1.130 becomes

(1.131)
 1.4 · Dynamic Fracture 53

For this case (Atkinson and Eshelby 1968), the value of the integral does not de-
pend on the choice of the contour Γ *, so there is no need to go to the limit Γ * → 0.
Now, to obtain the energy release rate G, recall that G is the rate of change of en-
ergy with crack length. Denoting this length by c,

(1.132)

Inserting Eqs. 1.115 and 1.118 for σαβ and u·α in Eqs. 1.130 and 1.132 finally yields
the result

(1.133)

where α, β and Ct were defined in Eq. 1.109, and v0 was replaced by v.

Equation 1.133 is sometimes written as

(1.134)

where G(0) is the velocity independent energy release rate (e.g., Eq. 1.87) and g(v)
(a universal function) contains the v-dependent part of Eq. 1.133. A good approxima-
tion for g(v) is (Freund 1972a, see also 1990)

(1.135)

where CR is the Rayleigh wave velocity – the velocity of a wave propagating on the
surface of the solid.
Equation 1.133 is the dynamic LEFM equivalent of the static case. It describes the
driving force for a crack to grow. If the dissipative energy per unit crack length, which
is needed for opening bonds, creating secondary damage (e.g., secondary cracks) ahead
of the crack tip, etc. is denoted by Γ (which can be velocity dependent), then the con-
dition for crack propagation is

G=Γ (1.136)

Eqations 1.134–1.136 are the ones used to analyze dynamic crack propagation. For
example, they imply that when V approaches CR, G(0) should go to infinity to sustain
the propagation, namely that CR is the crack’s maximal velocity. In experiments con-
ducted on brittle materials under constant load, the crack velocity invariably increases
monotonically in time, up to a certain terminal velocity. It can subsequently stay at this
velocity, which is then the maximal velocity or it can fluctuate around this terminal ve-
locity (Fineberg and Marder 1999), in which case we define the maximal velocity as the
time average of the fluctuations. In the literature, the terms maximal and terminal ve-
locities are therefore usually interchangeable. The energy release rate is proportional
to K 2(t). Thus, according to Eq. 1.134, the crack propagates as long as K(t) is larger than
54 Chapter 1 · Fracture Physics

a dynamic critical stress intensity factor called KID. The latter depends on the mate-
rial as before but also on the crack velocity (and probably on the temperature). By
Eq. 1.133 and Eq. 1.136,

(1.137)

which is usually called “crack resistance”.

If Γ and its dependence on v were known, Eq. 1.136, or alternatively

(1.138)

would provide a relation wherefrom the velocity of the propagating crack could be
obtained. Therefore, Eqs. 1.134–1.136 or Eq. 1.138 are commonly used by engineers for
these calculations. There exist, however, several problems with this approach. As men-
tioned earlier, the biggest discrepancy with measurement is that the maximal crack velo-
city is not CR. Experiments yield values of terminal velocities ranging between ~0.2CR
and 0.7CR. But even for small velocities where Eq. 1.136 and Eq. 1.138 are supposed to
be a good approximation to the exact situation, there exist several factors (not included
in Eq. 1.136 and Eq. 1.138) that influence crack speed. We enumerate here briefly some
of these:

1. The rate of loading (Liu et al. 1998; Arakawa and Takahashi 1987).
2. The geometry of the sample (Kobayashi and Mall 1978).
3. The non-uniqueness of the stress intensity factor in defining crack velocities (Dally
et al. 1985).
4. Additional effects (Ravi-Chandar and Knauss 1987).

The prevalent measuring methods of dynamic fracture parameters are described,

e.g., in Yang and Ravi-Chandar (2001), in Fineberg and Marder (1999) and in Knauss
(2000). They include both well-established methods that have been used for a very long
time and new developments of the last ten years.
We return now to the question of maximal crack velocity. Table 1.1 (adapted from
Fineberg and Marder 1999) shows that for homogeneous brittle materials, the maxi-
mal crack speed never exceeds 0.7CR. On the other hand, if we assume that K(t) under
a constant load σ0 is proportional to √⎯⎯c⎯(⎯t) σ0 (as is the case e.g., for an infinite plate)
where c(t) is the instantaneous crack length, Eqs. 1.134–1.136, with the help of Eq. 1.133
yield for a constant Γ

(1.139)

Table 1.1.
Maximal crack velocities in
some homogeneous brittle
materials
 1.4 · Dynamic Fracture 55

Fig. 1.31.
Data of Kobayashi et al. (1974)
in PMMA, showing the discrep-
ancy with linear elastic theory
(full line)

where q is a constant, or

(1.140)

where c0 = Γ / qσ 02 is the length where crack initiation occurred. Equation 1.140 can
be used to find the expected LFEM crack velocity

(1.141)

which is equivalent to Eq. 1.103, with CR as the terminal velocity. Figure 1.31 shows
the experimental results of Kobayashi et al. (1974) mentioned before, together with v
according to Eq. 1.141. This figure dramatically emphasizes the discrepancy between
the LEFM theory and the experimental results. This discrepancy obviously arises from
the processes occurring in the tip vicinity. These concern atomic nonlinear forces, not
included in the linear theory (e.g., i–iv above), which lead to instabilities in the crack
propagation and to the collapse of the LEFM theory.
Besides our simplistic semiphenomenological approach (see Sect. 1.4.1 and
Rabinovitch 1994), there exist at the moment two lines of thought that try to explain
these instabilities. They will be denoted by the “velocity threshold” and the “dissipa-
tion process” (Sects. 1.4.3.1 and 1.4.3.2, respectively).

1.4.3.1
Velocity Threshold, Microbranching

In a series of careful experiments, conducted mainly on PMMA and Homalite but also
on glass samples, J. Fineberg, M. Marder and their group (Fineberg et al. 1991, 1992;
Sharon et al. 1995; Sharon and Fineberg 1998, 1999) measured the accurate changes of
56 Chapter 1 · Fracture Physics

Fig. 1.32. A typical measurement of the velocity of a crack tip in PMMA. a as a function of time:
After an initial jump to about 150 m s–1 the crack accelerates smoothly up to a critical velocity vc
(dotted line). Beyond this velocity, strong oscillations in the instantaneous velocity of the crack de-
velop and the mean acceleration of the crack decreases (Fineberg and Marder 1999). b The steady-
state above vc and the corresponding power spectrum. Note the existence of a well-defined timescale
(from Hauch and Marder 1998)

crack velocities in those materials and the simultaneous appearance of secondary “pat-
terns” around the propagating cracks. It was found that in PMMA at and above a critical
crack velocity of vc = 0.36CR (which is independent of sample geometry, sample thickness,
and the applied stress or acceleration rate of the crack) velocity fluctuations appear
(Fig. 1.32) together with microbranching (Fig. 1.33) emanating from the main crack and
ending inside the samples. Note that vc is smaller than the final (or terminal) velocity. They
have concluded, therefore, that the fluctuations were due to these microbranching or
“branching attempts”. They have also identified these microbranching with the “mist” and
“hackle” patterns found in brittle materials (Fineberg et al. 1991; Sharon and Fineberg 1996).
An additional feature of these velocity measurements in PMMA (not observed in glass) is
a velocity jump at the beginning of crack propagation (Fig. 1.34) (see in this connection
Rabinovitch and Bahat 1979). In a series of “microscopic” calculations Marder and cowork-
ers (Marder and Gross 1995; Hauch and Marder 1998; Marder and Liu 1993) obtained re-
sults that support these experiments. Their models are based on a lattice of points con-
nected by springs whose performance is predetermined. Calculation results show
 1.4 · Dynamic Fracture 57

Fig. 1.33. Images of local crack branches in the x-y-plane in PMMA. The arrow, of length 250 µm,
indicates the direction of propagation. All figures are to scale with the path of the main crack in
white. Top: v < vc; center: v = 1.18vc; bottom: v = 1.45vc (from Sharon et al. 1995)

1. A velocity “gap” – the initial crack velocity is immediately greater than zero.
2. Instabilities arise whenever the crack velocity exceeds a certain threshold velocity –
above which the crack direction bounces and surface roughness is thereby created.

These results agree with some atomistic (e.g., Abraham et al. 1997) and finite ele-
ment (Xu and Needleman 1994) calculations in the identification of a threshold ve-
locity above which crack direction starts to “wiggle”. Such a velocity dependent insta-
bility was already predicted by Yoffe (1951), who showed that the stress field around
the crack tip was velocity dependent: For small crack velocities, the stress field at the
tip perpendicular to r, σθθ , had a maximum at the crack propagation direction (θ = 0),
while above a certain threshold velocity the maximum was shifted to non-zero θ values
(in Yoffe’s calculation θ = ±60°).
58 Chapter 1 · Fracture Physics

Fig. 1.34. Crack velocity vs. time in geometries designed to produce gradual crack arrest. In PMMA
and Homalite, cracks decelerate slowly towards zero velocity on the scale of µs, while in glass, cracks
are able to accelerate slowly from zero velocity and travel stably at very low velocities. These data
show that jumps in velocity from zero are not intrinsic properties of fracture dynamics in amor-
phous materials. Data of Hauch and Marder (1998) (from Fineberg and Marder 1999)

1.4.3.2
Dissipation Processes

In this model, instead of crack velocity, the controlling parameter is the energy re-
lease rate or the stress intensity factor. However, it is the static stress intensity factor
and not the dynamic one which is experimentally seen to cause the different changes
in crack behavior and the corresponding surface patterns (different roughness fea-
tures). It is assumed that the excess energy released from the elastic field is not “used”
solely to increase the kinetic energy of the medium. Rather, it goes into the creation
of microcracks or other damage mechanisms ahead of the propagating crack. Several
investigators in this school of thought are Kerkhof (1973), Johnson (1992), Ravi-
Chandar and Knauss (1984a–d). See also Ravi-Chandar (1998) for a thorough discus-
sion and comparison with the velocity threshold line of thought.
This line of thought is based on the fractographic investigation of fractured sur-
faces of brittle materials (glass, glass ceramic, ceramics and rocks; see Chap. 2 for a
detailed description) in which a universal pattern appears. Figure 2.1b is a represen-
tative schematic picture. It is seen that after the crack initiation (at ci or ccr; see
Sect. 1.4.1) in a sequence the following patterns appear: “Mirror”, “mist”, “hackle” and
“bifurcation”. The “mirror” is a rather flat surface, reflecting light (hence the name).
A careful atomic microscopy measurement, however, shows that “roughness” increases
across the whole range of the mirror, although it stays invisible to the naked eye. Crack
 1.4 · Dynamic Fracture 59

velocity increases gradually (Fig. 1.31) across the mirror and usually reaches its ter-
minal value there. The “mist” is a region of numerous secondary fractures, while the
“hackle” appears as a set of parabolic (or hyperbolic) secondary markings, the planes
of their propagations making acute angles with the primary crack plane. Bifurcations
(or branching) and attempted bifurcations appear only after the mirror-mist-hackle
sequence has occurred (there are however some conflicting observations; Sect. 2.2.3.14)
and ultimately induce a splitting of the crack into two or more branches.
The demarcation lines between the different patterns, i.e., the mirror-mist (rm in
Fig. 2.1) mist-hackle (r) and hackle-bifurcation (rb), are sometimes sharp and some-
times gradual. Many investigators (e.g., Bansal 1977; Johnson and Holloway 1966; Shand
1959) have found that these transition lines can empirically be obtained by relations
of the following form:

(1.142)

where ri can assume the values of rm, r, or rb; σ0 is the remote stress in the sample and
Ai (= Am, A, Ab respectively) are material constants. This relation was used, e.g., by Bahat
et al. (1982) to evaluate the shape of the mirror boundary when σ0(θ ) changed with the
polar angle and by Bahat and Rabinovitch (1988) to evaluate the paleostress in a rock
by the shape of its transition line into the hackle zone (mist was barely discernable there).
The empirical result embodied by Eq. 1.142 is very intriguing. It relates the static
stress intensity factor (which is proportional to σ0√ ⎯ri for an infinite plate) to the tran-
sition in crack behavior. This could mean either that dynamic effects, such as the
changes of KI with velocity (see above) do not count, or that all of these transitions
occur for the same (or nearly the same) velocity. In either case, Eq. 1.142 seems to in-
dicate that the controlling parameter is KI, at least for materials other than PMMA,
and not the crack velocity.

1.4.4
Fracture Pattern Analysis

A complete understanding of pattern formations on brittle crack surfaces is still miss-

ing, although more than a century has passed since they were detected (see Bahat
1991a). The identification of the mist zone as a zone studded by secondary microcracks
is perhaps due to Ravi-Chandar and Knauss (1984a–d) and as mentioned above
(Sect. 1.4.3.2), the relations (Sect. 1.142) were obtained by many investigators but with-
out a firm theoretical foundation. In a series of papers (Rabinovitch et al. 1994, 2000a,b),
we have found some properties of these patterns. These are presently discussed.
Based on the directions of the secondary cracks (Rabinovitch et al. 1993, see
Sect. 1.4.2), we have calculated (Rabinovitch et al. 1994) the lengths of these cracks as
a function of distance across the mist zone.
This calculation was based on two assumptions:

1. A flaw (or a microcrack), which is found ahead of the propagating crack, starts to
grow when the static stress intensity factor at its tip becomes equal to KIc.
2. The primary crack overtakes the secondary crack since the former moves faster,
thereby causing the secondary crack to halt.
60 Chapter 1 · Fracture Physics

The outcome was that the lengths of all secondary cracks should increase across
the mist zone by a common magnifying factor M which itself is growing monotoni-
cally as a function of the primary crack length. Although these results agreed quite
nicely with the roughness experiment by Ravi-Chandar and Knauss (1984c), this agree-
ment showed only a proportionality correlation. Indeed, a painstaking experiment
(Rabinovitch et al. 2000b) showed a rather surprising result. The experiment was con-
ducted on a glass bottle and the lengths of some 12 000 secondary cracks across the
mist zone were carefully measured under a microscope. It emerged that

1. The appearance of secondary cracks was abrupt (at the end of the “mirror plane”).
2. The length distribution of these cracks consisted of two parts – Gaussian for short
lengths (up to ~6 µm) and a negative exponential tail above 6 µm.
3. Length increase across the mist zone did behave like the predicted one with a com-
mon increasing magnification M, but only a small fraction (around 9%) of the sec-
ondary cracks were really magnified.

These features are not understood as of yet.

A more detailed calculation of the interaction between the primary crack and the sec-
ondary cracks in the mist zone was carried out in Rabinovitch et al. 2000a. A penny-shaped
flaw (microcrack) was assumed to start its propagation as a secondary crack when the
static stress intensity factor at its tip, caused by the primary crack stress field, reached KIc.
It was further assumed that the velocity of the secondary crack increased approximately
in an exponential manner, reaching ultimately the “terminal (maximal) velocity” of the
material. Each part of the secondary crack overtaken by the primary one ceased to grow,
since stress there was relaxed. In this way, the two-dimensional shapes of the secondary
cracks could be calculated. Results (Fig. 1.35) showed that these shapes should change along
the way from the mirror to the hackle zones from being circular to elliptic and to “tear-
drop”-like. Since the stress field ahead of the primary crack increases along its way (re-
call that KI ~ σ0 √
⎯2⎯π⎯R, where R is the primary crack radius), the secondary flaws start
to grow as cracks at distances further and further ahead of the primary one. Finally,
they acquire such a “forestart” that the primary crack cannot overtake them anymore.
This point is assumed mark the beginning of the hackle zone. A secondary crack shape
calculation in this zone yields for it a hyperpolic form (Fig. 1.35). Experimental sec-
ondary crack shapes (Fig. 1.36) are seen to agree well with the calculated ones.
These results constitute a step in understanding the mirror-mist-hackle-bifurca-
tion sequence of surface patterns.

Fig. 1.35. Calculated shapes of “final” secondary cracks in the mist and hackle zones, for different radii
R of the primary crack (flaw size c0 = 0.002 mm). x and y scales are in mm. a R = 0.3 mm, the beginning
of the mist zone (oval); b R = 0.4 mm, in the mist zone (elliptic); c R = 0.495 mm, towards the end of the
mist zone (tear drop); d R = 0.6 mm, in the hackle zone (hyperbolic) (Rabinovitch et al. 2002a)
 1.4 · Dynamic Fracture 61

Fig. 1.36.
Experimental results for soda
lime fracture. a The beginning
of the mist zone (oval), scale
bar 2.5 µm; b towards the end
of the mist zone (teardrop),
scale bar 10.0 µm; c at the mist
hackle transition (two semipa-
rallel lines), scale bar 10.0 µm;
d in the hackle zone (neighbor-
ing hyperbolic forms, e.g., A and
B), scale bar 20.0 µm (Rabino-
vitch et al. 2000a)
62 Chapter 1 · Fracture Physics

1.5
Subcritical Cracking
1.5.1
General

This section concerns situations for which the Griffith criterion is not fulfilled, i.e.,
situations where K < Kc, hence the name “subcritical”. Experimentally this area is di-
vided broadly into two parts called “fatigue” and “creep”. The term “fatigue” is nowa-
days used specifically for a situation where stress (or K) is periodically changed be-
tween two values, while the term “creep” is reserved for the description of a gradual
rupture under a constant low stress. The fatigue case, which is very important for
metallic samples, will not be discussed here except to mention the empirical, so-called
“Paris law” for the rate of crack length, c, increase with the number of cycles N (Paris
and Erdogan 1963)

(1.143)

where ∆K = Kmax – Kmin, Kmax and Kmin being the maximum and minimum K values
between which the load is cyclically changed and C and p are material constants. Since
this relation does not always agree with experiments, more sophisticated relations have
appeared (e.g., Suresh 1998).

1.5.2
Creep

This phenomenon of rupture under low stress conditions is also called by many other
names, such as corrosion, stress corrosion, corrosion fatigue, static fatigue, etc. It is
directly related to the question of flaw and crack nucleation, see Sect. 1.2. Since ex-
periments can only be carried out within a finite temporal span (of a maximum dura-
tion of the order of 1 year), geological results can not be directly verified in the labo-
ratory. As we shall show, these results demonstrate an approach to fracture, which is
different than the one governing the behavior of shorter times (higher stresses).
The usual accepted result of creep is that of Wiederhorn (1967) or Evans and
Johnson (1975). See also Wiederhorn (1978). In this approach, the governing param-
eter of rupture is again the stress intensity factor K. Results are given in the form of
crack velocity vs. K-value. This approach applies to the scenario (see Sect. 1.2.3.3) where
a single flaw grows under the external stress field. The whole range of K values
(0 ≤ K < Kc) is customarily divided into four parts. In the lowest K region, K < KIsc, no
crack increase is detected. A controversy exists in the literature whether KIsc = 0 or
constitutes a real (>0) lower bound. The other three ranges of K (Fig. 4.34) are termed
regions I, II and III, respectively. In region I, a very fast increase of fracture velocity
with KI is observed. It is usually empirically described by

(1.144)
 1.5 · Subcritical Cracking 63

where A and n are material constants. A compilation of such constants for rocks can
be found in Atkinson (1984) and Atkinson and Meredith (1987a,b). Region II is called
the stress corrosion zone, where crack velocity changes slowly with KI, the change being
strongly dependent on the properties of the environment, i.e., temperature, water con-
tent of air, material contaminations, etc. Region III is again a transition range (to the
supercritical region), where velocity is initially increased very rapidly and eventually
saturates to the “final” constant (maximal) velocity of the material.
However, this behavior of crack velocity can not be used for fracture under ex-
tremely low stresses. The latter is important especially for geologic situations, where
processes occur very slowly and presumably under extremely low stresses (or strains).
One of the most important manifestations of such processes is the abundant single
layer jointing that has been extensively studied. In the next section we present in de-
tail our contribution to this study and show that the fracture probability under these
extreme conditions is not proportional to KI but rather to σ itself. This result is in line
with a recent experiment by Yoon et al. (2000).

1.5.3
The Dependence of Jointing in Rock Layers Solely on Stress

1.5.3.1
General

We show that the criterion for fracture for the jointed layers is really based only on the
level of stress (σ0) they have endured during the elapsed geological times. It is there-
fore conceivable that such a fracture criterion based on σ0 alone is the prevailing one
for extreme below-threshold-extended-time fractures for all brittle materials. This re-
sult is shown to be in line with results obtained in current experiments on “lengthy” labo-
ratory creep. Note that we are dealing only with joints that were formed quasi-statically. It
is observed that dynamic joints are abundant in granites, whereas quasi-static ones are
prevalent in sedimentary rock layers (Bahat et al. 2003). Joint spacing in such layered rocks
is considered to be originated by two main processes and possibly by intermediate ones.
The first process envisages the layer denoted by A, as being “sandwiched” by “upper”
and “lower” layers, each of which is made of different material than that of A. The neigh-
boring layers, because of their higher compliance, are usually less jointed than A. But they
tend to stretch the latter when they themselves are stretched, thus transferring to it the
stress field. This process is referred to as the Cox-Hobbs case according to the two authors
who analyzed it (Cox 1952; Hobbs 1967). In this process, the three layers are treated as
perfectly connected (glued) to each other and the shear stress transferred at the interlayers
is transposed into tensile stress in the middle layer. The Cox-Hobbs model can be divided
into two parts. The first part, based on Cox (1952), calculates the stress transfer from the
outer layers to A, and the second part treats the process of joint “in filling”, i.e., the addi-
tion of more and more fractures to the layer A as a function of strain (or stress) increase.
The second process by which a layer is stretched and jointed is when it is completely
separated from its neighboring layers and is itself pulled upon, either by geologic (tec-
tonic) occurrences or by pore pressure, or both. This process is called the Pollard-Segall
(Pollard and Segall 1987) process after the two authors who analyzed it. These two
64 Chapter 1 · Fracture Physics

extreme processes of layer fracturing actually appear in nature, as do intermediate

modes in which the “gluing” of the layers is less than absolute and friction exists between
layer A and its neighbors, thus transferring to A only part of the stress in them. We em-
phasize again that we are considering here extremely slow phenomena, which no labora-
tory measurements can exactly simulate. The experiments that have the closest similarity
to them are those related to creep or fatigue (Nomenclature). Although the durations of
both experiment types are of course much smaller than geologic ones, we shall use results
of creep experiments to glean some information about geologic jointing.

1.5.3.2
The Model by Rives et al. (1992)

We start by analyzing the shapes of the joint spacing distributions obtained experimen-
tally. It such an experiment, a “measuring” line (scan-line) is drawn parallel to the layer;
the joints cut it at intersection points; and the distribution of the spacings between these
points is analyzed. It was observed that the shape of these distributions is gradually
changing with the increase of the density of cracks (number of joints per unit length).
The latter is evidently due to the stress magnitude in the layer. The higher this magni-
tude is, the higher the joint density is and vice versa. The prevailing conjecture regard-
ing this change of shape was that of Rives et al. (1992), who claimed that with increas-
ing joint density the distribution passed from a negative exponential one through a log-
normal one to a normal distribution (for very high densities). This conjecture was “sup-
ported” by a numerical experiment where a line segment representing the spacing
measuring line was cut by randomly applied points (representing the intersection points
on the line) with the provision that a new joint could not be closer to an existing one
than a certain distance, d. The latter was used as a representative of the “shadow” area,
namely, the area around an existing crack where it is highly improbable, due to stress
release there, that an additional crack would appear. The interpretation of the numeri-
cal experiment apparently showed that the joint spacing distribution actually behaved
as predicted. However, Rabinovitch et al. (1999a) considered two difficulties in this in-
terpretation of the numerical experiment. They showed that the numerical result of Rives
et al. (1992) did not fit the exact analytical solution of their numerical experiment, which
can be obtained from a 1D car parking process model (CPP) (Widom 1966). Moreover,
the exact CPP results (Fig. 1.37b,c) are shown not to be compatible with existing joint
spacing distribution results, as explained below.

1.5.3.3
The Exact Shadow Model

One of the problems with the CPP model is the analysis of the “shadow”. It is not true
that a new crack is completely prevented from occurring in a close neighborhood of an
existing crack (Bai and Pollard 2000a, see below). The probability of such an event is
only decreased as we approach the existing crack. We denote this probability by q(x),
namely, q(x)dx is the probability that a new crack can appear at a distance between x
and x + dx from an existing crack (examples are shown in Fig. 1.38). Note that for the
CPP model q(x) = 0 for say |x| < b, 2b denotes the excluded area around an existing crack,
while q(x) = λ say, for |x| > b, where λ measures the crack density as we will show later.
 1.5 · Subcritical Cracking 65

Fig. 1.37.
The calculated probability dis-
tribution function p_ of normal-
ized crack spacing l obtained
under “random sequential ad-
dition” of “cracks” onto a scan-
line (CPP). Three curves are
shown for: a Very low crack
density (ρ = 0.02), b an inter-
mediate density (ρ = 0.5), and,
c a density very near jamming
(ρ = 0.74) (Rabinovitch
_ et al.
1999a). Here l = (l / d) – 1,
where l is the intercrack dis-
tance; ρ = (N / L)d, where N is
the total number of cracks and
L is the total length of the scan-
line. “Jamming” or saturation
here means that no additional
cracks can be added
66 Chapter 1 · Fracture Physics

We next calculate the joint spacing distribution, which we denote by p(l) (Fig. 1.39),
namely p(l)dl is the probability that the distance between neighboring joints would reside
exactly in the interval between the distance l and the distance l + dl (Rabinovitch and Bahat
1999). The calculation of this distribution is based on the “shadow” probability q(x), in
the following way. The probability p(l)dl is equal to the product of two terms as fol-
lows: First, the probability that no fracture has occurred between x = 0 and x = l, and
second, the probability that it actually did occur between x = l and x = l + dl. The sec-
ond probability is of course q(l)dl, while the first one is somewhat more difficult to
obtain. Let us denote this “negative” probability by y(l). We presently show that

Let us divide l into N intervals each of length ∆x (= l / N). As mentioned, the prob-
ability of an additional fracture appearing in ∆x at a distance x from the origin (where
a crack resides) is q(x)∆x. Therefore the opposite probability, namely that no crack
appears in ∆x is (1 – q(x)∆x). Since the occurrence probabilities in all intervals are
mutually independent, the ensuing probability that no fracture has appeared in all
N intervals is the product, denoted by “Π”, of the individual ones, namely

(1.145)

Taking the logarithm of both sides we have

(1.146)

where

denotes the sum between i = 1 and i = N.

Passing to the limit as ∆x → 0, we get

(1.147)

where use has been made of the relation

(1.148)

And therefore
 1.5 · Subcritical Cracking 67

Hence, the product of probabilities mentioned before yields

(1.149)

Equation 1.149 is an interesting result. It states in fact that to obtain p(l) one has
only to know the shadow probability q(x). A nice feature of Eq. 1.149 is that it is nor-
malized for any q(x). This feature is proven as follows. We denote

then

and

(1.150)

In Rabinovitch and Bahat (1999), several possible qualitatively different shadow dis-
tributions were considered:

(1.151)

where µ is a parameter measuring the shadow range and α is an exponent defining

the qualitatively different distributions.
For each such q(x) (Fig. 1.38), the related p(l) is given in Fig. 1.39. These possibili-
ties of q(x) were applied to several pure experimental distributions (“pure” here means
that no addition of distributions from different positions was carried out; each distri-
bution originated from a specific layer which went through the same stress history).
The results (Rabinovitch and Bahat 1999) showed that only two types of shadow dis-
tributions appeared, one (Eq. 1.151) with α = 1 and one with α = 3. This fact reminds
us of the two main processes mentioned above for layer fracturing, namely the Cox-
Hobbs process and the Pollard-Segall process. It was shown (Rabinovitch and Bahat
1999) that in fact, the α = 1 distribution was related to rocks for which the Cox-Hobbs
process was the preferred jointing method, while the α = 3 distribution agreed with
the Pollard-Segall one. This result led us to the conclusion that q(x) was proportional
to the stress field σ (x), where x is the distance from an existing crack. This conclu-
sion is discussed in the remainder of this chapter. First we show the agreement of
α = 1 and 3 with the two σ(x) behaviors; in Sect. 1.5.3.4 we discuss the theoretical im-
plications of this conclusion and show their agreement with a field measurement; while
in Sect. 1.5.3.5 a recent creep experiment is shown to lend support to this conclusion.
(a) For the Cox-Hobbs case, σ(x) is given by
68 Chapter 1 · Fracture Physics

Fig. 1.38.
The complying shadow – the
probability q(x) (Eq. 1.151) of
a second crack, a distance x
away from an existing one,
for µ = 5 cm, λ = 0.1 cm–1,
for a α = 1, and b α = 3

(1.152)

where (Cox 1952; Hobbs 1967), δ = (d / 2) (E / GN)1/2, d is the layer thickness, E is its
Young modulus, GN is the shear modulus of the neighboring layers, and σ0 is the re-
mote stress. For short distances, namely x << δ, the exponent can be developed in a
Taylor series to yield σ(x) ≈ σ0 x / δ, which is proportional to x. (b) For the Pollard-
Segall case, σ(x) is given (Pollard and Segall 1987) by

(1.153)

and for small values of x, x << d, (x / d)2 can be neglected with respect to 1 and
σ(x) ≈ 8σ0 (x / d)3 which is proportional to x3. The powers of x, the α, for the two cases
are therefore α = 1 and α = 3, which are exactly equal to the powers of x in the
q(x) relations obtained experimentally.
 1.5 · Subcritical Cracking 69

Fig. 1.39.
Simulated fracture spacing
distributions, p(l) (Eq. 1.149,
1.150), for B = 100, µ = 5 cm,
λ = 0.2 cm–1, for a α = 1, b α = 3,
and c α = 10
70 Chapter 1 · Fracture Physics

1.5.3.4
The Relation between the Shadow and Stress

Hypothetical KI dependence. The assumption that q(x), the probability of the ap-
pearance of a new crack at a distance x from an already existing one, is proportional
to the stress field σ (x) itself (and not say, to KI!) is a very unusual one. To see what
happens for “regular” assumptions let us calculate q for a hypothetical case where the
Griffith criterion is used, the layer is under a constant stress σ0 and the flaw length
distribution is an inverse exponential. Denoting the flaw length by c, it is thus as-
sumed that KI ∝ σ √ ⎯c and that the flaw length distribution p(c), where p(c)dc is the
probability that the flaw length is between c and c+ dc, is given by

(1.154)

where c0 is the average flaw length. The probability that a flaw grows under these stress
conditions is the probability that KI ≥ KIc or that KI2 ≥ KI2c. The latter is equivalent to
the condition that σ 2c ≥ A, where A ∝ KI2c. Thus, the probability that a flaw would grow
is equal to the probability that c ≥ A / σ 2, or

(1.155)

The ensuing q is evidently not linearly proportional to σ(x) and leads to a joint
spacing distribution that is not observed experimentally.

Verification of the q = γ σ· , Relationship. We establish our proposed relation, q = γσ,

in two ways. In the first one we assume this relation to hold, and calculate the exact p(l)
that should arise. This procedure is carried out for the Cox-Hobbs case, and the re-
sults are compared with the field data of Becker and Gross (1996) and shown to agree
quite well with the experimental distribution obtained there. The second verification
of the q = γσ· relation is derived from a recent extended creep experiment which agrees
with this relation for low σ values. The whole procedure will be elaborated here, since
several interesting results, besides the verification of the q(x) vs. σ(x) relation, were
obtained. According to Hobbs (1967), joint spacing should continuously decrease as
long as stress (or strain) continues to act on the fractured layer. Experiments, how-
ever, display the phenomenon of “saturation”, where the density of joints apparently
does not grow above a certain critical value. This value is deemed by some investiga-
tors (see Bai and Pollard 2000a,b for a review of previous results) to depend on the
layer thickness. Bai and Pollard (2000a,b) also challenged the Cox-Hobbs model. This
challenge consists of two parts. In the main part, they carry out a finite-element method
calculation of a layer containing four joints under strain. In contrast to the Hobbs’
assumption that the stress in the layer is always tensile, their numerical simulations
indicate that between the inner pair of joints there might appear compressive stresses.
The latter are obtained for values of the ratio of inter-fracture spacing to layer thick-
ness that are smaller than a certain threshold, which in turn depends weakly on the
elastic parameters of the media and on the overburden stress. Such compressive stresses
 1.5 · Subcritical Cracking 71

would obviously prevent the formation of additional joints in the region where they
operate. In the second part of their challenge Bai and Pollard (2000a,b) show that if
they extend the basic assumptions of Hobbs to the limit, the results do not fulfill the
elasticity theory equations.
We take a different approach and consider Hobbs’ (1967) calculation of the stress
transfer from the outer layers to A to be a valid approximation. As such it does not
have to strictly fulfill all the elasticity equations. On the other hand, we develop a dif-
ferent approach to the infilling process than that of Hobbs and show that a modified
form of saturation does exist within this approach (see below). This would explain the
“closely spaced fractures” for which Bai and Pollard (2000b) invoke an additional
mechanism (flaw size, length and position distribution, possible pore pressure).
We now return to our statistical model for the Cox-Hobbs case. Assuming that
q(x) = γσ (x) where γ is a proportionality constant, we have, by Eq. 1.152.

(1.156)

By Eq. 1.149 we therefore obtain

(1.157)

where z = l / δ and η = γσ0δ.

Equation 1.157 is the joint spacing distribution for the Cox-Hobbs case, according
to our assumption that q(x) is proportional to σ (x). Some remarks are in order regard-
ing the different parameters that appear in Eq. 1.157 (or Eq. 1.160, see below):

1. Lengths are always to be measured in dimensions of δ = (d / 2)(E / GN)1/2. The latter

has dimensions of length and, in a sense, normalizes bed thicknesses. Normalized
spacings z = l / δ can then be compared between different beds.
2. The proportionality constant γ = q / σ ) has dimensions of length · (stress)–1 or of
(surface tension)–1. It can probably be deduced from creep experiments (see be-
low) if the latter can be seen as applicable to geologic processes.
3. The parameter η = γσ0δ is dimensionless. A crude interpretation for it is that it
provides the approximate probability that a second crack would appear within a
distance δ of an existing fracture. Higher η values therefore imply higher joint den-
sities.
4. The parameter γσ0 = η / δ) has dimensions of (length)–1. It measures the probabil-
ity per unit length that a second crack appears a long distance away from an exist-
ing fracture. As such it is a direct measure of joint density.

Equations 1.157 (or 1.160) can be viewed as the one and only joint spacing distribu-
tion both for low and high densities of joints. It is thus neither a “transition” between
exponential, log normal to normal distributions, nor is it the CPP distribution of
Fig. 1.37. The exact distribution, Eq. 1.157, is shown in Fig. 1.40 for several joint densi-
ties. The latter are represented by the parameter η. Note that the shapes of these distri-
butions could easily mislead one to assume that the formerly mentioned “transitions”
72 Chapter 1 · Fracture Physics

Fig. 1.40. Theoretical spacing distributions, p(z) (arbitrary units) for the Cox-Hobbs case (Eq. 1.157).
Lengths z are measured in dimensions of δ: z = l / δ. a for η = 0.1, b for η = 1.0, c for η = 10.0. Note that
η is proportional to the remote stress in the layer and to δ. d for η = 0.5, 1, 3, 5, 10; here distributions are
shown on the same scale to emphasize the fast change of the peak with η between 0.5 and 3 and the
slow change for η between 5 and 10

does occur. Now, before comparing this result to creep measurements, we want to pause
a little at this point in order to derive some interesting features of this distribution
and discuss the significance of the only two parameters that appear therein, namely δ
and η. First, let us remark that since all lengths are scaled by δ, any statistic such as
the average or the median should also be proportional to δ. We next discuss the be-
havior of the distribution median, its peak and the related interesting question of joint
“saturation”.
We start with the question of what “saturation” is and what is its origin. Saturation arises
by the following two opposing influences. On the one hand, a larger stress results in smaller
joint spacing. On the other hand, the “shadow” prohibits spacing from going to zero. To
better understand this question, consider Eq. 1.157 and Fig. 1.40. It is seen that the distri-
bution becomes more and more concentrated as η (or σ0) increases. Yet the position of
the peak and the mean of the distribution only slightly decrease with η (for high η val-
ues), and it seems that they would never become zero. Actually, as σ0 increases, the addi-
tional amount of stress needed for the same incremental decrease in spacing greatly in-
creases. To calculate the position of the peak of the distribution as a function of η, we equate
to zero the derivative of p(l) with respect to l. This yields exp(–z) – η (1 – exp(–z))2 = 0 or,
denoting ν = exp(–z) and solving for the quadratic equation, we get
 1.5 · Subcritical Cracking 73

Fig. 1.41. Distribution peaks, zmax, and medians Z (both measured in dimensions of δ) as functions
of η. Note the fast change of zmax and Z values for small η values, where new cracks are relatively easy
to form for even a small change of stress; and the very slow (logarithmic) change of these values for
larger η (effective saturation) where increasing the crack density becomes harder and harder

(1.158)

hence, the distribution peak is

(1.159)

where only the negative sign was chosen, in order to agree with the fact that exp(–zmax)
should be less than one. Figure 1.41 shows the change of zmax as a function of η. It is
seen that although zmax tends to zero with η (spacings decrease with stress increase),
this tendency is very slow, logarithmic in fact (but not hyperbolic). Therefore, although
no real saturation occurs, zmax presents an “effective saturation” behavior, (which is in
line with experimental results (quoted by Becker and Gross 1996, Fig. 2) on polysty-
rene plates and cross-ply glass where the mean fracture spacing was measured as a
function of stress (strain) and deemed to be hyperbolic (Rives et al. 1992)). In these
experiments, excessive additional stresses are required for the attainment of additional
cracks. Note that the change of zmax and median values (see below) are gradual and
the transition from the range of “easy” crack density increase to the range of “effec-
tive saturation” also occurs gradually, (say, in the range 5 < η < 15) rather than through
a threshold value of η (σ0) (as suggested by Bai and Pollard 2000a).
74 Chapter 1 · Fracture Physics

The median of the distribution L = δ Z can easily be calculated numerically. Results

are shown in Fig. 1.41, where Z is shown as a function of η. It is seen that the behavior
of the median is similar to that of the mean, as expected. Since Z = L / δ, which is pro-
portional to L / d, Z is actually proportional to the FSR parameter (Gross 1993) (called
by Bai and Pollard the “fracture spacing to layer thickness ratio”). The parameter Z is
(within the Cox-Hobbs’s model) the “correct” normalized fracture spacing ratio. If Z
would have been a constant, then the slope of L as a function of d (more accurately
of δ ) would have been constant, leading to a constant value of the FSI (Narr and Suppe
1991) which is the slope of the best fit line for points depicting L vs. d for layers of
different thicknesses. According to Fig. 1.41, this is not so. The median is proportional
to δ only as long as η does not change significantly between different layers. This im-
plies that the straight line obtained for FSI measurements from different layers (Narr
and Suppe 1991) is somewhat fortuitous. For an almost “exact” linear dependence,
both σ0 and δ would have to be the same for the different layers, and the spread of
layer thickness for the different layers should not be too large and should lie in the
larger η (“effective saturation”) region.
This study shows that there are only two parameters that can be obtained by the
analysis of spacing distributions, i.e., δ and η (or δ and γσ0). The value of δ can also
be calculated by the layer thickness and the elastic properties of the rock. Therefore,
η becomes the most important measurable parameter. Its value

1. Determines the position of the fracture population of a particular layer with re-
spect to effective saturation.
2. On comparing different layers, can provide ratios of remote stresses causing frac-
ture in them. It is seen (Fig. 1.40 and 1.41) that spacing distribution depends on the
maximum stress that has acted in the rock layer for a significant period of time.

We now show that the distribution of Eq. 1.157 is indeed compatible with the his-
tograms of joint spacing measured by Becker and Gross (1996, Fig. 9) of the Turonian
Gerofit Formation, southern Israel. The comparison is outlined here for their Sect. I,
and the results of all sections are given subsequently. Consider the histogram of Sect. I
(Fig. 1.42a). The l-axis stretches between zero and 80 cm, in increments of 3 cm. The
bars have heights (number of points) as follows (obtained from Becker and Gross 1996,
Fig. 9): 1, 4, 9, 14 etc. (since we would like to treat bars with widths of 1 cm each (unit
length), each number should actually appear three times with 1 / 3 height). The sum
of all spacing is 110. Now, in order that the distribution be normalized (the sum of all
bars be equal to one), each number should be divided by 3 × 110 = 330. These num-
bers (* in Fig. 1.42a) are compared to Eq. 1.157 written in the form

(1.160)

in a “least square” fashion (such as “curve fit” in Matlab). Here, the two parameters
sought are γσ0 and δ. Results for all four sections are shown in Fig. 1.42 and the values
of γσ0, δ and η = γσ0 δ are summarized in Table 1.2.
As seen, the agreement of the data to Eq. 1.160 is very good, especially the shapes
of the distributions. However, the parameters γσ0 , δ and η of sect. II are seen to be far
 1.5 · Subcritical Cracking 75

Fig. 1.42. Analysis of results of Becker and Gross (1996) spacing measurements. Sections I–IV are
given in a–d, respectively. * are measurements and full lines are fits (Eq. 1.160). Note that spacing is
not normalized to δ (measured in cm) and that p(l) is normalized (see text)

Table 1.2.
Physical parameters of joint
spacing (based on data from
Becker and Gross 1996)

off from all the rest. As discussed by Becker and Gross (1996), this section experienced
extreme stress conditions and the data thereof is therefore exceptional. Moreover, these
data are composed of several subsections, and such composition is not advisable (see
below). Some remarks are in order.

1. The “normalized layer thickness”, δ, of sections I, III and IV is ~15.3 ±4 cm and

the stress field in sect. III is somewhat higher than these in sect. I and IV (which
are rather similar). In order to better understand the results of sect. II, the histo-
gram of sect. IIa was analyzed separately in Fig. 1.43 (the histogram of sect. IIb is
dubious since (a) this is the section with the largest spread of σ, and (b) the num-
ber of events does not add up to the quoted 120). As seen from Table 1.2, even sect. IIa
76 Chapter 1 · Fracture Physics

Fig. 1.43.
Section IIa of Becker and Gross
(1996). Signs as for Fig. 1.42

shows already much higher parameters than the other three, suggesting uneven-
ness of layer thickness, stresses and/or elastic parameters.
2. As mentioned, Table 1.2 shows that δ for sect. I, III and IV has an average of
15.3 ±4 cm. The latter value could be used as the range of error in the measurement.
To check this result, we have calculated δ by its definition δ = (d / 2)(E / GN)1/2. The
values of E and GN for limestone and shale, respectively were taken from Engelder
and Peacock (2001) to be 66 ±10 and 6.5 ±0.9, respectively. For a layer thickness of
19 cm (Becker and Gross 1996) the value of δ, calculated by its definition, becomes
29 ±4.2 cm. The discrepancy between δ = 15 cm and δ = 29 cm can be interpreted
to represent the deviation of the elastic properties of the rocks in the Gerofit For-
mation from those borrowed from Engelder and Peacock (2001).
3. It is seen that the only genuine length-normalizing coefficient is δ, being the in-
herent physical variable of the Cox-Hobbs spacing. Nevertheless, regarding the
question of the addition of distributions from different layer positions and dif-
ferent lithologies, it is clear (from Fig. 1.40) that such a procedure is not advis-
able. For example, an addition of two, even normalized, distributions of types ap-
pearing in Fig. 1.40a and 1.40c would create a “distribution” having a fictitious
double peak.
4. The values of η obtained for sect. I, III and IV are all around 1.5, placing these sec-
tions deep inside the “unsaturated” region in Fig. 1.41. The high η value of sect. II
indicates that this section has “suffered” a much higher stress (strain) during its
history, bringing it (or at least the IIb part) into the saturated region.
5. A simple test of the self-consistency of the results can be carried out. Consider, e.g.,
sect. III. A direct measurement from the histogram yields the median to be
L ≈ 11.5 cm. Hence, Z = L / δ ≈ 11.5 / 11.37 = 1.01. By Fig. 1.41, this Z value leads to
η ≈ 1.7, which is close to the 1.41 obtained by the least square method.

1.5.3.5
Creep Experiments, a Close Analogy to Slow Geological Fracture

We turn now to creep experiments (see e.g., Poirier 1985). These experiments are usu-
ally conducted at high temperatures and low stresses. Due to the limited period of
 1.5 · Subcritical Cracking 77

measurement, however, stresses are usually not as low as those prevailing in geologi-
cal layers. In creep experiments, the measured quantity is the creep (strain) rate, ε· ,
measured by a scanning laser technique. Three stages of creep are usually discerned
(primary, secondary and tertiary) by the magnitudes of the respective ε· . But since in
geological processes, the stresses generally involved in single-layer jointing are as-
sumed to be small (see below), we restrict ourselves here to the primary stage. Usu-
ally for creep in ceramics, the dependence of creep rate on stress and temperature is
represented by the so-called Norton equation:

(1.161)

where B is an empirical constant, Q an activation energy and n the stress exponent,

which normally ranges between 1 and 2. However, as was pointed out by Yoon et al.
(2000) this empirical formula is inadequate to describe low stress creep both in ten-
sion and in compression. Although the experiments of Yoon et al. (2000) were carried
out only for silicon nitride, we assume that they hold for other ceramic materials as
well as for polycrystalline rocks. Their results show that although the creep processes
within a material under compression were different from those operating under ten-
sion, in both cases the first part of ε· (i.e., for small strains, in the “primary” stage) be-
haved in a similar way with respect to stress: ε· was in fact proportional to the stress σ.
Note that the obtained behavior for tension was actually

(1.162)

but for low stresses, exp(ασ) ≈ 1 and ε· is again proportional only to σ.

To recapitulate, according to Yoon et al. (2000) for a brittle material both under ten-
sion and under compression, the response in creep is that ε· , the rate of change of strain,
is proportional to the stress σ, when σ is small enough. Since for many joints fracture
processes have continued for very long times, it is conceivable that the stresses present
were quite small; otherwise, all jointings would have been in the “extreme saturation”
state (see above, Sect. 1.5.3.4), which is not the case. We therefore assume that the layer
creep rate is given by

ε· = σ a (1.163)

where a is a constant. Consider now a layer in which a single joint exists and is under
a stress σ (t), which is quite small but can change with time. Hence, after eons have
passed, the present strain in the layer is given by

(1.164)

_
where σ is the average stress.
This relation should also be true for each segment ∆l around the point x. The strain
in the vicinity of x, ε(x) can be approximated by
78 Chapter 1 · Fracture Physics

(1.165)

where ∆N(x) is the number of fractures in the segment ∆l, ∆ is the opening of a crack,
and it is assumed that the increase in length of the segment is due only to crack open-
ings. Measuring x from an existing joint, q(x) is evidently proportional to ∆N(x) / ∆l
and hence proportional to ε (x), which is in turn proportional to σ(x).
This argument is obviously somewhat crude, since we have neglected the changes
in σ by the newly added joints. If, however, the density of cracks is not very high, this
approximation is reasonable.

1.5.3.6
Summary

Several important results have been achieved regarding joint spacing.

1. The Rives et al. (1992) conjecture that the joint spacing distribution (JSD) is chang-
ing with joint density from negative exponential to log-normal and to normal was
shown to be incorrect.
2. The 1D car parking process distribution, which is a derived distribution supposed
to simulate the joint spacing one, is shown to be inadequate since it completely
ignores the shadow effect.
3. The JSD is shown to be dependent only on q(x), where q(x)dx is the conditional
probability that if a crack exists at x = 0, another crack would appear between x
and x + dx away from it.
4. The JSD is exactly calculated from q(x) (Eq. 1.149).
5. An empirical method to analyze experimental joint spacing results is presented.
Using the method for several experimental cases revealed that only two processes
lead to jointing, the so-called Cox-Hobbs and Pollard-Segall. The method is shown
to possess the ability to find out which process was operative in a specific layer.
6. A bold assumption is made, partly based on indication from a recent creep experi-
ment under extremely low stress loads, that q(x) is proportional exclusively to the
stress level σ (x). This assumption is verified by comparing the analytical JSD cal-
culated under this assumption with field data (Becker and Gross 1996). The JSD
analytical calculations were carried out only for the Cox-Hobbs case, while the
treatment of the Pollard-Segall process is to be carried out in the future.
7. For the Cox-Hobbs case, a complete JSD is obtained, valid both for high joint den-
sities as well as for low ones. Contrasting a recent work by Bai and Pollard (2000a,b),
the obtained JSD does not discard Hobbs calculation. It just amends his infilling
approach and leads to a new understanding of the saturation phenomenon.
8. Saturation is shown not to be a certain lower level threshold below which joint
spacing would never occur. It is rather an infilling process in which joint spacing
always decreases when strain (or stress) increases, but when joint spacing is high,
the rate of change of spacing with stress is high, while, when it becomes lower, the
rate of change decreases logarithmically, i.e., it “costs” higher, and higher stress
increments for the same amount of joint spacing decrease. There is a clear distinc-
 1.5 · Subcritical Cracking 79

tion between unsaturated and saturated jointing, although the transition from the
first to the second occurs gradually.
9. It is shown that the only two parameters governing the JSD are δ, which is a nor-
malized bed thickness and η, which is proportional to the highest stress level pre-
vailing in the layer. Experimental results can yield only values of these two param-
eters. The parameter η can be used to compare stress levels at different layers as
well as to locate the position of the joint spacing in the measured layer with respect
to saturation (that occurs for η values above ~10).
Chapter 2

Elements of Fracture Geology

2.1
Introduction
It is often quite a challenge to distinguish between fractures induced in rocks by shear
or by tension when examined in micro scales (features that are unseen or almost un-
seen with the naked eye) or in outcrop scales (between centimeters and tens of meters
in size). We take this challenge and concentrate in this chapter on fractures that are
predominantly induced by mode I. Thus, fracture geology is limited in the present con-
text to various phenomena of fracture growth, primarily by tension (or extension). This
is a constraint that we essentially impose on the other chapters in this book, i.e., Chap. 2
stems from a treatment of principles of fracture physics in Chap. 1 and provides a base
for the treatment of fracture provinces on regional scales in Chap. 3 and 4, also con-
centrating on tensile fracture.
The issues of fracture nucleation and growth in rocks will be raised first, followed
by descriptions of fourteen experimental investigations of fractography in technologi-
cal materials that have bearing on fractography in rocks. The presentation of the data
will emphasize the distinction of morphological features in the mirror plane from those
in the fringe (Fig. 2.1a–d). Observations on fracture surface morphology from sedimen-
tary rocks as well as from granites will come next. The treatment of two outstanding
questions in structural geology and in rock mechanics, namely, conjugate hybrid joints
and longitudinal splittings, will end Sect. 2.2. The classification of systematic joints into
the burial, syntectonic, uplift and post-uplift four genetic groups (Bahat 1991a) pro-
vided reference for subsequent studies on fracture in rocks, for us and for other au-
thors (e.g., Engelder et al. 1993; Ghosh 1993; Bankwitz et al. 1994, 1997). We expand on
the nomenclature of uplift and post-uplift joints and introduce the class of “surface
joints” in Sect. 2.3. The fascinating multi relationships between primary joints/faults
and secondary fractures are far more understood today than one or two decades ago
due to recent publications on this issue (Peacock 2001a and references therein). This
topic is addressed in Sect. 2.4.

2.2
Fractography and Tectonofractography
2.2.1
Introduction

Fracture initiates at stress concentrators in rocks and in synthetic, brittle and semi-
brittle materials (as well as in some non-brittle materials, Bahat 1991a, p. 67). The frac-
82 Chapter 2 · Elements of Fracture Geology

Fig. 2.1. Schematic representation of various fractographic elements on a fracture surface and distin-
guishing the mirror from the fringe. a The mirror radius, r (arrow) is measured from the critical flaw to
the inner boundary of the en echelon fringe. b The mirror radius, rm (arrow) is measured from the
critical flaw to the inner boundary of the mist. Branching may initiate either from the outer rim of
hackles at the right or from the inner rim of hackles, at the left (forming hackle and branching initia-
tion together). c A fracture surface in bitumene (from De Freminville 1914). Note a few radial striae
within the mirror plane and a multitude of hackles beyond it. Many hackles appear as echelon cracks,
extensions of the striae. d Schematic fracture surface of brittle materials showing idealized initial flaw
length 2cci and depth ai, critical flaw length 2ccr and depth acr. The three mirror radii, rm (mirror-mist
boundary) r (mist-hackle boundary) and rb (initiation of microscopic crack branching) are shown as
well (modified from Mecholsky and Freiman 1979)

ture then propagates through a great number of disturbances caused by material het-
erogeneities (e.g., flaws) and inhomogeneous stress fields, emitting sound waves. The
sound waves interact with the advancing fracture front, and the results are recorded on
the fracture surface by fracture surface morphologies, also termed fracture surface to-
pography, fracture markings, or in short, fractography (e.g., Kerkhof 1973; Hull 1999).
Every individual fracture is characterized by its specific fractography. Woodworth (1895,
1896) set the stage for the science of fracture surface morphology by astute observa-
tions of the morphology of joints in geologic exposures. He also noted the essential
morphologic similarities between fracture markings in glass samples and rock out-
 2.2 · Fractography and Tectonofractography 83

crops. Fractography is that branch of science that analyses fracture surface morphol-
ogy and related features and their causes and mechanisms in technological materials.
Fractography also contributes significantly to the understanding of fracture processes
that form joints. The fractographic analysis helps to define on the fracture surface the
following (Fig. 2.1a–d):

1. Fracture origin.
2. Properties of the initial flaw and of the critical flaw.
3. Modes and directions of fracture propagation.
4. Stress configurations.
5. Failure types (including, fatigue, brittle, ductile, impact and thermal influences).
6. Fractographic symmetric elements.
7. The distinction between the mirror plane (the parent joint) and the fringe.
8. Estimating the local paleostress.

This analysis establishes whether the surface features reflect a single or multistage
fracture process and provides tools for fracture mechanics analysis, fracture toughness
properties and related parameters of joints. From all the excellent photographs of mir-
ror planes found in the literature (e.g., Holloway 1986), we select to show the ingenious
discovery by De Freminville (1914, Fig. 2.1c). Tectonofractography is a branch of tec-
tonics and tectonophysics, which applies fractographical analysis to rock fractures and
to regional fracture systems (joint sets). The objectives of tectonofractography are to
determine the fracture paleostress directions and the mechanical conditions involved
and identify the tectonic processes that produced the fractures. Both fractography and
tectonofractography were discussed extensively in Bahat (1991a). This discussion is
expanded in subsequent sections of this volume, while minimizing repetitions from the
previous volume to mere essentials.

2.2.2
The Origin: Crack Nucleation via Fractographic Techniques

2.2.2.1
Different Applications of the Term Nucleation

The terms crack (or fracture) origin, crack initiation, nucleation and “Griffith flaws”
are often used interchangeably in the literature, relating to vast scale ranges. Lawn
(1993, p. 307) distinguishes two ranges of “Griffith flaws” dimensions from which fail-
ure of brittle materials initiates. In homogeneous materials like monocrystalline quartz,
flaws range from 1 nm (in pristine fibers and whiskers) to 1 µm (in aged, as-handled
solids) and usually occur at the surface. In heterogeneous polycrystalline materials
like ceramics, flaws range from 1 µm (in high-density, fine grained, polished materi-
als) to 1 mm and above (in refractories and concrete) and occur in both the surface
and the bulk. Flaws vary in size, shape and orientation such that the coexistence of
more than one flaw type often gives rise to a bimodal or even multimodal flaw popu-
lation. Lawn (1993, Table 9.1) provides a rather liberal-size classification of “Griffith
flaws”, inversely correlative to strength. In very strong materials (10 GPa for whiskers)
the flaw size is in the nm range, whereas in very weak materials (1 MPa for the Earth’s
84 Chapter 2 · Elements of Fracture Geology

crust) the flaw size is in the m range. An experimental study of fracture in a transpar-
ent glass ceramic reveals two oppositely propagating striae from an origin that can-
not be precisely determined in a spot. An area of 1–2 × 1–2 mm is measured instead
and termed “nucleation zone”, where the bilateral fracture originated (Sect. 2.2.3.7).
See a further elaboration on nucleation in micro scales in Sect. 1.2.

Joints. There are earth scientists who search fracture nucleation in the rock microcrack
(Simmons et al. 1975) grain size scale (e.g., Segall and Pollard 1983). In the Devonian
layered siltstones and shales of the Appalachian Plateau, central New York, and in the
Eocene chalks from the Beer Sheva syncline, many joint initiation points are located
along bedding boundaries (Bahat and Engelder 1984; Helgeson and Aydin 1991;
Lacazette and Engelder 1992; Weinberger 2001a). These initiations occurred at stress
concentrators such as free surfaces, fossil inclusions, pyrite concretions, voids, cusps,
and burrows along the boundaries. Many other joint initiations in these two provinces
occur from stress concentrators within the rock layer, remote from its boundaries
(Weinberger 2001a). The exact location of joint nucleation within granite bodies can
be fractographically identified on the Borsov joints that formed by slow propagation
(Sect. 4.4.2) but not on the Mrákotin joints that formed rapidly (Sect. 4.4.3). On a larger
scale, early fan-shaped cracks that were marked by plumes ranging in size from a few
meters to tens of meters on surfaces of exfoliation joints interacted with each other and
coalesced into composite fractures on vertical cliffs of granitic rocks at Yosemite Na-
tional Park, California (Sect. 4.5.3.3). The orientation and arrangement of the fans in-
dicate that they initiated in “nucleation zones” at certain heights, from where they propa-
gated upwards and downwards parallel to the cliff of El Capitan monolith.
The treatment of joint initiation has been idealized from a Griffith internal crack
model in glass (either elliptic or slit shapes). In moving from a model of a slit (Griffith
1924), the shape of the crack has been thought to be an oblate (Sneddon 1951, p. 486)
or a prolate spheroid. The “empty” crack model (that does not have a mass) was modi-
fied to one in which fluid pressure is activated from the interior on the crack walls.
Secor’s theory (1965, 1969) suggests that the growth of an extension fracture at depth
in the Earth’s crust is a slow process, consisting in detail of numerous short episodes
of crack propagation that alternate with longer periods of quiescence. Pore fluids from
the surrounding rock percolate into the crack during quiescence and then wedge it
open. Secor (1969) considers that the failure criteria for tension fracturing of rocks
with internal pore pressure P in plane strain is

(2.1)

where σe3 is the least effective principal stress and –S is the tensile strength of the rock
(tension considered negative) for the general pore pressure condition σej = σj – P, (Secor
1969), where c, σej, σj, γ and ν are the penny-shaped crack radius, the effective princi-
pal stress, the principal stress, the fracture surface energy and the Poisson ratio, re-
spectively. Fracture initiates when a fraction of P exceeds σ3 and becomes effective in
wedging open the crack. This will occur when fluid in the crack reaches a “critical
volume” (Secor 1969):
 2.2 · Fractography and Tectonofractography 85

(2.2)

The cancellation of the traction parallel to the crack is done for convenience. The
resemblance of Eq. 2.1 to the known Griffith equation (Eq. 1.21) implies fracture by
extension that according to Secor can form joints at considerable depths. Field obser-
vations support Secor’s model of a fracture process by alternating episodes of crack
propagation and periods of quiescence (Bahat and Engelder 1984). A remaining out-
standing question, however, is what the range attained by KI in jointing is. Slow frac-
ture processes in sedimentary rocks cannot be well characterized by K (Sect. 1.5.3),
while in granites, in some cases there are indications of jointing under KI ≥ KIC condi-
tions (Chap. 4).

Faults. Segall and Pollard (1983) investigated nucleation and growth of strike slip
faults cutting granite. They suggested that faults nucleated on pre-existing joints that
acted as weak surfaces capable of accommodating later shear deformations and mod-
eled fault nucleation from sub parallel arrays of joints. Triaxial tests have been used to
study the nature of the “damage zone” that propagates ahead of the growing fracture
(Hoagland et al. 1973). While the “damage zone” describes a zone at the tip of an exist-
ing fracture that consists of multiple cracks oriented with respect to the fracture, “nucle-
ation zone” (see above) relates to an area where fracture nucleation unrelated to an
existing fracture occurs. These two terms are occasionally combined. Lockner et al.
(1991) and Reches and Lockner (1994) applied triaxial tests to simulate nucleation of
faults by the coalescence of axial microcracks into an inclined “process zone” under
mixed mode conditions. The latter triaxial experiments were performed on intact gran-
ite cylinders 190.5 mm long and 76.2 mm in diameter, and planar quasi-circular “pro-
cess zones” were suggested to be the nucleation sites. In both the outcrop scale and the
triaxial test (Segall and Pollard 1983; Reches and Lockner 1994), the nucleation zones
consisted of arrays of cracks, and in the two models, the fault growth continues along
the orientation of the array.
Studies of nucleation of earthquakes (e.g., Rice 1979; Rudnicki and Olsson 1998;
Gudmundsson and Homberg 1999) are carried out on seismograms (e.g., Umeda 1990;
Ellsworth and Beroza 1995). Based on friction parameters, Scholz (1998) distinguishes
two regimes, stable and unstable, and suggests that earthquakes can nucleate only in
those regions of a fault that lie within the unstable regime. During the inter-seismic
period, the fault is loaded by steady slip on the deep, stable portion of the fault. Just
before the earthquake, a pre-seismic phase occurs. In this phase, known as nucleation,
slip accelerates until instability results in coseismic motions. The nucleation phase is
sometimes associated with precursory phenomena, such as foreshocks. A model of
earthquake nucleation is considered for the weak crust of California by Bahat et al.
(2001a). They suggest that the multi fault rupture pattern in California shows greater
resemblance to multi fracture patterns of weak chalk cylinders broken in triaxial tests
than to a single process zone in fractured strong granite. The rupture of the chalk cyl-
inders does not start from a localized nucleation; it rather develops from many dis-
tinct sites by a coalescence process. Possibly, the nucleation of earthquakes in a weak
terrain occurs from many centers rather than a single site.
86 Chapter 2 · Elements of Fracture Geology

2.2.2.2
Fractography

We restrict the treatment of crack initiation into a rigid framework, which is limited
by visible and measurable fractographic parameters. A key feature is the “critical flaw”
(Fig. 2.1a,b), which is measured as 2ccr (Fig. 2.1d). When 2ccr is attained, crack growth
occurs spontaneously, whereas below this length nucleation is unstable and the crack
is expected to heal (in glass in vacuum). Thus, the critical flaw is the upper boundary
of the crack nucleation stage. The relationship in Fig. 1.12 is analogous to the maxi-
mum in free energy plotted against the radius of a nucleated new crystal. Below a “criti-
cal radius size” the nucleus tends to be unstable and dissolve, and beyond it crystal
growth occurs spontaneously (Carmichael et al. 1974, Fig. 4–14). This analogy should
be useful particularly for demonstrating the distinction between rates of nucleation
and growth (Bahat 1970).
The “critical flaw” can be identified experimentally, as it grows from the “initial
flaw” 2ci (Fig. 2.1d and 2.2), showing that 2ccr may reach sizes around 10 mm for an
opening mode fracture in glass. It was found (Mecholsky and Freiman 1979) that the
ratios of the mirror radii (rm, r and rb, Fig. 2.1d) to the critical flaw radius have corre-
sponding constants. However, fracture results obtained by various investigators of glass
and ceramics have yielded considerable discrepancies in these ratios, leading to the
suggestion that a poor distinction between 2ci and 2ccr was possibly an important factor
contributing to these discrepancies (Bahat 1991a, p. 109). Fortunately, occasional dif-
ficulties in defining the boundaries of 2ccr in geological outcrops may be trigonometri-
cally overcome (Fig. 6.18).

Fig. 2.2. Microphotograph of an initial flaw (white arrow pointing upward) growing to concentric
ripple marks A, and radial scars in glass. These markings reveal that the growth of the initial flaw is
controlled by a heterogeneous stress field. Curve B marks the transition to the mirror plane at the tip
of the critical flaw (black arrow pointing downward). Lengths of diameters of initial flaw and critical
flaw from ends to ends are 0.9 mm and about 1 cm, respectively (after Bahat et al. 1982)
 2.2 · Fractography and Tectonofractography 87

Thus, in an attempt to provide a common denominator for comparison, where pos-

sible, we make the following distinction. The transition from sub-visible defects or
singularities (undetected in the optical microscope) to the initial flaw 2ci is at 2 µm
(Lawn 1993, p. 13). The initial flaw may reach several mm (Segall and Pollard 1983).
Its transition to the critical flaw 2ccr varies in the mm range but can reach the cm scale
on joints (Bahat and Rabinovitch 1988). These two transitions belong to the nucle-
ation stage, while propagation from 2ccr is taken to mark the initial growth (see also
Doremus and Johnson 1978; Rabinovitch 1994). Occasionally, initial flaws may devi-
ate from a single plane and can lead to different parts of the flaw acting essentially
independently, creating more than one critical crack, such that each one would gener-
ate its own part of a mirror plane. The latter effect seems to increase with the decrease
of the average angle of the initial flaw relative to the tensile axis (Bahat 1991a, Fig. 2.33d
and 2.35c), resulting in the increase of shearing and the reduction of relative tension
on the growing crack during the nucleation stage (Fig. 2.2).

2.2.2.3
Fracture Initiation in Stratified Muddy Sediments

In this and in the next section we cite studies by Weinberger (1999, 2001a,b) on frac-
ture initiation in sedimentary rocks. These systematic studies improve our understand-
ing of fracture origin in rocks and help to focus on the difficulties involved in differ-
entiating joint nucleation from joint initial growth.
Persistent desiccation and contraction of muddy sediments give rise to mud cracks.
The fracture characteristics of mud cracks were studied in dehydrating mud puddles
in the Dead Sea region, Israel (here we cite extensively Weinberger 1999). The mud
consists mainly of carbonate and clay particles typically displaying fining upward.
There is a distinct surface discontinuity between the upper desiccated layers (that tend
to contract and crack) and the uncracked lower ones. This surface is referred to as the
bottom of the polygons. Mud cracks with delicate plumes formed several days after a
rainstorm and ceased to propagate a few days to weeks after initiation. Several poly-
gons of convenient size and weight (mostly containing a pair of desiccated layers) were
then systematically lifted out of the dried puddles and their walls examined under
optimal oblique illumination. In this way, the surface morphology of hundreds of mud
cracks were studied in precise detail, based on fractographic techniques. The mud
cracks form networks of interconnected tension fractures arranged in polygonal pat-
terns. Because tensile stress due to drying declines downward through the sediment,
mud cracks have generally been theorized to nucleate near the surface, propagate
downwards, and terminate at depth.
Weinberger (1999) found that systematic nucleation at the bottom of the polygons
and upward propagation of mud cracks are much more prevalent than previously
postulated. The mud-crack initiation points are often located along or near the base
of the desiccated sediment, where potential flaws (e.g., long grain boundaries) are more
abundant because of the natural fining-upward sorting of grains (Fig. 2.3a). Conse-
quently, these large flaws become critical before small flaws at the top (e.g., short grain
boundaries), even though the tensile stress caused by drying declines downwards
through the mud (see below). On growing, the cracks predominantly rupture the des-
88 Chapter 2 · Elements of Fracture Geology

Fig. 2.3. a Grain-size distribution in the studied sediment displaying fining upward (sorting effect)
and stratification. This distribution was obtained by gently disintegrating the grain aggregates of
each layer and sieving the grains through screens of decreasing mesh size (from Weinberger 1999)

Fig. 2.3. b Oblique view of a polygon, a photograph above and a drawing below, explaining propaga-
tion directions by arrows. The plume shows that the crack nucleates at depth, ruptures the desiccated
layers, and subsequently propagate bi-laterally away from the origin along curved paths. Minor dam-
age caused to the polygon wall during recovery (left corner in photograph) is not presented in the
surface-morphology drawing. Note in the photograph the dendrite-like plume that is divided into
branches that are further sub-divided into minute barbs (from Weinberger 1999)
 2.2 · Fractography and Tectonofractography 89

Fig. 2.3. c Schematic illustration of stress distribution throughout a polygon wall. The least compressive
principal stress σ3 is everywhere perpendicular to the crack plane, but the greatest compressive principal
stress σ1 and the intermediate compressive principal stress σ2 vary in orientation from place to place
on that plane. Next to the crack origin, where the crack extends predominantly vertically, σ1 is vertical and
perpendicular to bedding, on account of the weight of the overlying sediment. Away from the crack origin
at the region where the crack extends predominantly laterally, the in-plane horizontal stress due to
contraction seems to be more significant than the in-plane vertical stress due to the weight of the sedi-
ment, and σ1 and σ2 switch directions. Plume formation and segmentation along layer boundaries are
associated with the indicated rotations of the local principal stresses (from Weinberger 1999)

iccated layers before they significantly extend laterally and sequentially form the po-
lygonal patterns (Fig. 2.3b,c).
Thus, the consistent location of crack origins at the bottom of the polygons strongly
suggests that stress concentration due to flaw discontinuities and layer boundaries play
a fundamental role during mud-crack nucleation. Consequently, the stress gradient
due to drying (possibly related to capillary forces) may be less important for mud-
crack nucleation than has been previously postulated. Returning now to the Griffith
model, Weinberger explains that mud-crack propagation consumes energy in the form
of surface energy for the creation of a new crack surface. This energy comes from the
release of elastic strain energy within the drying mud. In this mechanism, the only
mechanical energy available to drive a crack is the change in elastic strain energy
(Eq. 1.19), which must decrease while the surface energy increases (Eq. 1.20) during
crack growth. Adhesive forces along the bottom of the polygons resist the horizontal
contraction of the mud. This resistance gives rise to stresses along the bottom and causes
the elastic strain energy to be stored. Since crack growth strongly depends on the stored
energy, the boundary effect probably plays a key role not only for crack nucleation but
also for crack propagation. Hence, Weinberger (1999) describes nucleation and growth
processes where actual ci and ccr (Fig. 2.1) are not identified. It is, however, quite pos-
sible that nucleation in the mud took place at sites that ranged between ci and ccr. Note
that the large grains along the bottom of the polygons exceed the size of 0.062 mm
(Fig. 2.3a). Thus, his interpretation of both the nucleation and growth processes is quite
convincing. See the experiment on profiles of “mud cracks” in Sect. 2.2.3.9.
90 Chapter 2 · Elements of Fracture Geology

2.2.2.4
Fracture Initial Growth in Dolomite Layers

Weinberger (2001a) investigates the well-developed surface morphology of joints in

the Soreq Formation, Judea Group (Arkin and Hamaoui 1967; Upper Cretaceous, cen-
tral Israel) that uniquely records the history of fracture initial growth in layered rocks.
Weinberger (2001a) uses fractography to elucidate the role of macroscopic flaw dis-
tribution in the shape of cavities, on the jointing mechanism in natural conditions,
and to examine the effect of bedding interfaces on joint growth and arrangement. The
formation of the cavities is related to the dissolution of earlier anhydrite nodules in
the dolomite by marine-derived water. The study by Weinberger (2001a) is very de-
tailed and is cited extensively below. We differ from Weinberger in using the term nucle-
ation. For us, joint initiation from 2ccr is growth rather than nucleation (as mentioned
above). There are cases where a distinction between the nucleation and initial growth
stages can be made reasonably well (e.g., Fig. 2.2b), whereas in other instances this
distinction is unclear. It could be that nucleation started from certain 2ci sites within
the cavities while the anhydrite nodules still resided in them (Weinberger 2001a). It
appears more likely however that both nucleation and initial growth occurred from
2ccr locations around the peripheries of the cavities after the removal of the anhydrite
nodules.

Fig. 2.4. Graphic representation of the outcrop west of Givat Ze’ev, displaying arrays of joints with
diagnostic surface morphologies. Joints are selectively confined to certain layers, commonly with a
differing fracture surface morphology. Layer 2 and layer 3 are not fractured by a systematic joint set.
Gray ellipses represent non-critical cavities; black ellipses represent critical cavities. Cross-joints and
definition of “overlap” and “offset” are indicated at the top of layer 7 (from Weinberger 2001a)
 2.2 · Fractography and Tectonofractography 91

Fig. 2.5.
a A photograph that illustrates the
three-dimensional structure of
cavity-driven joints in layer 1.
Concentric arrest marks close to
the critical cavity become ellipti-
cal away from it. Some of the nearly
radial plumes become coarse in as-
sociation with the elliptical arrest
marks (arrows) (see scale in b).
b Drawings of joints in the Soreq
Formation. Joints seldom cross
bedding interfaces. Upper views of
type 1 (layer 5) and type 2 (layer 7).
Joint initiation points at layer 5 are
commonly located within the layer
and at layer 7 along the upper part
of the layer. A small fault is younger
than the joints. Middle, side view
of type 3 plumose structure in
layer 4. Joint initiation points are
located within the layer. Lower,
view of type 1 in layer 1. Three
adjacent joints and their associ-
ated critical cavities are indicated.
Numbers near critical cavities
refer to cavities in Fig. 2.6. Arrows
indicate propagation directions.
See also legend in Fig. 2.4 (from
Weinberger 2001a)

Cavities associated with joint initial growth in the dolomite layers are termed here-
after “critical cavities” and their respective fractures are termed “cavity-driven joints”.
A cavity in the present context approximates the “critical flaw” (Fig. 2.1a,b,d): The
cavity’s horizontal semi-axis, Ch (corresponding to ccr) and vertical semi-axis, Cv (cor-
92 Chapter 2 · Elements of Fracture Geology

Fig. 2.6. Representative distribution and dimensions of cavities in layer 1 (Fig. 2.4 and 2.5b) along an
8 m E-W traverse. The small steps along the traverse (Fig. 2.4) are not indicated. Bars represent the
in-plane axes of the ellipsoidal cavities. Cavity centers are located at the intersections of the bars.
Numbers denote critical cavities that are associated with joint initial growth (from Weinberger 2001a)

responding to acr) are measured from the center of a cavity to its periphery, defined
by the intersection of the respective fracture plane and the cavity rim. The cavity
“depth”, Cd, is measured from the center of the cavity to its rim normal to the fracture
plane. The characteristic size of a cavity, Z, is defined as the mean of Ch and Cv.
Fourteen adjacent critical cavities in layer 1, one in layer 5, and five in layer 7 and
their associated joints were carefully measured for the analysis (Fig. 2.4–2.5). For
comparison, all cavities along an 8 m long traverse in layer 1, including six critical
cavities and 58 non-critical cavities, were also studied in detail (Fig. 2.6). An additional
interdependent variable, the degree of isolation, can be derived from these measure-
ments. These variables are analyzed below in an attempt to characterize features of
flaw criticality.

Shape. Critical cavities are approximately spherical, with C h = 45 ±20 mm,

Cv = 43 ±20 mm, Cd = 45 ±23 mm, and in-plane Ch / Cv ≈ 1 ratios. Non-critical cavities
are also approximately spherical, with C h = 35 ±27 mm, C v = 30 ±23 mm, and
Cd = 32 ±23 mm (Fig. 2.7a). Noticeably, several eccentric cavities (Ch / Cv ≈ 2) appear-
ing close to the base of layer 1 are non-critical, e.g., an ellipsoidal cavity at a distance
of 4.40 m along the traverse (Fig. 2.6). If a cavity deviates from a perfect sphere, the
major semi-axis commonly coincides with Ch. This is particularly the case for critical
cavities, where only in one case the difference between Cv and Ch is larger than 3 mm.
 2.2 · Fractography and Tectonofractography 93

Fig. 2.7. Analysis of cavity shape, size, and distribution in layer 1. a In-plane horizontal vs. vertical semi-
axes of critical (gray) and non-critical (black) cavities along an 8 m E-W traverse in layer 1 (Fig. 2.6).
b Histograms of sizes of critical (gray) and non-critical (black) cavities in plane 1. c Origin location of four-
teen cavity-driven joints with respect to the local layer thickness. d Histograms of isolation indices of critical
(gray) and non-critical (black) cavities along an 8 m E-W traverse, layer 1 (Fig. 2.6) (from Weinberger 2001a)

Size. The mean characteristic size of all critical cavities (Z = 44 mm) is larger than
that of non-critical cavities (Z = 33 mm). More details are gained from histograms of
cavity sizes in layer 1 (Fig. 2.7b). The frequency of critical cavities is different from
that of all cavities; relatively small and very large cavities are not associated with joint
initiation. Indeed, no critical cavity is among the top five largest and the fifteen small-
est cavities.

Location within layer. To test the null hypothesis that cavities (critical and non-criti-
cal) are randomly distributed with layer 1, the section presented in Fig. 2.6 was di-
vided into 48 area units of 0.5 × 0.5 m each, and the number of cavities in each unit
counted. This procedure provided the observed data for a standard Chi-square test on
this cavity population and helped to calculate the expected data based on a Poisson
distribution (Davis 1973), with λ equal to the mean size of all cavities. The test con-
cluded that the null hypothesis should be rejected at 95% confidence. Thus, cavity
distribution is non-random, reflecting the tendency of cavities to be more abundant
toward the layer base. Additionally, of all cavities, critical cavities are generally lo-
cated closest to the top of the layer. Only 5 out of 58 non-critical cavities are located
more closely to the top of the layer than the critical cavities (Fig. 2.6 and 2.7c). In
addition, all five are smaller in size than the smallest critical cavity (Cb = 15 mm,
Cv = 15 mm and Cd = 10 mm).
94 Chapter 2 · Elements of Fracture Geology

Isolation. For each cavity, the distance, Dj, was measured from the center of cavity j to
the center of the in-plane nearest neighbor, and the number of neighbors Sj within a
radius of 0.5 m around the center of cavity j was counted. The dimensionless index of
in-plane isolation for cavity j, Ij, is defined by

(2.3)

A high isolation index implies a large distance to the nearest neighbor and a lim-
ited number of neighbors. The results indicate that the isolation indices of critical
cavities are typically one order of magnitude greater than those of non-critical cavi-
ties (Fig. 2.7d), i.e., critical cavities are much more isolated than most of the non-critical
cavities. Noticeably, several isolated cavities appearing close to the base of layer 1 are
non-critical, e.g., a small cavity at a distance of 6.00 m along the traverse (Fig. 2.6).

Initial growth of joints. Weinberger (2001a) found that the cavity shape plays a minor
role in governing the joint initiation points in the Soreq Formation. On the other hand,
the cavity size has a noticeable influence on the joint initiation points, because the
frequency of critical cavities differs significantly from that of all cavities (Fig. 2.7c) and
the mean size of critical cavities is larger than that of non-critical cavities. Noticeably,
for two closely located cavities of similar shape, degree of isolation and distance from
the layer base, the largest cavity is the one associated with joint initiation (e.g., compare
critical cavity 2 and its nearest neighbors to the east at a distance of 0.7 m along the
traverse, Fig. 2.6). These observations suggest a cavity size effect, whereby larger cavi-
ties preferably fail before smaller ones. This size effect is complicated by post-fractur-
ing dissolution processes that might obliterate the exact cavity size at initiation. None-
theless, this later dissolution might enlarge all cavities, including the non-critical cavi-
ties. Although theoretically the stress concentration factor (SCF-q in Eq. 1.22) of circular
hole is 3.0, independent of the size (Lawn 1993, p. 3), actual strength measurements in
rock mechanics experiments indicate that sample strength decreases with increasing hole
size (Anderson 1995, p. 339). Therefore, the SCF effects of the larger hole act over a wider
distance, indicating that the volume over which the stress acts is important for hole failure
and rock strength. Other studies have also shown that rock strength decreases with in-
creasing flaw size and flaw density (Wong et al. 1996; Wong and Chau 1998; Bazant and
Planas 1998). These size effects, however, are obscured by the fact that they depend on the
deformation processes, which in turn depend on the loading conditions (Li et al. 1999).
Indeed, in the Soreq Formation, only one critical cavity is among the top five largest cavi-
ties, indicating that other factors play a more significant role than the size during jointing.
The distribution of cavities in layer 1 is non-uniform, with cavities clustered to-
ward the layer base. This, in turn, leaves relatively isolated cavities near the middle of
the layer, many of which are critical (Fig. 2.6). For two cavities of similar shape, size
and distance from the layer base, the more in-plane isolated cavity is the one associ-
ated with joint initiation (e.g., compare cavity 6 and non-critical cavity at a distance
of 3.45 m along the traverse, Fig. 2.6). The analysis of the stress in the vicinity of a hole
in an elastic material indicates that the stress distribution is markedly affected by the
hole only within an area of about three radii from the center of the hole (Timoshenko
and Goodier 1951). Weinberger (2001a) arrives at the interesting implication that
 2.2 · Fractography and Tectonofractography 95

closely located cavities may inhibit joint initiation because of the stress perturbation
induced by adjacent cavities. On the other hand, in more plausible configurations,
cavity-driven joints in less dense regions near the middle of the layer might commu-
nicate with other cavities, facilitating joint growth caused by stress enhancement.

Three types of ripple marks. Weinberger (2001a) distinguished three types of ripple
marks (Fig. 2.5). Type 1 consists of circular-to-elliptical marks and is formed in layers
with plentiful cavities. Type 2 consists of semi-circular-to semi-elliptical marks and is
formed only in “cavity-free layers”. Type 3 consists of marks with approximately simi-
lar curvature and is formed in relatively thin layers (<0.4 m).

Summary: Joint growth after initiation from ci and ccr sites. In summarizing the obser-
vations from the study by Weinberger (2001a), the analysis of cavity geometry and distri-
bution reveals that joint initiation at ccr locations is governed by isolated, relatively large
cavities, preferably located close to the bedding top and by stress gradients during joint-
ing. In the absence of macroscopic cavities, sites of joint initiation vary in size from ci to ccr
along bedding interfaces (layer boundaries). Consequently, joints typically show several
forms of growth, depending on the abundance and spatial distribution of cavities within
the layers. In layers with plentiful cavities, joints preferably initiate at critical cavities in the
layer interior and propagate vertically upward and downward and horizontally and form
circular to elliptical fractures that abruptly end (Fig. 2.4) or more commonly arrest either
at layer boundaries or at surfaces of previous joints (Fig. 2.5b). In layers free of cavities,
joints commonly nucleate at layer boundaries and propagate downward toward the layer
base and adjacent joints (Fig. 2.4, layer 7). Alternatively, they may nucleate at surfaces
of previous joints or contacts between joints and layer boundaries (Bahat 1991a, Fig. 4.8).
The various forms of growth in cases where cavities are not apparent are summa-
rized schematically in Fig. 2.8a,b A–H, first defining joint dimensions (Fig. 2.8a).
Figure 2.8b A–D show fractographic manifestations of various initiation and growth
styles without fringes in single-layer joints. These joints often form fringes along one
or two layer boundaries (Fig. 2.8b E, F). Less common are en echelon fringes along the
distal continuations of joints (Fig. 2.8b G), possibly because the fracture slanting mecha-
nism (Sect. 2.2.5.2) does not promote segmentation there. Very rare are hackle fringes
or cuspate hackle fringes (Fig. 2.42) along the distal continuations of single-layer joints
in sedimentary rocks (Fig. 2.8b H). In a rare case of a hackle fringe in a sedimentary
layer (Fig. 2.30a), the fringe does not occur in a distal position. This general observa-
tion, based on thousands of inspections of joint fractographies, provides a strong ar-
gument in suggesting that the joints under consideration mostly form under low or very
low effective stresses (Sect. 1.5.3).
The division into two forms of fracture growth by Weinberger (2001a) corresponds
quite well to an analogous division into two types of fracture growth in technological
materials. Breakage in good quality glass bottles most often initiates at the outside wall
and propagates inside (towards the center of the bottle), forming semi-circular frac-
ture (Fig. 2.1d), because stress concentrators (e.g., cavities) inside the body are absent.
On the other hand, in low-quality materials that contain discontinuities (e.g., cavities/
voids), fracture may start from inside the glass body (e.g., Bahat 1991a, Fig. 2.34d) and
propagate along directions complying with constraints imposed by the boundary con-
ditions, much as they control fracture in sedimentary layers (Fig. 2.8).
96 Chapter 2 · Elements of Fracture Geology

Fig. 2.8. Several styles of joint growth in a rock layer. a Definition of thickness, width and length dimen-
sions (a–c, respectively) of a single-layer joint. b Several growth forms of joints, as described by plumes
and ripple marks, from left to right: A: An initial circular joint; B: A full joint growth to an elliptical
mirror, limited by the layer boundaries; C: An advanced quasi-elliptical joint that arrests along an ear-
lier fracture; D: An advanced joint that initiates at a corner formed by a layer boundary and a previous
joint; E: A semi-elliptical joint initiation at the upper layer boundary and propagation downward creat-
ing an en echelon fringe; F: A full joint growth to an elliptical mirror, forming en echelon fringes along
the layer boundaries; G: A joint like F that forms en echelon segmentation at its distal ends; H: A hypo-
thetical joint F that forms hackles at the distal ends (modified after Bahat 1998a and Weinberger 2001a)

2.2.3
Fractographic Experiments

This section brings together a series of fractographic experimental studies (mostly

since 1991) that concern fracture areas beyond the critical flaw (Fig. 2.1). The first stud-
ies (Sect. 2.2.3.1–2.2.3.7) pertain to fractures that are contained within the boundaries
of the mirror plane, while the other studies (Sect. 2.2.3.8–2.2.3.13) relate to fractures
that reach the fringe beyond these boundaries. The fractographic results of these stud-
ies have implications regarding various tectonofractographic issues, which are dis-
cussed in subsequent sections.
 2.2 · Fractography and Tectonofractography 97

2.2.3.1
Experimental Arrest Marks

According to Guin and Wiederhorn (2003), Michalske (1977) carried out on soda lime
silicate glass microscope slides the following experiment (Fig. 2.9a–e). First, he propa-
gated a crack in water at a stress intensity factor KI of 0.375 MPa m1/2, which was clearly
above the static fatigue limit K0 (Fig. 4.34). Then, he reduced the applied stress inten-
sity factor to 0.225 MPa m1/2, a value clearly below the static fatigue limit, which was
inferred from crack growth studies on soda lime silicate glass (Wiederhorn and Bolz
1970). He held the stress intensity for a period of 16 h. Finally, he increased the stress

Fig. 2.9.
Optical micrographs obtained
by using the height mode, to
illustrate the crack deflection
along the crack front of crack
arrest marks left on fracture
surfaces as a consequence of
holding the crack for a period of
time at a stress-intensity factors
that lies below the “fatigue limit”.
a KIh = 0.24 MPa m1/2, th = 67 h,
KIr = 0.37 MPa m1/2;
b KIh = 0.12 MPa m1/2, th = 48 h,
KIr = 0.37 MPa m1/2;
c KIh = 0.12 MPa m1/2, th = 1 h,
KIr = 0.35 MPa m1/2, where KIh,
th and KIr are the hold KI below
the static fatigue limit, the hold
time and the hold KI above the
static fatigue limit, respectively.
The white arrows on the figure
give the direction of crack pro-
pagation. d and e images by
atomic force microscope from
the circled region of the crack
arrest mark shown in b using
the deflection mode to show the
change in slope along the crack
front. The crack approaches the
arrest mark on a common plane.
On repropagation, the crack has
discontinuously split, some ar-
eas along the crack front lying
above (white) the incoming sur-
face, some areas lying below
(dark gray) the incoming sur-
face. In e the bright or dark hands
indicate a sharp alternate change
in crack growth (from Guin and
Wiederhorn 2003)
98 Chapter 2 · Elements of Fracture Geology

intensity factor to the value used for crack propagation, 0.375 MPa m1/2. The time to
repropagate the crack was about 2 000 s. Michalske found crack arrest marks on the
fracture surface at the site of the arrested cracks that he attributed to restarting crack
growth. No such marks were observed for a crack that had its stress intensity factor
momentarily reduced and then increased again.
Following up on Michalske’s (1977) study and duplicating his experiment, Wieder-
horn et al. (2002) and Guin and Wiederhorn (2003) found that the arrest marks ap-
peared not to be the result of blunting, but represented what we term a “split-discon-
tinuity” in the crack surface so that after the hold period below the crack growth thresh-
old, the crack developed waviness that was not there when the crack was arrested
(Fig. 2.9b,d,e). By matching both halves of the fracture surface, these authors showed
that the split upper and lower surfaces matched across the marks left at the crack front.
Within the experimental accuracy of the technique, ±2 nm, crack blunting did not
occur. The discovery by Wiederhorn et al. (2002) and Guin and Wiederhorn (2003) of
the split-discontinuity shows a subtle but significant mixed mode I and III loading,
additional to mode II along arrest marks that form when the crack restarts growing
(Fig. 2.9e). The implication of this section is discussed in Sect. 2.2.4.2.

2.2.3.2
Experimental Undulations

Richter and Kerkhof (1994) used a method of superposing shock waves or continuous waves
onto a static or quasi-static tensile field, causing a crack to propagate in soda lime silicate
glass from a starter crack (edge-notched) induced into the glass by thermal or mechanical
techniques. This procedure allowed them to create visible, concentric stress wave mark-
ings (undulations) on fracture surfaces. The undulations formed in successive positions
of crack fronts at definite points in time by applying the stress waves at different modu-
lated frequencies, and this enabled them to calculate the velocity of crack propagation from
10–6 m s–1 up to the maximum velocity of 1 540 m s–1 (Kerkhof 1975). This method has been
successfully applied to other materials as well (Richter and Kerkhof 1994).
Figure 2.10 shows a fracture surface of a single edge-notched tensile specimen gen-
erated at a stress wave frequency of 5.02 MHz. The fracture surface displays a blunt,
starter crack, SC and zones of mist, M, hackle, H, and crack branching, B. The undula-
tions convex in the direction of propagation; they are superimposed by thinner and
sharper Wallner lines at different orientations. Also visible are a few straight striae in the
mist and hackle zones. In this method, an increase in distances between undulations indi-

Fig. 2.10. Fracture surface of a single edge-notched tensile specimen showing zones of mist, M, hackle, H,
and crack branching, B. Stress wave frequency: 5.02 MHz. SC = starter crack (after Kerkhof and Richter 1969)
 2.2 · Fractography and Tectonofractography 99

cates increase in crack velocity. In this example, a relatively high load had to be applied to
start propagating the blunt starter crack. Consequently, the crack acceleration was high,
leading to wide intervals between the undulations almost throughout the fracture surface,
and the maximum crack velocity of 1 540 m s–1 was attained. The crack branched beyond
the hackle zone after having passing a certain distance at almost a constant crack velocity.
After a phase of crack branching, a region of smoother fracture surface occurred.

2.2.3.3
Oscillating Cracks

Yuse and Sano (1993, 1997) applied thermal stress along thin borosilicate cover glasses
that varied in size around 100 × 20 × 0.1 mm by heating them to high temperature, Th
(maximum 400 °C), in a hot region. The cover glasses were then moved slowly down
into a cold region, Tl (18 °C) at velocity v (maximum 50 mm s–1), while the distance
between the upper hot region and the lower cold region was fixed. If v was slow enough,
the part of the sample in the hot region was at temperature Th and the part in the cold
region at temperature Tl, imposing a temperature difference ∆T ≈ Th – Tl, and maximum
thermal stress between the two regions (Fig. 2.11a). Thus, the control variables of the
quasi-static crack propagation were ∆T, v and W, the sample width. Various types of
crack patterns were formed by different control parameters. Generally, increasing the
values of the control parameters changed the crack morphology from simple to com-
plex, falling into three types with increasing complexity:

1. Straight cracks (Fig. 2.11b).

2. Oscillating cracks (Fig. 2.11c).
3. Branched cracks (Fig. 2.11d,e).

Fig. 2.11. a Configuration of the experiment. A thin sample plate moves slowly across the temperature
gradient gap in the direction of the arrow with velocity v (after Yuse and Sano 1993). b–e Several crack
patterns. b Straight cracks; c two oscillating cracks; d, e branched cracks (after Ferney et al. 1999)
100 Chapter 2 · Elements of Fracture Geology

Fig. 2.12.
The morphology of oscillating
crack changes from simple to
complex with increasing con-
trol parameters. Arrows show
initial cusps. The width is
scaled by two (after Yuse and
Sano 1993)

We shall concentrate on the oscillating cracks. While both undulations (Sect. 2.2.3.2)
and oscillating cracks bend along the crack propagation direction (Fig. 2.11 and 2.12,
respectively), undulations appear to superpose as small, secondary bends on primary
crack surfaces, while oscillating cracks are larger and exhibit the actual shape of the
primary crack. For v and ∆T near the onset of oscillation, the oscillating cracks were
almost sinusoidal (Fig. 2.12a). With increasing v or ∆T, the shapes resembled a series
of semi-circles (Fig. 2.12b). After that, the crack became asymmetric in the propaga-
tion direction. For large v or large ∆T the crack propagated in an almost horizontal di-
rection at a very high speed until it stopped at some distance from the side edge, then
grew slowly upward, forming a cusp (Fig. 2.12c). For larger parameters, additional os-
cillations appear quasi-periodically or double periodically with cusps (Fig. 2.12d).
Yuse and Sano (1997) demonstrate that under severe thermal stresses, cusps replace
oscillations within the range where they become asymmetric. Murgatroyd (1942) ob-
served symmetric cusp marks in fractured glass that was caused by a low-energy im-
pact (Bahat 1991a, Fig. 2–20a). He suggested that the crack propagated upward before
coming to rest and resumed its course in a slightly downward direction when it re-
commenced. An asymmetric cusp-like profile was induced in glass by thermal stresses
(Bahat 1977). These observations suggest a need to reconcile the occurrences of asym-
metric ripple marks and cusps in both slow and fast fracture propagation.

2.2.3.4
The Distinction of Tensile from Shear Fracture Surfaces in Laboratory Tests of Chalk

A full understanding of developments in fracture geology is often hampered by our

limited ability to distinguish between “pure” tensile loading (mode I) and processes
controlled by shear stresses and mixed mode loading (of modes I, II and III in differ-
ent combinations), particularly in small sizes (mm–cm). This section relates to this
problem. A detailed fractographic technique that was applied on experimentally frac-
tured chalk enabled the establishment of a clear distinction between tensile and shear
surfaces (Fig. 2.13–2.17). In their study, Bahat et al. (2001a) tested chalk cylinders
 2.2 · Fractography and Tectonofractography 101

Fig. 2.13. Various types of brittle fracture developed experimentally in chalk. a, b Profile sections.
a Longitudinal splitting by uniaxial stress in continuation of initial conic fracture induced by boundary
effects at the end of the sample. b Shear fracture by a triaxial test, showing formal straight conjugate cut
(dashed line) and actual curved fracture. c, d Parallel sections. c Derived by uniaxial test, showing a
shear surface above, separated by a boundary from a tensile surface subdivided into domains below it.
d Meso-fracture, cutting throughout the sample, formed by a triaxial test, showing adjacent domains
marked by ridges and grooves in different orientations (from Bahat et al. 2001a)

Fig. 2.14. Photographs of fractured chalks under triaxial stresses (cm scale). a A shear surface divided
into distinct domains in which ridges and grooves slightly change their orientation. b A section shows
three parts from bottom to top, the original cylinder, a tensile crack (gray) and a large shear crack
(white), respectively. The shear part displays conjugate Lüders’ bands (from Bahat et al. 2001a)
102 Chapter 2 · Elements of Fracture Geology

Fig. 2.15. Photographs of fractured chalks under uniaxial stresses (cm scale). a A longitudinal split
dominated by an axial tensile fracture. The two tensile parts are displaced in order to show a late
inclined shear fracture (between upper and lower boundaries) that crossed the split. Sub-axial plumes
mark the split. b Reconstruction of an original cylinder (at upright position, below) and the quasi-
cone (axis parallels to a) (from Bahat et al. 2001a)

100 mm long and 52 mm in diameter under both uniaxial and triaxial loading. At the
end of this section, we present results of a recent study on the pore radii of this chalk
(Rabinovitch et al. 2003a).

Summary of results. The results of the above study are condensed below as follows.

1. The drying and desiccation procedures of the chalk used in the study resulted in
high compressive strength results (σ1–σ3) ranging from 37.7 MPa to 52.6 MPa.
2. With the exception of a few cylinders that fractured either by a single shear frac-
ture or by strongly dominated longitudinal splittings, most samples under both
triaxial and uniaxial loading failed by a combination of shear and tensile fracture
(Fig. 2.13–2.15).
3. The authors identified several “key surface morphologies” according to shear and ten-
sile categories (Fig. 2.13). Shear features include a sugary texture, ridges and grooves
(RG) and steps (Fig. 2.14). The RG appear on sheared white sugary, ragged surfaces
that form curved topographies within boundaries of individual domains (of about 3 to
10 cm2 in size). RG in neighboring domains may form various angular relationships
with the cylindrical axis and are either parallel or sub-parallel to the fracture dip. An
individual domain on a shear surface is often subdivided into sub-domains (of about
 2.2 · Fractography and Tectonofractography 103

Fig. 2.16.
a A display of half of the total
number of fracture surfaces
(one of each matching couple)
from a sample. Shear surfaces
are marked by ‘–’ and tensile
surfaces by ‘+’. Scale bar is 4 cm.
b Relationship between the
number of fractures (one from
each two matching fractures)
and total area of all fractures.
c Relationship between the ra-
tio of (fractured area by shear) /
(fractured area by extension),
and lateral compression, σ3
(from Bahat et al. 2001a)

0.5–1 cm2 in size) that occur at different topographic elevations. RG maintain the same
orientation in passing from sub-domain to sub-domain, indicating the continuity of
fracture. A sub-domain consists of a set of microcracks (≤ 0.1–4 mm long) that are
aligned along the RG that cross sub-domain boundaries within individual domains.
No wing cracks (Sect. 2.4.2) were found on the fracture surfaces.
Tensile features include plumes (or striae) and stairs (Fig. 2.15). The plumes
appear as braids on planar tensile surfaces of a matt white-grayish color and on the
micro-scale the barbs are sub-parallel to the cylindrical axis. In tensile domains
(that are of approximate sizes of shear domains), “stairs” result in a division of
large surfaces into smaller planar sub-domains at different elevations, ranging in
size from 0.5 cm2 to 2 cm2. Sub-domains maintain the general plume orientation
within a domain. Domains of all tensile surfaces are planar and smooth, either
parallel or sub-parallel to the cylindrical axis.
104 Chapter 2 · Elements of Fracture Geology

Fig. 2.17.
Schematic summaries of six
modes of crack propagation
from two experiments on weak
rocks. a, b In uniaxial compres-
sion and c, d in biaxial compres-
sion. Thick inclined lines are pre-
cuts, zigzag inclined lines are new
shear fractures and subvertical,
slightly curved thin lines are wing
cracks (from Bobet and Einstein
1998, Tables 3 and 4). e In
uniaxial compression, and f in
triaxial compression from the
present study (without pre-cuts):
Shear fractures show ragged pro-
files and are inclined, and tensile
fractures are straight and paral-
lel to σ1, not showing wing cur-
vature (from Bahat et al. 2001a)

4. Whereas tensile longitudinal splitting (Sect. 2.2.13) can stretch continuously on a

parallel tensile plane or on adjacent parallel planes, shear failure into meso-frac-
tures (that divides the cylinder into two pieces) combines a series of distinct do-
mains on separate surfaces that slightly differ in curvature and orientation. Thus, a
key feature of single shear fractures (Fig. 2.14a,b) is their strong deviation from
planar into concave and convex topographies (Fig. 2.13b). Therefore, surfaces of
shear fractures have considerably larger areas than the “nominal” ones, whereas in
tensile fractures “actual” and “nominal” surfaces are similar.
5. The reconstruction of the original cylindrical samples was accomplished by as-
sembling the pieces (Fig. 2.15b), which enabled the determination of the fracture
sequences that resulted in our experiments by the cross cutting relationships. Four
fracture sequences were distinguished:
a several stages of longitudinal splitting,
b initial quasi-cone and subsequent longitudinal splitting,
c longitudinal splitting that initiated from around the sample peripheries and
subsequent inclined shear fracture, and
d longitudinal splitting that initiated from around the sample center and subse-
quent inclined shear fracture. Generally, the sequence on shear surfaces is in-
cipient shear fracture displayed by RG that occurs before total loss of cohesion,
which is then followed by displacement shown by steps.
6. Conic and incomplete “quasi-cones” (Fig. 2.15b) may be obtained under both
uniaxial and triaxial tests. The patterns of the tensile fractures often vary with the
shape of the quasi-cones from two large matching longitudinal splits to repeated
subdivisions into radial longitudinal splitting.
7. The tensile and shear areas that formed were quantified on each fractured sample
(Fig. 2.16a). The total fracture area increases with the number of cracks (and frag-
ments) (Fig. 2.16b). The ratio of “fractured area by shear” to “fractured area by
tension” tends to increase with confining pressure σ3 (Fig. 2.16c). Triaxial failures
resulted in a few fractures (some 1–12 pieces) and in ratios of total shear area to
tensile area >1, while uniaxial failure resulted in many fractures (some seven to
eighteen pieces) and in ratios of total shear area to tensile area <1 (Fig. 2.16c).
 2.2 · Fractography and Tectonofractography 105

8. There is a distinct trend of length increase of the longitudinal splits with the de-
crease of the confining pressure, as shown below by three samples: From 1 cm (when
σ3 = 5 MPa) through 2 cm (when σ3 = 2 MPa) to 4 cm (when σ3 = 1 MPa). This cor-
relation shows that tensile longitudinal splitting occurs also under triaxial stresses,
where the extent of the tensile fracture is inversely controlled by the confining
pressure.
9. Most large fractures are composites of fracture domains. Fractures formed under
both uniaxial and triaxial compression show that there is a strong correlation be-
tween the orientation of cracks (small fractures) and their mode of propagation:
Shear cracks are inclined to σ1 (Fig. 2.13b), while tensile cracks are parallel or sub-
parallel to σ1 (Fig. 2.13a and 2.17) The transitional angle between the two is αc = 8 ±2°.
This correlation possibly supports previous suggestions that early microcracks are
linked by inclined shear cracks (e.g., Hoek and Bieniawski 1966) rather than by ten-
sile cracks (see implications of this observation in Sect. 2.2.12).
10.The authors correlated variations of the electromagnetic radiation (EMR) with
changes of crack sizes in the fractured chalk throughout the investigated stress
range (see elaboration in Chap. 5).
11.The multi-fracture complexes exhibited by the chalk samples bear a greater resem-
blance to surface ruptures that display wide belts of shear zones (associated with
earthquakes that occurred in weak crusts) than to single slip surfaces simulated in
strong rocks. The possible implication is that fracture in weak crusts is simulated
better on weak rocks than on strong ones.

Microcrack populations and orientations. Stress induced microcracks that consti-

tute RG on shear fractures generally are parallel or sub-parallel to the fracture dip,
i.e., they may be tilted up to about 10° right or left with respect to the fracture dip (on
parallel sections) (Fig. 2.13c–d). Likewise, barbs that constitute plumes on tensile sur-
faces are sub-parallel to the fracture dip. Whereas microcrack lengths can readily be
measured between their tips on RG, barb lengths on tensile surfaces are not clearly
defined, because they are continuously connected to form plumes. Microcracks may
vary from sub mm to several cm in length, where the upper sizes are limited by the
sub-domain and domain boundaries. The cm-long cracks and the microcracks com-
monly maintain the same sub-axial orientation.
All the meso-fractures that cut through the chalk cylinders (and have surfaces on
the order of 50 cm2 in size) display fracture-composites of domains and sub-domains
consisting of microcrack coalescence. These assemblages are more common on the
three-dimensional shear fractures (Fig. 2.14a) but also occur on planar tensile surfaces.
Note, for example, that even the “uniform” large tensile fracture shown in Fig. 2.15a is
divided into a “lower” surface marked by a delicate plume and an “upper” surface
(partly shaded) marked by a coarser one (black arrow). Both plumes have the same
orientations on the two sub-domains, suggesting that the planar fractures propagated
in “sets”, much like joint sets in the field.
The large EMR pulse populations at low stresses are obtained from many micro-
cracks. Some of these microcracks possibly reflect dilatancy under both uniaxial and
triaxial loadings. Dilatancy is generally understood to occur between one-third and
two-thirds of the macroscopic fracture stress and is associated with an opening of
mirocracks throughout the sample normal to σ1 (Brace et al. 1966). Sprunt and Brace
106 Chapter 2 · Elements of Fracture Geology

Fig. 2.18.
a Photomicrograph of a Mid-
dle Eocene (Horsha Forma-
tion) chalk (width of picture is
1.2 mm). Large white spots are
plankton skeletons, small white
spots are microsparites, and
gray patches are micrites (cal-
cite). Black spots are pores (the
one shown by arrow is 84 µm
in diameter) and iron oxide
grains. b Measured pore radii
distribution in a, fitted to
P'(r) = a2rexp(–ar); best fit ob-
tained for a = 0.105 ±0.004 µm–1
(R2 = 0.98) (a and b after
Rabinovitch et al. 2003a)

(1974) observed that stress-induced microcracks are distinguishable from the initial
defects in the rock in their shapes and orientations. The stress-induced microcracks
are sharp-ended and elongated sub parallel to σ1, whereas the initial defects are blunt
and are oriented in all directions. Our results show that EMR pulses can be correlated
with microcracking throughout most of the stress range, i.e., much earlier than one-
third of the peak stress (Chap. 5). This includes pulses at the lower range of σ1, prob-
ably in association with the deformation of “initial defects” (pores). Note that since
EMR pulses are created solely by cracks and not by existing defects, this deformation
must consist of some cracking. The measured pore radii distribution in the chalk is
presented here for reference (Fig. 2.18).

Ridges and grooves and incipient shear before failure. Ridges and grooves (RG) are
elongated lineations (slickenlines/slickensides) parallel to the direction of relative slip
on fault surfaces. However, their origin is not clear (Davis and Reynolds 1996 p. 277).
It is not understood why continuous lineations are longer than the known displace-
ment along faults (Hobbs et al. 1976, p. 303) and how these lineations can be formed
when the displacement is negligible (Means 1987). The results obtained by Bahat et al.
(2001a) only exacerbate this dilemma. They obtained RG on fractured surfaces that
had practically zero displacement. These surfaces did not cut through their respective
cylinder peripheries; instead, they were arrested within the cylindrical samples
(Fig. 2.14b). The close resemblance of the RG in the latter samples to RG that occur on
 2.2 · Fractography and Tectonofractography 107

surfaces that were relatively displaced (e.g., Bahat et al. 2001a, sample 16) is also in-
triguing. The only fractographic difference is that steps occur on the fracture surface
of sample 16 but not on that of sample 15 in which the cut was arrested within the
sample. The implication is that RG develop by the shear process during a decrease in
sample cohesion due to microcracking prior to failure. The clear morphology of the
plumes in these cases is a compelling indication of the lack of relative displacement
on them because such displacements would have erased the delicate plumes.
It is possible therefore to distinguish an incipient shear when some but not total
cohesion is lost. This is the time when the grooves and ridges are formed. The displace-
ment that occurs subsequently is recorded by steps that can be assigned to a total loss
of cohesion. These observations partly resemble results obtained by Wibberley et al.
(2000), who identified an “incipient breakdown zone” at the immediate tip of the mi-
cro-fault. This zone consisted of early, long, pervasive tensile micro-cracks, oriented
parallel to σ1 that were later cross-linked by a set of perpendicular short tensile micro-
cracks (unlike the results by Bahat et al. 2001a), leading to a subsequent failure.

Inclined shear and axial tensile fractures. The strong correlation between crack ori-
entation and mode of propagation mentioned above relates to both cracks several mm
to several cm in length and fractures up to about 10 cm in length. This correlation
does not automatically apply to microcracks below 0.1 mm in size, for which Bahat
et al. (2001a) did not have accurate fractographic observations. If, however, this corre-
lation could be confirmed for micro-sized cracks, it might have important implica-
tions regarding the transition from microcracks to meso-fractures close to the triaxial
peak stresses. This would imply that the initial axial tensile fractures (longitudinal
splits) were connected by oblique shear microcracks. It would correspond to the ob-
servations by Shen et al. (1995) and Bobet and Einstein (1998) on fractures in gyp-
sum, and by us on inclined shear fractures in chalks (Fig. 2.17). Note that Hoek and
Bieniawski (1966) also suggest that early microcracks are linked by inclined shear
cracks, with the difference that the latter authors assumed the early microcracks to be
inclined to σ1. The “connection” by oblique shear microcracks would differ from the
models by Horii and Nemat-Nasser (1985) and by Reches and Lockner (1994), who
advocate that such a connection is carried out in granite by tensile microcracks. Addi-
tional experiments backed up by close fractographic examination of induced tensile/
shear fractures should help to ascertain whether the application of results on
microcrack linkage obtained on weak materials (gypsum and chalk) to strong ones
(e.g., granite) is justified.

2.2.3.5
Experimental Fracture on Five Sedimentary Rocks

Berea sandstone, Marianna limestone, Hasmark dolomite, Repetto siltstone and Muddy
shale have been subjected to triaxial compression tests, measuring the confining pres-
sures, the compressive strength (maximum ordinate of the stress-strain curve) and
the dihedral angle 2α (measured relative to the maximum principal stress direction)
by Handin et al. (1963). They show that both the compressive strength and dihedral
angle results varied considerably in the five tested rocks. Generally, these rocks were
weak (not stronger than 6 MPa), and the dihedral angles range was large (ranging
108 Chapter 2 · Elements of Fracture Geology

between 0° and 80°) (see Table 2.2). No correlation between the dihedral angle and
the confining pressures can be detected for these rocks. We are not aware of signifi-
cant deviations from the results presented here that have been found for similar rocks
by additional investigators (e.g., Friedman and Logan 1973).

2.2.3.6
Experimental Fracture on the Chelmsford Granite

The compressive strength of the Chelmsford granite ranges from 106 MPa to 200 MPa
(Peng and Johnson 1972). Peng and Johnson (1972, Fig. 27) observed strong preferred
orientation of the growing microcracks under stress along the splits. Characteristi-
cally, about 70–80% of the microcracks were oriented within 10° off the maximum
principal stress direction σ1. Some of the microcracks were straight, while others were
partly curved. Holzhausen and Johnson (1979) also observed that a characteristic mode
of failure for uniformly stressed specimens of Chelmsford granite was evidence of
longitudinal splits that were inclined up to about 10° from σ1. We interpret the straight
splits to be essentially extensile (mode I) even if they form small angles 5 ±5° with
the load direction (Peng and Johnson 1972, Fig. 18), whereas curved microcracks re-
flect local growth under shear. This interpretation follows the criteria derived from
experimental observations by Bahat et al. (2001a) that shear cracks are inclined (also
curved) to σ1, while tensile cracks are (straight) parallel or sub-parallel to this direc-
tion. However, since these criteria concern the behavior of microcracks in a weak rock
(chalk), the analogy to a strong rock (granite) should be presently qualified until more
evidence becomes available to support it.

2.2.3.7
Experimental Fracture on Glass Ceramic

The fracture properties of glass ceramic induced by compression were investigated

by combined electromagnetic radiation (Chap. 5) and fractographic methods (Bahat
et al. 2002a). The study of a transparent sample enabled the fractographical elucida-
tion of the sequence of crack nucleation, growth and interaction, and the ultimate lon-
gitudinal splitting under the incremental increase of uniaxial stress in five stages.
Sample failure by longitudinal split occurred under uniaxial compression of 112 MPa
in the fifth stage. A split into two slightly unequal parts (termed larger and smaller)
took place along a straight, smooth plane parallel to the sample axis such that differ-
ent fracture events were identified on the surfaces of the two parts (see more about
the sample shape and experimental procedure in Bahat et al. 2002a). Experiments on
transparent samples (Germanovich and Dyskin 2000 and references therein, and
Sect. 2.4.2.2) were aimed at different objectives.
Striae (plumes) stretch continuously along the fracture surface of the large part al-
most from end to end of the sample (Fig. 2.19a). They start from an origin that is lo-
cated about 18 mm from the F end (see below) and 86 mm from the B end, and occur
at about mid distance (10 mm) from the two lateral edges of the main fracture. The striae
radiate in two directions towards the two ends of the specimen. The bulk of the striae
propagate towards the B end such that with the increase of distance from the origin
 2.2 · Fractography and Tectonofractography 109

Fig. 2.19. Longitudinal split in a transparent glass ceramic. a Striae initiate above vertical arrow and
propagate bilaterally (see inset above photograph) toward B and F ends of specimen (horizontal ar-
row shows striae at the F end) in one part of the split. Striae initiate within the specimen. They propa-
gate along straight trajectories towards the B end at first. When gradually growing in size they ap-
proach the upper surface of specimen and curve towards that boundary. b Two successive cracks that
formed after the specimen failure and split formation, when stress has relaxed. Two arrows mark the
size of the first, successive crack (after Bahat et al. 2002a)

gradual coarsening of the striae morphology occurs. Only three-four short (6–7 mm
long) striae propagated towards the F end (inset in Fig. 2.19a). Accordingly, the exact
location of the origin, that is, the focus where the radial striae originated, is within an
area of 1–2 × 1–2 mm, which we term the “nucleation zone”. A microscopic examina-
tion indicates that the nucleation zone is not associated with any morphological
discontinuities (wing cracks, glass defects etc.). Moreover, the transparency of the sample
enables us to verify that no microcracks occur underneath the origin. Various shades
in axial and transversal-concentric orientations around the origin are reflections from
the outer surface of sample and do not belong to the main fracture surface. Hence, a
longitudinal split along almost the entire length of the sample was partly straight and
parallel to the compression axis before being deflected by the sample boundaries in the
strong glass ceramic material. The second part of the split contains two successive cracks
that formed after the specimen failure and split formation, when stress has relaxed, as
indicated by the electromagnetic radiation results (Sect. 5.3.4) (Fig. 2.19b).
110 Chapter 2 · Elements of Fracture Geology

2.2.3.8
Inducement of En Echelon Fractures by Mixed Mode I + III in PMMA

Cooke and Pollard (1996) experimented on rectangular blocks (15 × 15 × 5 cm, ASTM
E399-83) of polymethyl methacrylate (PMMA or Plexiglas), resembling the concep-
tual fracture mechanics models of mixed mode loading by Lawn (1993, Fig. 2.3) and
Broek (1991, Fig. 1.3). They built a system that enabled them to load cracks in mixed
mode I + III. Initially, a starter crack was grown from a chevron notch at a rate of about
1 cm in 10 s up to the peak load. Then, when the crack has reached the desired length,
the sample was unloaded in order to begin the mixed mode portion of the experiment.
They made the distinction that if mixed mode I + II loading is changed abruptly on a
quasi-static fracture, the propagation path follows a sharp kink (Erdogan and Sih 1963)
(Fig. 2.20a). On the other hand, if the KII / KI ratio is smoothly changed on a continu-
ously propagating crack, laboratory fractures follow a continuously curving crack path
(Cotterell and Rice 1980; Thomas and Pollard 1993) (Fig. 2.20b). A similar relation-
ship between the loading history and the smoothness of the path is expected for mixed
mode I + III loading (Fig. 2.20c,d). Cooke and Pollard (1996) considered that fracture
propagation paths depend not only on the load ratio applied but also on sample ge-
ometry, loading configuration, and interaction among growing fractures.
Cooke and Pollard (1996) found that both the angle of twist ω of the en echelon
cracks (Fig. 2.21a) and the number of segments formed increase with the increase of
the ratio KIII / KI. They however observed that the increase of ω with the ratio KIII / KI
falls below theoretical predictions (Cooke and Pollard 1996, Fig. 14). They propose that

Fig. 2.20. Mixed mode fracture propagation paths for various loadings, when the maximum princi-
pal stress σ1 is horizontal. a If a crack is loaded abruptly in mode I + II, a sharp kink is produced,
forming the α angle at the distal part of the parent joint; the plume on the parent joint discontinues
and a new plume forms on the kink. b If mixed mode I + II loading is increased gradually, as the
fracture propagates, a path that follows a curve results and the plume on the parent joint continues to
the curved kink that forms. c An abrupt I + III loading produces twisted echelon segmentation at ω
on fringes that form along one or two layer boundaries, the plume on the parent joint discontinues
and new plumes form on the fringe. d If mixed mode I + III loading is increased gradually, en echelon
segments that gradually curve result, and the plume on the parent joint continues to the curved seg-
ments (modified after Cooke and Pollard 1996)
 2.2 · Fractography and Tectonofractography 111

Fig. 2.21.
a Twist angle between parent
crack and incipient fracture
after break-down vs. stress in-
tensity ratio for nine samples. A
linear best fit through all mea-
surements shows a trend of in-
creasing twist angle with KIII / KI.
For each sample a similar trend
can be observed as KIII / KI var-
ies with echelon fracture loca-
tion along the breakdown front.
Data from sample 22 are la-
beled with letters correspond-
ing to digitized fractures in
Fig. 2.21b and show a trend
of increasing twist angle with
KIII / KI. b Digitized data from
an experimental result showing
breakdown geometry. Fracture C
is sandwiched between frac-
tures B and D. While the over-
lap with B is minor, there is
considerable overlap between
fractures C and D. The mechani-
cal interaction between adjacent
cracks may act to decrease the
angle of twist in experimental
and natural mixed mode I + III
fractures. c Breakdown front
shape for four samples com-
pared with half ellipse with
major axis of 5.0 cm and half
mirror axis of 0.92 cm. The
elliptical shape matches best at
the center of the sample where
most of the data are collected.
The difference between the half
ellipse approximating the crack
front and a best fit through the
crack front data is less than
0.1 cm (12%) in the region
–1 cm < z < 1 cm where 86% of
the data are collected (from
Cooke and Pollard 1996)

interaction among growing fractures (Fig. 2.21b) may contribute to the discrepancy
between theoretically predicted twist angles and those observed in these and other
mixed mode I + III experiments. Cooke and Pollard (1996) noted that the majority of
the en echelon fractures form at the center of the sample where the ellipse best ap-
proximates the parabolic breakdown fronts (Fig. 2.21c). Such breakdowns often oc-
cur in nature as well (e.g., center of Fig. 2.35c), suggesting that their results are appli-
cable in geological outcrops.
Cooke and Pollard found that the crack front advanced further than expected com-
pared to the trailing edges at the sample sides (Fig. 2.21c). They suggested that when
the sample sides are oppositely displaced in a tearing motion by mode III, a compo-
nent of mode II stress intensity KII is applied to the curved crack front near the sides of
112 Chapter 2 · Elements of Fracture Geology

the sample. This is because the direction of shear is not perfectly parallel to the frac-
ture front. Thus, a local mode II is induced in the system loaded by mode III. Similar
load superpositions probably occur around the tips of all normal and reverse faults.
This experimental finding is supported by a similar study that was carried out on the
behavior of the crack front in fractured glass bottles. “Edge effects” caused the mist and
hackle zones to advance more at the front center compared to the trailing edges of these
zones at their intersection with the bottle’s outside surface (Bahat et al. 1982, Fig. 4).

2.2.3.9
Fractography of Tensile Cracks in Starch-Water Mixtures

We briefly cite below a study on desiccation cracks in starch-water mixtures of

50–100 mm in diameter and 2–40 mm in thickness that were investigated by Müller
and Dahm (2000) and Müller (2000). Fracture velocities were measured from videos
and estimated from photos. Fractures covered the range from spontaneous nucleation,
dynamic cracks with velocities of 100–200 mm s–1 to quasi-static cracks with veloci-
ties of 0.1 mm s–1 and less. Plumes that formed on the fracture surface gave the frac-
ture direction (Fig. 2.22a), and their relation to fracture velocity was similar to the
relation between seismic rays and seismic wave velocity. A ray-tracing method from
seismology was used to calculate plume lines for depth-dependent rupture velocity.
Moreover, an inverse method, based on finite difference travel times and conjugate
gradients, was developed to invert a set of measured plume directions into a rupture-
velocity distribution, which could also depend on the horizontal coordinate on the
fracture surface. The main results and conclusions of this study are as follows.

Fig. 2.22.
Fracture surface in starch.
a A coarse plume propagates
on the parent joint from the
nucleation point NP. A fringe
zone shows an angular tilt
from the parent joint at the
lower right. b A coarse fringe
zone, where fracture velocity
was lowest, formed beyond a
mirror boundary. Length (di-
ameter) of specimen is 65 mm
(after Müller 2000)
 2.2 · Fractography and Tectonofractography 113

1. Plume lines can successfully be inverted into relative fracture velocity.

2. In thin starch layers (thickness less than 0.2 times the diameter), fracture velocity
decreases from top to bottom by a factor of 2–5, following a decrease of tensile
stress due to the increase in water concentration.
3. Horizontal variation of fracture velocity reflects horizontal variation of stress, in-
cluding stress relaxation due to the propagating crack, and ranges from dynamic
to quasi-static velocities.
4. In thick starch layers (thickness about 0.5 times diameter), fracture is predomi-
nantly quasi-static.
5. Starch cracks sometimes have a fringe zone where topographic amplitudes of the
en echelon segments are higher and rupture velocities lower than on the main part
of the crack (Fig. 2.22b); this probably also applies to joints in rocks and their fringe
zones.
6. Starch-water mixtures at rupture have a Poisson ratio close to 0.5. Cracks in starch
are closest to subsidence or diagenesis joints in sedimentary racks.

There is some reservation regarding the interpretation of the fringe shown in

Fig. 2.22b, because its morphology deviates from the common perfect shingling of en
echelon segmentation, often seen in nature (e.g., Fig. 2.35). This deviation should pos-
sibly be assigned to the material properties and the difficulties involved in its matur-
ing during the fracture process. There is no doubt, however, regarding the distinction
between the mirror and the fringe in Fig. 2.22b.

2.2.3.10
Formation of Fringe Cracks in Silicon Single Crystal

Cramer et al., (1999, 2000) conducted dynamic fracture experiments on silicon single
crystal plates that were loaded to force a {110} cleavage crack in a <110> direction.
The fractographic features that resulted vary with the changes in the average crack
velocity vav and the energy release rate G. The fracture surface is smooth and mirror-
like at the lowest fracture stress, corresponding to G = 7 J m–2 (circles in Fig. 2.23a). At
G ≤ 14 J m–2 and above vav = 3 000 m s–1 = (2 / 3)vR, the Rayleigh wave velocity (triangles
in Fig. 2.23a). Atomic Force Microscopy (AFM) reveals that the crack partially devi-
ated from the initial (110) plane and displays hills and valleys extending in the crack
propagation direction. Facets appear on {111} planes next to the mirror with further
increase in G. The size of the facets increases with increasing G. Instability, exhibited
by an abrupt onset of rough hackle zones, starts at G > 40 J m–2 and at velocities close
to V (or vter), the terminal crack velocity (squares in Fig. 2.23a). The instabilities are
manifested by the cracks on the {111} facets to “overgrow” each other, displayed by
the simultaneous propagation of two crack fronts and in secondary shear fracture of
the material placed between them (Fig. 2.23d).
Thus, there are three modes that show how the excess energy formed with the frac-
ture growth is spent. Initially, the excess energy contributes to the increase in crack
velocity. More excess energy results in faceting. Finally, great excess energies lead to
path instabilities, resulting in the propagation of multiple cracks. The path instabili-
ties in the silicon plates bear similarities to the bifurcation instability found in PMMA
by Sharon and Fineberg (1999). The important difference, however, between the latter
114 Chapter 2 · Elements of Fracture Geology

Fig. 2.23. a Dependence of the average fracture velocity vav on the steady state energy release rate Gs.
At the lowest Gs (open circle) the fracture surface is smooth. A faceted fracture surface is observed at
higher Gs (triangles). The fracture surface is very rough at the highest Gs (squares). The solid line
follows their Eq. 1, which is given by our Eqs. 1.134–1.136, using Γ = 3 J m–2. b This equation is used
to determine Γ as a function of G, and c as a function of vav. d SEM micrograph, showing smooth en
echelon segments cleaved from top to bottom at {111} planes (light) that are bridged by steps of shear
cracks (dark) (from Cramer et al. 2000)

and the result by Cramer et al. (2000) (Fig. 2.23d), is that the bifurcation in the isotro-
pic amorphous PMMA occurs along various alternative fracture paths, whereas in the
silicon crystal instability develops by tensile cleavage on particular planes that are
essentially en echelon cracks, which are connected by shear fractures, often termed
steps or bridges (Fig. 2.39). Hence, under constraints of anisotropy, en echelon seg-
mentation may occur at fracture terminal velocities.

2.2.3.11
Simulation of En Echelon/Hackle Fringes and Longitudinal Splits in Glass

A brief description of the investigation. We introduced new loading techniques

(Fig. 2.24a–c) in investigating the formation of transitional en echelon/hackle fringes
and longitudinal splits in soda-lime glass samples. Previous investigators combined
sequential loading of mode III after mode I for obtaining en echelon fringes (e.g.,
Sect. 2.2.3.7). In the new experimental setting, a remote uniaxial compression was applied
on cylinders (10 cm in length and 3 cm in diameter) that contained pre-cuts along the
sample walls. Therefore, the compression was transformed into local mixed mode I + III
 2.2 · Fractography and Tectonofractography 115

Fig. 2.24. Two pre-cut types. a The first type consists of two pairs of pre-cuts (PC1 and PC2) parallel
to the sample vertical axis. The PC1 pair is normally oriented to PC2. PC1A and PC1B are the first and
second pre-cuts, respectively, of the first pair PC1, while PC2A and PC2B are the first and second pre-
cuts, respectively, of the second pair PC2. PC2A and PC2B are interconnected at the top and at the bot-
tom ends of the samples. Depth of the interconnection zones at each end of the sample is 30 mm.
b The second type consists of two pre-cuts inclined (“X”-wise) to the sample vertical axis. Both pre-
cuts create approximately the same inclination angles α to the vertical axis. c Measurement of α is
carried out between the pre-cut and the vertical axis of the sample

Fig. 2.25. A photograph of a fracture surface of glass that resulted by failure of the second type
(135 MPa). Sample length is 100 mm. The sample is rotated 90° clockwise into a horizontal position
and thick arrows show the direction of the uniaxial compression. Note the plumes that extend close
to the two tips in continuation with the axial split at the center

loading, resulting in segmentation into two fringes along the pre-cuts. Subsequently, a
longitudinal split of complex fractography formed along the fringes (Fig. 2.25).

Experiment. Two types of crack pre-cut geometries were introduced into the sample
walls. The depth of all pre-cuts was 10 mm and aperture 2 mm. The first type consists
of two pairs of pre-cuts parallel to the sample vertical axis (Fig. 2.24a, with one
pair (PC1) normally oriented to the other (PC2)). Two pre-cuts, PC2A and PC2B, of the
second pair were also interconnected at the top and bottom ends of the sample. The
depth of the interconnection zones was 30 mm at both end parts of the sample. The
116 Chapter 2 · Elements of Fracture Geology

second type consists of two pre-cuts inclined (X-wise) to the vertical sample axis
(Fig. 2.24b), creating a series of pre-cut angles α, while maintaining about the same α
(±2°) for the two pre-cuts (Fig. 2.24c). Note that for the first pre-cut type, α = 0°. Glass
cylinders in the size tested are very strong and generally resisted stresses beyond
200 MPa that we could achieve with our instrument. The introduction of pre-cuts to
the samples weakened them. Another method of reducing the effective strength of the
material is applied by using truncated elliptic cone shaped samples (Bahat et al. 2002a).

Fig. 2.26.
Fracture parameters of the
cylindrical glass. a The relation
between the pre-cut angle α
and sample strength. b A look
at the sample along the axis.
c The relation between α and
the twist angle of longitudinal
splits γ. d Schematic picture of
a failure envelop, showing an
increase stress, corresponding
to the increase in the fracture
angle αf created by joints/faults
(modified from Twiss and
Moores 1992, Fig. 9.9)
 2.2 · Fractography and Tectonofractography 117

Fig. 2.27. A flat 2D projection of the 3D (twisted splits) surface created during failure of glass samples
of the second type. The fracture process consists of three main stages. a Nucleation of en echelon
segmentation at the edges of both pre-cuts at about the middle of sample height. b An increase of the
en echelon dimensions and number, towards the sample center and along both edges of pre-cuts
towards sample tips. c Propagation of axial splits from en echelon fringes (at about the mid sample
height) to sample tips (see arrows) along the sample vertical axis. d Schematic arrangement of three
zones, pre-cut, en echelon fringe and longitudinal split in the sample

Results. Generally, longitudinal splits do not start from the sample center, because
this is a zone of great strength. They often initiate from the sample tip peripheries
and propagate axially to the center. Weakening of the sample center can however ini-
tiate a longitudinal split there due to local tension induced by buckling (Holzhausen
and Johnson 1979) or as in the present study by en echelon segmentation. The intro-
duction of the second type of pre-cuts to the samples enables one to study the system-
atic dependence of strength on α (Fig. 2.26a), showing minimum glass strength at
6.5–7 ±2°, and then a rise in α requires an increase in load. The enlargement of α
augments the angle produced between the en echelon fringes that form along the pre-
cuts in the “X”-wise structure (Fig. 2.26b). The longitudinal splits that form along the
fringes interact with each other creating a twist along the cylindrical axis. The twist
measured by γ (Fig. 2.26b) increases with α, and the shape of the curve in Fig. 2.26c
produces a mirror image of Fig. 2.26a. It is quite intriguing that the minimum and
maximum in Fig. 2.26a and c, respectively, are close to the maximum deviation of
splits from the sample axis in conventional tests (in granite, Sect. 2.2.3.6, and in chalk,
Sect. 2.2.3.4). The fracture sequence is diagrammatically shown in Fig. 2.27a–c.

Discussion. Contrary to previous experiments and many field exposures in sediments

that exhibit transitions from parent fractures to fringes, the present study simulates a
reverse transition. We demonstrate a transition from fringes to parent fractures by
changing local mixed mode I + III to a local single mode I. Monitoring of the electro-
magnetic radiation (FMR) that was induced by the fractured samples enabled us to
determine in real time the sequence of events, the fracture size and the fracture ve-
locities along the various failure stages (Sect. 5.3.4). Very often geological descrip-
tions trail behind descriptions of experimental results. Presently, the fracture sequence
described here may be reasoned to be an “unlikely geological process”. However, new,
118 Chapter 2 · Elements of Fracture Geology

“unexpected” fracture phenomena are repeatedly discovered in granites (Chap. 4),

and who knows what else is still concealed from us there? The stress analysis of these
unexpected results is currently under investigation.

2.2.3.12
Hackle Formation in Polymers

Arakawa and Takahashi (1991) compared fractographic parameters to several me-

chanical ones in polymers. Fractography was measured by two methods, pit density
and surface roughness λ. They scanned the fracture surfaces with a scanning instru-
ment at an interval of 0.5 mm using a needle (10 µm in tip radius) in the direction
perpendicular to the direction of crack propagation and defined surface roughness
using the term “root mean square roughness” (RMS):

(2.4)

where L is the scanned length and f(x) is the roughness height at the point x. Arakawa
and Takahashi (1991) found a good correlation between changes in pit density and
surface roughness. They employed the shadow optical method of caustics to evaluate Kd
and Kc, the dynamic tensile stress intensity factor during crack propagation and the
tensile stress intensity factor for the arresting crack, respectively, where a Cranz-
Shardin type high speed camera was used. Kd was evaluated by

(2.5)

where φ is the caustic diameter, Z0 is a distance between the specimen and the image
plane, d is the specimen thickness, c* is the stress-optical constant for dynamic con-
ditions and η is a convergence factor for incident light rays.
Arakawa and Takahashi (1991) show that the curve linearly correlating crack
velocity, v with Kd, may be divided into three regions of different slopes in Hommalite-
100 (Fig. 2.28a). There is a gradual increase in Kd with v in the low velocity region A,
where λ remained relatively smooth. In region B, there is a rapid increase in Kd along
very short intervals of v and a corresponding increase in λ. The curve slope in region C
exhibits an additional rapid increase in Kd and an extremely rough surface that reached
its peak prior to branching. The latter slope clearly shows that branching did not occur
at the peak crack velocity but at the maximum value of Kd, i.e., branching is Kd depen-
dent. The dotted curve connects the peak velocity points, separating the accelerating
and decelerating areas in the diagram (see below).
Arakawa and Takahashi (1991) show that maximum λ is best correlated with Gv. The
physical meaning of Gv is energy per unit crack width per unit time, and it is consid-
ered to relate to the flow rate of energy into the crack tip region. A very good fit among
the maxima of the curves of λ, G and Gv occurs for epoxy (araldite D), while it is shown
that crack lengths giving peak values of λ and v differ from each other (Fig. 2.28b). A
slight discrepancy between crack lengths also seems to exist, giving peak values of λ
 2.2 · Fractography and Tectonofractography 119

Fig. 2.28. a Dependence of stress intensity factor Kd, fracture velocity v and fracture acceleration ä
in Homalite-100 on fracture length (after Arakawa and Takahashi 1991). b Surface roughness λ, spe-
cific crack extension resistance G and the product of Gv for epoxy as a function of crack length a
(after Arakawa and Takahashi 1991)

and G: For the same λ values indicated by points A and A', the corresponding values
of v (at points B and B') or G (at points C and C') slightly differed. On the other hand,
a good agreement is shown between crack lengths giving the peak values of λ and Gv,
and the values of Gv are almost equal at points D and D', which correspond to points A
and A', respectively. These results support the suggestion that λ is dependent on Gv.
Hence, the implication is that in the examined polymeric materials, λ behavior (pre-
sumably along the traverse through the mirror-mist-hackle-branching, Fig. 2.1) is cor-
related with Gv , Kd, G, and v in a decreasing order. The results by Arakawa and Takahashi
(1991) were confirmed by Arakawa et al. (2000), using different experimental conditions
on PMMA. These studies demonstrate that maxima in Fig. 2.28a and b (particularly the
latter) show strong fluctuations in the various parameters along the crack length. The
association of v with the formation of hackle in polymeric materials shows, however,
the subordinate dependence of this morphology on v. On the other hand, these results
indicate the important influence that Gv and Kd have on the formation of hackles.
120 Chapter 2 · Elements of Fracture Geology

Fig. 2.29. Primary and secondary cracks. a A schematic picture of the geometrical setup. A second-
ary crack of length 2cs at a distance a from the origin of the primary crack of a semicircular radius R
starts to grow at velocity vβ(t) when t = 0. b Side view of the primary crack, T its tip, while θ and α are
the starting growth angle of an existing flaw and the direction of its growth into a secondary crack,
respectively (after Rabinovitch et al. 2000a)

2.2.3.13
Hackle Formation in Glass

Previous views about the mist and hackle zones maintain that they are “identical in ap-
pearance but different in scale” (e.g., Mecholsky 1991). More recently, however, Rabinovitch
et al. (2000a) suggested a fundamental difference between these zones, as explained be-
low. The secondary cracks (SC) start to grow from an existing “Griffith flaw” (Sect. 1.1.2
and 2.2.2) when the critical stress intensity KIC is attained in front of the tip of the pri-
mary crack (PC) cutting the glass (Fig. 2.29 and Sect. 1.4.4). The SC move under a chang-
ing stress field caused by the PC, whose distance from the SC is continuously changing.
Hence the SC start from zero velocity, and asymptotically reach vt. In the mist zone, the
PC overtakes the SC almost instantly after they begin to grow. Therefore, the SC in this
zone are short (for details see Rabinovitch et al. 2000b). However, towards the end of the
mist zone, the growing SC become longer before the PC overtakes them, and they manage
to attain larger sizes and rougher morphologies. Apparently, the transition to the hackle
zone is related to the inability of the PC to catch up with the SC. In this case, the SC grow
separately compared to those SC that are to a large extent “swallowed up” by the PC and
form the mist (Rabinovitch et al. 2000a). Previous authors made the (erroneous) assump-
tion that the SC move from their very incipience, with the same vter of the PC, resulting in
parabolic or hyperbolic shapes throughout. Such shapes actually appear in the hackle zone
but not before. Hence, hackle formation is associated with the distinct attainments of vter
by both the primary and the secondary cracks. This association however does not imply a
dependence of hackle formation on vter. It has been suggested that hackle formation is rather
stress intensity dependent (e.g., Bahat et al. 1982). This dependence seems to be more in-
tricate than the latter, as is shown in Sect. 2.2.3.12.

2.2.3.14
The Sequence of Mirror, Mist, Hackle Fringe and Branching on Fracture Surfaces of Glass

It has been accepted by many authors (e.g., Rice 1984, Fig. 1; Hull 1999, Fig. 5.3 and 5.31) that
fracture growth from the origin to its termination is in the following “conventional order”:
 2.2 · Fractography and Tectonofractography 121

The mirror plane, mist, hackle and branching (branching at the right side of Fig. 2.1b and
Fig. 2.10). Branching involves the creation of two secondary fractures, at least one of which
is non-coplanar with the primary fracture. Accordingly, a distinction has been made be-
tween the radii from the origin to the mist boundary, the hackle boundary and to branching
(Fig. 2.1d). This view is quite surprising, considering the existence of many observations
that are unlike the conventional order. Rice (1984, Fig. 66b), Ravi-Chandar and Knauss
(1984d IV, Fig. 11), Holloway (1986, Fig. 13), Hull (1999, Fig. 5.4) and Bahat et al. (2001b,
Fig. 2b) obtained results of a different “fractographic sequence”. They show that the mist zone
is followed by a transition zone from mist to small hackles, beyond which a drastic change
of superposing large hackles on new branches occurs. This indicates simultaneous growths
of these two features, i.e., branching occurs before large hackles (branching at the left side of
Fig. 2.1b). The same fractographic sequence occurs on joint surfaces in chalks and geologi-
cal exposures of granites (Fig. 2.30a,b and 2.56). These observations should equate the radii
from the origin to the hackle boundary and to branching. What could be the reason for
the discrepancy between the “conventional order” and the “fractographic sequence”?
Possibly, the results of repeated four cycles from the mirror plane to branching shown
below can be used for explanation. A soda lime silicate glass bottle was fractured by the
internal pressure technique (Bahat 1991a, p. 121). The saddle-shaped fragments that
characteristically result in such fractures (Fig. 2.30c), which show repeated branching
in profile (Fig. 2.30d), were investigated. A saddle-shaped fragment shows the initia-
tion of four cycles from mirror to successive branchings in plan view (Fig. 2.30e). A
primary mirror M1 passes on to mist m and then to hackle h, ending in fracture-branch-
ing that initiates in the outer boundary of the hackle, at h. The second cycle starts in
secondary mirrors that form a shingle-like en echelon segmentation M2 that transforms
into secondary mist m and hackle h, where a new branching occurs. The third cycle starts
by a series of tertiary mirrors M3 in a disturbed shingle-like segmentation. On one
mirror, there are undulations that convex in the direction of fracture propagation. The
fourth cycle (only part of it has been retrieved) is “disorganized” such that tertiary and
quaternary mirrors M4 have no distinct boundaries.
Most intriguing is the change from cycle to cycle when each branching starts with
new mirrors. The first branching occurs at the outer boundary of the first hackle
(before M2), following the “conventional order”. However, the second hackle is already
divided into two sub-zones, such that the first one (before M3) belongs to the second cycle,
while the second hackle sub-zone rests on the early parts of the third cycle mirrors, M3.
The next hackle zone occurs beyond a mist zone at the ends of M3 but also appears to
have initiated the fourth cycle, M4. Thus, one observes superposing hackles on the new
branch, indicating simultaneous growths of these two features according to the above
“fractographic sequence”. Hence, both “conventional order” and “fractographic se-
quence” occur adjacent to each other, forming during the early (second cycle) and late
(fourth cycle) stages of the same experiment, respectively, while the fractography of the
third cycle, M3, shows a transition between the two.
The occurrence of the “fractographic sequence” in geological exposures (Fig. 2.30a
and 2.56) encourages efforts to comprehend the meaning of cycling fractography that
may help to improve the understanding of the paleostress conditions leading to dynamic
jointing (Sect. 2.4.6.1 and Chap. 4). A possible future discovery of a “conventional or-
der” in geological exposures may add important information for detailed interpreta-
tion of the fracture process.
122 Chapter 2 · Elements of Fracture Geology

Fig. 2.30.
a A geological exposure showing
the hackle zone on the fringe f,
that forms an angle with the mir-
ror plane m, along a sharp mir-
ror boundary b. b A mirror m,
and a hackle fringe f, forming an
angle along the mirror bound-
ary b, on a fracture formed by a
blast (the “Safad” and “Ma’alot”
fractures, respectively, after Bahat
and Rabinovitch 2000). c A frac-
tured bottle. The boundaries of
the saddle-shaped fragment are
fractures 2rB (the approximate
length of the mirror plane and
bottle split), L and the curved
cracks that connect them. d A
microphotograph of a profile of
a saddle-shaped fragment that
shows two cycles of branching at
arrows, starting from the bound-
ary of the primary mirror plane
M1. Various M, m, h and rB in
d and e denote mirror planes,
mist, hackle and radius of the
primary mirror plane, respec-
tively. e A drawing of a saddle-
shaped fragment in plan, show-
ing four generations of mirror
planes M, that represent three
cycles of branching from the
boundary of the primary mirror
(e after Bahat et al. 2001b)
 2.2 · Fractography and Tectonofractography 123

2.2.3.15
Fracture Interaction at Contacts along Layer Boundaries by a Numerical Experiment

The nature of interaction between adjacent fractures has been investigated for decades
both in the laboratory and in geological outcrops (e.g., Yokobori et al. 1971; Olson and
Pollard 1989, 1991; Bahat 1991a, p. 317). This section relates to fracture interaction with
layer boundaries, which is fundamental to fracture geology. Furthermore, within lay-
ered sedimentary rocks, the termination of fractures at layer boundaries can limit
vertical flow (particularly in the unsaturated zone, Sect. 6.9) and produce highly tortu-
ous flow paths (e.g., Tsang 1984). In contrast, fractures that propagate straight-through
layer boundaries (vertically and at various oblique angles) (e.g., Bahat 1991a, Fig. 4.5)
provide well-connected pathways for vertical fluid flow (Cooke and Underwood 2001).
A potential intermediate case for fluid flow is a fracture that jogs or steps over a few
centimeters along successive layer boundaries (see the profile in Fig. 2.31). Compos-
ite joints, such as those sketched in Fig. 2.31 are believed to form by repeated step-
over of a propagating fracture across the layer boundaries (Helgeson and Aydin 1991;
Cooke and Underwood 2001).
A numerical experiment was carried out by Cooke and Underwood (2001) explor-
ing the influence of deformation at contacts along layer boundaries, in the form of local
sliding and/or “debonding” and subsequent opening, on the fracture intersection with

Fig. 2.31.
Sketch of composite joint within
interbedded shale and sand-
stone sequence in New York
State. The overall vertical joint
trace has a discontinuous nature
(after Helgeson and Aydin 1991)
124 Chapter 2 · Elements of Fracture Geology

Fig. 2.32.
a Inferred variations in geom-
etry of fracture-bed contact
intersection for bedding con-
tacts with different strengths.
b Hypothetical pattern of frac-
tures within a series of layers
separated by different strength
interfaces. Circles highlight
occurrences of step-over frac-
tures due to opening along the
moderate-strength interfaces
and circle size correlates with
distance of fracture step-over.
The spacing of fractures is
approximated from the layer
thickness. Apparent step-overs
along the weak interface (indi-
cated with arrows) result from
the coincidental alignment of
fractures among layers. The
distinction between coinciden-
tal alignment and step-over is
difficult to assess without char-
acterization of fracture surface
textures (i.e., signatures of
propagation direction), relative
interface strength (to guide
probable intersection mecha-
nisms), and consistencies in
fracture pattern along bedding
contacts (Fig. 2.32a,b from
Cooke and Underwood 2001)

layer boundaries (see also Thomas and Pollard 1993). Their conclusions are cited be-
low. Fractures propagating toward layer boundaries may either

1. Terminate at the boundary.

2. Propagate straight through the boundary.
3. Step-over to the left or right at the layer boundaries (Fig. 2.32a).

Based on the model results, Cooke and Underwood 2001 suggest that local interface
opening, rather than sliding, is primarily responsible for the termination and step-over
of fractures. Furthermore, the model results suggest that the strength of contacts con-
trols the type of resulting fracture intersection. Fracture termination is favored at very
weak contacts, whereas fractures propagate straight through strong contacts (Fig. 2.32b).
Most sedimentary contacts will fall between these two end members. Such moderate-
strength contacts may develop step-over fractures due to a local opening along the con-
tact, or if the stresses are not great enough to produce new fractures, the parent fracture
will terminate at the moderate-strength contact. Fracture termination is more likely un-
 2.2 · Fractography and Tectonofractography 125

der conditions of shallower burial depth, lower effective layer-parallel tension, and fluid-
driven propagation, rather than equivalent remote layer-parallel tension. Thicker layers
and greater effective layer-parallel tension may produce greater amounts of step-over
than thinner layers and more compressive layers. Fractures aligned within several cen-
timeters across a contact may be coincidentally aligned or result in fracture step-over.
Careful characterization of the fractures and analysis of the pattern may distinguish
whether the fractures are the result of coincidental alignment or fracture step-over.

2.2.4
Joint Mirror Planes

Two end members of mirror planes are characterized in this section, drawing on both
laboratory and field observations.

2.2.4.1
Two Types of Mirror Planes

We distinguish between two types of fracture mirror planes. One type characterizes
fractographies that form dynamically by rapid fracture when high stresses are exerted.
This type has been studied for decades in the laboratory on glass, ceramics, metals
and polymeric materials (e.g., most of Chap. 2 in Bahat 1991a), and has recently been
characterized in outcrops from the South Bohemian Pluton (Chap. 4). The second
mirror type characterizes fractographies that form quasi statically by slow fracture
when low stresses are exerted. This type is far less known from laboratory investiga-
tions than the first type, but it can be identified on innumerable surfaces of system-
atic joints that cut sedimentary rocks.

The dynamic mirror plane. When fractured primarily by high tensile stress the mir-
ror plane exhibits characteristic morphological features (Fig. 2.1b–d), including the
fracture origin at an initial flaw that develops into a larger critical flaw, concentric
undulations, striae, Wallner lines, and the mirror boundary. This mirror plane is
smooth, and often distinguished as a polished-like surface. Undulations (note below a
distinction from ripple marks) develop on this surface by local mixed modes I + II
due to stress fields that progressively bend the crack front around the axis normal to
the direction of propagation when the two larger principal axes are on the fracture
surface. On the other hand, striae form by local mixed modes I + III in a stress field
that rotates the fracture front about the axis of the direction of crack propagation
(e.g., De Freminville 1914; Smekal 1953). The distance between the origin to the mir-
ror boundary is the mirror radius. Three mirror radii are legitimate, rm, r and rb, at
the inner boundaries of the mist zone, the hackle zone and branching initiation, re-
spectively (Fig. 2.1a,b,d). In multi-crystal ceramics and rocks, the micro-features of
the mist are “camouflaged” by the grain boundaries and only the hackle zone may be
recognized. Practically, all the above-mentioned fractographic features are generally
restricted to the mirror plane. There is a distinction between primary mirrors that are
related to the parent fracture and secondary mirrors that form on subsequent fringes
(Fig. 2.30d,e, 4.9c,d and 4.11a–c).
126 Chapter 2 · Elements of Fracture Geology

Fig. 2.33a.
Two matching halves of a frac-
ture in a ceramic body made of
BaTiO3, showing a smooth mir-
ror plane around the fracture
origin (dark gray), developing
towards right into sharp hackle
ridges which are termed cus-
pate-hackles (from Rice RW
1974)

Fig. 2.33b,c. Photograph and drawing of a “wall” consisting of four joints (1–4), that propagate to vari-
ous directions as shown by arrows, arrest marks a, occasionally show “box” profiles b, and striae s, in
Zion National Park. A camera bag 38 cm in length is used for scale

The quasi-static mirror plane. When fractured under low stresses, propagation is slow
and sub-critical (Sect. 1.5), and a different mirror plane forms. On this mirror, the location
of fracture origin can be identified by the convergence of plumes at the center of concen-
tric ripple marks, but the distinction between an initial flaw and the critical flaw is vague.
Additional morphological features are arrest marks and the mirror boundary. Experi-
mental sub-critical surfaces are briefly described in Sect. 2.2.3.1. Single-layer joints that
cut sedimentary rocks often fall into this category. Quasi-static mirror planes form matte,
 2.2 · Fractography and Tectonofractography 127

Fig. 2.33d.
An elliptical mirror plane deco-
rated by concentric arrest marks
and radial striae in Ordovician
pencil slate, Thuringia, Germany.
Many short striae superpose the
arrest marks, and some striae
merge into a single, large stria
(from Bankwitz 1965). Recall
that those striae are products
of mixed modes I and III, and
the arrest marks are products
of mixed modes I, II and III

Fig. 2.33e.
Photograph of a mirror plane
and fringe on an exfoliation
joint from Zion National Park.
The somewhat hackled en ech-
elon segments display different
lengths. Note the exact, curved
mirror boundary. A page at left
is used for scale (after Bahat
et al. 1995)

rather than polished-like surfaces. While rapid fracture through grainy ceramic material
or rock form transgranular, smooth mirror planes (Fig. 2.33a and 4.11a, respectively), slow,
intergranular fractures (Sect. 4.5.2.2) maintain a grainy, matte surface morphology. On
such surfaces, arrest marks often replace undulations, plumes replace striae, and minute
features including Wallner lines on the mirror plane and mist along the mirror boundary
would be unrecognizable. Distinctions between arrest marks and undulations are given
below, and distinctions between striae and plumes follow.
128 Chapter 2 · Elements of Fracture Geology

2.2.4.2
Fractography on Mirror Planes

We extend below the treatment of fractography on mirror planes from Bahat (1991a)
and make some changes along the following lines:

1. Ripple marks on mirror planes.

2. Arrest marks and striae on geological exposures.
3. Undulations on geological exposures.
4. Striae and plumes.

Ripple marks on mirror planes. Ripple marks are systematized on mirror planes ac-
cording to two categories, type and shape. Two types of ripple marks are distinguished:

1. Arrest marks that signify crack arrest.

2. Undulations that form on the fracture surface while it propagates.

Undulations may form in different fracture velocities, often starting when propa-
gation is slow, and continue at greater velocities (Fig. 2.10), or even maintain slow
propagation. Ripple marks occur in various shapes, including curved cross sections
(Fig. 2.10), cusps (Fig. 2.12), and angular cross sections that may attain “box” (defined
by four kinks) and “semi-box” (defined by three kinks) structures (see below). These
distinctions are important for the quantitative interpretation of the mechanism of frac-
ture in geological exposures by using v and K parameters (Sect. 4.8.3).

Arrest marks and striae on geological exposures. The distinction between undula-
tions and arrest marks has been obscured in the past, seriously hindering the applica-
tion of fractography in engineering materials and structural geology for the interpre-
tation of fracture processes. Particularly, it was impossible to distinguish between ripple
marks that formed by rapid fracture (undulations) and ripple marks that reflected slow
propagation (arrest marks). Here are a few examples and rules that can help in identi-
fying arrest marks both in experimental specimens and in rock outcrops:

1. Arrest marks may occur in all shapes and sizes, including angular cross sections
(Fig. 2.33b,c). Asymmetric cross sections are particularly common.
2. Following the results obtained by Guin and Wiederhorn (2003) (Sect. 2.2.3.1), an
arrest mark is associated with an inducement of mixed mode I + III loading at its
front (Fig. 2.9), such that striae would be induced on the mirror while commencing
from this front. Thus, an arrest mark is frequently associated with intense striae mor-
phologies, characterized by either multiple striae that superpose arrest marks or by
the increase of the intensity of individual striae, or both (Fig. 2.5a and 2.33b–d).
3. Quite often striae on the mirror plane and particularly in association with arrest
marks resemble fringe cracks and may erroneously be interpreted as en echelon
segments or hackle cracks. The distinction between these two features may be pri-
marily based on two striae types. First, many short striae start when superposing an
arrest mark but they fade away after a small distance from it (Fig. 2.33b–d), com-
pared with fringe cracks that maintain their lengths, or even increase it (Fig. 2.34,
 2.2 · Fractography and Tectonofractography 129

2.35, 2.33e). Second, when striae start between arrest marks they merge into a single
stria of intense morphology at the arrest mark, compared to en echelon segments
that maintain their number or hackles that even multiply (Fig. 4.11a). Figure 2.33d
shows quite well the two morphological types. Both morphologies signify slow frac-
ture propagation, and an immediate transition from an arrest mark to a hackle fringe
that signifies rapid fracture would be an unlikely occurrence (this criterion may be
used for examining this theory in future observations).
4. Arrest marks may form in the shape of “box”-shaped structures that superpose mirror
planes (Fig. 2.33b,c) or as “semi-box” structures that occur even on fringes (Fig. 2.30a). In
the latter case, the right side of the semi-box on the fringe indicates reduction in fracture
velocity, compared to its left side, which is marked by hackles that indicate rapid frac-
ture. Note that the striae and the “semi-box” on the fringe are significantly coarser than
the striae and undulations on the mirror (see more in Bahat and Rabinovitch 2000).

Finally, many, but not all arrest marks may be recognized. Some more arrest marks
that can be identified on rock fractures are presented in Bahat (1991a, Fig. 3.42c, 4.2a,b,
4.4 and 4.11), they all exhibit striae according to the above rules. Occasionally, transi-
tions between areas of multiple striae into mirror areas that are not imprinted by ar-
rest marks and do not contain striae (Fig. 4.36) possibly suggest a reduction in frac-
ture velocity but not quite an arrest.

Undulations on geological exposures. How do undulations differ from arrest marks in

geological exposures? Two examples of undulation propagation are described and inter-
preted below. As mentioned above, when distances between undulations increase, fracture
velocities increase (Fig. 2.10). The undulations in Fig. 3.24c display fracture “diffraction”
along boundary A, such that they change orientation and trail behind along boundary B.
At some stage, however, drag is reduced, resulting in velocity jump (at point C) to conform
to a uniform velocity in area D. There is a difference between the shapes of the upper ripple
mark and the lower ones at the right side of Fig. 3.24d. The upper one is convex towards the
direction of propagation, which is faster midway between the two bounding sub-vertical
surfaces, and trailing behind along the surfaces. This is a characteristic feature of undula-
tion propagation on experimental fractures (e.g., Bahat 1991a, Fig. 2.1c). The lower ripple
marks are straighter, implying an approach to equal velocity along the fracture front. The
ripple marks on the left side of Fig. 3.24d are inclined as they come closer to their arrest at
the older horizontal joint, suggesting that these are arrest marks. Note that the undula-
tions are approximately symmetrical and the arrest marks are asymmetric.

Striae and plumes. Striae that form by rapid fracture in glass or glass ceramic pen-
etrate into the material and propagate as sharp, continuous cuts along straight trajec-
tories, unless influenced by boundaries that may deflect them. For instance, there is a
change from approximately straight striae within the material in their early part (closer
to the F end) and a gradual curving in their continuation towards the B end as they ap-
proach the sample upper surface (closer to the viewer, Fig. 2.19a). They may lose some of
their sharpness while cutting granular rocks when moving fast (see delicate straight strie
and hackle in Fig. 2.30a). They may completely lose their sharpness and sub-divide into
dendrite-like plumes, such that every dendrite branch would be further sub-divided into
minute barbs (Fig. 2.3b) when propagating slowly (Fig. 2.22). This may explain why the
130 Chapter 2 · Elements of Fracture Geology

topography of plumes on mirrors that form dynamically either in the laboratory (Fig. 2.10),
by explosives (Fig. 2.30b), or in geological outcrops (Fig. 2.33e, 2.56, 4.11a) is often delicate
(or hardly visible), whereas on slowly forming joints, or when they curve (Fig. 4.22), plumes
may cover significant areas and attain coarse morphologies (e.g., Fig. 4.7b).

Summary. Two types of mirror planes are distinguished. One type characterizes
fractographies that form dynamically by rapid fracture when high stresses are ex-
erted. Such mirror planes are smooth, often distinguished as polished-like surfaces
that reflect light. The second mirror type characterizes fractographies that form quasi
statically by slow fracture when low effective stresses are exerted. This mirror type
characterizes surfaces of many systematic single-layer joints that cut sedimentary
rocks; they maintain grainy, matte surface morphologies.
We divide ripple marks in geological exposures into two categories. First, “undula-
tions” generally deviate slightly from symmetrical cross sections (Fig. 3.24c) that sig-
nify fracture propagation in various velocities. Second, “arrest marks” often are cut
characteristically by striae and deviate considerably from symmetrical shapes, with
the occasional tendency to form “box” or “semi-box” cross sections (Fig. 2.33b,c). These
marks signify slow propagation down to arrest.
Striae that form by rapid fracture in glass or glass ceramic penetrate into the mate-
rial and approximate sharp, continuous cuts (Fig. 2.19a). However, when striae propa-
gate slowly in granular rocks, they may completely lose their sharpness and form den-
drite-like plumes (Fig. 2.22a, 2.45).

2.2.5
Joint Fringes

2.2.5.1
Introduction

We start with a first order distinction between two end member fringe types, en echelon
fringes (Fig. 2.1a) and hackle fringes (Fig. 2.1b). En echelon segmentation from parent
fractures generally occurs in reduced velocities (Kulander et al. 1979; Müller and Dahm
2000), standing for the “quasi-static end member”, at a velocity of ≤10–1 m s–1 (Wieder-
horn et al. 1974) in the “fringe spectrum”. Hackles form at high velocities (e.g., Kerkhof
1975), representing the “dynamic end member” at a velocity of ≥10–1 m s–1 (Wiederhorn
et al. 1974). Between these two end members, there are two distinct fringes that bound
mirror planes on joint surfaces. One of them consists of somewhat hackled en echelon
segments (Fig. 2.33e) that display different lengths, resulting in no uniform width of
the fringe (Fig. 2.34a). The other one consists of en echelon segments of equal lengths
that are not hackled (Fig. 2.35a,b), often resulting in a uniform fringe width (Fig. 2.34b).
The need for this distinction has been previously raised, but it has been based on
limited data derived from rock exposures in sedimentary rocks (Bahat 1991a, Table 3.3;
Bahat 1997). Recent studies on fracture in granites enriched the information on the
properties of geological hackle fringes (Chap. 4) so that a sound distinction between
the two fringe types becomes possible. Understanding the full mechanical significance
of the spectrum between hackle fringes and en echelon fringes around mirrors is quite
a challenge. This understanding would improve the tectonofractographic understand-
 2.2 · Fractography and Tectonofractography 131

Fig. 2.34.
Diagrams of two types of
fringes. a A no uniform width
of the fringe of hackly en ech-
elon segments that have differ-
ent lengths. b A uniform width
of the fringe of en echelon seg-
ments of approximately the
same lengths (after Bahat 1998b)

ing of joint formation, including the quantitative estimation of paleostress conditions

leading to fracture, as well as applications to other research fields, such as petrology
(Chap. 4 and 6). It is appreciated, however, that the above distinction between the two end
member fringes is not always simple, and the question whether a particular marking is to
be classified as a hackle fringe or en echelon fringe may arise in some cases. Occasion-
ally en echelon segments and hackle cracks populate the same fringe (Fig. 2.1c). Also,
there is an experimental finding of fast propagation of en echelon cracks (Sect. 2.2.3.10).
Thus, an awareness of occasional difficulties in the distinction between the two end
member fringes is required. En echelon fringes are better known to geologists than
hackle fringes. Also not well known are cuspate hackles, but these form a small sub-
group that is not central in this chapter. Accordingly, the next two sections (Sect. 2.2.5.2
and 2.2.5.3) present a recap on en echelon fringes and on the sub-group cuspate hack-
les, respectively (partly from Bahat 1991a, pp. 170–183). The issues of hackle fringes and
transitions between the two end member fringes are expanded in Chap. 4.

2.2.5.2
The En Echelon Fringe

General. En echelon segmentation in non-coplanar planes is ubiquitous on various scales

from micrometers to kilometers in geological exposures occurring in different terrains.
This implies that breakdowns may be the consequences of various mechanisms (see a
comprehensive description of en echelon distribution in Pollard et al. 1982). Interac-
tion of non-coplanar cracks was observed by De Freminville (1914) in bitumene. Fol-
lowing that, Preston (1931) modeled the process leading to striae and en echelon seg-
mentation in glass. He pioneered the explanation that when a change in the direction
of the principal tension occurs, the crack must break down into a number of segments
so that each segment readjusts its orientation to maintain normality with respect to the
new direction of the principal tension. This mechanism has since been reaffirmed (e.g.,
Sommer 1969) and was formulated into a sophisticated fracture mechanics theory of
mixed mode I + III loading (Lawn and Wilshaw 1975a; Ryan and Sammis 1978; Pollard
et al. 1982; Cooke and Pollard 1996; Hull 1999), essentially following the rules leading
to the formation of the plumes by the same mixed mode (Fig. 2.20c,d).
Starting with some of the earlier works (e.g., Woodworth 1895, 1896; De Freminville
1914; Murgatroyd 1942; Hodgson 1961a,b; Bankwitz 1965, 1966; Sommer 1969;
Kulander et al. 1979; Bahat 1979a, 1986b; Pollard et al. 1982), the study of en echelon
fringes involved various fracture aspects that have been outlined by Bahat (1991a).
132 Chapter 2 · Elements of Fracture Geology

These include fracture geometry, breakdown styles, en echelon segments and connect-
ing bridges (or steps), fracture markings on segments, breakdown mechanisms, the
different roles of the dilatant en echelon cracks vs. their bridges, and the distinction
between the influences of remote and local stresses on the evolution of the fringe.

Fig. 2.35a,b. Styles of en echelon segmentation. a Photograph (from Bahat 1997) and b drawing of dis-
continuous breakdown of en echelon fringes both above and below the parent joint. The sense of rota-
tion of the segments in the two fringes is counter clockwise. Note similarity in shape and size of seg-
ments in both fringes. Scale is 15 cm

Fig. 2.35c,d. Styles of en echelon segmentation. Photograph (c) and drawing (d) of the curved-joint
with an intense discontinuous breakdown into many segments, below the elliptical perimeter of the
early “embryonic” joint at center, and a continuous breakdown at the distal sides. The joint undulates
along the strike. The upper numbers are measured azimuth degrees (from 029° to 046°) on the parent
joint. The lower numbers are azimuth degrees (from 043° to 047°) on en echelon segments. The delicate
barbs on the joint surface represent a bilateral plume. Scale bar is 50 cm (after Bahat 1997)
 2.2 · Fractography and Tectonofractography 133

Several aspects of en echelon segmentation are expanded below, making the distinc-
tion that fracture markings on the parent joint record events from the early history of
the joint, while fringes reveal processes that were associated with advanced and final
stages of fracture.

Fig. 2.35e.
Styles of en echelon segmenta-
tion. “Double fringes” beyond
the boundary of the mirror
plane on a joint cutting lime-
stone. Note the reverse sense of
stepping on the two fringes.
The first fringe shows en ech-
elon segmentation on which
there is an “upstairs climbing”
from right to left, while the
second fringe shows climbing
in the opposite direction, im-
plying change in the stress
field between the formation
of two fringes (a non-in situ
Cenomanian sample from
Iben-Dib, south Jordan)

Fig. 2.36.
a A joint from outcrop R show-
ing a division into two parts. An
upper semi-ellipse parent joint
concave downward is separated
by a curved shoulder from a
fringe of en echelon segments
below. There is a (hardly vis-
ible) division of the parent joint
into three concentric zones (A–
C, see text for more informa-
tion). Note the steps of different
sizes at a discontinuous break-
down along a curved shoulder.
Scale is 1 m (after Bahat 1997).
b Schematic block diagram
showing primary surface mark-
ings of a systematic joint:
1: Main joint face; 2: Fringe;
3: Plumose structures (plumes);
4: F-joints (b-planes); 5: Cross-
fractures (c-fractures); 6: Shoul-
der; 7: Trace of main joint face.
Note that cracks 5 are steps
that bridge adjacent en echelon
cracks in the fringe. Note in a
and b vertical and lateral frac-
ture fronts, respectively (modi-
fied by Roberts (1995) from
Hodgson (1961a))
134 Chapter 2 · Elements of Fracture Geology

Breakdown styles. Two breakdown styles are distinguishable: Discontinuous (Fig. 2.35a
and b) and continuous (Fig. 2.35c and d). In layered rocks, the discontinuous break-
down starts along a shoulder at the edge of the parent joint from where the steps start
and approach the layer boundary. When the shoulder is straight and parallels the adja-
cent layer boundary, steps of a given fringe would reach approximately the same maxi-
mum size (Fig. 2.35b). However, a discontinuous breakdown with curved shoulders
may have steps of different sizes (Fig. 2.36a). Note that in Hodgson’s diagram (Fig. 2.36b),
echelons appear in a discontinuous style and the plumes initiate on them at the fringe
of the parent joint (not on the parent joint itself). A discontinuous breakdown is often
characterized by a “root zone” (Bankwitz 1966). This zone is a narrow strip of small en
echelon cracks along the joint shoulder to which larger en echelon cracks are added as
the fringe develops. The continuous breakdown starts on the joint plane at various
distances from the layer boundaries and the steps have no shoulder as a starting line.
Such breakdowns often display fringes with steps of quite different shapes and sizes
(Fig. 2.35d). On one continuous breakdown, Bankwitz (1966) shows plumes that ini-
tiate on the parent joint and continue to propagate along the en echelon crack.

Abrupt and gradual loadings. It is inferred from the experimental results by Cooke
and Pollard (1996) (present Fig. 2.20c,d) that generally, breakdowns that are mani-
fested by discontinuous en echelon segments reflect processes of abrupt twisting. These
are associated with abrupt mixed mode I + III loading that occurs between the devel-
opment of the parent joint and the fringe and produces en echelon twisting (e.g.,
Fig. 2.35a,b). On the other hand, continuous breakdowns are associated with gradual
twisting, responding to mixed mode loading as the joint propagates and segmenta-

Fig. 2.37.
Mineralization on en echelon
cracks (dark color) but not on
the steps that connect them
(alternating white fractures)
demonstrates two separate
phases of failure. The sequence
of events in this chalk sample
(from the Middle Eocene) was
fracture of the parent joint
(note delicate plumes), en ech-
elon cracking, mineralization
of iron oxides that was fol-
lowed by dendrites of manga-
nese oxides on the parent joint
and en echelon cracks, and
finally, delayed step formation,
after flow of solutions had
seized (sample length from
top to bottom, about 12 cm,
after Bahat 1986b)
 2.2 · Fractography and Tectonofractography 135

tion occurs (Fig. 2.35c,d). The gradual mixed mode loading probably does not involve
any time-break during the process. It is, however, possible that time breaks do occur
during discontinuous processes, but their extents are not known (Fig. 2.35e). Quite
possibly the different fracture modes of the en echelon cracks and steps that cause
propagation vertically downward and laterally, respectively, may be associated with a
time interval between them (Fig. 2.37). Two fracture paths are manifested by part D of
the composite fracture (Fig. 2.38, see also Fig. 3.40a) along Wadi Secher (Fig. 3.33c).
An abrupt clockwise breakdown occurred in part D, striking into a path of twisted
segments (azimuths 080–083°). However, as the loading increased while these segments
propagated downward beyond the arrest mark, a curved path took over, and a gradual
counter clockwise rotation of the segments occurred (into azimuths 072–076°).
In outcrop Y (Fig. 3.33c) continuous and discontinuous breakdowns occur in the
same layer (Bahat 1986a). Furthermore, the “curved joint” was first segmented by a
discontinuous process underneath the early elliptical fracture on the curved joint, and
the second breakdown was continuous along the distal edges (Fig. 2.35c,d). The close
proximity of these two breakdown styles suggests that their fracture conditions were
not drastically different. The curved joint (Fig. 2.35c,d) and the composite fracture
(Fig. 2.38 and 3.40a) demonstrate that the continuous breakdown may have two mani-
festations. First, a gradual growth of en echelon segments from distal locations on the
parent joint (Fig. 2.35d) and second, a gradual curving of growing segments (D in
Fig. 3.40a). These two fracture manifestations are associated with relatively small twist
angles (0° ≤ ω ≤ 11°). On the other hand, discontinuous breakdowns in the curved joint
and in other joints from stations Y and R (Fig. 3.33c, 2.35d, 2.36a) are connected with
larger angles (12° ≤ ω ≤ 25°) (Bahat 1997).

Fig. 2.38. Photograph of a composite fracture from Wadi Secher (see explanation in the text and in
Fig. 3.40a) (after Bahat 1997)
136 Chapter 2 · Elements of Fracture Geology

Breakdown mechanisms. We discuss below two fracture mechanisms that form en

echelon fringes by local mixed mode loading. The “breakdown” model by Pollard et al.
(1982) concentrates on the mixed mode I + III that develops by an abrupt process dur-
ing the transition from the parent joint to the fringe (Fig. 2.39a–c and 2.40a,b). The
“fracture slanting” model (Bahat 1991a, 1997) has some resemblance to the “break-
down” model and supports its basic principles. An alternative model by Ramsay and
Lisle (2000) that applies the mixed mode I + III by remote extension and shear stresses
is presented in Sect. 6.10. See also reservations regarding the “breakdown” model by
Lazarus et al. (2001a,b).
Pollard et al. (1982) examined the mechanism of a fracture breakdown into ech-
elon cracks, and this section basically is cited from their work. They use three ideal-
ized block diagrams (Fig. 2.39a and 2.40a,b) and show that

(2.6a)

and

(2.6b)

Fig. 2.39.
a Block diagram illustrating the
geometry and mutual relations
between parent crack and en
echelon cracks (after Pollard
et al. 1982). b, c Schematic
sections of a discontinuous
fringe normal to en echelon
crack lengths. W = crack width,
w = part of the width which is
not overlapped, 2 m = the over-
lapped portion of the width,
when W = w + 2 m; 2s step, the
normal distance between two
adjacent cracks, 2k = the dis-
tance between centres of two ad-
jacent cracks parallel to cracks,
2c = the distance between cen-
ters of two adjacent cracks
along the shoulder, 2d = the
distance between initiation
points of two adjacent cracks
along the shoulder, and the
twist angle ω, between the
shoulder and en echelon cracks.
b ω = 20°; c ω = 30°. A compari-
son between b and c shows that
overlapping increases with ω.
At low ω, W ≅ w (after Bahat
1991a)
 2.2 · Fractography and Tectonofractography 137

where ω is the twist angle, 2c and 2k are different distances between centers, and 2s is
a step. Pollard et al. (1982) adapt Sommer’s (1969) explanation on fracture breakdown,
but add the internal pressure P to the normal stress σ yy ∞ acting on the fracture plane

(tensile stress is positive) (Fig. 2.40a,b).

Pollard et al. consider an elastic body subject to remote principal stresses σ1∞ and σ2∞
acting in the y-z-plane, and a fracture of width 2b subject to pressure P lying in the
x-z-plane with straight tips parallel to z (length and width of parent fracture by Pollard
et al. are reversed in the present text, to be consistent with other descriptions). Break-
down into echelon segments is independent of the third remote principal stress, which
acts in the propagation direction x. A uniform rotation Ω of the remote stresses about
the x-axis introduces a normal stress σ yy ∞ and a shear stress σ ∞ acting in the fracture
yz

Fig. 2.40.
Block diagrams illustrating
relations among applied stress-
es and cracks. a Parent crack of
width 2b and infinite length in z
is loaded by pressure P and
remote principal stresses σ1, σ2.
Rotation Ω of principal stresses
about x introduces stresses σyy
and σyz acting on the crack
plane. b Small element cut from
near parent-crack tip (small
arrowed square in a with incipi-
ent echelon cracks of width W
whose normal, the maximum
local tension σ1 making angle ψ
to the y-axis. Polar coordinates
centered at parent-crack tip
are r, θ. Heavy arrows indicate
relative motion of parent crack
walls when subject to mode I
and mode III deformation
(modified after Pollard et al.
1982)
138 Chapter 2 · Elements of Fracture Geology

plane. According to Pollard et al., breakdown of the parent fracture depends upon Ω
and the applied stress ratio

(2.7)

The inequality relating P and σ1∞ is required for the parent fracture to be open. The
numerator in Eq. 2.7 is the difference between the internal pressure and the remote
mean stress or pressure, –(σ1∞ + σ2∞) / 2. The denominator in Eq. 2.7 is the remote
maximum shear, (σ1∞ – σ2∞) / 2.
Near the crack tip the stress field is proportional to the stress intensity factors.
Pollard et al. resolve that the remote normal stress and internal pressure contribute
∞ + P)(Π b)1/2, and the remote shear stress intro-
to a mode I stress intensity, KI = (σ yy
∞
duces a mode III stress intensity, KIII = σ yz (Π b)1/2. The tendency for breakdown in-
∞ and σ ∞
creases as the ratio of stress intensities KIII / KI differs from zero. Writing σ yy yz
in terms of Ω and the remote principal stresses by resolving stress in the x-z-plane,
Pollard et al. obtain

(2.8)

Inequality of R and cos 2Ω is required for the parent fracture to remain open. For a
particular rotation angle, the stress intensity ratio increases as the stress ratio R de-
creases. For R ≅ 1, which is reasonably close to the range of R in nature, KIII / KI ≅ tan Ω,
and Pollard et al. find that relatively small rotations of the principal stresses could
initiate breakdown.

The fracture slanting mechanism. Various fracture properties of both edge and cen-
tral cracks were investigated in metallic plates as a function of their thickness (Bank-
Sills and Schur 1989). It was found that for both geometries, as plate thickness increases
the ratio of the critical stress for the slant crack divided by the critical stress for the flat
(unslanted) crack increases at failure, and the crack propagation rate for the slant crack
is slower than that for the flat crack (Fig. 2.41). Bank-Sills and Schur (1989) observed that
for both edge and central slant cracks, the KIII / KI ratio increases as plate thickness de-

Fig. 2.41.
Diagram showing fracture
mode transition from flat frac-
ture under plane strain condi-
tions (left) to slant fracture
under plane stress conditions
(right) (after Hertzberg 1976)
 2.2 · Fractography and Tectonofractography 139

creases and that the greatest value of this ratio occurs at the plate surface. Furthermore,
slant cracks in thinner plates will propagate more rapidly than those in thicker ones.
Thus, as thickness increases the tendency for slanting decreases, and the tendency for
slanting becomes greater in thinner plates as plane stress conditions are attained.
These experimental results reaffirm previous observations (Hertzberg 1976; Broek
1984) that cracks in tensile fields under fatigue loading propagate out of their plane in
thin plates, and they also show that this property is thickness dependent. The tendency
for slanting that is associated with the increase in the KIII / KI ratio becomes greater with
the decrease in thickness of the plate and from the center of the plate towards its edges.
Cooke and Pollard (1996) investigated crack segmentation in PMMA under various
KIII / KI conditions. They found that for ω of the en echelon cracks and the number of
segments formed, both increase with the increase of the ratio KIII / KI (Sect. 2.2.3.8,
see also Pollard et al. 1982). Hence, the main resemblance between the “breakdown”
and “fracture slanting” models is in the dependence of en echelon segmentation on
the KIII / KI parameter (Sommer 1969). The fracture slanting model, however, links (and
limits) the increase in KIII / KI to plane stress conditions, i.e., to fringes along joints in
layered rocks, and it does not apply to en echelon segmentation in massive rocks like
granite, a limitation not articulated by Pollard et al. 1982. An extension to other quasi-
plane stress settings of stratified structures, such as segmentation of large igneous dikes
near the ground surface (Anderson 1951) or en echelon fracturing on the ground along
branches of major strike-slip faults (e.g., Allen et al. 1972), should not be ruled out.

Non fracture slanting mechanisms. Fringes along curved boundaries often exhibit
changes in the senses of segment stepping, implying slight deviations from orthogo-
nality of the maximum extensile direction with respect to the parent joint. This mecha-
nism does not operate along straight layer boundaries, but it may operate in connec-
tion with en echelon segmentation that often occurs on large joints in sandstone at
remote locations from straight boundaries (e.g., Fig. 3.23a). Scarcity of hackles on
such surfaces seems to suggest sub-critical conditions. Another type of en echelon
segmentation occurs along fracture fronts when critical stress intensity KI ≥ KIc and
rapid crack propagation conditions are attained (Chap. 4). En echelon fringes around
exfoliation joints may (Fig. 2.33e, 3.24b and 3.26c) or may not (Fig. 3.24c, 3.25a) attain
critical fracture conditions. On fringes along curved surfaces see Sect. 4.5.3.2.

2.2.5.3
The Cuspate Hackle Fringe

Cuspate hackles are distinguishable from en echelon and hackle fringes. Cuspate hack-
les have sharp ridges of approximately constant width that radiate in a fringe around
the main fracture. They are common on blasted fracture surfaces (Fig. 2.30b) and rare
on joints (Fig. 2.42). The fracture marking from the Charing chalks in Kent, England
displays a curious type of fringe. The echelon cracks occur above and below the plume,
whereas cuspate hackles continue the trend of the plumes along the distal direction
(left side of joint). This is interpreted as greater stress intensity along the lateral frac-
ture propagation as compared to the upward and downward directions. The common
occurrence of cuspate hackles in quarries (Bahat 1991a, Fig. 3.60a–c), imply fracture
by high energy conditions (fast crack propagation).
140 Chapter 2 · Elements of Fracture Geology

Fig. 2.42. Cuspate hackles. A plume developed into right-stepping and left-stepping echelons at the
upper and lower parts of the marking, respectively, but transformed laterally towards left into cus-
pate-hackles, suggesting a lateral cracking under conditions of greater stress intensity. From an Up-
per Cretaceous chalk in south England (after Bahat 1991a)

The sharp peaks of cuspate hackles are distinct from the flat surfaces of en echelon
cracks. Crack overlapping, which is a typical feature of en echelons, does not occur in
cuspate hackles. They have not been identified on fracture surfaces of technical glasses.
When they occur on joint surfaces they seem to serve the role of hackles on fracture
surfaces of glass. There is a strong resemblance between cuspate hackles on joints and
fractures that form by blast to fracture surfaces of some rapidly fractured ceramics
(Fig. 2.33a), even if size scales are an order of magnitude different. As mentioned above,
hackle fringes are described in Chap. 4.

2.2.6
Mirror Boundaries

We characterized in the preceding sections the properties of the mirror plane and the
fringe. Another important parameter is the boundary that separates these two parts
of the joint. The importance of finding well-defined critical flaws when coupled with
well-defined mirror boundaries provide the necessary conditions for estimating the
local tensile paleostress on a joint surface (Bahat and Rabinovitch 1988) (Sect. 6.5). In
plutonic rocks, there are occasionally favorite conditions for finding clear mirror
boundaries (e.g., Bahat et al. 1999, 2003, see present Chap. 4), no such clarity has been
so far obtained in sedimentary rocks. We now consider the criteria for elucidating mir-
ror boundaries of single-layer joints. At least one of the following two criteria must be
identified on a joint surface for determining a mirror boundary, while the third and
fourth criteria can increase the credibility of the definition:
 2.2 · Fractography and Tectonofractography 141

1. There is a stepwise morphological increase in the size of the cracks across the bound-
ary from plumes on the mirror to fringe cracks (Fig. 2.22b and 2.33a,e). This has to
be consistent along at least 50% of the suspected mirror boundary of the investi-
gated joint.
2. The tilt angle φ is formed between the imaginary continuation of the mirror plane
and the fringe (Bahat and Rabinovitch 2000). This angle is better seen in profile
than in plan view (Fig. 4.11c).
3. The ratio of mirror radius (Fig. 2.1d) to the radius of the critical flaw should be
15 ±5 (Bahat and Rabinovitch 1988).
4. Apart from size difference between plumes and fringe cracks, there are occasion-
ally changes in the sense of stepping in the transition from the plume to the en
echelon segmentation on the fringe (Fig. 3.40b).

2.2.7
Recent Studies on En Echelon Fringes from the Beer Sheva Syncline

This section relates to the occurrence of fringes on joint surfaces from sedimentary
rocks in the Beer Sheva syncline, based on the study by Bahat (1997) that we cite be-
low. Here we concentrate on fractographic relationships, while the geological impli-
cations of this study are presented in Sect. 3.4. The Eocene chalk layers from the Beer
Sheva syncline (Fig. 3.33) are only slightly deformed platform cover sediments. Chalk
is a weak rock with considerable change in properties when exposed to aqueous solu-
tions. Therefore, fracture may be readily induced in this material, even in very low
differential stresses. The focus here is on bed restricted uplift joints (Bahat 1999a).
Jointing in Eocene chalks around Beer Sheva is enriched with various phenomena of
en echelon segmentation that bear relationships to diverse fracture styles and mul-
tiple joint systems (Chap. 3).

2.2.7.1
The Relationship of En Echelon Length to Layer Thickness

Lengths of en echelon segments cutting layers of various thicknesses were measured

along stations Y, R and Q in Wadi Naim (Fig. 3.33c). In these locations, en echelon
fringes occur only below the parent joint (Bahat 1986a). The length of the longest
segment for each joint was measured from its initiation point to the lower layer bound-
ary. The layer thickness at that location was measured as well (estimated error for
both measurements is ±2 cm). Five continuous and fourteen discontinuous seg-
mentations were studied, and the results are summarized in Fig. 2.43. There is a posi-
tive linear fit (R2 = 0.928) of the readings for discontinuous segmentation, which
is given by l = 0.270T + 0.829, where l is segment length and T is bed thickness. R2 is
reduced to 0.836 when one reading (T = 210 cm, l = 55 cm) is omitted. The readings
for the continuous segmentations scatter above the curve. There is no evidence of
any layer-parallel slip, along either chalk-chalk or chalk-chert boundaries linked
to en echelon fringes in the stations investigated in this study. On the other hand,
there are many inter-layer radial markings on horizontal layer boundaries that re-
flect tensile vertical strains perpendicular to layer boundaries in the chalk (Bahat
1991a, p. 298).
142 Chapter 2 · Elements of Fracture Geology

Fig. 2.43.
A plot of layer thickness vs.
length of en echelon cracks
in chalk outcrops along Wadi
Naim. The large fourteen open
squares and the five solid
squares designate respectively
discontinuous and continuous
breakdowns. The solid and
dashed plots are the products
of 14 and 13 readings of dis-
continuous breakdowns, re-
spectively. The last reading
(T = 210 cm, l = 55 cm) is omit-
ted from the latter. The former
is considered to be a better rep-
resentation of the results (after
Bahat 1997)

2.2.7.2
Similar Fringes below and above the Parent Joint

En echelon segments maintain counter clockwise rotation in fringes close to both the lower
and upper boundaries in outcrop X (Fig. 2.35a,b), and the two fringes, above and below the
parent joint, have similar geometries and shapes. This coupling implies that en echelon
segmentation occurred simultaneously in the two fringes. It took place after the end of the
required diagenetic process for the equalization of the lithological properties along both
the lower and upper layer boundaries. The fractographic features of the joints in outcrop X
are quite distinct from the fractographies of all other joints in the investigated area.

2.2.7.3
Opposite Senses of Segment Rotation in a Single Fringe

The observation of the close proximity of both clockwise and counter clockwise rota-
tions of en echelon cracks in a single fringe is quite intriguing. Based on experimental
results (Cooke and Pollard 1996), it follows that the clockwise twisted segments in fringes
reflect right-lateral shear, and the counter clockwise twists disclose left-lateral shear.
There is a need to distinguish between two cases of opposite senses of segment rota-
tion in a single fringe: First, in a curved fringe, and second, in a straight fringe.
Opposite senses of segment rotation in curved fringes are quite common in frac-
tured engineering materials and in association with discoid joints. Segment rotation is
counter clockwise on the left side of a rounded parent fracture in steel and is clockwise
on the right side of the fracture (Fig. 2.44a, after Yukawa et al. 1969). Exactly the same
pattern occurs around a discoid joint in the Frauenbach quarzite from the Thuringian
Schiefergebirge (Fig. 2.44b, after Bankwitz and Bankwitz 1984). In fact, Bankwitz and
Bankwitz (1984, Fig. 2) observed a symmetry that was created by a reverse segmenta-
tion. The same segmentation pattern is repeated on the surface of a fractured glass rod
that failed under combined operation of modes I and III (Sommer 1969, Fig. 2).
 2.2 · Fractography and Tectonofractography 143

Fig. 2.44. Drawings of fringes around circular parent fractures. a Segment rotation is counter clock-
wise on the left side of a fracture in steel and is clockwise on the right side of the fracture (after Yukawa
et al. 1969, Fig. 1). Scale bar is 12 mm. b A repeated pattern from a, around a discoid joint in the
Frauenbach quarzite from the Thuringian Schiefergebirge (after Bankwitz and Bankwitz 1984, Fig. 3).
Scale bar is 5 mm

This fracture pattern is a result of a slight deviation of the tensile axis (by not more
than several degrees, Sommer 1969) from normality to the parent fracture. It caused
two segmentations advancing in opposite directions, rather than a single propagation
that surrounded the entire boundary of the circular parent fracture. The latter propa-
gation would occur under “pure tension”, which is probably difficult to achieve, and
therefore would be less common. On the discoid from Mt. Carmel (Fig. 3.40c), there
is no segment overlapping in certain parts of the concentric fringes. It is assigned to
“pure tension”, which was applied on the central part of the discoid but not on its right
and left parts, suggesting that even locally the stress field was not homogeneous. The
“concentric joint” displays opposite senses of segment rotation in a single curved fringe
(Fig. 3.40b) induced by a local stress field. A remote deviation of the minimum prin-
cipal direction would result in similar segment rotations in neighboring joints, typi-
cal of many joint fringes along Wadi Naim (Bahat 1986a), which is not the case shown
in this outcrop.

2.2.7.4
Opposite Rotations of En Echelon Segments in Fringes below
and above the Parent Joint

Significant dissimilarities in size and shape of different fringes of the same joint should
reflect a deviation from simultaneous en echelon segmentation. The two fringes in
Fig. 2.45 differ in the directions of segment rotation and in segment size and geom-
etry, quite possibly suggesting that the two fringes responded to different loading. It
appears that the fracture sequence of this joint was initially:

1. Propagation of the parent joint from left to right, as shown by the plume.
2. Segmentation of the lower fringe in a rational continuation with the plume.
3. Segmentation of the upper fringe follows a change in the loading direction.
144 Chapter 2 · Elements of Fracture Geology

Fig. 2.45.
A fracture cutting Lower Eocene
chalk in the Shephela syncline.
En echelon fringes occur on
both the upper and the lower
sides of the parent joint. There
is a clockwise rotation of the en
echelon segments with respect
to the parent joint in the upper
fringe and a counter clockwise
rotation in the lower one. How-
ever, cracks in the two fringes
differ considerably in shape.
Length of plume is about 30 cm
(after Bahat 1997)

2.2.7.5
Timing Relationships of Plumes, Arrest Marks and En Echelon Segmentation

The contact relationships of rib marks and barbs, combined with their interactions
with crack segmentation, are useful in estimating changes in local stresses and enable
the distinction of different stages in the fracture process. The plume on the “curved
joint” (Fig. 2.35c) is bilateral, and the segmentation shows a uniform clockwise rota-
tion, which corresponds to one direction of the plume but not to the other. The impli-
cation is that breakdown occurred after the completion of the growth of the bilateral
joint. Hence, the growth of the curved joint was in several stages:

1. Fracture into an early elliptical joint.

2. Growth along the upper part of the layer into a bilateral joint that curves along the
strike.
3. Breakdown of the joint into segments that propagated downward.

Some joint surfaces show divisions into concentric zones that are bounded by
arrest marks (Sect. 2.2.4.1). These arrest marks do not only represent the fracture
front at “rest” but they also mark the loci of stress redistribution (Murgatroyd 1942).
This is revealed by changes in the sense of rotation of overlapping barbs with respect
to the en echelon segments (Fig. 3.40b) or by changes in the segment azimuths (Fig. 2.38
and 3.40a, part D). Arrest marks (Sect. 2.2.4.1) can be useful in the interpretation
of the interrelationships of individual joints associated in a common composite
fracture. The series of sub-horizontal arrest marks that continue through parts B, D
and E (Fig. 3.40a) suggest an early synchronization of the downward propagation of
these parts.
What was the growth rhythm of the various segments along the “curved joint”?
There seem to be conflicting observations. On the one hand, the breakdown into seg-
ments along the curved joint occurred under various KIII / KI ratio conditions, which
may imply that segmentation was not synchronous along the joint, but it occurred at
the center before developing at the distal ends. On the other hand, it seems quite likely
 2.2 · Fractography and Tectonofractography 145

that the various segments reached simultaneously the lower part of the joint, as dem-
onstrated by an arrest mark that decorates all the segments at the same height and
practically appear to be a monotonous morphological feature. Thus, the possibility of
a non synchronous segment initiation but a synchronous completion of segmentation
is not ruled out.

2.2.7.6
The Spreads of Azimuths in Joint Sets versus En Echelon Fringes

The spread of azimuths in individual joints and in joint sets is often greater than the
spread of azimuths of the en echelon cracks associated with them. One example is
shown in Fig. 2.35d. Other joints in outcrop Y show similar patterns. A summary of
fracture orientations in outcrop X shows that the mean azimuth of the parent joints is
N 03° W with standard deviation ±12° (63 measurements). The corresponding result
for the en echelon cracks along the fringes of these joints is N 23° W ±6° (91 measure-
ments), i.e., considerably lower spread (Bahat 1986a).
Geologists often appreciate the difficulty in estimating paleostress directions because
of the uncertainties involved in determining the spread of the joint population into sets
and establishing the average azimuth of a given set. If the spread of azimuths of en ech-
elon segments in a set may be considerably smaller than the spread of azimuths of the
parent joints associated with that set, it should enable quite readily the distinction be-
tween neighboring joint sets and their associated en echelon segmentation.
This is well exemplified in Table 2.1. It would be difficult to distinguish between joint
sets X and O-Y3 because of the overlapping of ranges of fracture azimuths (350–014°
and 012–035°, respectively). It is, however, considerably easier to distinguish these sets
on the basis of their segmentation distinct ranges (338–342° and 026–053°, respec-
tively). Hence, the styles of en echelon segmentation and rotation can considerably
improve the confidence in the division of joints into sets and upgrade the analysis of
paleostress directions.

Table 2.1. Orientation of joints and associated en echelon segments in the Beer Sheva syncline and
corresponding compression
146 Chapter 2 · Elements of Fracture Geology

2.2.7.7
An Ascending Order of Fracture Morphologies on Joint Surfaces

It is useful to set the various fracture morphologies on joint surfaces in an ascending

order of compounded elements. These can be correlated with various stress scenarios,
from 1 to 12 (Fig. 2.46). In a crude way, scenarios 1 to 7 represent various manifesta-
tions of fracture morphologies that respond to changes in local stresses, whereas sce-
narios 8 to 12 show fracture morphologies that may help to evaluate changes in direc-
tions of remote stresses.
Scenarios 1 to 3 represent unilateral, bilateral and radial modes of joint propaga-
tion, respectively. Common to these morphologies is the perpendicularity of the mini-
mum principal stress with respect to the parent joint, with possible slight deviations
from it (Bahat 1997). Local constraints dictate the choice among the three, and a pro-
longed fracture in a thick layer would favor the third one (scenarios 1 and 2 are well
known, and scenario 3 is manifested by the parent joint shown in Fig. 2.36a, display-
ing three concentric plume series.

Fig. 2.46.
Fracture morphologies on
joint surfaces in an ascending
order of compounded elements
which correlate with twelve
stress scenarios; see text for
explanation (after Bahat 1997)
 2.2 · Fractography and Tectonofractography 147

Scenario 4 represents a composite fracture that developed by the merging of a se-

ries of sub-parallel smaller joints, propagating in the same approximate stress field
(Fig. 2.38). Scenario 5 describes the propagation of one part of the composite fracture
(Fig. 3.40a, part D), which is the only joint that deflects in response to a slight devia-
tion of the local minimum principal stress from perpendicularity. A deviation of the
remote principal stress from perpendicularity to the composite fracture would have
resulted in the deflection of additional parts of the composite fracture as well.
Scenario 6 results when the local minimum principal stress tilts somewhat from
perpendicularity to the parent joint, so that a left-lateral shear changes to a right-lateral
shear (Fig. 2.44a,b). It appears that segmentation without overlapping (Fig. 3.40d) signi-
fies a transitional stage between left-lateral and right-lateral shears. Scenario 7 concerns
alternating shears among neighboring joints that constitute a composite fracture. This
situation occurs in massive sandstones (Fig. 3.23b,c), but is not known in chalk.
Scenarios 8 and 9 are recognized by the abrupt transitions between the parent joint
and the en echelon straight fringe. These structures are good candidates for the iden-
tification of remote stress directions. Scenario 10 (Fig. 2.35a) may be considered as
an “upgrading” of scenarios 8 and 9, also leading to the identification of remote stress
directions. Scenarios 11 and 12 are related to opposite segment rotations in straight
fringes (Bahat 1997, Fig. 9 and Fig. 2.45, respectively). They also provide indications
for the distinction between different remote stress directions.

2.2.7.8
Local Stresses versus Remote Stresses

We summarize below the distinctions between the influences of local and remote
stresses on the formation of en echelon cracks (Pollard et al. 1982; Bahat 1986a, 1997).

Local stresses. Field observations indicate that much of the interrelationship of the
parent joint with the fringe and the behavior of the en echelon segments are influ-
enced by local conditions, as itemized below:

1. There is a correlation between the size of en echelon cracks formed by the discon-
tinuous breakdown and layer thickness (Fig. 2.43).
2. The various measurements of en echelon lengths vs. the layer thickness in the thick-
ness range of 55–65 cm (Fig. 2.43) are all from layer 5 (Bahat 1986a, Fig. 2) and
they show both continuous and discontinuous breakdowns that cannot be explained
by remote influences.
3. Occasionally, specific timing relationships exist among plumes, the vertical en ech-
elon cracks and the horizontal segments of arrest marks. These relationships may
occur on surfaces of individual joints or composite joints (Fig. 2.35d and 2.38) but
not in association with any of their neighbors along the same layer.
4. The sense of rotation of the segments in the fringe is clockwise, compared to counter
clockwise rotation of the barbs on the parent joint (Fig. 3.40b). It seems highly
unlikely that these two rotations reflect remote stresses. Quite possibly the counter
clockwise rotation of the barbs is in response to local stress.
5. The reverse segment rotation along the fringe of the concentric joint reflects local
conditions, rather than two distinct regional loadings.
148 Chapter 2 · Elements of Fracture Geology

Remote stresses. In spite of the important influence of local conditions on the behav-
ior of en echelon segments, there are certain parameters that can quite clearly be cor-
related with remote stresses:

1. The decoupling of the fringe from the bilateral plume described above (Fig. 2.35c,d)
suggests that these two morphologies were formed in response to distinct stress
conditions. Here, the preference of the plumes to propagate bilaterally rather than
unilaterally demonstrates a response to local conditions. On the other hand, the
azimuth uniformity of the en echelon segmentation in the fringe and other fringes
of neighboring joints in a set conforms to the influence of a remote stress. The
decoupling of a fringe from a unilateral plume is as much an indication that the
segmentation in the fringe represents a remote stress (as in the previous case re-
lated to the bilateral plume). A good example is provided by Younes and Engelder
(1999). The uniformity in the sense of rotation of en echelon cracks with respect to
their parent joints over great distances is a manifestation of fracture induced by
remote stresses.
2. The conformity of clockwise rotation of en echelon segments with the NNE joints
in station Y, on the one hand, and the strict counter clockwise rotation associated
with the N-S joints in outcrop X on the other hand suggest two sets responding to
distinct remote stresses (Table 2.1).
3. Specifically, fractures in outcrop X differ from all other fractures in the investi-
gated area by the uniformity of their en echelon cracks in terms of their sense of
rotation, their size and shape, and in the fringes that occur both below and above
the parent joint. This distinct assemblage of properties categorizes a distinct re-
mote stress field.
4. When a set of joints demonstrates a gradual azimuth rotation, the en echelon
cracks at the fringes of these joints rotate in covariance with the joints (Bahat
1986a).
5. Discontinuous breakdowns are distinct from continuous breakdowns by often
having a series of small en echelon cracks along the shoulder distributed among
the larger en echelon cracks. Such series of small cracks have been named the “root
zone” (Bankwitz 1966). They generally maintain their azimuths in alignment with
their larger neighbor en echelon cracks, responding to remote stresses.

Spatial and temporal misalignments. The reader should be aware of differences

in terminology. There are investigators that prefer to use different terms in their
distinction between the remote and local categories. Younes and Engelder (1999)
assert that if an advancing joint enters a volume of rock subject to a stress field
misaligned relative to that guiding the parent joint, transverse adjustments in crack
propagation due to mixed-mode loading cause the joint to break down into fringe
cracks. If the misaligned stress field was present prior to joint propagation, the mis-
alignment is spatial. If the misalignment took place after propagation and arrest of
the parent joint, the misalignment is temporal. Due to occasional unclear cases re-
garding the propagation and arrest of the parent joint, we are not in favor of using this
distinction and prefer the terms remote and local according to the description pre-
sented above.
 2.2 · Fractography and Tectonofractography 149

2.2.7.9
En Echelon Segmentation under Constraints

Dilatational en echelon segmentation occurs under favorable conditions for mode III
operation, and it is limited when constraints are imposed on this mode. As was men-
tioned above, en echelon segmentation is very common in rocks. It is also frequent in
fractured polycrystalline metals (Hertzberg 1976; Tschegg 1983; Broek 1984 and ref-
erences therein) and in single-crystals (Gilman 1958). On the other hand, en echelon
segmentation is rare in (non-polymeric) glass, and the widely cited experiment by
Sommer (1969) is related to segmentation in glass under fluid pressure.
The explanation to these differences is that defects in the structure continuity, like
screw dislocations in single-crystals (Gilman 1958) and inclusions and grain bound-
aries in steel (Tschegg 1983) and in rocks, promote the development of local, hetero-
geneous stress fields, resulting in mixed mode loading that induces segmentation. The
much smaller population of such discontinuities in non-polymeric glass is possibly
the reason for the limitation of en echelon cracks in this material. En echelon segmen-
tation does occur in polymeric glasses like PMMA (e.g., Cooke and Pollard 1996).
However, fracture propagation in this material must involve an early strain and a break-
down of craze fibrils (Maccagno and Knott 1989). The effect of “mode III crack clo-
sure” (Tschegg 1983) imposes an important constraint on segmentation. Namely, en
echelon cracking along layer boundaries would be minimized under compression by
the overburden, and therefore, even if found in rocks in considerable depths, an early
failure at shallow depths should be suspected. On the other hand, high liquid pres-
sures could overcome the constraint imposed by the lithostatic pressure and induce
extensile en echelon segmentation at considerable depths, in a similar manner by which
they form joints at great depths by mode I (Secor 1965). The latter hypothesis still needs
to be verified.

2.2.7.10
The Influence of Ductile Conditions on En Echelon Segmentation

We have seen above that mode III loading is a major promoter of en echelon segmen-
tation. The plastic zone of mode III loading is several times larger than that for mode I
for similar materials and at the same stress intensity values (Tschegg 1983). Thus, seg-
mentation is associated with the increase in the plastic zone. This observation coin-
cides well with the plane strain/plain stress relationship in a plate, showing that the
plastic zone gradually increases from the plane strain size in the interior to a plastic
zone of a larger size under plane stress at the surface of the plate (Hertzberg 1976;
Broek 1984). Analogously in geological terms, conditions along the layer boundaries
are more conducive for segmentation than in the layer interior.

2.2.7.11
Limitations and Future Research

The study presented in this section has certain deficiencies that are addressed below.
First, most of the information presented here has been gathered by one researcher (the
150 Chapter 2 · Elements of Fracture Geology

first author), and some subjective interpretation is perhaps unavoidable. Therefore, the
various observations and subsequent rules that were determined in this study should
be considered tentative until they are verified in similar sedimentary fracture provinces.
Second, the present section essentially relates to a single lithology, chalk. Therefore, some
of the relationships that were established here may have to be modified when applied
to multi-lithological provinces (such as the Appalachian Plateau). Third, there is a need
to further investigate the nature of the continuous breakdown. What causes breakdown
to start from the boundary of the parent joint (shoulder in Fig. 2.35a) and when does it
to start from within the parent joint (e.g., Fig. 2.35c)? Fourth, alternative theories of
fracture mechanisms that exist today regarding the formation of fringes (Sect. 6.10) need
to be evaluated, in order to develop a strong framework into which additional field and
experimental information would be added and interpreted.

2.2.8
Straight Kinks and Curved Kinks

A kink is generally perceived as a relatively short crack that occurs at the termination
of a straight parent fracture that started as a joint. Due to mixed mode I + II loading,
kinks deviate from the strike of the parent fracture, pointing towards the continua-
tions of the extensional quadrants (Cotterell and Rice 1980) (Fig. 2.47a). Cruikshank
et al. (1991) used a mixed-mode linear elastic fracture mechanics model and applied
the maximum circumferential stress criterion (Erdogan and Sih 1963) in order to ac-

Fig. 2.47. a Relations between sense of shear and direction of kinking. For positive, right-lateral
shear, the kink angle is negative. For negative, left-lateral shear the kink angle is positive. b Examples
of idealized tail cracks for various far-field loading conditions. In all cases, pressure within the
fracture is adjusted to cause propagation. In 1 and 2, the stress parallel to the fracture is compressive
and in 3 it is zero. Tail cracks kink and then converge. In 4 the stress parallel to the fracture is tensile
and equal to that normal to the fracture. The fracture merely kinks. In 5, the stress parallel to
the fracture is a larger tensile stress than that normal to the fracture, and the tail rack diverges
(from Cruikshank et al. 1991)
 2.2 · Fractography and Tectonofractography 151

count for the orientation of cracks near the ends of small-displacement faults. They
assumed that the faults were subject to uniform driving stresses and that the cracks
opened perpendicular to the greatest circumferential stress (σϑϑ) near the fault tip (see
Lawn and Wilshaw 1975a, Fig. 3.4, for general conditions, and present Fig. 1.25
for ϑ = 0). In their model, the sense of kinking indicates the sense of shear during fault-

Fig. 2.48. a Definitions of kinematic properties of a crack. A Crack length is 2a. y-coordinate normal to
crack, x*- and x-coordinates parallel. Dot is a material particle split by the crack. B Displacement
components, U and W, of part of a material point along a crack wall relative to the other part of the
material point along the opposite crack wall. Mode I stress intensity equal to critical stress intensity for
mode I fracturing at the time a kink forms. C Displacement in mode I loading at the time the initial
fracture formed. Material point split and parts displaced only normally relative to one another. Mode I
stress intensity equal to critical stress intensity at the time the fracture stops lengthening. b Relations
between kink angle α and, A ratio of opening, W, to slip, U along fracture, and B relation between kink
angle and ratio of initial opening, Wi to subsequent slip, U, along fracture (from Cruikshank et al. 1991)
152 Chapter 2 · Elements of Fracture Geology

ing: A kink that turns clockwise with respect to the direction of the main joint is a
result of right-lateral shear, and a kink that turns counter clockwise is a result of left-
lateral shear. Furthermore, the kink angle α, the angle that the secondary fracture forms
with the extension of the primary fracture, is related to the ratio of the shear stress
responsible for the kinking to the normal stress responsible for the opening of the fault
(Fig. 2.47a).
Cruikshank et al. (1991) consider both sharp/straight and curved kinks and show
a span of orientations of a straight kink and curved kinks that can be digitally ob-
tained under various loading conditions (Fig. 2.47b). In their model, which limits
stresses to principal directions, pressure within the parent fracture is adjusted to cause
propagation. Thus, while Cruikshank et al. (1991) concentrate on stress differences
(Fig. 2.47b), Cooke and Pollard (1996) study the influence of loading rate on the shape
of kinks (Fig. 2.20a,b). Cruikshank et al. (1991) suggest that the amount of opening of
a joint at the time it faulted or even at the time the joint itself formed can be estimated
by measuring the kink angle and the amount of strike-slip at some point along the
faulted joint (Fig. 2.48a,b). They relate the ratio of fault aperture, W, to slip, U, at the
time of faulting to α. For a pure mode I fracture (W / U = ∞), α = 0°. For a pure mode II
fracture (W / U = 0), α = cos–1 (1 / 3) ≈ 70.5°. For angles between 35° and 55°, they cal-
culated that the ratio of opening to offset at the time of faulting was between 3.02
and 1.70, respectively. These angles hold no matter how large the relative displacements
of the fracture walls are, provided they are not zero.

2.2.9
Recent Studies from the Appalachian Plateau

2.2.9.1
General

This section relates to the occurrence of fringes and kinks associated with joints in
sedimentary rocks from the Appalachian Plateau, based on the study by Younes and
Engelder (1999) that we cite below. Here we concentrate on fractographic relationships,
while the wide geological implications of their study are cited in Chap. 3. Younes and
Engelder (1999) consider three types of fringe cracks in their study area: Gradual en
echelon fringes (that they term gradual twist hackles), abrupt en echelon fringes (that
they term abrupt twist hackles), and kinks. A description of these three types with a
subsequent discussion is presented below.

2.2.9.2
Gradual En Echelon Fringes

Gradual en echelon fringes occur within the same bed and lithology that hosts the
parent joint (Fig. 2.49a,b). The surfaces of the en echelon segments occasionally show
plumes that indicate an overall vertical propagation direction. Younes and Engelder
(1999) observe that this surface morphology is often continuous with that of the par-
ent joint, giving no indication that the parent joint arrested before the gradual en ech-
elon fringe propagated. In their model, however, the plumes on the fringe are sepa-
rated from the plume on the parent joint (Fig. 2.49a), as compared with our model
 2.2 · Fractography and Tectonofractography 153

(Fig. 2.20d) that shows plume continuity. Upon breakdown at the joint tip line (equiva-
lent to the shoulder in Fig. 2.36b), the en echelon segments twist away from the par-
ent. The segments realign normal to the local maximum tensile stress, propagating
perpendicular (i.e., vertically) to the overall lateral propagation direction of the par-
ent joint. This is clearly seen on the diagram in Fig. 2.49a (not as clear in the photo-
graph of Fig. 2.49b).

Fig. 2.49a,b. En echelon segmentation in the Appalachian Plateau. a Schematic of a gradual en ech-
elon fringe. The parent joint and en echelon fringes are carried within a single host bed. The number
of fringe cracks decreases away from the parent joint. The spacing of the cracks changes from small
at the tip line of the parent joint to large at the end of the fringe. b A photograph of a gradual en
echelon fringe. A parent joint and an en echelon fringe are carried within a single bed of siltstone of
the Ithaca Formation at Taughannock Falls State Park, New York. Propagation direction for the en
echelon fringe is upward. The sense of stress field rotation in this example is clockwise. The scale to
the lower right is divided into centimeters (from Younes and Engelder 1999, no changes are made in
the wording of original figures)
154 Chapter 2 · Elements of Fracture Geology

Fig. 2.49c,d. En echelon segmentation in the Appalachian Plateau. c Schematic of an abrupt en ech-
elon fringe. These fringe cracks can propagate downward or upward from a siltstone bed into adja-
cent bed. They have a characteristic sequence of large and small cracks, indicative of suppressed
growth where crack stress shadows limit the growth of some cracks. d A Photograph of an abrupt en
echelon fringe. This set of fringe cracks propagated downward into a thick shale bed from a thinner
siltstone bed hosting the parent joint at Taughannock Falls State Park, New York. These rocks are part
of the Ithaca Formation. The sense of stress field rotation in this example is clockwise. The scale is a
geologic compass with an 8-cm base (from Younes and Engelder 1999, no changes are made in the
wording of original figures)
 2.2 · Fractography and Tectonofractography 155

2.2.9.3
Abrupt En Echelon Fringes

According to Younes and Engelder (1999), abrupt en echelon fringes are commonly found
in shale and initiate from the tip line of dip joints in siltstone layers that are typically
less than 30 cm thick (Fig. 2.49c,d). Because abrupt en echelon segments abut the par-
ent joint at a point, they appear in map view to grow across the edge of the parent joint.
In contrast with gradual en echelon fringes that grow within the same bed as the parent
joint, abrupt en echelon fringes are separated from parent joints by a layer boundary
(Younes and Engelder 1999). These en echelon cracks are equally common, propagat-
ing upward, downward, or in both directions from siltstone beds.

2.2.9.4
Kinks

According to Younes and Engelder (1999), in the Appalachian detachment sheet kinks
occur exclusively in siltstone beds (Fig. 2.50), and no evidence for slip was observed
on any parent joint carrying kinks in the detachment sheet. They make the point that
kinks occasionally show a surface morphology of plumes that are continuations of the
plume morphology from the parent joint. Therefore, they suggest that kinks occur as
mode I cracks and the tilt in a crack path is driven by a shear traction superimposed
on the parent joint after arrest, following a finite amount of mode I propagation
(Erdogan and Sih 1963; Cotterell and Rice 1980).

Fig. 2.50. An example of fringe cracks in a siltstone bed of the Ithaca Formation found near Whitney
Point, New York. The shale above and below the siltstone bed carried an abrupt en echelon fringe. The
sense of stress field rotation in this example is counter clockwise, both for the kink and the abrupt en
echelon fringe. The parent joint is found in a bed roughly 20 cm thick (from Younes and Engelder
1999, no changes are made in the wording of original figures)
156 Chapter 2 · Elements of Fracture Geology

2.2.9.5
Tilt and Twist

Younes and Engelder (1999) observe that the angle of twist or tilt of a fringe crack set
changes according to the host rock, the type of the fringe crack and the sense of rota-
tion. For simplicity, they use the same angle sign (β ) for the twist angle (equivalent to ω
in this book), the tilt angle and the kink angle. Younes and Engelder found that the twist
angles for clockwise en echelon cracks are generally smaller than for counter clockwise
en echelon cracks. Gradual en echelon cracks have small twist angles, with a mode at 3°
for clockwise en echelon cracks and 6° for counter clockwise en echelon cracks. Abrupt

Fig. 2.51.
Change in the angle between
parent joint and the clockwise
(CW) gradual en echelon fringes
according to lithology. The maxi-
mum twist angle increases as
the clastic grain size becomes
larger (from Younes and Engel-
der 1999, no changes are made in
the wording of original figures)
 2.2 · Fractography and Tectonofractography 157

en echelon cracks have larger twist angles, with a mode at 13° for clockwise en echelon
cracks, and 22° for counter clockwise en echelon cracks. Kinks have tilt angles with a
mode of 16° for clockwise en echelon cracks and 14° for counter clockwise en echelon
cracks. The maximum twist angle for gradual en echelon cracks varies according to the
lithology hosting the parent joint. This is seen in a plot of the strike of fringe cracks vs.
the strike of parent joint (Fig. 2.51). In this plot, data for counter clockwise fringe cracks
would plot below a line with a slope of one, whereas data for clockwise fringe cracks
are plotted above this line. By drawing delimiting envelopes parallel to this line, Younes
and Engelder (1999) found the maximum twist angles in coarser beds. For example, en
echelon cracks in shale layers have a maximum twist angle of 4°, whereas those in sand-
stone layers reach 10° for en echelon cracks with a clockwise twist angle.

2.2.9.6
Discussion on Kinking

While Cruikshank et al. (1991) restrict kinking to faulting on existing joints, Younes
and Engelder (1999) consider kinking on joints without slip. The common gradual
curving of joints as they approach earlier joints (e.g., Fig. 3.36b) and curved mud cracks
that maintain continuous plumes along the strike (Weinberger 2001b, and pers. comm.)
imply that curved (gradual) kinking (Fig. 2.20b) is possible. A more difficult issue is
whether straight (abrupt) kinks (Cruikshank et al. 1991) exist on joints, i.e., whether
kinking occurs in continuation with the propagation of the parent joint (in one event)
or after parent joint arrest (in more than one event, Fig. 2.20a). If continuities of plumes
are observed on parent joints and straight kinks, single event processes may be sus-
pected. On the other hand, if discontinuities are observed between the plumes on the
parent joints and kinks, more than one event should be considered. The latter could
be the growth of a kink after a certain time interval following a fracture arrest. A careful
inspection of both the continuous and discontinuous plumes should be exercised on
suspected straight kinks, because a cutting relationship between two distinct joints of
different sets can be mistaken for straight kinking (see Sect. 2.2.10.2 below). It appears
to us that presently, no compelling evidence for straight kinking on joints exists, and
until such evidence is provided, straight kinking by faulting is the only known mecha-
nism. Finally, the important role of plumes in the interpretation of the genesis of kinks
has a certain similarity to their role in the study of pinnate joints (Fig. 2.68d).

2.2.10
On the En Echelon Fringe Terminology

2.2.10.1
The Parent Joint is a Part of the Mirror Plane

We note that the gradual en echelon fringe type of Younes and Engelder (1999) (Fig. 2.49a,b)
resembles the famous diagram of surface markings on a systematic joint by Hodgson
(Fig. 2.36b), where an abrupt transition from the parent joint to the fringe occurs along
the shoulder. Also, in both diagrams (Fig. 2.36b and 2.49a), new plumes start at the
fringes. The gradual en echelon fringe of Younes and Engelder (1999) is also similar to
the discontinuous fractography (Fig. 2.35a,b), which displays an abrupt transition of
158 Chapter 2 · Elements of Fracture Geology

the parent joint to the fringe. The above examples illustrate only part of the confusion
regarding joint terminology (see more in Bahat 1991a, Sect. 2.2).
The main deficiency of the existing descriptive schemes of en echelon fringe ter-
minology is that actually, they have no compelling limitations. A classification with
compelling limitations that would minimize confusions may be achieved by forming
a terminology based on superposing crack descriptions on fractographic features
that correspond to fracture mechanic rules (Sect. 4.8). We suggest that the key to a
comprehensive terminology should be the notion that the parent joint is part of a
mirror plane. This relationship will automatically define all the morphological fea-
tures that reside on the parent joint and on the fringe (Fig. 2.1). This is fairly clear
in granites (Chap. 4), but it is occasionally hindered when joints cutting sedimentary
rocks are considered, because in the latter rocks the parent joint is often a portion
of a mirror plane, which is constrained between layer boundaries and cannot achieve
full growth (Bahat 1998a and present Fig. 2.8). This terminology is a part of a wider
classification that distinguishes between en echelon fringes and hackle fringes (Fig. 2.1,
4.35 and 4.37–4.40).

2.2.10.2
A Fringe of a Parent Joint Can Only Form from “Continuous Material”

The study by Cooke and Underwood (2001) (present Fig. 2.32) seems to justify the re-
evaluation of previous discussions on three cases of en echelon segmentation at dif-
ferent lithological interfaces (Bahat 1997; Younes and Engelder 1999). The first case
concerns fracture in Devonian sediments from the Appalachian Plateau (Younes and
Engelder 1999) that relates to two joints in siltstone (relatively smooth surfaces) alter-
nating with shale beds (rough surfaces) (Fig. 2.50). While the joint surfaces in the silt-
stone layers change orientation along the strike (three sectors in the lower layer and
two sectors in the upper one) the fracture surfaces in the lower shale bed maintain
the same azimuth. We suggest that whereas the sectors of different orientation in the
siltstone represent two distinct joint sets, the fractures cutting the shales represent a
single, distinct joint set, i.e., the fractures in the shales do not form an “abrupt” en ech-
elon fringe to the parent joint in the siltstone. Possibly the set in the shales is Ia and
one of the sets in the siltstone is Ib (Engelder and Geiser 1980, the former set is clock-
wise with respect to the latter). This relationship is repeated in additional outcrops
(e.g., Fig. 2.49d). Quite intriguing is the decrease in spacing of the joints in the shales
as they approach the lower boundary of the siltstone bed. The reduction in spacing
implies an increase in the tensile stress concentration along the boundary between
the shales and siltstone. The multiplied tensile stress resulted in a repeated growth of
(short) joints in the shales that were added between the earlier long joints of an exist-
ing set (Fig. 2.49d and 2.50). Hence, the short joints are not en echelon segments in a
fringe and are unconnected to remote extension normal to the parent joint in the silt-
stone, but rather, they relate to the energy released at the contact between the two rocks
that affected fracture only in the shales.
The second case is about fracture in chalk adjacent to a fracture in a chert bed
(Fig. 2.52a). The interpretation here is the same as for the first case: The short frac-
tures represent a joint set cutting the chalk; they were not formed in a fringe of the
 2.2 · Fractography and Tectonofractography 159

Fig. 2.52. a A sandwich of two chalk layers and a chert bed in the middle from the Mor Formation
near Beer Sheva. En echelon segmentation occurs in the chalk only along contacts with joints cutting
the 15 cm thick chert bed, but not along contacts with unfractured chert nodules (after Bahat 1997).
b Plumes that developed within the limestone layers appear to have continued, forming a fringe of en
echelon segments in the interbedded shales underneath the limestone (after Roberts 1995, Fig. 5d)

parent joint in the chert. Quite possibly the interface contacts in the first and second
cases were not welded enough to act as a “continuous material”, which is a required
condition to form a fringe of a parent joint. In these two cases, separate fracture pro-
cesses took place in the two lithologies.
160 Chapter 2 · Elements of Fracture Geology

The third case concerns a study on jointing in interbedded limestones and shales
that appear to be welded together (Fig. 2.52b). While plumes developed within the lime-
stone layers, a fringe of en echelon segments occurs in the interbedded shales just
underneath the limestone in an apparent continuation from the plume in the lime-
stone. However, it appears that both fracture surface morphologies occur on a verti-
cal veneer of secondary mineralization that covers the two lithologies and acts as a
“continuous material”. Thus, we interpret the sequence under consideration to be: Early
vertical fracture through both lithologies, secondary mineralization, and then, late
fracture along the contact between the veneer and the early vertical fracture.
Our interpretation of the three cases needs to be verified in additional fracture
provinces. Particularly, the first case needs to be carefully examined (we take it as a
speculation that could be wrong). If this interpretation is proven correct, it would limit
or question the validity of the category of abrupt en echelon fringes in the Appala-
chian Plateau.

2.2.11
Summary of Five Categories of Breakdown Styles

The various extension paths from parent joints by tilt and twist joints to kinks and
fringes may be condensed into five categories that reflect different mixed mode load-
ings and propagation rates. Four categories are summarized in Fig. 2.20a–d. Note that
the scheme given in Fig. 2.20c,d concentrates on discontinuous breakdowns that ini-
tiate close to the layer boundary (Fig. 2.35a,b), excluding the continuous one that ini-
tiates between the layer boundaries (Fig. 2.35c,d). The continuous-gradual style may
be considered a variation linked to the gradual, the fourth category, and a full analysis
of joint fringes should concern this style as well.
While studying rare hackle fringes in sedimentary rocks, Bahat and Rabinovitch
(2000) observed angular relationships that differed from the above four categories. The
tilt angle φ produced between the mirror plane and the fringe varied from 0° to 30 ±5°
along the mirror boundary of a natural joint (Fig. 2.30a), φ , of an artificial fracture
that formed by blast varied from 30 ±8° to 50 ±8° (Fig. 2.30b). Thus, the tilt angle φ
appears to be larger on the artificial fracture that was dynamically blasted by high
stresses than on the joint surface that was probably created under moderate stress
conditions. The fractographic features on the mirrors and fringes of both fractures
seem to have been formed discontinuously, suggesting abrupt transitions. Hence, the
fractographic relationships in Fig. 2.30a,b suggest a fifth category. This category con-
cerns transitions from the parent fractures to the fringes that are influenced by mixed
I + II + III modes. The tilt angle φ is induced by mixed I + II modes, and the twist
angle ω (although can hardly be measure in the hackle fringes of Fig. 2.30a,b) is formed
by mixed I + III modes.
The fifth category is rare in sedimentary rocks but seems to be relatively common
in granites. Note that in Fig. 2.30a,b and 2.56a,b, which display semi-circular mirror
planes, there are no distinctions between fringes that tilt in the distal direction and
perpendicular to it, i.e., normal to the larger or smaller radii of the semi-circular mir-
ror planes. Thus, the fifth category that relates to a rim (consisting of quasi-fringes
and quasi-kinks) of secondary cracks that surround the mirror plane is clearly a prod-
 2.2 · Fractography and Tectonofractography 161

Fig. 2.53. Four styles of breakdown by mixed mode I + II + III. Breakdowns occur in the kink along
the distal continuation of the parent joint and in the fringes parallel to layer boundaries in sedimen-
tary rocks. a and b relate to abrupt and gradual tilted kinks, respectively. The segmented kinks are
described by α and ω angular relationships. c and d relate to abrupt and gradual tilted fringes, de-
scribed by φ and ω angular relationships. ω is formed by the edge of the mirror plane (normal to the
main plume) and the segment width, α is formed by the edge of the mirror plane (parallel to the
main plume) and the segment length (or its tangent), and φ is formed by the edge of the mirror plane
(normal to the main plume) and the segment length (or its tangent)

uct manifestation of mixed I + II + III modes. Nevertheless, we generalize and divide

this category in our scheme (Fig. 2.53a–d) into four corresponding groups that would
also fit hypothetical cases in sedimentary rocks. Groups a and b relate to abrupt and
gradual tilted fringes, respectively that form at the distal part of the parent joint; these
groups concern segmented kinks that are described by α and ω angular relationships.
On the other hand, groups c and d relate to abrupt and gradual tilted fringes, respec-
tively that form along one or two layer boundaries; these groups display φ and ω an-
gular relationships. These fractographies are treated in great detail by Bankwitz and
Bankwitz (2004) (see more in Sect. 4.4.4).

2.2.12
Recent Studies from the South Bohemian Pluton

2.2.12.1
Introduction to Joints from Two Granitic Quarries

The bulk of the existing knowledge on jointing is related to sedimentary rocks. Most
of these joints develop under anisotropic constraints imposed by rock discontinuities
such as layer boundaries and fabrics of different origins. In undeformed quasi-iso-
tropic plutonic rocks on the other hand, such discontinuities are rare, and even flow
fabrics mostly do not create loss of cohesion in the rock. Therefore, fracture in the latter
rocks would be expected to propagate in an unrestricted manner, following fracture
162 Chapter 2 · Elements of Fracture Geology

mechanic criteria: Initial elliptical cracks would tend to grow in isotropic materials
under tension into larger circular fractures, because the stress intensity factor is greater
for the portion of the curve that has a larger radius, resulting in the decrease of ellip-
ticity (Irwin 1962; Roy and Saha 1995).
To our knowledge the granites of the South Bohemian Massif in the Czech Repub-
lic exhibit one of the best spectra of joint fractography in the world by showing a great
fracto-graphic diversification. Most of these granites had fractured during the late cool-
ing stages of the pluton (Sect. 4.4.5) and were not deformed after solidification (apart
from later fracture) before or after their uplifts, at least during the Upper Cenozoic.
Therefore, they maintain the original shapes of early fractures from periods before
the uplifts. The unfissured properties of these rocks were beneficial during more than
a hundred years in the stone industry and construction purposes throughout Europe,
leaving behind large quarries, many of which are still active. These quarries display
large fractographic features on joints that cut the granites (Bankwitz and Bankwitz
1984). Therefore these granites are most suitable for studying joint fractography in
plutonic rocks.
Section 2.2.12.2 presents a description of characteristic fractographic features of
the mirror planes and fringes (Fig. 2.1) of ten selected joints from the Borsov granite
quarry in the South Bohemian Massif. An additional fractographic treatment that in-
cludes geological interpretation of these joints is presented in Chap. 4. Accordingly,
the fractographic descriptions are presented in two parts: Photographs are shown in
Chap. 2 (Fig. 2.54–2.56) and their corresponding drawings (Fig. 4.6–4.8) and expla-
nation, supplemented by Tables 4.1 and 4.2 are given in Chap. 4. A study on mirror
planes and fringes from the Mrákotin granite quarry in this massif is presented in
Sect. 4.4.3.1(the deferred treatment of this quarry is done in order to avoid its exces-
sive partition among sections). The Borsov and Mrákotin quarries were chosen from
many others in the massif since they best represent fractographic series between end
members from “quasi-static” to “dynamic” joints (Sect. 4.9.3.3).

2.2.12.2
Fractography in the Borsov Quarry

Fractographic diversity. In their full exposure, the Borsov joints are divided into four
groups:

1. Mirror planes without fringes (J1–J4).

2. Mirrors and simple en echelon fringes (J5–J7).
3. Mirrors and non-uniform/complex en echelon fringes (J8–J9).
4. A mirror and a hackle fringe (J10). They are briefly itemized below, citing Bahat
et al. (2003).

The simplest joint surfaces (J1 and J2) show a superposed single plume on an
array of arrest marks (Fig. 2.54a) and a single radial plume (Fig. 2.54b). More com-
plex textures are the series of superposed concentric arrest marks and a radial plume
that exhibits considerable changes in barb coarseness (width and height) on J3
(Fig. 2.54c), as well as the outcrop that reveals a set of joints (J4) that exhibit plumes
 2.2 · Fractography and Tectonofractography 163

and arrest marks of different styles and intensities (Fig. 2.54d). These four exposures
display incomplete mirrors without fringes (therefore they do not exhibit mirror
boundaries).
There are four additional examples of joints that display both mirrors and fringes:
First, the heterogeneous combination of arrest marks and plumes with the small fringe
on J5 (Fig. 2.55a); second, a series of elliptical arrest marks superposing a multi-tex-
ture plume and a triangular fringe on J6 (Fig. 2.55b), and third, a fully developed cir-
cular en-echelon fringe of a uniform width on J7 (Fig. 2.55c). The fourth example (J8)
consists of a circular mirror boundary with a non-uniform en echelon fringe
(Fig. 2.55d). Still more complex are the heterogeneous and noncircular mirror bound-
aries, which are surrounded by complex fringes on the “trefoil joint” (J9) (Fig. 2.56a).
The tenth joint, J10, displays a large fringe (relative to the exposed mirror) with tran-
sitions from en echelon to hackly morphologies (Fig. 2.56b). Fringe morphologies may
vary very considerably in comparison to the plumes on the mirrors. Whereas fringe
morphologies on J5–J8 (Fig. 2.55a–2.55d, respectively) are generally only slightly stron-
ger than the corresponding plume morphologies, on J9–J10 (Fig. 2.56a–2.56b, respec-
tively) the fringe morphologies are much more pronounced than the corresponding
plumes. Finally, note the coarse, cusp nature of some of the arrest marks that decorate
some of the “slow joints” (Fig. 2.54c, 2.55a, 2.55b), and lack of coarse ripple marks on
the “rapid joints” (Fig. 2.56a–2.56b). A full understanding of this difference requires
further studies (Sect. 2.2.4).

Fig. 2.54. Joints from the Borsov quarry (see corresponding drawings in Fig. 4.6–4.8 for explana-
tion): Mirror planes without fringes. a The “multi-arrest mark joint”, J1. b The “radial plume joint”, J2.
c The “quasi-core joint”, J3. d The “multi-joint exposure”, J4
164 Chapter 2 · Elements of Fracture Geology

Fig. 2.55. Joints from the Borsov quarry (see corresponding drawings in Fig. 4.6–4.8 for explana-
tion): Mirror planes that contain en echelon fringes. a “The divided arrest mark joint”, J5. b The “asym-
metric joint”, J6. c The “uniform fringe joint”, J7. d The “partly erased joint”, J8
 2.2 · Fractography and Tectonofractography 165

Fig. 2.56.
Joints from the Borsov quarry
(see corresponding drawings in
Fig. 4.6–4.8 for explanation):
Mirror planes with diversified
fringes. a The “trefoil joint”, J9.
b The “hackled fringe joint”, J10

Mirror ellipticities. Practically all the joints at Borsov with exposed boundaries show
mirrors that deviate from circularity into various ellipses. Some of these ellipses slightly
rotate, i.e., mostly maintaining the large axes of their ellipses at various acute angles
with respect to the horizontal orientation. Furthermore, often fracture ellipticity in-
creases (the ratio of long to short axis increases) with joint growth (Table 4.2).

Incremental crack growth displayed by arrest marks. Periodicity and incremental

crack growths are common. The arrest marks exhibit two common periodicity styles:

1. Uniform radius of curvature on all arrest marks (J1 in Fig. 2.54a).

2. Concentric growth of arrest marks (J6 in Fig. 2.55b).

Joint J6 displays a new type of incremental fracture of a strict growth of the arrest
marks radii along the inclined symmetric axis at a dip of 8°, without propagation along
the sub-vertical directions.

Central vs. eccentric fracture origins. Whereas certain joints initiate at the focus of
radiating plumes or approximately at the center of the smallest arrest mark (e.g.,
Fig. 2.54a,b and 2.55b), some other joints initiate close to one side of the elliptic arrest
mark (e.g., Fig. 2.54c and 2.55b). It is noticeable that the latter joints that display ec-
centric origins also show small angle plunge of their symmetry axes.
166 Chapter 2 · Elements of Fracture Geology

Variations in plume intensity. The Borsov joints mostly reveal three plume styles
(Table 4.1): First, plumes with a uniform intensity, that is, a monotonous barb coarse-
ness (relief); second, plumes that display alternating increases and decreases in their
intensities along their imaginary symmetry axis; third, plumes that display an increase
in their intensities as they curve away from their symmetry axis.
The first style is represented by J2 (Fig. 2.54b) and J7 (Fig. 2.55a). Whereas the plume
of J2 is relatively coarse, that of J7 is weak and is hardly visible. The multi-joint expo-
sure (Fig. 2.54d) displays adjacent joints decorated by plumes of various intensities.
However, plumes of individual joints generally maintain uniform barb coarseness. The
second style is represented by J3 (Fig. 2.54c), J5 (Fig. 2.55a) and J6 (Fig. 2.55b). On these
joints, plumes change their intensities along their lengths at various locations on the
mirror. Periodic changes in plume intensity are also known from sedimentary rocks
(Fig. 4.36). The large plume of J6 initiates at the edge of the en echelon “triangle” fringe
(Fig. 2.55b) and propagates horizontally towards the left at much lower intensities than
towards the right. A transition of a fringe into a plume on the parent joint is rare, i.e.,
in all other known descriptions the transition is from plume to fringe. The third style
is observed at the lower part of J1 (Fig. 2.54a), on the first and some subsequent ellip-
tical growth stages of J6 (Fig. 2.55b) and on J8 (Fig. 2.55d).

Deviation from orthogonality. Previous studies (e.g., Hodgson 1961a) reported that
plumes and ripple marks superpose each other orthogonally at every intersection on
a given joint, indicating a simultaneous development of the two features. We found a
deviation from the 90° pattern to smaller angles on J6 (see insert to Fig. 4.7b). While
plumes generally form by local mixed mode I / III, in the present case the local mixed
mode was possibly I / II / III.

2.2.13
Oblique Extension Joints and Longitudinal Splits (or Splittings)

2.2.13.1
Introduction

Conjugate hybrid joint sets. Occurrences of “conjugate hybrid joint sets” that form at
small angles (≤45°) in the field have been considered by many investigators (e.g.,
Muehlberger 1961; Price 1966; Stearns 1968; Ladeira and Price 1981; Etheridge 1983).
The same observation was termed differently by various authors, e.g., “oblique exten-
sion fracture” (Dennis 1972), “conjugate hybrid joints” (Hancock and Al-Kadhi 1978),
“hybrid extension/shear fractures” (Ladeira and Price 1981),“transitional tensile be-
havior” (Suppe 1985, p. 154), hybrid joints (Bahat 1989), and transitional-tensile frac-
ture (Engelder 1999). According to Hancock (1985), systematic joints can form an
angular continuum from a coaxial relationship with the maximum principal com-
pression (0°) to a maximum dihedral angle of c. 45°. Price (1977) and Hancock (1985)
interpreted the “hybrid-shear fractures” as belonging to a failure class transitional
between extension fractures and Navier-Coulomb shear fractures showing a conju-
gate set with a dihedral angle greater than 45°. However, explaining the structure of
“conjugate hybrid joints” has been an outstanding problem in structural geology. For
instance, Engelder (1999) suggests that based on assessments from laboratory (e.g.,
 2.2 · Fractography and Tectonofractography 167

Brace’s 1964 tests on a dog-bone-shaped sample), field and petrofabrics (e.g., a frac-
ture system from Price and Cosgrove 1990) and theoretical observations (e.g., Reches
and Lockner 1994), the transitional-tensile fracture propagation is unlikely to occur
in homogeneous, isotropic rock, and that it is not explained by a parabolic Coulomb-
Mohr failure envelope. We examine this suggestion below.
The outstanding question is whether oblique extensile fractures (joints) exist, i.e.,
whether a parent fracture (a vein) showing shear offset initiated as such within an iso-
tropic, homogeneous rock, and then propagated in its own plane while that plane was
subjected to a shear couple as well as an effective crack-normal tension (Engelder 1999).
Several observations suggest that under certain conditions such results are possible.
The occurrence of straight dilatational en echelon cracks at angular relations to the
remote compression (Pollard et al. 1982) implies that such a simulation is possible.
Willemse and Pollard (1998) observed this in nature and provided an explanation why
straight wing cracks (Sect. 2.4.2) are relatively common (Fig. 2.57a). Perhaps another
relevant case is bifurcation into two tensile branches that form small angles, bisected
by the imaginary continuation of the parent fracture. This is known to occur in vari-
ous scales, from small size crack bifurcation in glass bottles (Fig. 2.30c) to large ones
along faults in California (e.g., faults that branch from bifurcation 6 in Fig. 2.57b are
about 500 m long). We are aware that the above three examples can be “excused” on
different grounds (mixed mode and pore pressure, the straight wing crack is short,
and dynamic fracture, respectively), but they cannot be ignored.

Fig. 2.57. a Wing cracks near tip of a fracture that experienced mixed mode loading i.e., both sliding
and opening. Calcite cement (black) now fills the flaw and wing crack. Note the sharp angles at which
the (thick) wing cracks deviate from the parent fracture, the curved shape of the wing cracks, as well
as the (thin) straight one. Matching up of both surfaces of parent fracture allows determination of
the displacement discontinuity as shown. Example from Lower Jurassic limestones between East
Quantoxhead and Lilstock, Somerset, England (after Willemse and Pollard 1998). b Eight fault termi-
nations by branching (1–8) along segments of the Coyote Creek fault in south California. Branchings 3,
5, 6, 7 and 8 show approximate symmetries between the two branches (forming the angle 2α, that are
initially straight, reflecting local tensile stress. Branchings 1, 2 and 4 show deviations from symme-
try, implying significant mode II / mode I (modified after Clark 1972 by Bahat 1982)
168 Chapter 2 · Elements of Fracture Geology

Let us examine another set of observations. These are related to en echelon cracks,
which are equivalent to tension gashes and stepping joint segments at parent fracture
terminations (Petit and Barquins 1987, 1988; Petit 1988; Price and Cosgrove 1990; Cruiks-
hank et al. 1991) (Fig. 2.58a–d). Petit and Barquins (1987, 1988) induced a wing crack w
(that they term branch fracture), and a shear zone of en echelon cracks e in PMMA and in
sandstone under uniaxial compression, when applied on inclined pre-cuts (Fig. 2.58a,b).
The wing crack that showed an opening resulted from local mode I loading below peak
stress, and the en echelon cracks were induced by local mode II loading under maxi-
mal stress conditions. The sandstone revealed cataclastic deformation along the en ech-
elon cracks in the shear zone. The wing crack strongly deviated from the direction of
the pre-cut, whereas the shear zone was aligned approximately in continuation with it.
The various structures (Fig. 2.58a–d) show that the en echelon cracks do not main-
tain a uniform orientation; instead, they systematically rotate counter clockwise with
their distance from the tip of the parent fracture. While in the field where two adja-
cent parent fractures occur, this rotation could be viewed as a consequence of interac-
tion by the tips of the two fractures (Cruikshank et al. 1991) (Fig. 2.58d), i.e., caused
by the interaction of two stress fields. However, this en echelon rotation is unrelated
to such an interaction in an experiment where only one parent fracture is simulated
(Fig. 2.58a,b) and also in the outcrop shown by Fig. 2.58c (assuming an isolated struc-
ture), questioning the validity of the interpretation of interaction by the two fractures.
There is a systematic rotation of the en echelon cracks towards parallelism to the
maximum principal compression (Fig. 2.58a–c), and towards co-axiality with the par-
ent fracture (Fig. 2.58d). Thus, in a given remote stress field, a local predominant
mode II close to the tip of the parent fracture gradually changes to a local predomi-
nant mode I loading with the distance from the tip reaching the maximum rotation at
a particular distance from the tip, in a fracture termination zone. This is measurable
in Fig. 2.58a–d (see more in Sect. 6.6). Compared with the systems shown in Fig. 2.58a–
d, no significant rotation is discerned in the en echelon sets that occur beyond the tips
of joints along which no strain has occurred (Bahat 1991a, Fig. 4.22). Thus, w and e
formed by the same remote stress field (Fig. 2.58a,b) and a reorientation of the re-
mote stress field (Engelder 1999) was not necessary for their creation. They formed
by a stable process under conditions of increasing stresses (e.g., Brace and Bombolakis
1963; Jayatilaka and Trustrum 1978). Some of the en echelon tensile cracks in the ar-
rays formed oblique to the maximum principal compression in a transient rotation
(Fig. 2.58a–c). However, the last en echelon crack in the Garden Area (horizontal ar-
row in Fig. 2.58d) was formed at the end of the transient rotation, possibly suggesting
▲

Fig. 2.58. Rotation of en echelon cracks along their arrays, in different systems. a Induced wing crack w
(branch fracture), a shear zone of en echelon cracks e and uf in PMMA under uniaxial compression, when
applied on an inclined pre-cut p (from Petit and Barquins 1987, 1988). b En echelon superficial tensile
fracture formed from a pre-cut under biaxial loading in a plexiglas plate (after Petit 1988). c Rotation of
en echelon tension gashes at the tip of a tension gash (from Price and Cosgrove 1990). d Stepping joint
segments (long) and echelon cracks (short) in the Garden Area, Arches National Park, USA. Two ways
that joint segments terminate in the vicinity of neighboring joints. Tips of a segment may simply veer
toward a neighbor. Or, beyond the tips of a segment may be “array” of echelon cracks. The segments
step right and are twisted a few degrees counter clockwise relative to the trend of the zoned joint, whereas
the echelon cracks associated with the segments step left and are twisted a few degrees relative to the
trend of the segment. Note that in the sheared stress field shown here, the en echelon cracks rotate
counter clockwise along their arrays (modified after Cruikshank et al. 1991; see text for explanation)
2.2 · Fractography and Tectonofractography 169
170 Chapter 2 · Elements of Fracture Geology

Fig. 2.59.
a Longitudinal splitting along
wiggly curves (modified
after Twiss and Moores 1992,
Fig. 9.2B). b A model of a se-
quential fracture in a conju-
gate fault set (from Park 1983,
Fig. 9.4)

a general condition for forming joints oblique to the maximum principal compres-
sion. Hence, oblique extensile fracture is possible.
The problem of oblique extensile joints may be linked with another outstanding
enigma: The true nature of the longitudinal splitting. Some authors (e.g., Engelder et al.
1993, p. 46) contemplate the idea that longitudinal splitting is the mechanism leading
the formation of regional jointing. At the base of this idea is the rationale that uniaxial
compression is coaxial with longitudinal splitting (e.g., Griggs and Handin 1960;
Holzhausen and Johnson 1979) and analogically, the maximum horizontal principal
stress is coaxial with strikes of individual systematic joints (Engelder and Geiser 1980).
It is important to take into account that experimental splitting has been observed by many
investigators to be wiggly rather than sharp straight fractures (e.g., Jaeger and Cook 1979,
Fig. 4.5.1; Twiss and Moores 1992, Fig. 9.2B; Cosgrove 1995) (Fig. 2.59a). What is the mean-
ing of these deviations from straight fractures? Experimental results obtained from dif-
ferent materials under diverse fracture conditions as well as observations made in geo-
logical outcrops are presented in the next sections and compared, seeking to improve the
current understanding how oblique extensile joint sets and longitudinal splitting are
formed. These sections do not deal with situations of changes in magnitude of initial di-
hedral angles during fracture growth (e.g., Kalthoff 1973; Ramsay and Lisle 2000, p. 968).

Conjugate vs. contemporaneous, simultaneous and sequential fracture. There is a

need to make a few term distinctions. A convenient term in structural geology is con-
temporaneous. This term is applied when a field relationship reveals that in a given
research area, in some locations a joint of set A repeatedly arrests at a joint of set B,
suggesting that set A is younger than set B, and in other locations the sequence is
reversed. This is taken to indicate with reasonable confidence that the two sets were
formed contemporaneously during the same geological period, which could be “short”
or “long”. Thus, reverse sequences are implied by the term contemporaneous. More
challenging is the term simultaneous. It is quite possible to claim that a series of
fractures associated with a certain earthquake was formed simultaneously. It is more
difficult to prove that a series of joints of a particular (systematic) set was formed
together. Probably joints from some sets do not form simultaneously (Bahat 1991a,
Fig. 4.8). It is far more difficult to prove that joints of “two conjugate sets” formed
 2.2 · Fractography and Tectonofractography 171

simultaneously. Quite possibly, even two sets that exhibit the closest to ideal field rela-
tionships that would justify the term conjugate would be a product of a sequential
process. In a way this perplexity resembles uncertainties that are related to conjugate
fault sets: Do they really form simultaneously or possibly sequentially (Fig. 2.59b)?
We apply here the issue of simultaneously vs. sequentially to two topics, oblique
extensile joints and longitudinal splits, and to the problem of two contradicting inter-
pretations regarding the mode of formation of sets 309° and 344° in the Beer Sheva
syncline (Sect. 3.4.8 and 6.3.9.1).

2.2.13.2
Fracture Results from Different Sources

This section provides observations from geological outcrops, which are supplemented
by experimental results that are presented in Sect. 2.2.3.

Regional joint patterns from chalks in East England. Goodwin (1995) investigated
the natural fracture behavior of chalks in south England. He found two sets of re-
gional distribution, striking NW and NE. Both joint sets comprise steep conjugate
(ignoring the sequential issue) and vertical joints, everywhere consistent with bed-
ding (Fig. 2.60). In the southern investigated area (Norfolk), traces of the NW set on
the vertical NE-NW section typically show conjugate steeply inclined joints that form
dihedral angles between 25° and 45°. Vertical joints occur less commonly. In the north-
ern area (south Humberside/north Lincolnshire) by contrast, vertical NE-NW sec-
tions across the joint sets show mainly vertical extensional joints with few steeply
inclined planes occurring, i.e., the dihedral angle of the Humberside joints <20°. The
joints of Norfolk are interpreted by Goodwin (1995) as the conjugate hybrid joints of
Hancock (1985, see above). This would include “hybrid joints” that allow for loading
of both, shears along, and extension across the plane to take place.
Goodwin (1995) considers two alternative explanations for the differences in size of
the dihedral angle in the southern and northern areas. First, the dominance of exten-
sional joints in the north is correlated with “hard” chalk of lower porosity and greater
rock strength (dry density 1.78–2.27 g cm–3), compared with the conjugate characteris-
tics in the south that are correlated with “soft” chalk of high porosity and lower rock strength
(dry density 1.40–1.72 g cm–3). Second, the above differences reflect a possible northward
decrease in differential stress across eastern England. This dilemma was investigated
in our laboratory, and the results are in favor of the first explanation (Sect. 6.2.2).

Regional joint patterns from chalks around Beer Sheva, South Israel. Three types of
surface markings occur on the single-layer, cross-fold joint surfaces in the Lower Eocene
chalks from the Beer Sheva and Shephela synclines in south Israel:

1. Horizontal axial plumes.

2. Circular rib markings.
3. Transitional variations between the two.

The three types may occur together in the same outcrop (Bahat 1991a, Fig. 3.28).
About half (56%) of the 225 FSM examined on joints in four outcrops around Beer
 172
Chapter 2 · Elements of Fracture Geology

Fig. 2.60. Natural fracture behavior of chalks in south England, see text for explanation (after Goodwin 1995)
 2.2 · Fractography and Tectonofractography 173

Sheva bear circular markings, and these occur on joints of a set of average strike 328°
(N 32° W). Sixty five samples (29%) are axial plumes, and these occur on joints be-
longing to sets of average 309° and 344° strike. Transitional markings make up 12%,
and about 3% are other combinations. Joint sets with circular markings and joint sets
with plumes occur in separate rock layers, even in the same outcrop.
It was interpreted (Bahat 1987a, 1991a) that possibly the circular markings on the
328° joints have been formed along the direction of the horizontal principal stress, and
the axial plumes on sets 309° and 344° were formed on joints at low angles to that princi-
pal stress in a conjugate pattern (Stearns 1968). The circular markings developed when
the minimum principal stress σ3 was horizontal and normal to the joint surface, and the
maximum horizontal σ1 equaled the vertical principal stress σ2, i.e., exerting “pure exten-
sion” on the joint. The axial plumes represent the approximate conditions with respect to σ3
like for the circular markings, when the horizontal principal stress different from the
vertical, σ1 ≠ σ2, exerting some differential stresses on the joints and resulting in super-
posing a slight shear stress on the joint. The three sets were formed in the Lower Eocene
chalks during the burial stage while still immersed in water. Therefore, fracture occurred
while the chalk was weak, possibly having the tensile strength of about 0.5 MPa (Bahat
1991a, p. 243). Hence, a dihedral angle of 35° was formed by the strikes of the 309° and
344° sets that resulted from the regional maximum compression in the 328° direction, and
the plumes on the surfaces of the former two sets may reflect some mixed mode loading.

2.2.13.3
The Behavior of the Dihedral Angles from Different Sources

Plots of the dihedral angles. The dihedral angles obtained from the different sources
presented above are plotted in the range 2α = 0–65° vs. the compression strength and
dry density of the material (Fig. 2.61a and b, respectively, see Fig. 2.24c for the angu-
lar geometry). The decrease in 2α with the increase in compressive strength is cor-
relative for various materials. Longitudinal split, probably of an extensile nature, oc-
curs in the range 2α ~ 0–20° in “strong” glass ceramic, glass and rocks of medium
strength. We speculate that in weak rocks, slow (geological rate) fracture may result in
extensile fractures that attain 2α values up to 35° by an unclear mechanism. The value
of the dihedral angle may stretch up to 65° in “weak” rocks, involving the magnifica-
tion of the tendency of fracture by mixed modes I + II. There seems to be a general
trend of decrease in the dihedral angle with the increase of the dry density of chalks
in east England (Fig. 2.61b), resembling the quantitative curve (Fig. 2.61a). However,
due to the limited number of points on the latter curve, its value is only qualitative.

Fig. 2.61. Dihedral angle 2α plot vs. material properties. a Compression strength; b dry density
174 Chapter 2 · Elements of Fracture Geology

The reliability of the results. The comparison of the various dihedral angles obtained
from different sources (Table 2.2, Fig. 2.61) needs to be qualified, because the com-
parison relies on different degrees of certainties regarding the various results. Three
reliable results are the ones obtained experimentally on the chalk, glass ceramic and
glass (Sect. 2.2.3.4, 2.2.3.7 and 2.2.3.11, respectively), because direct fractographic ex-
aminations of their splits clearly suggest that these fractures are essentially extensile.
These results suggest that extensile longitudinal split maintains 0° deviation from the
compression direction in a strong material (glass ceramic) but may deviate up to
α = 8 ±2° from this direction in materials of reduced-strength. Included in the latter
category are the intensely dried chalk from the Beer Sheva syncline and the glass
samples to which pre-cuts were introduced. The results on longitudinal split obtained
in fractured Chelmsford granite (Sect. 2.2.3.6) support this suggestion.

Table 2.2. Dependence of the fracture dihedral angle on material/rock properties

 2.2 · Fractography and Tectonofractography 175

It is not clear how reliable the other results are that are not supported by direct
fractographic observations. One is convinced by Goodwin’s (1995) results that the
stronger chalks in east England have a greater tendency to fracture in extension than
the weak ones. The weak chalks (from Norfolk) on the other hand, provide a large range
of dihedral angles (25–45°). Fracture at Norfolk probably went far beyond extension,
but how dominant the superposed shear mode was in the failure of these chalks is
unknown. Muraoka and Kamata (1983) observed that the shear angle ranges between
conjugate faults in the Kusu Group were clearly lower in the generally more brittle silt-
stone (17–45°) than in the sandstone of lower competence (45–69°). Thus, the general
trend of the increase of the size of the dihedral angle with the material weakness (in-
competence) is also shown for faults.

2.2.13.4
Conjugate Extension Joints and Inclined Longitudinal Splits that Form by Mode I

The results that are summarized in Table 2.2 and ranked in Table 2.3 suggest a new clas-
sification that distinguishes four groups. The extension by mode I at 2α = 0° for a strong
material (item 10 in Table 2.2) is obvious. The interpretation of the results obtained for
the materials of reduced strength (items 7 and 9 in Table 2.2, second group in Table 2.3)
is that joints may form exclusively by mode I loading, even at small dihedral angles
2α ≤ 20°. This interpretation is supported by the observation made on the tendency of
tensile en echelon cracks to rotate, so that some of the cracks form oblique to the maxi-
mum principal compression under mixed mode I + II loading (Fig. 2.58a–d). With cer-
tain qualifications, this interpretation is possibly sustained by fracture in granite (item 8
in Table 2.2). The suggestion given here lacks a statistical support and stays tentative
until additional results will confirm it. In the meantime 2α ~ 20° will mark the bound-
ary between groups 2 and 3 (Table 2.3). The lack of fractographic criteria to distinguish
between these two groups remains a challenge for future research. Finally, the compos-
ite failure envelope formally distinguishes between groups 3 and 4 (Fig. 2.26d, B and C,
respectively). Additional studies may possibly identify fractographic criteria to distin-
guish between the latter two groups. The angular boundaries among the four groups
are not expected to be “critical angles”; they are likely to be transitional ranges that
depend on rock properties (competence/strength) and the rate of applied strain. But
the existence of four groups possibly suggests some “spectrum of discontinuities” of
dihedral angles, rather than a “continuous spectrum” (Hancock 1986).

Table 2.3. Four fracture groups with varying dihedral angles

176 Chapter 2 · Elements of Fracture Geology

2.2.13.5
The Mechanisms of Plume and Striae Formation on Various Fractured Surfaces

We can now extend the discussion on plume straightness and coarseness (Sect. 2.2.4.2).
Striae occur on fractured surfaces from groups 1, 2 and 3 that form under unlike
conditions and mechanisms (Tables 2.2 and 2.3). On the extension joints from
group 1 (fracture in glass ceramic) the striae are essentially straight, consisting of sin-
gle fractures, and close to their ends they curve towards one of the sample bound-
aries and slightly divide into secondary small microcracks (Fig. 2.19a). On the in-
dividual surfaces of the fractures from group 2, the micro-plumes are straight
(Fig. 2.15a). Straight plumes indicate straight fracture propagation predominantly
by mode I that occurred on the relatively small fracture surfaces in the chalk. Almost
no plume curving and division into smaller microcracks occurred, because fracture
propagation took place on surfaces that were isolated from influences of adjacent
sample boundaries. On the other hand, plumes cutting joints from group 3 were
strongly affected by layer boundaries, and their division into secondary branches
and barbs is quite pronounced (Fig. 2.22a); curving of plumes is more noticeable in
thick rock layers.
The plumes in the Lower Eocene chalks (Fig. 2.45) are considerably coarser
than the plumes in the Middle Eocene chalks (Fig. 2.35a,c, 2.36a and 2.38). Burial sin-
gle-layer jointing occurred in the Lower Eocene chalks under increasing overbur-
den stresses, whereas uplift single-layer jointing took place under increasing ten-
sile stresses in the Middle Eocene chalks (Bahat 1999a). These distinct conditions
quite possibly resulted in different fracture velocities, which were greater in the
chalks of the Middle Eocene than in the chalks of the Lower Eocene for several
reasons. First, it is quite likely that fracture velocity will be greater close to the ground
surface under tensile conditions, compared to conditions of increasing compression
with depth. Second, experiments by Müller and Dahm (2000) (Fig. 2.22) show the
correlation of increase in plume coarseness with the reduction in fracture velocity.
Third, we use a tentative assumption that still requires confirmation (Sect. 6.3.9.1)
according to which the plumes on the conjugate sets 309° and 344° from group 3
were formed under mixed modes I + II, which were associated with the remote
compression from 328° compared with the joints cutting the Middle Eocene chalks
that do not show a conjugate relationship and appear to have responded to rotating
co-axial compression. The possible implication is that while shear modes were
minimal on the latter joints, more significant shears were exerted on the conjugate
sets 309° and 344°. Possibly, the overburden constraints imposed on the Lower Eocene
chalks caused some of mode II to locally transform into mode III (Sect. 2.2.3.8),
leading to plume coarsening.
Even more intriguing is the fact that no fringes of en echelon segmentation occur
on joints cutting Lower Eocene chalks, which are marked by coarse plumes, while
fringes of the type shown in Fig. 2.35a are abundant on joints that cut the Middle
Eocene chalks, which are marked by delicate plumes. This shows that coarse plumes
are not always more likely to transform to en echelon segments than delicate ones. This
“paradox” supports the suggestion that a time interval may have occurred between
the fracture of the parent joint and its fringe of the type shown in Fig. 2.20c, and this
could involve significant changes in fracture conditions.
 2.3 · On Uplift, Post-Uplift, Neotectonic and Surface Joints 177

2.2.13.6
Conclusion

1. Our interpretation of the systematic rotation of the en echelon cracks towards par-
allelism to σ1 (Fig. 2.58a–d) is that in a given remote shear stress field, a local pre-
dominance of mode II gradually changes to a local predominant mode I loading.
In this setting, a reorientation of the remote stress field is not necessary for creat-
ing extensile cracks (mode I) oblique to σ1 in certain locations. Hence, extensile
cracks can form in inclined orientation with respect to σ1.
2. Based on experimental evidence (Table 2.2), extensile longitudinal splitting can be
extended from a dihedral angle of 2α = 0° in “strong” materials to 2α ~ 20° in ma-
terials of “medium strength”, supporting the interpretation in conclusion 1.
3. Field evidence possibly suggests that in “weak” materials the dihedral angle of con-
jugate hybrid joints (mixed modes I + II) may be extended to about 35°.
4. The wiggly longitudinal splits that have been observed by various investigators
represent composite fractures consisting of combined extensile and shear growths
that interacted with each other.

2.3
On Uplift, Post-Uplift, Neotectonic and Surface Joints
Joints in sedimentary rocks are genetically divided into four groups: Burial, syntectonic,
uplift and post-uplift (Bahat 1991a; Engelder et al. 1993; Ghosh 1993; Bankwitz and
Bankwitz 1994; Bankwitz and Bankwitz 1997). In this section we refer again to the differ-
ences between uplift and post-uplift joint groups and further elaborate on the latter.
We briefly discuss the diverse characteristics of “neotectonic joints”, “exfoliation joints”
(that are distinct from “sheet joints”),“cross joints” and introduce the “surface joints” group.

2.3.1
Neotectonic Joints

2.3.1.1
General

An inventory of neotectonic joint attributes was compiled by Hancock and Engelder

(1989) in various terrains containing flat-lying sedimentary sequences in the platform
covers of cratons from the Appalachian Plateau, southeast England-northeast France,
the Arabian platform, and the Ebro basin in Spain. Systematic neotectonic joints are
the youngest tectonic structures in all of these study areas. They are, however, older
than superficial non-systematic cross-joints of non-tectonic origin that abut and link
them. The next section is a citation from Hancock and Engelder (1989).

2.3.1.2
Fracture Styles of Neotectonic Joint Systems

The geometry of the systematic neotectonic joint systems is characterized by simplic-

ity, irrespective whether they cut previously intact or already fractured rocks
178 Chapter 2 · Elements of Fracture Geology

Fig. 2.62. Characteristic neotectonic joint systems. a Single set of systematic vertical extension joints
linked by non-systematic cross-joints. b A spectrum of systematic joints comprising vertical exten-
sion fractures and steep conjugate fractures enclosing a range of dihedral angles of less than 45°.
Two steep fracture directions are expressed by arrays of en echelon vertical joints. Non-systematic
joints link systematic joint. c A spectrum of systematic joints comprising vertical extension fractures
and vertical conjugate fractures (in plan) enclosing a range of dihedral angles of less than 45°. Non-
systematic joints link systematic joints; σ1, maximum effective principal stress; σ3, minimum effec-
tive principal stress. d Neotectonic joints cutting horizontal limestones. Line drawing from a photo-
graph, looking down on the double-fan joint spectrum. The symmetry axis of the spectrum trends
sub-parallel to the mean strike of all neotectonic joints in the investigated area (after Hancock and
Engelder1989)

(Fig. 2.62a). The commonest assemblage comprises vertical joints that are either par-
allel to each other or dispersed less than 10° about the mean orientation of the set
(Fig. 2.62b). Where the vertical set occurs in conjunction with other neotectonic joints,
it bisects the acute angle between conjugate sets (Fig. 2.62c). In all the above mentioned
four case-study terrains, a less abundant, but nevertheless typical neotectonic joint
pattern comprises steep conjugate joints striking parallel to a neighboring single ver-
tical set (Fig. 2.62b). Steep conjugate joints generally enclose dihedral angles (2α ) in
the range of 10–45°. The latter joint sets appear in some upper but muddier parts of
the Devonian succession of western New York State, whereas in the central part of the
state, lower silt and shale units are cut by a single set of joints.
Some conjugate joints strike parallel to locally developed late Cenozoic topogra-
phy. They occur in western New York and postdate systematic cross-fold joint sets
(Engelder 1989). Those joints with a dip direction pointing downslope become more
prominent in the near surface. This correlation of dip with Cenozoic topography is
the strongest evidence supporting a neotectonic age for the east-northeast joints of
the Appalachian Plateau.
 2.3 · On Uplift, Post-Uplift, Neotectonic and Surface Joints 179

At a few stations in the Arabian platform and Ebro basin, well-ordered conjugate steep
or vertical sets are replaced by a continuum of joint orientations with an angular disper-
sion of as much as 45° about a symmetry axis that is either vertical or trends parallel to
the mean strike of a nearby single joint set (Fig. 2.62d). Hence, Hancock and Engelder (1989)
distinguish several fracture styles in the systematic neotectonic joint systems: Vertical single
and conjugate sets, steep conjugate sets, conjugate joints which strike parallel to locally
developed late Cenozoic topography and joints that their dip orientation is characterized
by angular dispersion. In addition to the systematic neotectonic joint systems there are
non-systematic fractures, termed “cross joints” (distinct from systematic cross-fold joints)
that extend across intervals between earlier systematic joints (Fig. 3.36d) and may be cat-
egorized as products of the neotectonic stress field (Engelder and Gross 1993).

2.3.1.3
Neotectonic, Uplift and Post Uplift Joint Systems

According to Hancock and Engelder (1989), burial followed by unloading as a result of

denudation and lateral relief consequent of uplift are prerequisites for the formation
of shallow neotectonic joints. A major conclusion in their paper is that shallow (at depths
of less than 1 km) neotectonic joint systems are the most recent joint system. There-
fore, they are of potential value for tracking the orientation of the contemporary stress
field even in structurally complex terrains, although there may be a slight misalignment
between joint strike and the direction of the maximum horizontal stress. Some regional
joints, interpreted as neotectonic, strike parallel to or approximately parallel to the di-
rection of contemporary horizontal maximum compression known from in situ stress
measurements or fault plane solutions of earthquakes (Hancock and Engelder 1989;
Gross and Engelder 1991). This provides an important motivation for studying neotectonic
joints. We argue below that the application of these useful relationships needs to take
into consideration a sub-grouping within the broader group of neotectonic joints.
It appears that the “neotectonic joints” (Hancock and Engelder 1989) embrace two
distinct categories by Bahat (1988a, 1991a) of shallow fractures (depth down to about
0.5 km) without differentiating between them: The “uplift joints” and the “post uplift
joints”. The attributes that are common to both, uplift and neotectonic joint systems
(in addition to shallow depths) include multilayer and large size characteristics (i.e.,
larger spacing and apertures than in neighbor joint sets of older generations), signifi-
cant misalignments between joint strikes, and their occurrence in repeated failure events.
While many attributes are common to both neotectonic joints and uplift joints, post
uplift joints seem to be analogous to the conjugate joints that strike parallel to locally
developed late Cenozoic topography described by Hancock and Engelder (1989). The
latter two are the most recent joint systems. On the other hand, “uplift joints” need
not be the most recent joint system. Uplift joints may populate the same outcrop that
exhibits “post uplift joints” of a younger generation, and they entirely differ in their
strikes (e.g., Bahat 1991a, outcrop 3 in Fig. 5.29). Strike scattering of uplift joints (Bahat
1991a, Fig. 5.28) suggests that while some uplift joints may reflect the contemporary
(remote) stress field, others may not. The post uplift joints are geo-morphological
products of local stress fields (Bahat et al. 1999) and some of them do not reflect the
contemporary stress field. Hence, when neotectonic joints are characterized, their af-
filiation to either the uplift or post-uplift groups should be considered.
180 Chapter 2 · Elements of Fracture Geology

2.3.2
Exfoliation Joints and Sheet Joints

Exfoliation fractures (e.g., Sect. 3.3 and 4.5.3) and sheet fractures (e.g., Twidale et al.
1996) reflect fracture propagation in principal planes normal to the minimum princi-
pal stress and form approximately by the same mechanisms. However, they seem to
genetically and morphologically represent different categories as will be explained be-
low. Gilbert (1904) believed that sheeting is older than the topography. Various authors
(e.g., Dale 1923 and Holzhausen 1989) made comments on “generations” of sheet frac-
tures (e.g., preglacial and postglacial). Similarly, Holman (1976) observed hairline cracks,
which are evident only where weathering has etched them into relief that commonly
subdivide sheets into two to four secondary slabs. Hence, the term sheeting should be
applied to horizontal and sub-horizontal fractures that maintain approximately uni-
form spacing with their neighbor (lower and upper) sheets, but generally increase their
spacing with depth. Sheeting does not have to develop parallel to and close to a previ-
ous fracture, although sheet fractures often occur in sets. Commensurate with the above,
Holzhausen (1989) made the observation that sheet fractures must be products of states
of stress that are approximately uniform over distances of at least tens to hundreds of
meters in order to explain the regularity and size of the fractures.
Exfoliation joints, on the other hand, are generated by fracture parallel to existing
surfaces in all attitudes. They generally develop at short distances from the exposed
surface (tens of centimeters), in small spacing, and produce rock slices of both uni-
form thickness and lenticular shape. In a way, sheeting responds to unloading (before
the topography sets in), whereas exfoliation results from loading that takes place fol-
lowing the establishment of the topography. Hence, sheets and exfoliation fractures
correspond respectively to uplift and post-uplift fractures. We refrain here from fur-
ther addressing sheet joints and refer the reader to the literature on this subject (e.g.,
Johnson 1970). Concentric exfoliation cracks in basalts belong to a different fracture
category, which is not treated in this book.

2.3.3
Surface Joints

We introduce a new category of fractures that we term “surface joints”. These joints
mostly occur in the upper parts of rock outcrops. They may vary in length from sev-
eral centimeters (Fig. 2.63a) to several meters and can appear in various styles, from
straight shapes to twisted forms that transform into non-systematic anastomotic net-
works (Fig. 2.63b). Occasionally, they occur in chalks of Eocene age in the Beer Sheva
region, Israel and are filled with coarse sparitic gypsum. These veins are connected
with the loess-soil cover of the Upper Pleistocene, which overlay the chalks, possibly
concentrating by roots of sulfate tolerant trees and depositing from groundwater that
became over-saturated with CaSO4 (Issar et al. 1988). Particularly characteristic are
the sub-horizontal surface joints that occur in “transitional layers” that underlay the
upper, eroded soil layer and overlay an uneroded jointed chalk layer (Fig. 2.63a). These
surface joints possibly formed perpendicular to tensile stress that was induced by
overburden reduction due to the density decrease of the overlying soil, which formed
from the eroded chalk (see more in Sect. 6.8).
 2.3 · On Uplift, Post-Uplift, Neotectonic and Surface Joints 181

Fig. 2.63. Surface joints. a A section that consists of characteristic fracture elements that occur in
association with a “transitional layer” of surface joints (bounded by the two upper arrows) in Lower
Eocene chalks around Beer Sheva, Israel. The thickness of chert bed is 10 cm. b The “transitional
layer” bounded by the two arrows, that underlies the upper, eroded soil layer and overlay an uneroded
jointed chalk layer (after Issar et al. 1988)
182 Chapter 2 · Elements of Fracture Geology

Surface joints often fill spaces between previous joints and mimic the older joints.
Their identification is done by comparing the new, smaller joints that lose some of
their straightness and spacing to the larger and straighter, earlier joints that form
greater spacing in lower layers where the surface joints do not occur. When they cut
rectangular blocks that were formed by orthogonal joints, they further subdivide them
to into smaller blocks. Thus, surface joints can be distinguished by their relative ir-
regularities, reduced spacing, non-uniformity in spacing, exaggerated apertures and
by their containment of secondary mineralization, erosive rock and soil. These frac-
tures may concentrate along trenches and canals. Although they are essentially non-
systematic, they may occasionally appear systematic.
Surface joints have been mistaken for burial joints by some environmental special-
ists, and this justifies their distinct characterization. The recognition of the impor-
tance of surface joints may be critical in planning and evaluation engineering and
environmental projects. They are not only a subject of the past; these fractures may
develop in the future as well. Joint intensity (fracture area per rock volume) is a basic
parameter in geological engineering. The value of this parameter may change substan-
tially between two locations that are only a few meters laterally apart in the same rock
layer and differing in depths from the ground surface, along an eroded slope. The en-
gineer may like to consider the minimum depth that a survey trench should reach below
the depth of the “transitional layer” for obtaining “clean” information. The hydrolo-
gist may be misled by the attitudes and spacing of joints in this zone in predicting the
flow properties of water. Finally, surface joints differ from the sub-surface, non-sys-
tematic joints that form in association with rockburst in mines (e.g., Jaeger and Cook
1979, p. 505, see also present Chap. 5) and there should not be difficulties in distin-
guishing between these two fracture groups.

2.4
Primary and Secondary Fractures
2.4.1
Introduction

This section is about secondary fractures that form adjacent to primary fractures when
the latter are activated or reactivated by tensile or shear stresses. The geometric rela-
tionship of tensile secondary fractures in relation to a primary fracture on a principal
plane is relatively simple (the primary fracture is given for simplicity as 2r B in
Fig. 2.30c). More complicated is the system of a primary fracture in the shape of a
parent joint or a fault in the field, or a pre-cut in the laboratory (Sect. 2.2.3.11) under
remote compression at a small angle α to the primary, causing local shear and tensile
stresses that result in shear and tensile fracture (Fig. 2.64). In the latter diagram, we
show various tensile fracture types, omitting a certain type of joints from a fault ter-
mination zone (Fig. 6.19) and shear fractures known from Riedel experiments (e.g.,
Fig. 2.68). In most cases, these pre-cuts (occasionally also termed slits, out-of-plane
fracture, pre-existing crack, plane of weakness, initial crack or initial flaw) are intro-
duced to the material by design (Germanovich and Dyskin 2000). We therefore use
the term “pre-cut” in our descriptions. Pre-cuts are distinguished from initial flaws
(Sect. 2.2.2.2), which occur as unintentional defects in the material, commonly on sub-
 2.4 · Primary and Secondary Fractures 183

Fig. 2.64. A schematic, idealistic fracture pattern which forms by compression. Compression is applied
on the primary fracture or a pre-cut pr, at a small angle α. Several types of secondary fractures may
form, including, wing cracks wc, sets of horsetails (or pinnate joints) p, kinks k and coplanar joints j, at
the two extensional quadrants, and a tensile fracture T that forms at 45° along the primary. In addition,
en echelon sets e, as well as curved branching faults bf, may form at the two compression quadrants

millimeter scales. The geometric distinctions and the positioning of the secondary
fractures along the primary fracture and next to their tip are described below in rela-
tion to the various fracture mechanisms that are involved.
The literature is profuse with terms assigned to secondary fractures by various
authors, including wing cracks, curved branches, branch fracture, kinks, curved kinks,
pinnate joints, splay cracks, horsetail fractures, bifurcation fractures, overstepping
joints, en echelon cracks, and more. We concentrate below on several types of second-
ary fractures, including the wing crack, pinnate joints and horsetail fractures, frac-
ture polarization, termination-zone joints and fracture branching. Additional descrip-
tions of secondary fractures, such as kinks and stepping joints are presented elsewhere
in this chapter (Sect. 2.2.8 and 2.2.13).

2.4.2
The Wing Crack Model

2.4.2.1
Summary of the Conventional Concept

Different models were proposed to explain the phenomenon of crack growth in com-
pression. We cite below the study by (Germanovich and Dyskin 2000) on this subject.
The majority of works relate the crack growth to the action of a stress concentrator
situated in the tested sample. The role of such a stress concentrator is frequently played
by pre-cuts (Fig. 2.65). The pre-cuts with contacted faces can be considered as the stron-
gest source of the secondary crack growth, and the wing crack geometry has been a
convenient modeling tool in describing fracture in compression (Fig. 2.66).
We adhere to the convention of many experimentalists that the pre-cut of length
2c0 is inclined with respect to the compression direction at an angle α (Fig. 2.66)
(Fairhurst and Cook 1966). One notes, however, that this is the geometry often
used for the angle of hade θ , complementary to the dip angle α (e.g., Park 1983, p. 4),
a potential source of confusion. A rectilinear pre-cut is opened by a pair of concen-
trated forces simulating the action of the inclined contact area (Fig. 2.66). The value
of these forces (per unit length of the inclined contact area) is assumed to be equal to
the horizontal projection of the shear force, F, tending to displace the opposite faces
of the pre-cut:

(2.9)
184 Chapter 2 · Elements of Fracture Geology

Fig. 2.65. Mechanism of rock fracture near a free surface: a Loading begins, b crack growth initiates
and propagates stably while load is increased, c crack starts growing unstably when reaching a criti-
cal length (2l), while the influence of the free surface is increased, d repeated buckling and creating
new free surfaces when L > h, and e growth of initial cracks in the direction of compression in Ten-
nessee Sandstone (after Germanovich and Dyskin 2000)

Fig. 2.66.
Wing crack growth under
compression (modified from
Fairhurst and Cook 1966 by
Germanovitch and Dyskin
2000)

Here σ is the applied compressive stress acting along the free surface, q is the confin-
ing pressure and C is a factor accounting for the friction and influence of curvature of
the secondary cracks. Using finite clement modeling, Germanovich et al. (1995)
calculated C that happened to be of the order of unity. The propagating crack 2l is
governed by the intensity of the stress concentrations at the crack tip, which is ex-
pressed through the stress intensity factor K. As the crack model is symmetrical, only
KI is accounted for (under compressional stress, Fig 2.65a,b).

(2.10)

The influence of mode II loading in the case of actual wing crack (Fig. 2.66) is as-
sumed to be suppressed by the compressive stress acting along the crack. Hence, this
mode is neglected in the crack growth consideration. The direct numerical modeling
reported by Germanovich et al. (1995) confirmed this assumption to a sufficient de-
gree of accuracy.
The conventional criterion of crack growth is
 2.4 · Primary and Secondary Fractures 185

(2.11)

where KIC is the rock fracture toughness. It is seen that the crack growth is stable: As the
crack elongates, K decreases; hence, greater loads are required to maintain the propagation.
The presence of q significantly reduces K and hence, drastically hampers the crack
growth. The influence of the lateral pressure manifests itself in:

1. Reducing the crack opening.

2. Reducing the force, F, by decreasing the shear stress acting on the inclined crack.

The first type of influence is (l / c0) times stronger than the second one. Since the dis-
cussed model accounts for only the main asymptotic term with respect to l / c0 <<1, the in-
fluence of the lateral pressure on F should be neglected. Therefore, it will be assumed that

(2.12)

Note application of Fig. 2.65 in Sect. 4.5.3.3.

2.4.2.2
The Three-Dimensional Wing Crack Model

According to Germanovich and Dyskin (2000), the above wing crack model represents
the experiments and considerations that are essentially 2-D. The real fracture mecha-
nism is, however, 3-D. Numerous experiments on the growth of internal 3-D wing cracks
in various transparent and non-transparent brittle materials under uniaxial compres-
sion revealed that there is a strong limitation on the extent a single wing crack can grow:
Its wings can reach about the size of the pre-cut (Fig. 2.67a), which is insufficient to
cause failure. This limitation seems to be universal and independent of the shape and
orientation of the pre-cut. Therefore, the actual 3-D crack growth in uniaxial compres-
sion differs radically from simple 2-D mechanism crack growth. The main reason for
that limitation is in the shape of the wings as they grow, wrapping around the pre-cut.

Fig. 2.67. Three-dimensional wing crack growth. a A wing crack in PMMA. b Schematic diagram of
initial flaw in biaxial compression. c Extensive wing crack growth in biaxial compression (after
Germanovich and Dyskin 2000)
186 Chapter 2 · Elements of Fracture Geology

It is this “surrounding” that prevents the wings from extensive growth into plane verti-
cal fractures, such that the ability of the material itself to withstand load gets exhausted
for most brittle materials. Accordingly, once the induced stress exceeds the compres-
sive strength of the loaded material, the latter fails in an “inherent mode” regardless of
presence or absence of the pre-existing defect. However, under biaxial compression
(when one direction is still free of load) an extensive wing crack growth is formed
(Fig. 2.67b,c) at a load much below the material inherent strength limit. The reason is
that the action of compression in the intermediate principal stress direction prevents
wings from wrapping around the pre-cut, enabling them to grow extensively in a plane
perpendicular to the unloaded direction.

2.4.3
Pinnate and Horsetail Fractures

There is some uncertainty whether pinnate fractures that form adjacent to faults are
shear or tensile structures. Hancock (1972) describes several types of secondary frac-
tures that are associated with shear zones, including veins, arrays of tension gashes in
en echelon arrangements, Riedel shear structures and pinnate, antithetic conjugate shear
fractures at 60°to a fault (Fig. 2.68a,b). There are, on the other hand, descriptions of
tensile pinnate fractures (Fig. 2.68c). These tensile sets that form in small angles (<45°)
are different from the straight T tensile fracture that forms at 45° to the general direc-
tion of Riedel shear structure (Fig. 2.68a,b). Hancock et al. (1982) describe an intrigu-
ing outcrop (Fig. 2.68d), showing an en echelon array of tensile pinnate fractures that
are related to a small dextral strike slip fault, where many of the pinnate joints bear
plumes with nearly vertical axes aligned perpendicular to the bedding boundary.
The interpretation of the mechanism that created the en echelon array of tensile
pinnate fractures (Fig. 2.68d) follows from the model depicted in Fig. 2.68a. A primary,
remote compression created the stress field of the primary shear couple, which induced
the strike slip fault in Fig. 2.68d, i.e., the dextral wrench fault in the latter figure is
equivalent to Sc' in the former one. But the plume orientation cannot be explained as
a direct consequence of lateral shear by Sc' (that would have formed horizontal
plumes); it rather suggests two alternative explanations. First, σ 1'' was vertical (paral-
lel to the plume axes) rather than horizontal. Second, a vertical σ 2'' created the en ech-
elon set and plumes. In any event, it is quite clear that the vertical propagation direc-
tion of the en echelon cracks implies a local stress rotation. While according to the “frac-
ture slanting” model (Sect. 2.2.5.2) the en echelon fringe forms by a local rotation of
the maximum principal stress from horizontal direction to a vertical one, it appears
that in the pinnate joints case presented here, the local stress rotation was driven by
mode II exerted by the strike slip fault. The former rotation concerns the creation of a
fringe, whereas the latter seems to relate to the entire rock layer, i.e., to what should
have been a parent joint if no breakdown had occurred. A mechanical model depict-
ing the difference between the above two systems is a challenge for future studies.
Figure 2.68d possibly suggests that vertical plumes on pinnate joints are more com-
mon than previously thought: Perhaps they characterize many other conjugate shear
zones (e.g., Shainin 1950). If this would be shown, it should support the concept of
“tensile pinnate fractures”, suggesting that shear pinnate fractures (Fig. 2.68b) are
created from joints by a secondary process.
Fig. 2.68 a Conventional interpretation of en echelon and pinnate fractures with respect to primary and secondary stresses. T: Tension fracture at 45° to an array;
R1 and R2: Riedel shears at 15° and 75° to an array, Sc': Primary shear couple, Sc'': Secondary shear couple which may give rise to third order fractures;
σ1' and σ3': Primary maximum and minimum principal stresses; σ1'' and σ3'': Secondary maximum principal stresses; θ : Angle between shear fracture and σ1'. b Pinnate,
antithetic conjugate shear fractures at 60° to a fault (modified after Hancock 1972). c Pinnate fractures (or feather fractures) along a brittle fault. The acute angle
between the fractures and the fault points in the direction of relative motion of the fault block containing the fractures (from Twiss and Moores 1992). d Plumes on
pinnate joints at St. Ann’s Head, Dale, Dyfed southwestern Wales (see text for explanation) (after Hancock et al. 1982)
2.4 · Primary and Secondary Fractures
187
188 Chapter 2 · Elements of Fracture Geology

Horsetail secondary fractures are approximately straight, sub-parallel fractures near

the ends of primary fractures that form in response to shearing along primary joints, ap-
parently initiating at roughness elements along the primary fracture, where tension is de-
veloped as a result of slip (Cruikshank et al. 1991). Renshaw and Schulson (2001) propose
that bending in compression of slender microcolumns that form between horsetails could
result in instability and localized failure of the structure. It appears from these descrip-
tions that tensile pinnate joints and horsetail fractures relate essentially to the same struc-
ture. Thus, we focus here on tensile fracture sets that are confined (but unpolarized, Sect. 2.4.4)
to the extensional quadrants and form acute angles with the primary fault (Fig. 2.68c).
Martel (1997) points out that the uniform shear stress along faults applied by
Cruikshank et al. (1991) in simulating secondary fracture should result in a single frac-
ture to open precisely at the end of a fault trace. This opening will take place owing to a
singular, infinitely strong stress concentration there, and a new, secondary fracture
should extend from the fault trace at an angle of about 70° (e.g., Pollard and Segall 1987).
According to Martel (1997), this theory encounters two difficulties. First, real materials
cannot sustain infinite stress levels. Second, this theory is inconsistent with numerous
field observations. Often, multiple secondary horsetails open near, but not precisely at
the tip (end) of fault traces (Brace and Bombolakis 1963, Fig. 2). These fractures com-
monly form behind fault tips. They often have nearly straight traces and intersect at an
angle of 50° or less with the fault. Martel (1997) models a small fault along which the
shear stress is non-uniform when it slips. He postulates a “cohesive zone” where the shear
stress is elevated near the fault tip. This zone will help to eliminate the stress singular-
ity and would account for the observed orientation, shape, and distribution of horse-
tails. Martel’s concept is challenging and requires support through future studies.

2.4.4
Fracture Polarization in the Two Extensional Quadrants along the Fault

It has been observed that the occurrence of secondary fracture sets at various acute
angles with respect to some primary faults may apparently be polarized in the two
extensional quadrants (Sect. 6.7). One set of fractures assumes the form of a joint set
that dips approximately in the direction of the primary fault at the upper part of the
hanging-wall (Fig. 6.24c). They differ from pinnate fractures and from T fractures that
create small angles and 45°, respectively, with the fault (Fig. 2.68a,b). The other set
consists of twisted and tilted faults that resemble Riedel shear fractures in the lower
part of the foot-wall of a normal fault (1–3 in Fig. 6.24c). Hence, the appearance or
absence, the differences in the angular relationships, and the locations of secondary
fractures in the two extensional quadrants may vary, reflecting subtle differences in
the histories and dispositions of the respective stress fields along the fault.

2.4.5
Termination-Zone Joints

Termination-zone joints is the term assigned to a joint set that starts around the tip
of a strike slip fault and is associated with the secondary faults that form in its con-
tinuation along a termination zone (Sect. 6.6). This set curves in conformity with the
primary fault and secondary faults.
 2.4 · Primary and Secondary Fractures 189

2.4.6
Fracture Branching

2.4.6.1
Dynamic Joints

Dynamic jointing can be identified by systematic fracture branching (bifurcation). There

is a need to distinguish between systematic and non-systematic branching. Systematic
fracture branching implies successive branches that form in a single, continuous (rapid)
process in which the branches propagate in particular directions, away from the fracture
origin (Fig. 2.30c) (see more in Bahat 1991a, p. 112). Systematic dynamic branching joints
are very rare in sedimentary rocks, but they are known to have been formed in great
depths, probably by extreme fluid pressures in granites (Fig. 2.69a,b, see Chap. 4). Their
branching faces one direction, and commonly branching fractures are fairly straight.
Anastomose fracture systems are occasionally wrongly interpreted as bifurcations.
Anastomose fractures often form by long, discontinuous processes that contain inter-
mittent pauses (e.g., Fig. 3.36f). Numerous repeated “branching” that may be suspected
to represent dynamic fracture in such systems is fairly erratic and inconsistent in its
directions. This “rebranching” is often the result of adjacent existing fractures that
interact due to later stress release processes, rather than a single fracture process.
Anastomose fracture systems often display (on both horizontal and vertical outcrops)
erosional and colored surfaces recording repeated precipitations from liquid flows,
particularly in carbonate rocks.

Fig. 2.69a.
Joint bifurcation by sudden
stress release. Lausitz, Cadomian
biotite granodiorite, Klosterberg
quarry near Bautzen vertical
section; scale is 1 m (Fig. 4.3
and Lobst 2001, Fig. A-1)
190 Chapter 2 · Elements of Fracture Geology

Fig. 2.69b. Joint bifurcation by sudden stress release. South Bohemian Pluton. The large block at
right bifurcates towards left (note optical distortion)

2.4.6.2
Branching Faults

Branching faults are secondary fractures, often rectilinear in shape, which propagate
unstably from the tip of the primary fracture due to predominant mode I loading,
mostly under dynamic conditions. These secondary fractures may be symmetrically
displayed at both sides of primary faults in their two termination zones (Fig. 2.57b).
It has been suggested that such structures reflect predominant tensile mode even in
overridingly shear regimes (Bahat 1982, 1991a). However, deviations from rectilinear
to asymmetric “almost straight” or “curved” branching faults may occur when the pri-
mary fracture is influenced by significant mode II / mode I ratios. Rectilinear branch-
ing faults generally differ from curved splay faults (e.g., Cruikshank et al. 1991, Fig. 12;
Twiss and Moores 1992, Fig. 7.10) such that the former type develops unstably whereas
the latter may form stably.
Chapter 3

Four Fracture Provinces in Sedimentary Rocks

While Chap. 2 is devoted to principles of fracture geology on all scales from labora-
tory to outcrop (up to 1 km long), the present chapter deals with fracture provinces
on regional scales (tens of km in length). An ideal study of a fracture province should
present all the province’s fracture elements in comprehensive detail: Meso-structures
and regional structures complete with their spatial relationships, whether they are in-
terdependent or not. This would throw light on the strain and stress histories of the
province, besides providing useful data for applied projects. To date, such ideal
studies are practically unavailable; no large fracture province is sufficiently exposed
or probed to provide what may be regarded as a full structural picture. Every well-
investigated fracture province is “specialized” in certain structural features but nev-
ertheless keeps some of its secrets. The objective of this chapter is, rather than pro-
viding a comprehensive review of the literature, to present a brief account on four frac-
ture provinces in sedimentary rocks that display distinct fracture elements that would
complement each other. By mentally compounding the characteristics of these sepa-
rate provinces, a realistic picture may emerge on the nature of large-scale fracture
patterns and relationships.
First to be described is a Devonian fracture province of alternating siltstone and
shales in the Appalachian Plateau of eastern USA, where the classic work by Sheldon
(1912) has demonstrated the concept of regional jointing. This will be followed by a
study of the Somerset Buttress anticline in the Jurassic limestone along the Bristol
Channel in England, with references to fracturing in some other well-known anticlines.
Next we shall bring a description of exfoliation joints that cut Jurassic sandstone at
Zion National Park, Utah, in western USA. The fourth region of interest relates to the
Eocene chalks around Beer Sheva, southern Israel. A specific discussion is devoted to
each province, as well as, a comparative review at the end of this chapter.

3.1
The Devonian Fracture Province of the Appalachian Plateau, New York
3.1.1
General Geology

Engelder and Geiser (1980) mapped 20 000 km2 of systematic joint sets in the Upper
Devonian rocks of the Appalachian Plateau, New York (Fig. 3.1a). In this area, Wedel
(1932) mapped very gentle anticlinal folds, with limbs dipping less than one degree and
axes whose strikes vary over 30°, from azimuth 090°, in the east to 060° in the west of
192 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.1a. Geological maps from the Appalachians. Location map of the Finger Lakes district in New
York State. The extent of the Silurian salt is shaded and outlined by dashes in the states of New
York (NY), Pennsylvania (PA), Ohio (OH), and West Virginia (WV) (from Younes and Engelder 1999).
Dark lines depict the stress trajectories of SHmax according to the distribution of cross fold joints in
the Appalachian Plateau

the region. Engelder and Geiser (1980) discerned two different cross-fold joint sets (or
cross-strike sets, approximately perpendicular to the fold axis). These sets preserve their
approximate cross-strike position from east to west, with angles between the sets ap-
 3.1 · The Devonian Fracture Province of the Appalachian Plateau, New York 193

Fig. 3.1b. Geological maps from the Appalachians. Set Ia (from Geiser and Engelder 1983)

Fig. 3.1c. Geological maps from the Appalachians. Set Ib (from Geiser and Engelder 1983)
194 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.1d. Geological maps from the Appalachians. Orkan and Voight’s (1985) map of the trend lines
for regional joint sets within the central Appalachian fold-thrust belt. Sets A through E are those of
Nickelsen and Hough (1967). Regional joint sets are based on the data of Nickelsen and Hough (1967),
and Engelder and Geiser (1980) (from Engelder 1993)

proximating 18 ±2° in the east and 30 ±4° in the west. In many outcrops, one joint set
parallels the direction of maximum compressive strain (ef) as recorded by fossil dis-
tortion (parallel to the bedding). Joints of this set, termed Ia (Fig. 3.1b), are sometime
calcite-filled. According to Engelder and Geiser (1980) they are part of joint set I as
defined by Parker (1942), of the type called dip joints (i.e., parallel to the dip direction
of bedding) by Sheldon (1912) who often found them in more than one orientation.
Joints of the other set, termed Ib (Fig. 3.1c), strike at 16–34° counter clockwise from ef
and are parallel to eoc, the maximum in situ compressive strain as measured parallel
to bedding by overcoring techniques (in the Machias Formation). These correspond to
the compressive direction indicated by a penetrative fabric in outcrops in Allegany
County (Fig. 3.1b). Sets Ia and Ib are equivalent to the systematic joint sets E and A re-
spectively, described by Nickelsen and Hough (1967) (Fig. 3.1d). In addition, Engelder
and Geiser (1980) characterized joint sets II and III. Joints of set II strike sub-parallel
to the fold axes. They belong to joint set II of Parker (1942) and were called strike joints
by Sheldon (1912). They have the same orientation toward the fold axes as the non-sys-
tematic joints described by Nickelsen and Hough (1967). Set III, which belongs to set III
of Parker (1942), is not obviously related to either ef, eoc, or to the fold axes.
 3.1 · The Devonian Fracture Province of the Appalachian Plateau, New York 195

The above discrimination of joint sets illustrates how complicated the “regional joint
pattern” of a given area can be, and how complex the challenge is to try and form a
sensible tectonic picture from it. The above references and subsequent studies have
made the Appalachian Plateau into a classic arena of fracture studies, where many basic
concepts of jointing were first recognized.
A comprehensive updated account of the Devonian Fracture Province of the Appa-
lachian Plateau is presented by Younes and Engelder (1999) and their results and dis-
cussion are quoted extensively here. Fringe cracks on joint surfaces and kinks are con-
sidered in the context of individual outcrops, in Chap. 2. The application and conclu-
sions obtained from parent joints, fringe cracks and kinks and their relation to the
regional tectonics in the eastern United States are given in the present chapter.
Younes and Engelder (1999) document the spatial distribution and nature of joints
within the northeastern portion of the Appalachian Plateau detachment sheet (Fig. 3.1a).
The area, about 3 000 km2, includes the Finger Lakes district of New York State, which
extends from Broome County to Allegany County (Fig. 3.1a). The detachment sheet in
this area consists of clastic sedimentary rocks of the Devonian Catskill Delta, transported
from the Acadian highlands to the east (Ettensohn 1985). The delta prograded from east
to west, attaining its maximum thickness east of the study area. The delta complex con-
sists of packages grading from black and gray shales through siltstone to sandstone.
The clastic Hamilton, Genesee, Sonyea, and West Falls Groups (Fig. 3.2 and 3.3) are sepa-
rated by three black shale formations: Genesee, Middlesex, and Rhinestreet shales, which
reflect anoxic conditions at the time of deposition. The Tully limestone is an important
marker bed at the base of the Genesee Group.
The detachment sheet was deformed into a series of low-amplitude folds (mapped in
outcrop by Wedel (1932), and in the subsurface by Bradley and Pepper (1938) and Murphy
(1981)). Folds of the detachment sheet are broad, persistent, and can be traced farther to
the southwest in Pennsylvania and into West Virginia. In the study area, the fold axes trend
east-west in the eastern part and northeast-southwest in the western part. The detachment
sheet contains layer-parallel shortening structures of Alleghanian age above a decollement
within the Silurian Salina salt (Rodgers 1970; Engelder 1979; Murphy 1981). Much of the
layer-parallel shortening was accommodated by pressure solution and the formation of
solution cleavage (Engelder and Geiser 1979; Geiser and Engelder 1983). The amount of
slip on the decollement is as much as 22 km to the north-northwest as estimated at the
Allegheny front (Engelder and Engelder 1977). Mapping of the detachment uses strain
markers such as deformed fossils and subsurface data (Engelder and Engelder 1977;
Engelder 1979; Engelder and Geiser 1980; Beinkafner 1983; Geiser 1988; Hudak 1992). The
foreland limit of the detachment sheet and the region of folds in the post-Silurian rocks
coincides with the limits of the Silurian salt (Frey 1973).
Joints of Alleghanian age strike approximately normal to the axes of the folds within
the detachment sheet (Fig. 3.1b,c) (Parker 1942; Ver Steeg 1942; Nickelsen and Hough
1967; Engelder and Geiser 1980; Geiser and Engelder 1983; Lacazette and Engelder
1992). Note that clockwise rotation from west to east is observed both among indi-
vidual cross fold and within each set (Fig. 3.1b,c). Multiple sets of cross fold joints are
particularly well developed in interlayered siltstone-shale beds, where the earlier dip
joints favored siltstone beds. Clear fractographic distinctions can be made among the
various cross fold sets (Bahat and Engelder 1984; Bahat 1991a, p. 248), which more-
over are occasionally decorated with fringe cracks (Younes and Engelder 1999).
 196
Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.2. The stratigraphy and geography of the southern tier of New York State (from Engelder and Geiser 1980)
 3.1 · The Devonian Fracture Province of the Appalachian Plateau, New York 197

Fig. 3.3.
A simplified stratigraphic
column of major rock units
in the eastern and western
portions of the study area in
the Finger Lakes district of New
York State (modified from Van
Tyne 1983). The facies become
progressively coarser to the east
as the proximal parts of the
Catskil Delta are approached.
The unconformity at the base of
the Tully limestone extends up
to the base of the Cashaqua
shale in western New York (from
Younes and Engelder 1999)

3.1.2
The Tectonic Problem

Nickelsen (1979), in an analysis of the Bear Valley strip mine in the Appalachian Val-
ley and Ridge recognized a group of structures that testify to a clockwise rotation of
maximum horizontal stress (SHmax) during the Alleghanian orogeny (Younes and
Engelder 1999). Engelder and Geiser (1980) also found joint evidence on the Appala-
chian Plateau of a clockwise rotation of the Alleghanian stress field. Rocks with two
distinct cleavages or commonly displaying two dip joint sets led Geiser and Engelder
198 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

(1983) to conclude that the Appalachian Plateau was affected by two different tectonic
phases: The Lackawanna phase and the Main phase. However, outcrops in the anthra-
cite coal district of the Appalachian Valley and Ridge of Pennsylvania, including the
Bear Valley strip mine, indicate that the rotation of the Alleghanian stress field pro-
duced structures with a broad range of orientations. From these, Gray and Mitra (1993)
concluded that the Alleghanian orogeny was a continuous series of structural events
reflecting a gradual clockwise rotation in the Alleghanian stress field, not punctuated
by two tectonic phases as suggested by Geiser and Engelder (1983). Bahat (1991a,
Tables 5.1 and 5.2) and Evans (1994) showed structures in more than two orientations,
suggesting that the northeastern portion of the Appalachian Plateau detachment sheet
was also deformed by a more continuous clockwise rotation of the Alleghanian stress
field. If so, the tectonic evolution of the Appalachian Plateau detachment sheet is con-
sistent with that of the Valley and Ridge. These two conflicting concepts (discrete phases
vs. gradual stress rotation) are further discussed in Sect. 6.3.

Fig. 3.4. Abutting relationships west and east of Ithaca, New York. The diamond symbols above the
straight line indicate a clockwise abutting relationship, whereas those under the line are counter
clockwise rotations. The abutting of joints west of Ithaca is dominated by clockwise rotations, but
mixed rotations are found east of Ithaca
 3.1 · The Devonian Fracture Province of the Appalachian Plateau, New York 199

3.1.3
Sequence of the Dip Joint Development

Evidence for the timing of joint development is visible throughout the detachment
sheet (Younes and Engelder 1999). It consists of curving or kinking of younger joints
as they approach the pre-existing joints, or abrupt termination of younger joints against
older joints. Unambiguous cases of abutting are less common than mutual cross cut-
ting. Nevertheless, abutments between dip (i.e., parent) joints in the detachment sheet
show west-to-east constancy. The later dip joints, particularly in the western part of
the study area, strike a few degrees clockwise from earlier dip joints; younger dip joint
sets are defined by the horizontal clustering of data in Fig. 3.4. Where present, fringe
cracks are always counter clockwise from these parent joints (Fig. 3.5a–c). Clustering
is observed in abutting dip joints at strikes 342°, 351°, and 003°, suggesting that younger
dip joints have a consistent orientation throughout the region, even though their pre-
decessors show widely varying orientations. The earliest dip joint set, in the range of
320–330°, does not cluster. The same appears to be true for another dip joint set, striking
006° to 021° range in the easternmost part of the study area. These data indicate that
joints striking at 342° generally abut joints striking at 320–330°, and 351° joints gen-
erally abut 342° joints, and so forth. Nevertheless, an exception to this general rule of
clockwise sequence of younger joints is found at the eastern edge of the study area,
where 003° joints abut parent joints striking roughly 020°, indicating in this area a
counter clockwise rotation of the stress field with time. Thus, the sequence of dip joint

Fig. 3.5a. Arial distribution of gradual segmentation. The amount of rotation shown is the average of
all fringe crack sets at every station (modified after Younes and Engelder 1999)
200 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.5b. Arial distribution of kinks (after Younes and Engelder 1999)

Fig. 3.5c. Arial distribution of of gradual segmentation on 070° joints. The amount of rotation shown
is the average of all fringe crack sets at every station (modified after Younes and Engelder 1999)
 3.1 · The Devonian Fracture Province of the Appalachian Plateau, New York 201

development in the detachment sheet appears to be as follows (parentheses indicat-

ing sets of joints clustered about that orientation):

Other post-Alleghanian joint sets are also found throughout the detachment sheet. One
post-Alleghanian 070° set is best developed in black shale below the Rhinestreet Forrmation
(Loewy 1995). The 070° joints are cut by Middle Jurassic kimberlite dikes (Kay et al. 1983)
in the vicinity of Ithaca, and are therefore older than Middle Jurassic. Another post-
Alleghanian 000° set consists of short, curviplanar-to-planar joints that occasionally abut
earlier 070° joints. The 000° set striking parallel to the Mesozoic kimberlite dikes (Mar-
tens 1924; Parker 1942), is the youngest joint set found by Younes and Engelder (1999).

3.1.4
Progressive Development of the Regional Stress Field

3.1.4.1
Patterns of Parent Joint Sets

Data on the abutting of parent joints of the Appalachian Plateau detachment sheet show
that there are at least two joint sets, which according to Engelder and Geiser (1980)
extend several hundred kilometers along strike in a fan-shaped pattern. Four devel-
opment stages are discerned in the pattern of dip joints, as is indicated by both the
abutting of parent joints and the rotation angle of fringe cracks. The regional pattern
of parent joints is due to the overlapping of joint sets with different strikes and devel-
oped from west to east as progresively younger joint sets propagated within a clock-
wise rotating stress field. Throughout this development, the Alleghanian stress field
sweeps through roughly 40 degrees from 320° to 360°. To the southeast in the anthra-
cite region of the Valley and Ridge, the Alleghanian stress field rotated 30 degrees clock-
wise, from 336° to 006° (Gray and Mitra 1993). Although the stress fields of the Pla-
teau and of the Valley and Ridge differ slightly in magnitude, the correlation is close
enough to suggest that both areas underwent similar tectonic development, marked
by a clockwise rotation of the horizontal stress field.

3.1.4.2
The Orientation of Fringe Cracks Relative to Their Parent Joints

Fringe cracks occur in most lithologies that are found in the Finger Lakes district of
New York State. More than 225 sets of fringe cracks were mapped by Younes and
Engelder (1999) in the Finger Lakes district of New York State on the northeastern
portion of the Appalachian Plateau detachment sheet. Although there are millions of
joints throughout the detachment sheet, joints decorated with fringe cracks are rare,
202 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

and within individual outcrops their number varies from one to several dozen. Maps
in Fig. 3.5a–c show the orientation and aerial distribution of fringe cracks and their
parent joints as expressed by their average strikes. Younes and Engelder (1999) ob-
serve that the general orientation of all fringe crack sets throughout the detachment
sheet follows the systematic change in fold trend and strike of the parent joints from
east to west. In general, and depending on the type of fringe crack, the angle between
parent joint and fringe cracks remains about the same across the area. However, there
is a pattern to the sense of fringe crack rotation across the region. East of Ithaca, fringe
cracks strike counter clockwise from the parent joints, while to the west they strike
clockwise from their parents, and both types overlap in the Ithaca area. Because load-
ing conditions and the timing of propagation are different for each fringe crack type,
there are characteristic differences in twist and tilt angles (Fig. 2.20). Kinks are con-
centrated in the eastern half of the study area and tend to strike within a few degrees
of 002°, seemingly independent of the parent’s orientation.

3.1.4.3
Counter Clockwise Propagation of Fringe Cracks near the Eastern Boundary

According to Younes and Engelder (1999) the abutting geometries of parent joints,
coupled with fringe cracks, provide a record of the Alleghanian stress field as it evolved
across the Appalachian Plateau detachment sheet. The fidelity of the record of stress
field rotation as derived from the abutting geometries and fringe cracks can be evalu-
ated by comparing it with the rotation of the Alleghanian stress field and strain within

Fig. 3.6. Deformed-state grid given by the displacement field determined from finite strain as indi-
cated by deformed crinoid columnals and from cleavage (adapted from Geiser 1988). Dashed lines
are traces of the fold axes. The circled numbers indicate the trends of joint sets from the Appalachian
Plateau (Younes and Engelder 1999), and the direction of the Alleghanian stress field in the anthra-
cite belt of Pennsylvania (Gray and Mitra 1993) (from Younes and Engelder 1999)
 3.1 · The Devonian Fracture Province of the Appalachian Plateau, New York 203

the detachment sheet. An isostrain map for layer-parallel shortening of the detach-
ment sheet was constructed by Geiser (1988) from orientation data of cleavage and
fossil deformation. Layer-parallel shortening shows a radial fan pattern from west to
east (Fig. 3.6). The east-west lines display a uniform spacing (i.e., strain gradient),
except east of Ithaca to the eastern part of the study area, where the rectangular grid
is distorted and the east-west grid lines are more closely spaced. Greatest grid distor-
tion coincides with the region containing parent joints decorated with fringe cracks,
reflecting a counter clockwise rotation on the Alleghanian stress field (Fig. 3.5a). This
conformance of fringe cracks to strain distribution provides evidence for construct-
ing the tectonic history of the detachment sheet.
In the eastern part of the study area, evidence is strong for counter clockwise rota-
tion of the stress field after the formation of an initial set of dip joints striking be-
tween 006° and 021°. This is indicated by both abutting geometries and fringe cracks.
Relative to its orientation farther west (i.e., 342°), the 20° to 30° misalignment of the
early stress field to the east is significant. Younes and Engelder (1999) note that this
misalignment occurs in the pinchout area of the Silurian salt, which served as a con-
venient detachment surface. They attribute the misalignment of the stress field to rock
drag, produced by the lack of salt detachment east of Ithaca. Drag at the edge of the
detachment sheet sets up a local remote stress field, controlling the orientation of early
parent joints striking between 006° and 021° (see Geiser 1988). A later stress field pro-
ducing both the 352° and 003° sets reversed the sense of shear on the parent joints along
the eastern boundary of the detachment sheet. This causes the later fringe cracks to
form counter clockwise from parent joints.

3.1.5
Tectonic History of the Appalachian Plateau Detachment Sheet

Younes and Engelder (1999) maintain that any tectonic history of the Appalachian
Plateau detachment sheet must take into account the following observations:

1. Whenever observed in the same outcrops parent joints in the shale are oriented
clockwise relative to parent joints in the siltstone layers.
2. The sequence of formation of parent joints, as indicated by the abutting relation-
ships, shows a clockwise rotation of regional stress from west to east, except in the
eastern part of the study area.
3. Fringe cracks on dip joints show a clockwise rotation of the stress field in the west, a
counter clockwise rotation in the east, and both kinds of rotation near Ithaca (Fig. 3.5a).
4. The rotation angle for counter clockwise fringe cracks is generally larger than for
clockwise fringe cracks.

3.1.5.1
The Proposed Sequence of Stress Directions in the Northeastern Part
of the Detachment Sheet

The 320–330° parent joint set. The earliest parent joints in the study area are limited to
the western region, the area (Fig. 3.6) that is closest to the extension of the Juniata cul-
mination into the plateau. Joints belonging to this set range over about 10° (320–330°)
204 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.7. A model for the rotations of the Alleghanian stress field in the Central Appalachian Mountains.
The heavy black line is the boundary between the Valley and Ridge and Appalachian Plateau. The short
straight segments indicate the first and last stages of deformation as determined by the following au-
thors indicated by the corresponding number: 1: Engelder and Geiser (1980); 2: Younes and Engelder
(1999); 3: Nickelsen (1979); 4: Gray and Mitra (1993); 5: Nickelsen (1988); 6: Nickelsen and Hough (1967);
7: Dean et al. (1984); 8: Dean et al. (1988); 9: Evans (1994); 10: Nickelsen (1996); 11: Spiker and Gray
(1997); 12: Nickelsen and Engelder (1989). The thin arrows indicate the local sense of rotation, and the
thick arrows indicate the regional sense of rotation proposed by Younes and Engelder (1999). The heavy
dashed line marks the location of the Juniata culmination (from Younes and Engelder 1999)

(see explanation below). This same joint set is found west of the Juniata culmination
and in the detachment sheet as far south as West Virginia (Fig. 3.7). However, to the
west and southwest of the Juniata culmination, the earliest parent joints are charac-
terized by a counter clockwise sequence of development. Zhao and Jacobi (1997) re-
ported that west of the study area, joints striking 322° to 340° predate 312–320° joints
and a second group of 280–305° joints. Evans (1994) reported that west of the Juniata
culmination, the earliest joint set strikes 350° and is followed by the 320° to 330° joints.
In central New York State, the earliest structures include folding (042° fold axes) of
the Tully limestone (Younes and Engelder 1995). Because the shortening direction
indicated by the deformation of fossils is at a significant clockwise angle with the
earliest joints (Engelder and Geiser 1980), these earliest joints apparently predate layer-
parallel shortening. The earliest joints in the Valley and Ridge also predate appre-
ciable layer-parallel shortening (Nickelsen 1979). Early joints occur throughout the
stratigraphic section near the Juniata culmination, yet farther east they are found only
 3.1 · The Devonian Fracture Province of the Appalachian Plateau, New York 205

below the Tully limestone. The orthogonal relationship between the fold axes of the
Tully limestone and the early joints suggests synchronous formation of the two fea-
tures. At this stage, detachment within the salt decollement may have begun but was
not well developed. Farther southeast, early fold axes of the Lackawanna syncline
(Fig. 3.6) are parallel to those found in the Tully limestone. Thus, we suggest that
set 320–330° appears to be of the burial type which forms prior to the onset of intense
tectonics (Bahat 1991a, p. 240). The early age of this set fits that of the early fold axes
of the Lackawana syncline before the actual detachment. Possibly, the wide range of
strike in this set reflects small differential stresses, compared to relatively large differ-
ential stresses in the later, syntectonic sets, resulting in narrower strike ranges.

The 342° parent joint set. The propagation of the 342° set marks the initiation of sig-
nificant detachment along the Salina salt, as indicated by deformed fossils. Clockwise
rotation of the stress observed in the early stages of layer-parallel shortening was re-
corded as cleavage in the Tully limestone. As the amount of layer-parallel shortening
increased, the eastern edge of the detachment sheet began to drag where the eastern
salt pinch-out prevented easy detachment. As a consequence of this drag, a shear couple
developed within the eastern region of the detachment sheet so that the local stress
field was rotated considerably clockwise relative to that found in the rest the detach-
ment sheet (more about the association of shear couple and local stress field rotation
in Sect. 3.2). Joints in the eastern region propagate in directions ranging from 006°
to 021°. During this tectonic stage, folding continued in the section above the Tully
limestone. A clockwise stress field rotation east of the Juniata culmination is reflected
in the initial development of clockwise fringe cracks in the western region of the study
area. The overall uniform strike of set 342° suggests that it is of the syntectonic type.

The 351°parent joint set. A third stage is marked by the propagation of 351° joints. By
this stage, parent joints had propagated throughout most of the northern detachment
sheet, and drag along the eastern edge of the detachment sheet was at a maximum. At
this stage, counter clockwise fringe cracks began to be formed along the eastern edge
of the detachment sheet. Kinks equivalent to the 351° parent joints are also found
throughout the western parts of the sheet.

The 003° parent joint set. The continuing clockwise rotation of the Alleghanian stress
field is next indicated by 003° joints. It is possible that the kinks showing a counter
clockwise angle developed during this stage as a result of north-south compression.
The general north-south orientation of the kinks (Fig. 3.5b) and their constant angles
to the parent joint for both clockwise and counter clockwise sets (Fig. 3.8) indicate that
they are related to the same event. This is also supported by the mechanical require-
ment that kinks will only form after arrest of their parent joints (more in Sect. 2.2.9.6).

The 070° parent joint set. This stage of jointing is best developed in black shales of
the Catskill Delta (Loewy 1995) and is the most difficult to date. The 070° joints are a
result of tectonic relaxation before abnormal pressure within the delta shales could
leak off (McConaughy and Engelder 1999). The 070° set carries fringe cracks that are
always rotated counter clockwise (Fig. 3.5c).
206 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.8. Histograms of the magnitude of clockwise (CW) and counter clockwise (CCW) rotations for
gradual and abrupt fringe cracks and kinks. The number of fringe sets and the statistical mode of
orientation are shown inside each histogram. The amount of CCW rotation for fringe cracks is larger
than the CW rotation. Kinks have a more consistent orientation (mode is almost the same for CW
and CCW rotations) (from Younes and Engelder 1999)

The 000° parent joint set. The propagation of a second north-south joint set is post-
Alleghanian. These late joints are parallel to the kimberlite dikes cutting the detach-
ment sheet. These dikes were dated as Early Jurassic, and therefore this set may also
be early Mesozoic in age. This later 000° set may be associated with continuing slip on
faults of the Clarendon-Linden zone (R. Jacobi 1998, pers. comm. to Younes and
Engelder). Joints oriented 003° and 000° are also distinguishable by differences in length
and planarity.
 3.1 · The Devonian Fracture Province of the Appalachian Plateau, New York 207

3.1.5.2
Tectonics on Either Side of the Juniata Culmination

According to Younes and Engelder (1999), the Juniata culmination correlates with a
regional lineament known in central Pennsylvania from surface geology and magnetic
and gravity anomalies (Gold and Parizek 1976). The joint propagation sequence around
the Juniata culmination indicates that paleostress trajectories rotated away from par-
allelism with the culmination in opposite directions, leading to a clockwise sequence
of structures to the east and north of it and a counter clockwise sequence to the west
and south of it (Fig. 3.7 and 3.9) (Younes 1996). As mentioned, the trend of the Juniata
culmination is parallel to the oldest joints in the detachment sheet (320–330°).
Nickelsen and Hough (1967) mapped five sets of dip joints in the Pennsylvania portion
of the detachment sheet (Fig. 3.1d, sets A–E). Set A, oriented 330°, is parallel to the Juniata
culmination and perpendicular to the Lackawanna syncline. It is flanked by joint sets D
and E to the northeast, and joint sets B and C to the southwest. Based on the similarities
between the joint orientations in central New York State and those in northeastern Penn-
sylvania, and also on the similarities between those in southeastern Pennsylvania to those
in West Virginia (Dean et al. 1984, 1988), Younes and Engelder (1999) relate set D to clock-
wise stress field rotation in the northeast, and sets B and C to counter clockwise rotation in
the southwest. This interpretation agrees with Evans’ (1994) analysis of joints west of the
culmination. Zhao and Jacobi (1997) reported a clockwise superposition of joints to the
east of the Juniata culmination and a counter clockwise superposition of joints west of it.
Exposures of clockwise and counter clockwise fold rotations in the Antes shale of
the Sinking Valley fault zone, Pennsylvania, show a transport direction that is com-
patible with transport parallel to the Juniata culmination (Nickelsen and Engelder
1989). The direction of transport was estimated as 322°, similar to the oldest trends in

Fig. 3.9. Selected data on the sense of rotation of the regional stress field in the central Appalachian
Mountains during the Alleghanian orogeny (from Younes and Engelder 1999)
208 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

the study area. Deformation at the Allegheny front near WiIliamsport, Pennsylvania,
shows a single transport direction at 340° (Spiker and Gray 1997). Nickelsen (1988)
noted that the Jacks Mountain fault, farther south, is intersected by a series of small
wrench faults from which he was able to determine the directions of compression.
These directions fall into orientation about 322°, then 335°, and then 312°, which in-
dicates both senses of stress field rotation that might be expected near the culmina-
tion. Viewed on a larger scale, the Juniata culmination divides both the Appalachian
Plateau and Valley and Ridge into an eastern area of clockwise paleostress rotation
and a western area of counter clockwise paleostress rotation (Fig. 3.7).

3.1.6
Three Main Concepts of Regional Cross-Fold Joint Patterns

Based on the previous studies of fracture in the Appalachian Plateau, we distinguish

three main concepts relating to regional patterns of cross-fold joints. These concepts
apply to the different geometries of joint sets that form during various time spans. We
distinguish:

1. The “strict orientation pattern” of joint sets (adopted from Nickelsen and Hough
1967) (Fig. 3.10a).
2. The “change in orientation within a joint set” (adopted from Engelder and Geiser
1980) (Fig. 3.10b).
3. The “simultaneous fan-shaped pattern” (adopted from Younes and Engelder 1999)
(Fig. 3.10c).

Engelder (1993) notes that Nickelsen and Hough’s (1967) outcrop-to-outcrop cor-
relation of joints depends largely on the assumption that the Earth’s stress field is
“nearly homogeneous”, and therefore joints of one set have similar orientations over
large regions. If a set of joints at an outcrop differs by, say, 15° from a joint set at other
outcrops, then this local suite of joints belongs to a second joint set, regardless of its
orientation with respect to local structures. The implication is that joints of one set
do not change orientation even as fold axes swing and curve through a mountain range
such as the central Appalachian Plateau (superposition of Fig. 3.1d on anticlinal traces
of Fig. 3.1c). On a regional scale, the change in strike of fold axes is accommodated by
the overlap of joint sets of different orientations.
Engelder (1993) maintains that Nickelsen and Hough (1967) developed their cor-
relation strategy based on the observation that first-order folds in the Pennsylvania
Valley and Ridge are kink folds (e.g., Faill 1973) with straight axes. Nickelsen proposed
(personal communication to Engelder) that the Valley and Ridge developed as over-
lapping thrust sheets cored with duplexes moving toward the craton, with straight-
axis kink folds delimiting the thrust duplexes. The curvature of the central Appala-
chian Valley and Ridge is accommodated by abrupt changes in the orientation of first-
order folds. Strictly parallel joint sets are a kinematic feature of thrust sheets associ-
ated with straight-axis kink folds. Presumably, the motion of various sheets is inde-
pendent, thereby setting up homogeneous stress fields that produce parallel joint sets,
unrelated to others in time and space. If this is indeed the case, then stress trajecto-
ries cannot be correlated throughout the whole mountain belt (Engelder 1993).
 3.1 · The Devonian Fracture Province of the Appalachian Plateau, New York 209

Fig. 3.10. Three concepts of regional cross fold joint formation. a Three joint sets (A–C) form sequen-
tially in a rotational manner (similar to Fig. 3.1d). b Joints of a given set rotate sequentially from A to C
(Fig. 3.1b,c). c Joints form concurrently in a radial pattern (Fig. 3.12a). d Contours of principal stresses
strongly deviate from an orthogonal pattern in a Hertzian field induced by indentation. Half-surface
view (top) and side view (bottom) of stress trajectories, plotted for Poisson ratio u = 0.33. AA denotes
diameter of indenter contact (after Lawn and Wilshaw 1975b). e Hertzian fracture in granite from the
South Bohemian Pluton, showing the cone opening downward (see the α angle), note a person for scale

While remapping cross-fold joints on the Appalachian Plateau, Engelder and Geiser
(1980) developed the second concept, using the assumption that joint sets change ori-
entation gradually so as to remain roughly perpendicular to local fold axes. Their analy-
sis implicitly assumes that inhomogeneous stress fields accommodate stress trajecto-
ries that curve on a regional scale during an orogenic pulse.
Tracing joint sets in one lithology allows for the possibility that a single joint set
can change orientation along with local fold axes (Engelder 1985) (Fig. 3.1b,c). We
suggest that in the Appalachian Plateau, the distinction between the first and second
210 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

concept is based primarily on criteria of fracture surface morphology: Fractography

changes from set to set (Bahat and Engelder 1984), but it remains the same in a given
set, even if joints within the set display some change in strike.
While evaluating the concepts that were introduced by Nickelsen and Hough (1967)
and by Engelder and Geiser (1980), Younes and Engelder (1999) coined the third con-
cept, the simultaneous fan-shaped pattern. They suggested that if the early joints in the
Tully limestone at Taughannock Falls correlate with the 320–330° set found more than
100 km to the west, then the earliest 320–330° set does not fan, but rather is an indica-
tion that the same joint set appears in different lithologies in different locations. If, on
the other hand, the 342° set in the western part of the study area is simultaneous with
the 006° to 021° joint set in the east, then it reveals a fan-shaped pattern. Superimposed
on this fanning pattern are two more joint sets, each differing about 10° in strike. Mea-
surements show that individual sets are most recognizable in the abutment of parent
dip joints (Fig. 3.4) where later joints cluster at 342°, 351°, and 003°. This joint pattern
is consistent with data presented by Nickelsen and Hough (1967), which show how joints
of one set cluster about one orientation, not taking part in the fanning pattern.
The first and second concepts require changing stress fields, by rotation mecha-
nisms of SHmax. The two concepts complement each other in the joint pattern of New
York: The various joint sets were formed by the mechanism described in the first con-
cept, showing the rotation from set to set (Fig. 3.1d and 3.10a). The rotation within
each set (Fig. 3.1b,c) was effected by the mechanism outlined in the second concept
(Fig. 3.10b). Both rotations took place at intermittent time intervals. The third con-
cept, on the other hand, postulates the simultaneous formation of a fan pattern of ra-
dial joints over long distances (Fig. 3.10c). We agree with Younes and Engelder (1999)
that this concept cannot be rejected, though it would be better supported if fan pat-
terns were found to exist in other fracture provinces. The third concept can be ex-
plained by mechanisms that relate to “moderate” heterogeneous stress fields. The ra-
diating joints in Fig. 3.12a seem to be a good example. The formation of conjugate joint
patterns may also be connected with simultaneous growth of joints in different direc-
tions. For comparison, a “strong” heterogeneous stress field may be induced by Hert-
zian indentation (Fig. 3.10d). Such extremely heterogeneous stress fields can be pos-
tulated for Hertzian fractures (e.g., Fig. 3.10e) found in granites and other magmatic
rocks (Bahat 1979c) but rarely for fracturing in sedimentary rocks.

3.2
Fracturing in The Bristol Channel Basin
3.2.1
General Geology and Previous Investigation

The Bristol Channel Basin (BCB) extends on both sides of the Bristol Channel, from
west England to South Wales. Unlike jointing in the other three provinces described in
this chapter, which affects horizontal or slightly folded layers, joints of the Bristol Chan-
nel Basin occur in steeply inclined layers, some dipping more than 30°, due to asym-
metric anticlinal folding. Fracturing in this province was described by Rawnsley et al.
(1998) who worked on four edge locations of the basin (Fig. 3.11a), and by Engelder
and Peacock (2001) who concentrated on the eastern side of the channel (Fig. 3.11b).
 3.2 · Fracturing in The Bristol Channel Basin 211

Fig. 3.11a.
Map of the Bristol Channel.
Map showing the main geo-
logical features of the eastern
Bristol Channel, with positions
of locations studied (modified
from Rawnsley et al. 1998, based
on the Bristol Geological
Survey sheet 51° N–04° W,
Bristol Channel)

Fig. 3.11b. Map of the Bristol Channel. Geological map of south-western Britain, showing the loca-
tion of the Bristol Channel basin and Lilstock (from Engelder and Peacock 2001)
 212
Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.11c. Map of the Bristol Channel. Geological map of Lilstock Beach (see location in Fig. 3.11a, adaptation from Rawnsley et al. 1998). The locations of photos
in Fig. 3.13 are indicated (modified from Engelder and Peacock 2001)
 3.2 · Fracturing in The Bristol Channel Basin 213

The latter study focuses on the south-dipping beds at Lilstock (Fig. 3.11c), which these
authors regard as particularly informative. These two papers, particularly the one by
Engelder and Peacock (2001), are quoted extensively in the present section.
The Blue Lias Formation of Lilstock Beach consists of more than 50 limestone beds,
5–40 cm thick (Whittaker and Green 1983), interlayered with shales that range from a
few centimeters in thickness to over 300 cm. The order of layers is consistent and can
be followed even in separate fault blocks (Engelder and Peacock 2001). According to
Rawnsley et al. (1998) (Fig. 3.11a), the BCB is an exhumed E-W-trending Mesozoic
basin. Basin formation started as early as the Permian, with N-S extension causing
reactivation of south-dipping Variscan thrusts (Van Hoorn 1987; Brooks et al. 1988;
Roberts 1989). Extension and subsidence continued through the Triassic and Juras-
sic. To this stage belong E-W-striking normal faults and calcite veins, which are well
exposed on the Somerset coast (Peacock and Sanderson 1991, 1992, 1994). Evidence
for Alpine N-S contraction on the Somerset coast includes E-W-striking thrusts and
reactivated normal faults (Peacock and Sanderson 1992; Dart et al. 1995) and strike-
slip faults which conjugate about N-S (Peacock and Sanderson 1992, 1995). E-W-trend-
ing folds are related to wall-rock deformation around the E-W-striking normal faults
and were probably tightened during the N-S Alpine compression. NNW-striking faults,
e.g., the Sticklepath fault zone, appear to belong to the Variscan phase and may have
been active at various times (Holloway and Chadwick 1986; Lake and Karner 1987;
Van Hoorn 1987; Roberts 1989). Sinistral displacement occurred on these faults dur-
ing the early Tertiary, so σ1 was approximately NW-SE across the BCB at that time (Van
Hoorn 1987). Dextral displacement occurred during the late Oligocene to Miocene
(Arthur 1989), so σ1 was approximately N-S across the BCB (Van Hoorn 1987).
Rawnsley et al. (1998) found that the joint patterns at the four edge locations in the
BCB all post-date the high-density veins (HDVs) and the faults. The joints appear to
have been formed in five main phases:

Phase 1: These are sub-parallel to a regional Alpine compression direction (about

340–350°) and are almost unperturbed by faults. At Nash Point, faulting started to
perturb the stress during the development of phase 1 joints, i.e., there is a transi-
tion from phase 1 to phase 2 joints. Phase 1 joints were reactivated after phase 3 at
Lavernock.
Phase 2: These are joints that were perturbed by faults and by phase 1 joints. The
phase 2 joints follow the perturbed stress trajectories and curve from points of
stress concentration along fault surfaces, e.g., at fault bends. The transition from
non-perturbed (phase 1) to perturbed (phase 2) joints reflects a decrease in the
regional compression and an increase in residual tensile stress. Joints of this phase
are absent where no faults are present, as at Lavernock.
Phase 3: At Lavernock and Lilstock, phases 1 and 2 seem to have left no traces, and
the only joint set found there belongs to phase 3, which is parallel to the 290° strik-
ing fold axes. These joints cut the veins that formed during folding and result from
the relaxation of the Alpine compressional stresses within the folds. Phase 3 joints
typically rotated counter clockwise through time to become parallel to the fold
axes, perhaps as a result of diminishing the regional influence. This phase is absent
where there are no folds, as at Nash Point.
Phase 4: This phase consists of cross-joints.
214 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Phase 5: This phase consists of polygonal joint patterns. Phases 4 and 5 occur at all
the locations studied. They are related to relaxation or contraction of the rock at a
late stage. Rawnsley et al. (1998) show that their model applies well to the Liassic
exposures around the basin’s edge, although different patterns may be found in the
unexposed center of the BCB. Such may be the case for a different orientation of
joints linked to the regional Alpine compression, which e.g., at Lavernock are 010–
020° (Rawnsley et al. (1998, Fig. 7a).

Engelder and Peacock (2001) assume that the present-day lithospheric stress field
in the NW European platform has the same orientation as the Alpine stress field of
the Miocene (Brereton and Evans 1987; Zoback 1992). Furthermore, the youngest joint
set recorded throughout southern England correlates with the contemporary tectonic
stress field (Bevan and Hancock 1986; Hancock and Engelder 1989). Hancock and
Engelder (1989) argue that exhumation in a late-stage “Alpine” stress field is respon-
sible for these youngest NW-striking joints in northwestern Europe. Thus, the post-
Miocene tectonic event caused a clockwise rotation of the European stress field, only
to have it return to its “Alpine” orientation, as is testified by the latest jointing in north-
western Europe.
Engelder and Peacock (2001) emphasize that the Alpine deformation, or “Alpine in-
version”, was accommodated by reverse-reactivation of normal faults, new thrust faults,
new strike-slip faults, and by hanging wall buttress anticlines. Beck et al. (1993) de-
fine the “buttress effect” as resistance to displacement arising from a space problem;
a body of rock can move only if it has somewhere to go. Buttress folds of Eocene to
Oligocene age show a parallel geometry that is produced by flexural flow with bed-
parallel slip (Nemcok et al. 1995). It should be noted that reverse faulting was more
prevalent along the southern margin of the Bristol Channel Basin than on the north
margin (Nemcok et al. 1995). Only one case of minor thrust faults associated with fold-
ing was found along the north margin in the southern limb of the Trwyn-y-Witch
anticline in south Wales (Roberts 1974).

3.2.2
Fracture Description

Following are some fracture descriptions according to Rawnsley et al. (1998), special-
izing on phenomena that are less common in other provinces, including high-density
veins, perturbations of joints near faults, “ladder” patterns and polygonal joints.

High density veins (HDV). Thin, closely-spaced calcite veins appear to be present ev-
erywhere in the limestone beds at Lilstock. In the east they mostly strike 270–290°,
their common width is about 0.08 mm, and average spacing is less than 10 mm. They
are often only visible where differential erosion has enhanced the contrast between
the calcite HDVs and the limestone country rock, especially around the edges of joints.
In thin sections they can be seen to persist throughout the rock, often forming en
echelon patterns. HDVs either pass through the fault-related calcite veins or are slightly
deviated by them. The HDVs are not perturbed by joints, and some joints may follow
HDVs. In the west of the area, the HDVs strike 082–088° and have an average spacing
 3.2 · Fracturing in The Bristol Channel Basin 215

of about 20 cm. Most of the calcite has been removed, with oxidization fronts about
30 mm wide. In the cliff, the HDVs strike about 110° and dip about 80° more steeply
to the north than the bedding. HDVs strike 350–010° at Nash Point, which is the ap-
proximate Alpine compression direction. These are followed by 350–010° striking faults.
The 010° faults appear to be dextral and the 350° faults appear to be sinistral, so they
are conjugate about N-S. Thus, the HDVs represent a pre-joint phase that possibly was
associated with faulting that accompanied the Alpine compression.

Converging joints perturbed by the faults. At least three main zones of converging
joints occur along the 065° striking section of fault F5 at Lilstock (see also Swaby and
Rawnsley 1997, Fig. 6). They are elongated in the 320–360° range. Joint spacing de-
creases slightly towards the center of the convergence as the joints curve towards the
fault (Fig. 3.12a). These converging joints are the longest at Lilstock. For a detailed
description, see Rawnsley et al. (1992). According to Engelder and Peacock (2001) the
most plausible interpretation of the perturbed joints is that limestone-limestone con-
tact across the fault generated a local stress field during or after the period of high-
angle reverse faulting on reactivated normal faults, i.e., joint propagation followed
fault initiation but not necessarily total slip. Curved trajectories of phase 2 joints in
proximity to phase 1 joints are interpreted by Engelder and Peacock to mean that
phase 1 joints were open and also perturbed the stress field during stage 2 joint propa-
gation (Dyer 1988; Rawnsley et al. 1992, 1998).

Ladder joint patterns. “Ladder” patterns occur at Nash Point and at Lilstock. The joint
pattern at Nash Point varies across the area. In the east, the main set strikes 155–160°.
There is a two-level hierarchy, with 160° joints connected by orthogonal joints to form
a “ladder” pattern that crosses layer boundaries. Within these ladders, shorter 160° joints
and non-cross-cutting orthogonal joints form a smaller-scale system of ladders
(Fig. 3.12b), which mostly do not cross bedding boundaries. Oblique ladders occur at
Lilstock over a 095–135° range in all limestone beds within about 30 m of fault F3
(Loosveld and Franssen 1992). They usually rotate counter clockwise with time (i.e.,
135° joints are the oldest and 095° joints are the youngest), forming oblique ladders.
Left-stepping arrays spaced about 15 m trend 160–170° between two zones of conver-
gent joints. Multidirectional joints also occur between zones of convergent joints and
may be formed over extensive areas. Limestones appear to be more jointed than the
shales. Joints often pass from the limestones into the shales, suggesting that they ini-
tiated in the limestones.

Polygonal joints. These enigmatic joints are found throughout the BCB, typically sur-
rounding blocks. They are believed to indicate horizontally isotropic effective ten-
sion. On the other hand, they may represent thermal contraction following uplift, which
may also be responsible for non-cross-cutting perpendicular joints between closely-
spaced joints. The latest, isotropic joints, are related to biaxial effective tension in the
plane of bedding, especially in “shadow zones” between convergence points. Between
the convergent zones, an almost isotropic joint pattern was produced by a low hori-
zontal differential stress. Impedance of later joints on earlier joints indicates that the
latter were always open. Deciphering the origin of these joints remains a challenge.
216 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.12a. Line drawing of fracture patterns at Nash Point (see Fig. 3.11a). Perturbations in joints
(modified from Rawnsley et al. 1998)

Fig. 3.12b.
Line drawing of fracture patterns
at Nash Point (see Fig. 3.11a).
Joints displaying a ladder pattern
of joints (modified from
Rawnsley et al. 1998), note a
certain resemblance to cross-
joints (Gross 1993)

Engelder and Peacock (2001) note that at Lilstock Beach joints within limestone
beds are generally contained within these beds and rarely pass into adjacent shales.
They also note that patterns of systematic jointing differ from one limestone bed to
 3.2 · Fracturing in The Bristol Channel Basin 217

the next, which explains the different descriptions of the joint sets at Lilstock Beach
in earlier studies. All that can be said about the sequence of jointing at Lilstock Beach
is that the joints postdate early normal faulting and vein development. Commonly,
joints propagate after normal fault development, as is indicated by abutting and cut-
ting relationships (e.g., Fig. 3.13d,e). Some curving joints radiate from points of stress
concentration on faults (e.g., Fig. 3.13c), which are a natural consequence of slip on an
irregular fault. These joints could have propagated during a later stage of slip (Engelder
and Peacock 2001). By way of comparison, perturbed curving joints that are associated
with the Naim fault in the Beer Sheva syncline were formed under a secondary stress
field, after cessation of movement along the primary strike-slip fault (Sect. 6.6).

Faults and folds. The structures on Lilstock Beach include a set of four north-dip-
ping, high-angle faults that are offset by later strike-slip faults (Fig. 3.11c). Engelder
and Peacock (2001) divided Lilstock Beach using three of these high-angle faults
(faults 3, 4, and 5 of Rawnsley et al. 1998) and two strike-slip faults. The area north of
fault 5 is called block 5 (or “the bench”, in recognition of Loosveld and Franssen’s
(1992) mapping there). The other three blocks are hanging walls to faults 4, 3 and 1.
Engelder and Peacock (2001) refer to the eastern, central, and western portion of each
“normal” fault block, depending on outcrop position relative to two prominent strike-
slip faults (Fig. 3.11c).
Engelder and Peacock found that separation at any point on the faults can be mea-
sured with confidence by juxtaposition of bedding (Fig. 3.11c). Throw on at least one
of the high angle faults (fault 4) was reversed during Alpine inversion. Reverse slip may
have occurred on the other faults, but the total throw was not reversed. The steep dip
of fault 4 and its parallelism with normal faults in the area indicate that it originated
as a normal fault, which inversed together with the other reverse faults in the region
(e.g., Whittaker and Green 1983; Dart et al. 1995). Net throw (the combination of pre-
Alpine normal slip and Alpine reverse slip) on the four faults with throw >2 m is:
Fault 1 ~ 18.7 m (normal), fault 2 ~ 8 m (normal), fault 4 ~ 8 m (reverse), and fault
5 ~ 14 m (normal). The outcrop is also cut by numerous high-angle faults with
<2 m throw, which mostly exhibit normal slip and are parallel to the strike of bed-
ding. Strike-slip faults displace the high-angle faults and the limbs of the Lilstock
buttress anticline. The strike-slip faults are cut off at fault 4 (Fig. 3.11c), indicating that
strike-slip faulting and some high-angle faulting may have been contemporaneous
(Kelly et al. 1998).
Block 4 consists of a large buttress fold, the Lilstock buttress anticline, with a gen-
tly dipping north limb and relatively steep south-dipping beds (Fig. 3.13a,b). The
Lilstock buttress anticline originated as a reverse-drag fold (Hamblin 1965) during the
Mesozoic normal faulting and was tightened during reverse slip on fault 4. Engelder
and Peacock focused their study on joints in the south-dipping limb of this anticline.
Beds in the northern half of block 4 dip gently to the north (<5°), whereas beds in the
southern half of the outcrop dip to the south, and in places the dip is more than 30°.
Along the south side of the Bristol Channel, dips of more than 30° are restricted to the
flanks of buttress anticlines. Diagrams of the tight folding typical of buttress folds in
the Mesozoic strata of the Somerset coast are presented in Peacock and Sanderson
(1992, Fig. 7a) and Dart et al. (1995).
218 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.13a–d.
Outcrops from Lilstock. a View
looking NNW from the eastern
cliff along faults 3 and 4 with the
hinge of the buttress anticline
shown in the background.
b View looking NNW from the
eastern cliff along the hinge of
the Lilstock buttress anticline.
Bed 0497 is visible in the fore-
ground. A control sample from
bed 0497 was taken across the
hinge of the anticline and
Fig. 3.13h was taken at the
point where bed 0497 rolls over
into the limb of the anticline.
c Curving joints in bed 1848 of
fault block five. Photographer
is standing on fault 5 (location
shown in Fig. 3.11c). d J2 joints
cutting two normal faults within
bed 2429 of fault block 1 (from
Engelder and Peacock 2001)
 3.2 · Fracturing in The Bristol Channel Basin 219

Fig. 3.13e–h.
Outcrops from Lilstock.
e J3 joints abutting a small
(throw <1 m) normal fault in
bed 1921 in the western portion
of fault block three. A sinistral
strike-slip fault offsets bed
1921 (in the foreground) by
2 m. That fault is internal to
the western portion of fault
block 3. f Joint set development
within bed 1848 on the bench
north of fault 5. Joint set J2 is
parallel to the direction of view.
Sets J4 and J5 are oriented
anticlockwise from set J2, with
J6, anticlockwise from J4. The
view is looking south-east to-
ward the cliff from where the
photographs for Fig. 3.13a,b
were taken. View across the
bottom of the photograph is
about 3 m. g J3 joints cutting
veins and transecting beds dip-
ping in a relay ramp developed
during early extension in the
western portion of fault block 3
(coin diameter = 27 mm). h The
counter clockwise relationship
between J2 and J3 in bed 0497
where it rolls over into the south
limb of the anticline (from
Engelder and Peacock 2001)
220 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

3.2.3
Data Collection and Joint Identification

Engelder and Peacock (2001) studied joint development in 25 out of 46 carbonate beds
in a 40 m thick Liassic section (British Geological Survey stratigraphic column,
Whittaker and Green 1983) (Fig. 3.14). To indicate position in the vertical section, they
labeled beds by height (in cm) above an arbitrary datum. For example, bed 1848,
18.48 m above the datum features prominently in the studies of Loosveld and Franssen
(1992) and Rawnsley et al. (1998). Engelder and Peacock sampled systematic joints at
Lilstock Beach in a volume of Blue Lias measuring 40 m thick by 200 m long (E-W).
The repetition of the section by high angle faults permitted sampling up to 50 m along
the N-S direction of some beds. Focusing on the effect of Alpine inversion on joint
development, they assumed that joints of this stage are likely to be found in the south-
dipping limb of the Lilstock buttress anticline in block 4, since the relatively steep dip
of these beds indicate they are the most strained rocks at Lilstock Beach. To confirm

Fig. 3.14. Stratigraphic section at Lilstock Beach showing the position of 46 carbonate beds above an arbi-
trary datum. The beds are labeled according to their position relative to the datum. The British Geological
Survey (Whittaker and Green 1983) designation for each bed is also given (from Engelder and Peacock 2001)
 3.2 · Fracturing in The Bristol Channel Basin 221

this, they measured the joint patterns on sub-horizontal surfaces in block 5 as well as
in the hinge area of the Lilstock buttress anticline of block 4, to serve as control samples.
The joint patterns in dipping and sub-horizontal limestone beds were compared. Stan-
dard scan line techniques (LaPointe and Hudson 1985) were then used to record the
location and orientation of each systematic joint. The joint sets in each bed, based on
abutting and orientation characteristics, were then correlated from bed to bed by
means of subsequent data reduction. The most subjective assignments concerned
curving joints that on the base of orientation might be included in more than one set.
A collection of parallel systematic joints counts as a set (Hodgson 1961b). Accord-
ing to Engelder and Peacock (2001), however, the poles to joints of a systematic set
may cluster through several degrees, because individual joints are never perfectly pla-
nar and measurements cannot be precise (Fig. 3.15). Therefore, joint sets that differ
by less than 10° may have overlapping clusters in a stereographic plot, and not each
individual joint can be assigned to a particular set even by post-fieldwork analysis.
Hence, Engelder and Peacock rely on field notes, where joints were presorted into sets

Fig. 3.15. Lower hemisphere stereographic projections for beds and joints in fault block 1. a Bedding
traces and poles for bed 2429. b Present-day poles to joints in bed 2429. c Poles to joints in bed 2429
with bed dip removed. d Present-day poles to joints in bed 3883 (from Engelder and Peacock 2001)
222 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

within each particular bed. The grouping of joints into distinct sets was most difficult
where several beds are involved and where joint sets tend to curve, as they do in the
south-dipping limbs near the faults of Lilstock Beach. In these cases the control samples
(see above) served as a guide. Using this procedure, six joint sets were distinguished,
falling in the following ranges: J1, (330–310°), J2 (310–295°), J3 (295–285°), J4 (285–275°),
J5 (275–265°), and J6 (<265°)(where 265° is equivalent to 085° in the upper hemi-
sphere). It should be noted that joint distribution in the western part of the BCB dif-
fers somewhat from the findings in England. Between Mumbles Head and Rhossilli
Bay in south Wales, Roberts (1979) found six sets striking 360°, 340°, 290°, 270°, 240°,
and 200°, as perhaps can be expected over such distances and changes in the regional
structure. The challenge to the investigator is how to sieve out homogeneous domains
from a heterogeneous province for deciphering the tectonic history of the region.
Engelder and Peacock (2001) used a simplified AVTD plot (see Wise and McCrory
1982) for obtaining a qualitative sense of relative joint density by normalizing the data
on joint occurrence and comparing joint development in sub horizontal beds with beds
that have been tilted by Alpine inversion. This procedure is not given here, and the
interested reader is referred to the original paper.

3.2.4
Joint Sets in the Blue Lias

3.2.4.1
Joint Sets in Steeply-Dipping Beds

Joints in the south-dipping limb of the Lilstock buttress anticline dip to the north as
expected for beds that dip to the south (Fig. 3.15a–d). These commonly carry more
than one joint set, but no beds carry as many sets as found in bed 1848 of fault block 5
(Fig. 3.16a). To correlate joints in the south-dipping limb with the control sample,
Engelder and Peacock (2001) rotated joints to their attitude in horizontal bedding. After
rotation, the joint-frequency data for steeply-tilted beds were normalized in the same
manner as for sub-horizontal beds. The population of joint orientations in the south-
dipping beds of Lilstock Beach is indeed different from the population in the sub-hori-
zontal beds. The most prominent joint set in the south-dipping beds, J3, is nearly par-
allel to the strike of bedding and the hinge of the Lilstock buttress anticline.

3.2.4.2
The Orientations of Joints as Related to Faults

Joints of set J3 (strike 290°) transect the fold and veins in a relay ramp in the western
portion of fault block 3 (in bed 1735) (Fig. 3.13g). This relay ramp is evidence that
J3 joints post date the early extension of the Bristol Channel. Various relationships
between J3 joints and normal faults are observed at Lilstock Beach. For example,
some J3 joints cut through incipient high-angle faults without being displaced by
them (Fig. 3.13d). Other J3 joints abut high-angle faults without appreciable deflec-
tion (Fig. 3.13e). Both types of fault are early calcite-filled normal faults with no evi-
dence of reverse reactivation during Alpine contraction. This is like the relay ramp in
bed 1735.
 3.2 · Fracturing in The Bristol Channel Basin 223

Fig. 3.16.
a Joint strike vs. distance for a
scan line corrected to a scan
line azimuth of 040° for beds
1848 and 1921 on the bench
north of fault 5. b Joint strike
vs. distance for a scan line azi-
muth of 044° for beds 0169 and
0497 in eastern fault block 4.
c Joint strike vs. distance for
bed 1921 for a scan line azi-
muth of 040° for the bench
north of fault 5 and 110° in
fault block 3 (from Engelder
and Peacock 2001)

The orientation of the joint population varies from fault block to fault block. For
example, bed 1921 in western block 3 dips up to 43° to the south compared with its
slight northward dip in the bench of block 5. Within western block 3, J3 and J4 appear
without J1, which is the only set in bed 1921 on the bench (Fig. 3.16c). This example is
typical of the difference in joint patterns within the same bed from block to block.
Engelder and Peacock (2001) conclude from these frequent differences that jointing
took place after beds became isolated within individual fault blocks following at least
some fault displacement. The widespread occurrence of systematic joint sets within
fault blocks suggests that the internal portions of fault blocks were not affected by stress
reorientation during fault slip. They also suggest that, alternatively, these differences
mean that different mechanisms operated in the south-dipping limb of the Lilstock
buttress anticline and in the sub-horizontal beds.
224 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

3.2.4.3
The Sequence of Joint Development

The abutting sequence within sub-horizontal bed 1848 indicates that J2 propagated
before J4, which propagated before J6 (Fig. 3.13f). Assuming that joints are created par-
allel to regional SHmax directions, this sequence shows the same counter clockwise re-
orientation of the remote stress field as seen throughout the eastern part of the Bristol
Channel (Rawnsley et al. 1998). The same counter clockwise sequence may be traced
from the sub horizontal bed 0497, from block 4 into the south-dipping limb of the
Lilstock buttress anticline (Fig. 3.13h). This is a rare case of J3 joints abutting joints
of J2. Where bed 0497 steepens, J2 is absent and J4 abuts J3. Another case of this counter
clockwise joint formation is found in bed 3883, which dips up to 30° in the central and
western parts of fault block 1 (Fig. 3.15d). Other beds also show a clear counter clock-
wise age sequence, although this is found in other joint sets (Table 3.1).

Table 3.1. Joint development: All stations in south limb of buttress anticline
 3.2 · Fracturing in The Bristol Channel Basin 225

Fig. 3.17.
Joint frequency vs. joint strike.
Joint count normalized to a
250 m scan line with the data
binned into 1° intervals. a Sub-
horizontal beds in blocks 4
and 5. b Dipping beds of the
Lilstock buttress anticline, in
blocks 1, 3 and 4. c Binned data
without normalizing joint data
for either scan line length or
scan line orientation, in blocks 1,
3 and 4 (from Engelder and
Peacock 2001)
226 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

The counter clockwise age sequence found in south-dipping beds is more clearly ob-
served on sub-horizontal surfaces (Table 3.1). Determination of sequence requires two
or more joint sets abutting at a station, but most exposures are too small to display the
required joint contacts (e.g., Fig. 3.13d,e,h), although beds may contain two groups of
joints with different orientations on different parts of the exposed bedding surface along
strike. In such cases, it is difficult to determine whether the groups belong to different
sets or whether they belong to a single curving set, as is seen near faults. More than half
the samples from the south-dipping beds display only one set (Fig. 3.17), showing no
time sequence. As a rule, beds containing J1 or J2 joints lack J3 joints. Where J3 joints
are found, they are the oldest joints and strike closest to bedding strike (Table 3.1). Thus,
the evidence for the age of J3 relative to J1 and J2 is rather too meager to be definitive.
The distribution of joints in the south-dipping limb of the Lilstock buttress anticline
is also correlative to stratigraphic and structural position. J1 and J2 joints (found only
in block 1) are high in the section. Sets J3 and J4 are most common in the middle of the
section (blocks 3 and 4), with the greatest development of J3 joints near the reverse fault
that created the Lilstock buttress anticline. J5 and J6 joints are commonest in the lower
parts of the section and nearest the anticline-hinge, where dips are shallow.

3.2.4.4
The Attitude of Joints in Relation to Bedding

Joints in competent beds are typically perpendicular to the bedding (Hancock 1985).
However, this is not the case for sets J1 to J5 in the south-dipping beds of the Lilstock
buttress anticline. When bedding is restored to the horizontal (e.g., J3 joints in
Fig. 3.15c), these joints consistently dip to the north, with their poles plunging south,
towards the European Alpine hinterland. The joint-pole plunge angle is used to de-
fine the “tilt towards the hinterland” for the joint sets (Fig. 3.18). The tilt increases
from 3° for the older J1 joints to 7° for younger J5 joints. J3 joints occur only in the
south-dipping limb of the anticline, and are therefore of particular interest for inter-
preting the geometry and mechanics of fold and joint development; their geometric
characteristics are used as a guide in the search for a cause of joint tilt.

Fig. 3.18.
Box plot for the tilt of the vec-
tor mean pole to joints from
44 joint populations in south-
dipping beds of the Lilstock
buttress anticline. The tilt is
measured downward in the
direction of the hinterland
(i.e., southern Europe) relative
to bedding when it is restored
to horizontal. The center line
of each box represents the
50th percentile of the data
while the left and right edges
of the boxes represent the 25th
and 75th percentile of the data,
respectively (from Engelder
and Peacock 2001)
 3.2 · Fracturing in The Bristol Channel Basin 227

3.2.5
Stresses in the South-Dipping Limestone Beds during Alpine Inversion

3.2.5.1
J3 Joints in the Lilstock Buttress Anticline

According to Engelder and Peacock (2001), the J3 joints in the south-dipping beds at
Lilstock Beach were formed during the development of the Lilstock buttress anticline,
reflecting Alpine inversion. They base this conclusion on six observations:

1. J3 joints are rare in the sub-horizontal beds of Lilstock Beach but are the most common
set in the steep limb of the Lilstock buttress anticline. The mechanism driving J3 jointing
must involve higher strain on the limbs of folds and little strain in the hinge area.
2. J3 joints are most common in the vicinity of fault 4, which was reversed to form the
Lilstock buttress anticline. This spatial association suggests that the formation of
set J3 with a high-angle reverse fault was synchronous with Alpine inversion.
3. Poles to joints in the southern limb of the Lilstock buttress anticline consistently
plunge toward the hinterland, their azimuth parallel to the inferred direction of
compression (019°) during Alpine inversion (Kelly et al. 1998). The consistent rela-
tion of tilt to bedding suggests that joints developed by shear couple on bedding.
4. The strike of J3 joints is nearly parallel to the strike of beds in the south-dipping limb
of the anticline. Strike-parallel joints are commonly associated with outer-arc exten-
sion by bending of beds during folding (e.g., Van Hise 1896; Hancock 1985) (Fig. 3.19a),
but outer-arc extension by tangential longitudinal strain folding is greater in the hinge
area and absent at inflection points in the limbs (Price and Cosgrove 1990).
5. J3 joints cut completely through beds and thus differ from joints that form by outer-
arc extension above a neutral fiber (Van Hise 1896).
6. J3 joints are not present in beds with J1 or J2 joints, and vice versa (Table 3.1). What-
ever mechanism was responsible for J3, it is clear that they propagated neither in
sub-horizontal beds, nor in previously jointed beds.

Alpine inversion was driven by a remote SHmax directed about 019° (Kelly et al. 1998).
This presents a mechanical paradox if J3 joints propagated during parallel folding, be-
cause they opened in a direction that nearly coincides with that of SHmax. If controlled
by the remote stress, J3 joints would open in a plane that contains SHmax. Engelder and
Peacock (2001) therefore suggested that the opening must be controlled by a local stress.
Accordingly, they sought a set of boundary conditions that generate a local tensile stress
on the limestone-shale package when a regional compressive stress is superimposed of
the Lilstock buttress anticline.
To understand the tilt of joints in the south-dipping beds of the Lilstock buttress anti-
cline and their ability to open sub-perpendicular to SHmax, Engelder and Peacock apply a
simple finite element model, which involves the development of a shear couple on lime-
stone beds during parallel folding by flexural flow of the shale interlayers. If there indeed is a
genetic relationship between the orientation of J3 joints and the shear couple on limestone
beds, the model proposes a possible orientation for SHmax that fits these boundary condi-
tions. To anticipate their result, remote SHmax controlling J3 joints should be the same direc-
tion as the remote SHmax that caused the conjugate strike-slip faults during Alpine inversion.
228 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.19. Fold mechanisms. a Outer-arc (buckling or tangential longitudinal strain), showing outer-
arc stretch, inner-arc contraction and no distortion on limbs. b Flexural slip, successive layers are
displaced upwards towards the anticline crest with respect to the layer below. Individual layers are
unstrained. c Flexural flow or flexural shear, showing no shear on hinge and shear within beds on the
limbs (Fig. 3.19a–c from Park 1983, p. 72). d A three-bed model of the limb of the Lilstock buttress
anticline contoured for σ3 (compressive stress is negative). The top surface is loaded as indicated
with the bottom boundary fixed and the left and right boundaries free to shear. The thin middle bed
is a 30 cm limestone with elastic properties as indicated. In close-up inset the orientation of the local
principal stresses are indicated with σ1 = –27.4 MPa and σ3 = 0.4 MPa (limestone), and σ3 = –4.8 MPa
(shale). Tensile stress is indicated by a solid line and compressive stress is indicated by a dashed line
(after Engelder and Peacock 2001)

3.2.5.2
Stress Configuration in Limestone Beds of the Lilstock Buttress Anticline
during its Development as a Flexural-Flow Fold

According to Engelder and Peacock (2001), the Lilstock buttress anticline, a sequence
of interbedded limestones and shales, should display a significant contrast between
the elastic properties of the limestone and shale (e.g., Hatheway and Kiersch 1982).
 3.2 · Fracturing in The Bristol Channel Basin 229

This lithological pair is commonly associated with flexural slip folding (Fig. 3.19b) in
which the more brittle limestone beds slip on ductile shale interlayers (e.g., Ramsay
and Huber 1987). Displacement also occurs by layer-parallel shearing of the shale
during flexural-flow folding (Fig. 3.19c) (flexural flow is also termed flexural shear,
e.g., Twiss and Moores 1992, Fig. 12.7). Often, the shale interlayers are thin as com-
pared to the carbonate beds, but at Lilstock the opposite is the case. Engelder and Pea-
cock suspect that thick shale layers obscure all evidence for slip by simple shear strain
within the shale, if present, largely because the shear strain could be relatively small
and because such strain markers as stretched ammonites are rare. Nevertheless, the
regional SHmax that tilted the beds of the anticline induced a shear couple in them (as-
suming for simplicity that SHmax and Sv are principal stresses). This assumption of a
remote horizontal principal stress is supported by horizontal slip on the conjugate
strike-slip faults active during Alpine inversion and by slickenside lineation orienta-
tions on the inverted normal faults (Kelly et al. 1998).
Engelder and Peacock (2001) mapped the stress fields induced in an elastic me-
dium under loading configurations consistent with a shear couple developed during
layer-parallel shear, applying 2D finite element analyses, using the interactive fracture
analysis program FRANC (Wawrzynek and Ingraffea 1987). They used a model mesh
of quadratic, isoparametric elements (Fig. 3.19d) applying boundary conditions to
simulate a three-layer model of a limestone bed of the Blue Lias between two shale
layers. Figure 3.19d shows a contour map of the maximum tensile stress (i.e., σ3 with
tensile stress positive) generated in beds dipping at 17° and subject to a stress ratio
(i.e., R = SHmax / Sv) of 1.66. Their model has a length to height ratio of 5, which is suf-
ficient to carry end conditions away from the central part of the model, which por-
trays a 3 m2, centrally situated square element. The initial versions show that the elas-
tic contrast between the limestone and shale beds leads to a tensile σ3 within the lime-
stone layers for values of R between one and four and νsh > νls (Poisson ratios of shale
and limestone, respectively). Tensile stress is generated via the flexural flow as per-
mitted under the tractions specified in Fig. 3.19d. Fixing the ends of the model to pre-
vent layer-parallel shear strain would of course be inconsistent with folding by flex-
ural flow (Price and Cosgrove 1990) and it would not permit the generation of tensile
stresses in the stiff beds.
Not only is the local σ3 tensile in the stiff beds under conditions of flexural-flow
folding, but σ3 cants downward relative to bedding in the direction of the horizontal
stress (Fig. 3.19d). When local σ3 is tensile, joints propagate parallel to local σ1, which
is always compressive in the models (the dashed lines in Fig. 3.19d). Local σ1 refracts
slightly at the boundary between the limestone and the shale beds, becoming steeper
in the stiffer limestone bed.
Engelder and Peacock (2001) ran several versions of the model to test the effect
of R on the tilt of the local σ1, and hence the predicted attitude of J3 joints with re-
spect to bedding. As R is increased from 1 to 4, the local σ1 tilts at an incrementally
larger angle toward the foreland in both the shale and the limestone beds. Tensile σ3
in limestone beds becomes larger (0.5 MPa to 5.2 MPa) as R increases from 1 to 3. At
R ≈ 4, σ3 in the shale near the limestone bed also becomes tensile. When running the
same series of tests for beds dipping at 3°, 5° and 9°, the tilt of the joint plane toward
the hinterland (i.e., local σ1) increases for increasing R, with higher rates of increase
230 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

in the more steeply dipping beds. Engelder and Peacock also ran tests (not cited here)
to determine the effect of the Poisson ratio on the model. They concluded that ν was
smaller in limestones than in shales because more joints would otherwise occur in the
shales than in the limestones and because νls > νsh is inconsistent with in situ geophysi-
cal logging data.

3.2.5.3
Joint Propagation at Depth Not by Means of High Fluid Pressure

Joint development at depth is often attributed to high fluid pressures (Secor 1965; Bahat
and Engelder 1984) except in cases of composite layers with different elastic proper-
ties (Hobbs 1967; Gross et al. 1995). The above described model for the generation of
tensile stresses during flexural flow is a variation on the composite layer model first
proposed by Hobbs (1967). An important consequence of the model is that compres-
sive stresses may generate local tensile stresses in the stiff layers of a composite stack
where bounding surfaces are subject to shear traction, as would be required for flex-
ural flow folding. This mechanism, acting at depth, can produce jointing also in the
absence of high fluid pressure. Engelder and Peacock favor the flexural flow mecha-
nism over hydraulic fracturing because: (1) joints with tilt occur only in dipping rocks
of the fold limbs and (2) as predicted by their model, joints are not orthogonal to bed-
ding. Engelder and Peacock’s (2001) model for the Lilstock buttress anticline deter-
mines that joint (and vein) growth in the limbs of flexural-flow folds should have sev-
eral distinct characteristics. The joints should extend through the entire bed, should
be non-orthogonal to the bedding, and dip toward the anticlinal axial plane, while the
poles tilt away from it.

3.2.5.4
Stress Ratio and Frictional Slip

Laboratory values of friction (Zoback and Healy 1984) predict that fault slip occurs
when R = 3, where R is the ratio of the remote principal compression SHmax divided by
the global vertical stress Sv. Engelder and Peacock (2001) use this stress ratio and the
hinterland tilt of joints to predict the dip of bedding at the time of the onset of joint-
ing. Their model predicts that J3 joints with a tilt of about 4° should have propagated
by the time beds had been upfolded to have a dip of 3°. If in the course of Alpine in-
version faults slipped under low friction, R could have been <3. For example, if R = 1.33
due to low-friction faults, a hinterland tilt of 4° could cause jointing in beds dipping
as much as 17°. They suggest that the flexural flow model may be applied outside the
immediate area of the Lilstock buttress anticline. During the extensional phase of the
Bristol Channel tectonics, large blocks throughout the basin were slightly tilted (Dart
et al. 1995), entering the Alpine inversion phase when already possessing a modest dip.
The model allows tensile stress to be generated at dip angles smaller than 2°, particu-
larly since the Poisson ratio of limestone is lower than that of shale. Many of the jointed
strata cropping out on the north and south shores of the Bristol Channel have a gentle
dip that would favor joint formation if submitted to inversion tectonics. Joints not
orthogonal to the bedding have been recorded at Lavernock Point, South Wales, by
Rawnsley et al. (1998).
 3.2 · Fracturing in The Bristol Channel Basin 231

3.2.6
Conclusions by Engelder and Peacock (2001)

Engelder and Peacock (2001) established the presence of six joint sets in the limestone
beds of the Jurassic Blue Lias, numbering them J1 to J6. They all cluster about the axial
strike of the 290° striking Lilstock buttress anticline, including J1 (330–310°), J2 (310–295°),
J3 (295–285°), J4 (285–275°), J5 (275–265°), and J6 (<265°). In sub-horizontal beds, in-
cluding those near the hinge surface of the Lilstock buttress anticline, J3 joints are rare.
Abutment of joints belonging to J2, J4, J5 and J6 indicates a counter clockwise sequence
of development. In the south-dipping beds of the anticline, J3 joints are most frequent
in the vicinity of the reverse fault responsible for the anticline. J3 joints are absent
where J1 and J2 joints are present and diminish with distance from the anticlinal hinge
and higher in the stratigraphic section. J3 joints in the south-dipping beds are not
perpendicular to the bedding but have poles that plunge several degrees towards the
south when bedding is restored to horizontal.
According to Engelder and Peacock, the mechanism that formed the J3 joints must
explain:

1. Their tilt relative to bedding on the steep south flank of the anticline.
2. Their greater frequency in the south-dipping limb as compared to the axial region.

These two characteristics strongly suggest that the J3 joints were formed during
folding, but before beds reach their present tilt. These conditions are met in their shear
model for interlayered limestone-shale beds during flexural flow folding (Fig. 3.19c),
where local tensile stresses are generated within a regional stress field capable of driv-
ing inversion tectonics. Such tensile stresses will create through-cutting joints rather
than arresting at a neutral surface associated with outer-arc extension (Fig. 3.19a). In
this model, joints can open against the direction of the maximum compressive stress
under conditions where external tractions are compressive. This indicates that at least
some of the joints at Lilstock formed during the main Alpine inversion (cf. Rawnsley
et al. 1998). It also shows that joints that are parallel to a fold hinge may form on the
limbs during flexural-flow folding and not in the hinge region as a consequence of
outer-arc extension during tangential longitudinal strain folding.

3.2.7
Discussion by Engelder and Peacock (2001)

3.2.7.1
The Problem of Delayed Fracture

Both Rawnsley et al. (1998) and Engelder and Peacock (2001) accept that jointing in
the Jurassic (Blue Lias) rocks at Lilstock is associated with Late Oligocene to Miocene
Alpine compression, i.e., that there was a long time interval between deposition of the
limestones and their eventual fracturing. Very often single-layer joints form in sedi-
mentary rocks during their burial stage. There must be a reason why jointing at Lilstock
skipped the burial and did not develop during this long time interval. A similar ques-
tion applies to the “delayed fracturing” in the Middle Eocene chalks of the Beer Sheva
232 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

syncline, Israel (Bahat 1999a), where the time interval between deposition and fractur-
ing, all within the Tertiary, was much shorter than at Lilstock Beach. At Lilstock Beach,
fracturing began in the syntectonic stage (or during relaxation of tectonic loading). With
few notable exceptions (e.g., Becker and Gross 1996; Weinberger 2001a) systematic joints
are rare in limestone and dolomite strata in anticlines in Israel (unpublished data by Bahat).
What is special about these exceptions, for instance, the Tully limestone (e.g., Younes
and Engelder 1999) that contains systematic joints? Why are systematic joints uncom-
mon in the Upper Cretaceous chalks of South England (Bahat 1991a, Sect. 5.3.3), while
they are very abundant in Santonian and Eocene chalks in Israel? Research combining
limestone petrology and structural geology may shed more light on these questions.

3.2.7.2
Opening of Joints Counter to Remote Compression, and Deviation
from Perpendicularity to Layer Boundaries

Engelder and Peacock (2001) found that J3 joints open against the remote compressive
SHmax instead of the expected opening in a plane that contains SHmax. They consider this
to be a “mechanical paradox” and constructed a flexural flow model to explain it. We
propose an alternative mechanism of local stress field rotation that would produce frac-
ture patterns similar to those described by Engelder and Peacock.
Previous investigators have also reported on opening of joints against the remote com-
pression. Dieterich and Carter (1969) investigated a numerical model of the stress field in
two-dimensional, large-amplitude folding of a viscous layer in a less viscous matrix, using
a modification of the finite element method. They found that large magnitudes of σ1 act
parallel to the layer throughout the early stages of folding. With increasing amplitude,

Fig. 3.20.
a A sketch showing geometry
of folded quartzite layer. Dots
indicate locations of specimens
used for dynamic analysis of
quartz lamellae. Long arrows
show orientation of σ1, and
short arrows show orienta-
tion of σ3 (all principal stres-
ses are compressive and posi-
tive) (after Dieterich and Carter
1969). b A schematic symmetric
fold, showing basic four joint
systems, F1–F4. Each system
ideally consists of three sets.
Fractures from the systems F1
and F2 are more common than
others, and in the latter two,
the sets co-axial with the prin-
cipal stresses are more com-
mon than the inclined sets
 3.2 · Fracturing in The Bristol Channel Basin 233

σ1 decreases in magnitude along the limbs and rotates into increasingly higher inclina-
tions to the layer. In the upper hinge area, σ1 decays rapidly with increasing amplitude,
and the deviatoric stresses parallel to the layer rapidly become low-magnitude tensile
stresses. Their analyses generally show that in the limbs of open folds, σ1 is inclined at low
angles to the layer, σ2 is parallel to the fold axis, and σ3 is nearly normal to the layer
(Fig. 3.20a). Figure 3.20a also implies that the stress axial cross may deviate from orthogo-
nality with respect to the bedding. Thus, the fractures that may develop during the local
rotation of the principal stresses need not be normal to the layer boundaries.
According to Dieterich and Carter (1969), similar results were obtained by other stud-
ies (Hansen and Borg 1962; Carter and Friedman 1965; Scott et al. 1965), which focused
on intragranular flow of calcite and quartz during folding. Burger and Hamill (1976,
Fig. 14A) performed petrofabric stress analysis on the Dry Creek Ridge anticline in
Montana and also found rotation of principal stress during folding. Initially, the entire
multilayer package was subjected to compression with σ1 parallel to layering and or-
thogonal to the fold axis. As shortening continued, a neutral surface formed, dividing
the hinge and upper limb areas into domains. The compression domain implies
σ1 parallel to layering and orthogonal to the fold axis, and the extension domain im-
plies σ3 parallel to layering and σ1 orthogonal to layering.
Reches (1976) identified five joint sets (J1–J5) that cut two asymmetric anticlines
in southern Israel, the Hathira and Hazera monoclines. None of the sets display sym-
metry with the fold axes, and as such they differ from the idealized joint geometry
suggested by Stearns (1968) for symmetric folds (Fig. 3.20b). Four sets (J1–J4) are
normal to the bedding and are thought to have developed prior to or at the beginning
of the folding, and their present orientation reflects their rotation by folding. On the
other hand, joint set J5, considered to be related to advanced folding, is inclined to the
bedding. Its average strike is 030°, which is sub-parallel to the trend of the monoclines,
parallel to the bedding and normal to the axes of the monoclines. The deviation of
the five joint sets (J1–J5) of the Lilstock Beach anticline from normality to the bed-
ding resembles that of the J5 joint set from Hathira and Hazera. Tilted joints were
observed in the Santonian chalks from the “Arad syncline, Israel. These are apparently
related to an intense deformation phase of the chalks (Bahat 1991a, p. 259). It appears
that all the above-mentioned tilted sets in both anticlines and synclines were formed
during advanced folding, while rotation of the local principal stresses underwent ro-
tation. Hence, the local rotation of the principal stresses during advanced folding in
Dieterich and Carter’s model and in field observations by others point to an alterna-
tive mechanism to the flexure flow model (Fig. 3.19c) proposed by Engelder and Pea-
cock (2001). This mechanism is a more general one and may operate in formations
that do not necessarily consist of alternating lithologies (e.g., limestone and shale). It
may also explain why there is no actual evidence of slip by simple shear strain, as ex-
pected in the flexure flow model within the shales at Lilstock. Finally, Engelder and
Peacock (2001) argue that in their model, tensile stresses will drive joints completely
through beds, and will not be arrested at a neutral surface in the outer-arc extension
model (Fig. 3.19a). Chinnery (1966), on the other hand, maintains that a tensile frac-
ture, once started, could continue to propagate even where entering a region of com-
pression. Thus, the flexure flow model is not the only mechanism to explain the for-
mation of set J3 at Lilstock. For a discussion on fracture in symmetric folds and in
slightly folded structures, see Sect. 3.4.
234 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

3.3
Post-Uplift Exfoliation Joints and Arch Formation in Sandstone
at Zion National Park, Utah, USA
3.3.1
Introduction

3.3.1.1
General

Exfoliation in the massive sandstones of the Colorado Plateau, and at Zion National Park
in particular, has drawn the attention of many investigators (e.g., Gregory 1950; Holman
1976). Exfoliation in rocks has been debated since Gilbert (1904) proposed a connection
between sheeting and the expansion of the rock body in response to unloading by ero-
sion. Several natural mechanisms have been suggested, including contraction of magma
during cooling, regional tectonic stresses and residual stresses, fluctuations of ground
and surface temperature and differential weathering (Johnson 1970; Holzhausen 1989).
Various stress conditions have been proposed to explain the formation of exfolia-
tion joints. The stress conditions close to the ground surface, particularly the high
compressive stresses that parallel the exfoliation, are of great importance (Dale 1923,
p. 34; Holzhausen 1989). Fracturing by residual compressive stresses (Bradley 1963;
Holman 1976) develops when critical stress difference occurs between the principal
stresses parallel to the surface (e.g., White 1946) and if these are sufficiently high so
that additional stress will fracture the rock (Cadman 1969) (for further discussion on
exfoliation, see in Holzhausen (1989). Although it is clear that exfoliation in Zion Na-
tional Park generally parallels earlier fractures, it has been debated whether exfolia-
tion joints curve to remain faithful to the new valley sides, independent of the earlier
regional jointing (Bradley 1963), or they rather follow the trends of prominent earlier
vertical prominent joints (Gregory 1950).
Sandstone arches in Zion National Park were considered by Gregory (1950) to be
primarily the result of erosion. Cruikshank and Aydin (1994) explained the arches of
Arches National Park in southeastern Utah as products of localized erosion that oc-
curred at zones of intense fracturing due to shear on existing discontinuities. Robinson
(1970) explained the formation of large arches in Zion Canyon by modeling the arch
as a plate with a semicircular hole, which is subjected to uniaxial vertical compres-
sion. When the rock above the center of the arch is under tension, it would tend to fail
along curvilinear stress trajectories.
Apart from scientific interest, exfoliation must be reckoned with in engineering and
architecture. Exfoliation results from a rock’s tendency to fracture at the appropriate
stress, parallel to a previous surface, like cliffs or the topography. It is therefore an
important consideration in the construction and maintenance of roads, canals, dams
etc. Quarrying and pit mining deal with it daily. Carefully planned excavation, taking
account of exfoliation, can overcome some acute safety problems and prevent economic
losses (Niles 1871; Cadman 1969; Holzhausen 1989). Knowing a terrain’s exfoliation
features is essential to hydrological projects of waste disposal and remediation (e.g.,
Cohen 1993). A distinction is made between the terms “exfoliation” and “sheeting” (see
end of chapter). Section 4.5.3 deals with post-uplift exfoliation joints in granites.
 3.3 · Post-Uplift Exfoliation Joints and Arch Formation in Sandstone at Zion National Park 235

3.3.1.2
Geologic Setting

Zion National Park, a fracture province marked by a multitude of exfoliation joints and
arches, is located in southwestern Utah, between the Hurricane and the Sevier fault zones
(Fig. 3.21a). This area occurs within the transitional region between the Colorado Pla-
teau in the east and the Basin and Range province in the west (Grater 1945; Kurie 1966;
Stewart and Taylor 1996). Mesozoic rocks of the Colorado Plateau in southern Utah bear
structural evidence of three major regional tectonic events (Davis 1999). The first de-

Fig. 3.21. a Simplified structural map of the Colorado Plateau (thick lines). Thin lines represent ma-
jor faults; ZP designates Zion National Park (modified after Hunt 1956). b Stratigraphy of the Navajo
sandstone in the Zion National Park area (modified from Gregory 1950)
236 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

formation was during the Laramide orogeny, from ca. 90–50 m.y., when an ENE-WSW
convergence between the North America plate to the east and the Farallon plate to the
west generated horizontal compression that contracted the Colorado Plateau region.
The second deformation took place during the Miocene, from ca. 25 to 19 m.y., when
the rate of plate-tectonic convergence had slackened from ~100 mm yr–1 during the
Laramide orogeny to ~50 mm yr–1, and the angle of subduction was becoming steeper.
The third deformation was due to the Basin and Range extensional faulting, which be-
gan at about 15 m.y. and is still active today. This extension produced three major high-
angle normal faults along the southwestern margin of the Colorado Plateau. They are
(from west to east): The Hurricane, Sevier and the Paunsagunt faults.
Two regional joint sets striking about NW-SE and NE-SW are known from the Colo-
rado Plateau in northeast Arizona and in southeast Utah (Hodgson 1961b). Three Neo-
gene sets striking 335° (N 25° W), 355° and 030° were identified by Bergerat et al. (1992)
in the massive sandstones of the Colorado Plateau. Gregory (1950) observed two series
of regional joints striking NNW and ENE in the Zion Park region. No clear correlation
can be established between the various joint sets and any of the above major deforma-
tions. The present section concentrates on a series of exfoliation joints in the Navajo
sandstone associated with the NNW and ENE sets observed by Gregory (1950). The
Navajo sandstone is about 400–600 m thick in the park area, and is thought to be of
Middle Jurassic age (Fig. 3.21b). The rock is quartz-rich, massive, friable, fine-grained,
and equigranular (Grater 1945; Gregory 1950). Various investigators (e.g., Gregory 1950;
Bradley 1963) have studied the exfoliation joints in the massive sandstones of the Colo-
rado Plateau and at Zion National Park in particular. Bradley considered the exfolia-
tion to be of Pleistocene age. Based on studies by Bahat et al. (1995, 1999) and Grossen-
bacher et al. (1996), we characterize below the initiation and growth stages of exfolia-
tion joints in relation to previous regional joints. The role of exfoliation fractures in
arch formation is explored, including a mechanical model of arch formation.

3.3.2
Fracture Characteristics at Zion National Park

3.3.2.1
Exfoliation Joints at Zion National Park

Two types of exfoliation joints were discerned at Zion National Park. The first type con-
sists of large fractures that form vertical cliffs. These joints occur mainly along straight
tributaries, mostly striking NNW (Fig. 3.22). The tributaries trending NNW are sub-
parallel to each other. Most of them run for 2 000 m or more, and many of them are
deep and narrow canyons only tens of meters across with straight walls hundreds of
meters high (USGS 1 : 24 000 topographic maps of the Zion Park area, 1980), displaying
extensive exfoliation (Fig. 3.23a). Many narrow canyons are elevated from the Virgin
River and form large stairs down to the main canyon. They belong to the more gener-
alized regional pattern of prominent fractures that mostly strike NNW or NE at Zion
National Park (e.g., Gregory 1950). The second type of exfoliation joints occur in vari-
ous sizes and orientations that conform to present-day surface morphology (e.g.,
Fig. 2.33e). Joints of this type occur in various parts of the main canyon as well as along
the tributaries.
 3.3 · Post-Uplift Exfoliation Joints and Arch Formation in Sandstone at Zion National Park 237

Fig. 3.22a.
Tributary map in Zion National
Park, Angle Arch, Red Arch,
Lady Mountain, Weeping Rock,
Hidden Canyon, Springdale and
Zion Lodge are designated re-
spectively, by A.A., R.A., L.M.,
W.R., H.C., Sp. and Z.L. Dash-
lines represent tributaries (after
Bahat et al. 1995)

Fig. 3.22b. Histograms of parameters on exfoliation joints along approximately straight tributaries
in Zion National Park. A Lengths along N-NNW tributaries. B Azimuths and cumulative lengths of
N-NNW tributaries. C Azimuths and cumulative lengths of NE tributaries. D Spacing of N-NNW tribu-
taries (after Bahat et al. 1995)
 238
Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.23. a The lower part of a vertical NNW exfoliation joint in Hidden Canyon, showing two vertical remnants of exfoliated slices (marked A), several meters high
and up to 1 m thick. Such slices occasionally show en echelon relationship. Right-stepping en echelon segments cut the exfoliation at upper center (B). At the center is
another third en echelon array (C) with inclined stepping, tens of cm long and tens of mm thick, spreading upward from a single center. Note sub-horizontal bedding
of the sandstone, which does not interfere with en echelon cracks and exfoliation. b Interaction of adjacent quasi circular/elliptical cracks of different size that grew
independently and impinged on each other on an exfoliated joint. c Diagram of b. Fracture A is ornamented by several long vertical striae that arrest at the en echelon
fringe below A. Fracture B is a combination of two smaller fractures and the en echelon fringe below B. Three smaller fractures, x, y, z, lie along the contact between
A and B. Additional fractures around A and B are marked by letters m–p. The height of ellipse B is on the order of 4 m (after Bahat et al. 1995)
 3.3 · Post-Uplift Exfoliation Joints and Arch Formation in Sandstone at Zion National Park 239

3.3.2.2
Circular/Elliptical Cracks and Composite Fractures

Joints of the first exfoliation type are often marked by quasi-elliptical or quasi-circu-
lar cracks (QCC), which mostly range from one meter to several tens of meters in di-
ameter, occasionally interacting with each other by partial penetration (Fig. 3.23b,c).
They grow approximately coplanar to each other and often merge in a variety of styles,
producing composite fractures (Bankwitz and Bankwitz 1984; Helgeson and Aydin
1991; Bahat 1991a, p. 268). Some composite fractures form either partial or full ellipses
or circles, occasionally rimmed by en echelon fringes (Fig. 3.24a,b).

Ripple marks. Ripple marks are of two types: Undulations and arrest marks
(Sect. 2.2.4.2). Occasionally, undulations occur on large exfoliation joints unrelated to
QCC. A spectacular series of more than forty undulations (Fig. 3.24c) occurs on the
western wall of the “Hidden Canyon” (Fig. 3.22a), where they cover a cliff surface tens
of meters wide, and about as high. This feature marks the propagation of a large, con-
centric, single-fracture front that is convex to the north. The undulations show that the
joint grew horizontally from south to north. They also show how rock anisotropy influ-
ences fracture propagation: The fracture cuts across bedding, but one boundary was
resistant, and refracted the fracture front into another direction. Arrest marks propa-
gate in all directions but seem to advance particularly downward (Fig. 3.24d). Arrest
marks also occur on fractured side-walls of arches (Fig. 3.24b).

Fig. 3.24a,b. a The Angel Landing Arch. b Scaled photograph of the same arch, showing detailed mea-
surements. Many circular/elliptical cracks are surrounded at left by a fringe of en echelon radial cracks.
Note vertical discontinuity (at left of points 1 and 4), and arrest marks on left side-wall (three arrows).
(after Bahat et al. 1995)
240 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.24c,d. c A central part of a nearly planar, vertical NNW-striking exfoliation joint. Undulations sev-
eral tens of meter long convex towards right (north). The fracture front is interpreted to have changed its
propagation direction at a resistant bedding boundary (inclined arrow). The occurrence of two superpos-
ing exfoliation joints with similar ripple marks is shown by two sub-horizontal arrows. Letters A–D relate
to locations of changes in fracture velocity. d Different characteristics of ripple marks, undulations – ver-
tical arrow, and arrest marks – inclined arrow (Sect. 2.2.4.2) (after Bahat et al. 1995)
 3.3 · Post-Uplift Exfoliation Joints and Arch Formation in Sandstone at Zion National Park 241

Radial plumes (striae). Quasi-circular cracks may be decorated by radial striae that
indicate the sites of fracture initiation and illustrate the fracture growth mode. Occa-
sionally, radial striae decorate a niche (see below) and constitute the only part re-
maining of the QCC. Striae are not common. They occasionally occur either as straight
individuals, often in radial setting, or as straight strands bent in places (Fig. 3.25a,b).

Fig. 3.25. a A series of large en echelon cracks on a NE striking cliff at Lady Mountain. Note previous
exfoliation slices at left. b Scale diagram of a, showing crack dimensions. A and B are series of upward
convex niches. Radial striae are marked at B. C and D are two sub-horizontal, straight niches formed by
a series of en echelon fractures, propagating leftward and producing partial arches. c Exfoliations on a
cliff, exhibiting a series of small niches (at center), with a large niche or a small arch about 20 m long
above (arrow). A series of partial arches is seen at lower-center of picture (after Bahat et al. 1995)
242 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

3.3.2.3
En Echelon Cracks, Niches and Arches

En echelon cracks. Many fractures deviate a few degrees from the cliff plane, and ap-
pear as en echelon segments, ranging in size from tens of cm to tens of meters. The en
echelon segments occur in various shapes and orientations. Occasionally three or more
en echelon sets of different sizes and orientations occur in close proximity (Fig. 3.23a).
Other cliffs exhibit large en echelon fractures next to niches (Fig. 3.25a,b) or along the
mirror boundary (Fig. 3.24a,b).

Fig. 3.26. a The Red Arch, showing sharp fracture contours in the upper parts (x). Fringe of en echelon
segments (two arrows) marks the lower boundary of the quasi-circular arch and at lower-left side some
remnants of exfoliated slices. Scale bar is about 15 m. b Scale diagram, showing measured dimensions
of a. A, B and C contours describe earlier large and small “attempts” of arches, while D follows the con-
tours of the main, Red Arch quasi-circular contours, surrounding the large, flat back-wall at the center.
This back-wall is a mirror plane. Numbers I, II and III mark the upper boundary of the lower fringe of
the mirror, and IV and V occur on the lower boundary of the fringe. x in a and b depict the same spots.
c The fringe in close-up (after Bahat et al. 1995)
 3.3 · Post-Uplift Exfoliation Joints and Arch Formation in Sandstone at Zion National Park 243

Niches. Occasionally, individual QCCs or composite fractures cut into the rock and
deviate several degrees from the cliff face. At these localities, receding bays formed in
the cliff are termed niches (Fig. 3.25a–c). The niches commonly appear in crescent
shapes that concave downward (distinguished in the outcrop by their shades). Occa-
sionally, they tend to be straight. Sizes vary from several meters to several tens of
meters. Two types of niches, A and B, are distinguished. Niches of type A occur close
together at about the same topographic elevation (Fig. 3.25b). Type B niches are
grouped one above the other. Adjacent niches in an outcrop may show shape and size
transitions to “small arches” (Fig. 3.25c). The majority of QCCs, composite fractures,
en echelon segments and niches occur within the middle third of the cliff face
(Fig. 3.24a,b, 3.25a,b, 3.26a,b and 3.27).

Fig. 3.27.
Side-wall of a partial arch with
convex ripple marks indicating
downward fracture propaga-
tion. The upper mark coincides
with a bedding boundary, which
influenced its location (below
arrow). See a rappelling person
(arrow) for scale (after Bahat
et al. 1995)
244 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Arches. Arches are found by the hundred in Zion National Park in a variety of sizes
(Table 3.2) and shapes (Fig. 3.26–3.28). Arch sizes range from several meters to many
tens of meters. Some are full arches (Fig. 3.26a,b and 3.28a), either symmetric or asym-
metric (Fig. 3.28a); others are incomplete arches (Fig. 3.27 and 3.28b). A full arch ex-
hibits a ceiling and two side walls. Height is measured from base to the highest point on
the ceiling, and the width is measured between opposite side walls at the base. In some
arches, width is larger than the height (Fig. 3.28a), in others this relationship is re-
versed (Fig. 3.24a,b). Sub-circular arches (Fig. 3.26a) are rare. A partial arch consists of
a side wall and a fraction of ceiling (Fig. 3.27 and 3.28). Many ceilings are not fully
concave, but partly coincide with straight, sub-horizontal niches or other bedding ori-
ented boundaries (right side of Fig. 3.24a and 3.32b). Occasionally vertical fractures
cut ceilings of full arches or form boundaries of partial arches (left side of Fig. 3.24b).

Fig. 3.28. a Diagram showing elements of an asymmetric arch as follows. Side-walls S, ceiling C of
the arch (shown between two arrows that indicate thickness on both sides of the arch), the vertical
fracture V cuts the arch at the ceiling, and exfoliations e divide the rock into slices. The height of the
arch H, and the width, W. N1, N2 and N3 are niches associated respectively, with external exfoliation,
at the ceiling, and cutting the back-wall of the arch. Two QCC, marked E, occur on back-wall, bw. For
simplicity, niches are mostly crescent and not straight. b Diagram of a side-wall of a partial arch

Table 3.2. Measured parameters of arches in Zion National Park

 3.3 · Post-Uplift Exfoliation Joints and Arch Formation in Sandstone at Zion National Park 245

We interpret the external exfoliation above the arch (N1 in Fig. 3.28a) to be a relict
of an earlier cliff face. N3 marks the back wall of the arch, which is the youngest,
internal exfoliation joint (Bahat et al. 1999) (Fig. 3.26a and 3.28a). Side walls and
ceilings of arches penetrate into the cliff, perpendicular to the external and internal
exfoliations and connect them. Side walls and ceilings of arches are generally angular
toward the external exfoliation, indicating that the side-walls and ceilings were formed
by brittle fracture rather than by erosion. Also, concentric arrest marks on side-walls
(Fig. 3.24b, 3.27) are interpreted to indicate that the arches were formed as mode I frac-
tures (Bahat 1979a). The lower parts of arches have various shapes. In some arches,
the side walls reach close to the ground surface, whereas in others (Fig. 3.26a) they
merge into the cliff. The Red Arch is a sub-circular arch, and the lower boundary
of the mirror plane (that coincides with the back-wall) is marked by a concave-up
en echelon fringe (Fig. 3.26a–c). The rock-slice that separates between the external
and internal exfoliation joints has remained partly in place, closing the lower left side
of the arch.

3.3.3
Fracture Mechanisms at Zion National Park

3.3.3.1
Circularity of QCCs on Principal Planes

When tensile stress operates normal to an elliptical crack, it tends to grow into a cir-
cular fracture because the stress intensity KI is greater at the end of the minor axis of
the ellipse. This trend toward greater circularity is apparently the cause of the com-
mon occurrence of QCCs and some composite circular fractures (Fig. 3.23c and 3.24).
The planarity of both isolated and linked QCCs is consistent with mode I fracturing,
with the fracture planes essentially normal to the least compressive stress.

3.3.3.2
Fractography of Exfoliation Joints

The QCCs at Zion National Park resemble the fan-shaped early cracks of the El Capitan
granite at Yosemite National Park, which merge into exfoliation joints (Fig. 4.27). The
merging fracture styles at the two localities in different rock types is characteristic of
post-uplift joints and was first described as such from Eocene rock strata near Beer
Sheva (Bahat 1991a, p. 301). The merging style is unknown on surfaces of earlier burial,
syntectonic or on uplift regional joints (Bahat et al. 1999). Hence, one good criterion
for distinguishing exfoliation joints from other joint types cutting sandstone is their
specific fractography of QCC merging (Fig. 3.23b,c and 3.24b).

3.3.3.3
The Growth of Quasi Elliptic Cracks to Composite Exfoliation Fractures

The growth of QCCs into composite exfoliation fractures proceeds through four steps
(Fig. 3.29). They begin as QCCs (as shown in Fig. 3.23b,c and upper part of Fig. 3.29a),
or as early fans in granodiorite of El Capitan at Yosemite National Park (Fig. 4.27 and
246 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.29a–d. The development of composite fractures from early cracks (see text for explanation)

lower part of Fig. 3.29a). These cracks grow into elliptic fractures, with their long di-
ameter either horizontal or vertical (Fig. 3.29b). The elliptic fracture grows into a cir-
cular shape (Fig. 3.29c) by the mechanism described above (Sect. 3.3.3.1). Finally, the
fractures grow into a composite exfoliation joint (e.g., Fig. 3.24b and Fig. 3.29d), where
they may attain critical conditions (around KIC, Fig. 4.34 and 4.35), leading to the de-
velopment of an en echelon fringe (see lower left side of composite in Fig. 3.26). The
latter is also seen at Half Dome at Yosemite National Park (Bahat et al. 1999) (Fig. 6.18).

3.3.3.4
The Evolution of Arches through Niche-Forming Exfoliation

We favor Robinson’s (1970) explanation of arch formation by a mechanical process.

Figure 3.30 shows the stress trajectories calculated around a circular hole in a plate of
elastic rock subjected to uniaxial compression (Means 1979, p. 118). The σ1 stress tra-
jectories surrounding the hole resemble magnetic force lines around a “circular” mag-
net or hydraulic streams around a circular obstacle. When a vertical plate with a cir-
cular hole is subjected to a biaxial stress with vertical maximum compression σ1 ex-
erted by the weight of the overlying rock, the resulting strain may lead to an arch. In
an isotropic rock, tensile stress will concentrate on point A at the apex of the circular
hole, resulting in a vertical fracture normal to the stress trajectories of σ3 and to the
cliff. However, on many full arches (Fig. 3.24) and partial arches (Fig. 3.27) at Zion Na-
tional Park, vertical fractures (Fig. 3.28a,b) occur at locations like B and C, which are
not at the exact apex of a circular hole (Fig. 3.30). Due to the imperfect isotropy of
natural rock, arch ceilings are often not rounded at their tops, but are partly shaped
by straight niches that follow essentially horizontal sedimentary bedding. Such influ-
ence of bedding on fracture growth is seen in many outcrops: Undulations deviate from
smooth curvature in Fig. 3.24c, the location of at least one arrest mark coincides with
bedding in Fig. 3.27, and niches that follow bedding are shown in Fig. 3.25a. That the
modification of arch curvature is caused by unequal strength of the rock is supported
by experiment. Robinson (1970) found four values for tensile strength of Navajo sand-
stone cores determined by uniaxial tension tests. The values found for two cores par-
allel to the bedding were 3.0 MPa and 1.2 MPa, while values for two cores normal to
the bedding were significantly lower, 0.5 MPa and 1.0 MPa. This shows that bedded
sandstone is not isotropic. Resistance to fracturing is greater across the bedding than
parallel to it.
Arch formation proceeds through several stages of tensile fracture (Fig. 3.31a–d).
Vertical fractures (Fig. 3.28a,b) form in zones of tensile stress, which is usually concen-
 3.3 · Post-Uplift Exfoliation Joints and Arch Formation in Sandstone at Zion National Park 247

Fig. 3.30.
Stress trajectories near a circu-
lar hole in a plate subjected to
biaxial compression. A, B and C
are possible locations of tensile
stress concentration (modified
after Means 1987) (see text for
explanation)

Fig. 3.31. The evolution of an arch from initial niches. a Two incipient niches. b A vertical fracture
forms in the apical part of a growing niche. c Vertical fractures grow at the apex and off apex, in
analogy to A and B from Fig. 3.30. d The two curves meet and create an arch circle. The area 2a can
form a mirror, rimmed by a fringe, like the back-wall of the Red Arch

trated at apices of niches or along their sides at their lateral continuations (Fig. 3.31a,b).
The vertical fractures propagate downward along the curvilinear stress trajectories σ1
(Fig. 3.30) almost completely separating area 1a from area 1 on the cliff (Fig. 3.31c).
This curvilinearity varies in shape between the quasi-circular geometry of the Red
Arch (Fig. 3.26) and the more elongated shape of the Angle Arch (Fig. 3.24), depend-
ing on local conditions. Such a fracture would also propagate into the interior of the
rock, perpendicular to the cliff surface, even where entering a region of compression
(Chinnery 1966). The deeper it propagates into the rock and the more exfoliation joints
it traverses, the thicker will become the arch. The thickness of the arch is therefore
independent of the depth of the initial niche. Arch thickness is also independent of
niche shape, whether semi-circular or straight, because the fracture follows the σ1 stress
trajectories. The ceiling, the side walls and the back wall are fractures, which together
with the cliff face separate the arched block from the surrounding sandstone, leading
to its collapse by gravity. An arch like the Red Arch (Fig. 3.26a,b) is formed when a
rock slice, 2a, is detached from the inner exfoliation and collapses (Fig. 3.31d). This
description means that arches may be formed entirely by fracturing, and undercut-
ting by erosion is not required in these cases.
248 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

The common occurrence of arches in anisotropic sandstone formations in Zion

Canyon and in other provinces (e.g., Fig. 3.32a,b) and their relative rarity in isotropic
granitic rocks (Sect. 4.5.3.5) favor the role of rock anisotropy in arch formation by ten-
sile fracturing (Fig. 3.31). Cruikshank and Aydin (1994) explained arch formation
in Arches National Park in southeastern Utah by localized erosion along zones of in-
tense shear fracturing. Such fracture zones are not found near arches at Zion Canyon,
so that this mechanism probably does not apply to these arches. We did not identify
the Zion exfoliation joints and other arch characteristics on the two arches found at
Timna in south Israel. These arches are poorly shaped compared with many arches at
Zion Canyon, probably because the Timna arches were formed by erosion and not by
fracturing.

3.3.3.5
Parallelism between Exfoliation Joints and Previous Regional Joints

The position of the QCCs at Zion Canyon strongly resembles that of the fan-shaped
early cracks which appear about halfway up the cliff of El Capitan in Yosemite National
Parks (Bahat et al. 1999 and Fig. 4.27). This suggests also similarity in the fracture
mechanism. At both localities, earlier cliff-parallel joints provided the free surfaces
needed for buckling and longitudinal splitting, which lead to joint exfoliation (Holz-
hausen and Johnson 1979). Exfoliation at Zion Canyon displays two directions, NW-
NNW and NE, occasionally producing square blocks (Fig. 3.32d). These directions corre-
spond to trends of earlier, regional, vertical joints (Hodgson 1961b; Bergerat et al. 1992).

Fig. 3.32a. Arch from thick Paleozoic sandstone, near Petra, south Jordan
 3.3 · Post-Uplift Exfoliation Joints and Arch Formation in Sandstone at Zion National Park 249

Fig. 3.32b,c.
Arches from: b An Arch and
niches from thick Jurassic
sandstone, in Chapada dos
Guimaraes, about 50 km NEE
of Cuiaba, central Brazil. Note
approximately straight ceilings
that conform to bedding;
c Sub-orthogonal jointing at
Zion National Park. The sur-
faces marked A and B strike
NE-ENE and NNW, respec-
tively. Note arrest marks that
convex downward (arrows)
250 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

3.3.4
Discrimination between Sheeting and Exfoliation

The terms exfoliation (Cadman 1969) and sheeting (Holman 1976; Twidale et al. 1996)
have been used interchangeably by many authors. Both sheets and exfoliation frac-
tures reflect fracture propagation in principal planes, normal to the minimum princi-
pal stress. However, they belong to genetically and morphologically different catego-
ries. Citing Gilbert (1904), who believed that sheeting was older than the topography,
we can add that exfoliation follows topography. Therefore, sheeting and exfoliation
correspond, respectively, to uplift and post-uplift fractures.
The term sheeting should be applied to series of horizontal and sub-horizontal
fractures that maintain approximately uniform spacing but generally with a trend
to increase their spacing with depth (Johnson 1970, p. 356). Various authors (in-
cluding Dale 1923 and Holzhausen 1989) distinguished “generations” of sheet frac-
tures (e.g., preglacial and postglacial). Holman (1976) further distinguished two to
four secondary sheets or slabs. Holzhausen (1989) notes that the sheet fractures
must be products of states of stress that are approximately uniform over distances
of at least tens to hundreds of meters in order to explain the regularity and size of the
fractures.
Exfoliation joints, on the other hand, are generated by fracturing parallel to exist-
ing surfaces at various orientations, particularly vertical. These joints generally start
to develop at a short distance (tens of centimeters), from the exposed surface and form
closely spaced sets in small spacing (e.g., Fig. 4.29), producing rock slices that vary
from prismatic to lenticular. Thus, whereas sheeting responds to unloading of the
overburden, exfoliation results from loading of the overburden.

3.3.5
Summary

The joints at Zion National Park, USA initiated as circular cracks and grew to di-
mensions as great as several tens of meters. Interaction between these cracks pro-
duced composite exfoliation joints tens to hundreds of meters across. Large con-
centric undulations on some of the large exfoliation fractures indicate a single
rather than a composite origin. Generally, exfoliation joints develop parallel and
adjacent to joints formed by earlier tectonics. Where the exfoliation joints bend
into the cliff, niches are formed, often having crescent shapes that concave down-
ward. Such niches are sites of stress concentration that leads to arch formation.
However, the essentially stratified structure of many sandstones causes horizontal
extension of the niches, displacing arch formation laterally from the point of
stress concentration at the original apex of the niche (Fig. 3.30). Thus, an arch-
bounding joint often starts tangent or vertical to a niche and propagates down-
ward, creating a curved surface that approximates the trajectory of σ1 that is induced
by the weight of the overlying rock. The propagating curviplanar fracture also
cuts into the rock body, traversing parallel joints that divide the rock into slices paral-
lel to the cliff. Fracture-bounded blocks that detach and fall from the rock body leave
an arch-shaped hollow.
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 251

3.4
Jointing and Faulting in Eocene Chalk Formations
in the Beer Sheva Syncline, Israel
3.4.1
The Beer Sheva Syncline

The Beer Sheva syncline is an asymmetrical fault-fold basin within the Syrian Arc
(Krenkel 1924), a sigmoid fold system that stretches from Syria in the north, through
Israel to Egypt in the south (Fig. 3.33a). The Beer Sheva syncline, approximately 100 km
long, is oriented about 050° (Fig. 3.33b,c). The width of the syncline is not well defined
on its western side due to a Cenozoic cover, but is estimated to be about 35 km or less.
The structural geology of the Beer Sheva syncline, particularly its northern part, which
is rich in Eocene chalk outcrops, was studied by Bahat (1985–2001). Research was also
carried out in the Shephela syncline north of Beer Sheva (Fig. 3.33b,c). Fractures, i.e.,
joints and faults associated with joints, were investigated in two formations: The Mor
Formation and the Horsha Formation, lower and lower-middle Eocene, respectively,
each of them about 100 m thick. The present account uses data from about one hun-
dred stations (Fig. 3.33d), quoting extensively publications by Bahat.
The Lower (or Early) Eocene Mor Formation, also termed Adulam (Fig. 3.34), is
exposed along the peripheries of the syncline. It consists of thin (40 to 90 cm) chalk
layers alternating with beds of chert nodules, up to 10 cm thick (Bahat and Grossmann
1988). Two types of joints are common: Single-layer (s.l.) joints and multilayer (m.l.)
joints, i.e., that cut through many layers (Bahat 1988a). The cross-fold s.l. joints of
set 328° (normal and sub-normal to the synclinal axis) arrest at the boundaries of the
chalk layers with chert beds. At one locality, a normal fault that displaces these joints
contains in its fracture zone fragments of chert that were broken during early offset-
ting of the fault, and unfractured nodules that cumulated after fault movement. The
solidification of chert (Knauth and Lowe 1978) is associated with the burial phase that
includes sedimentation, downwarping and diagenesis, all pre-dating syntectonic de-
formation and uplift. Chert does not occur in the overlying Middle Eocene strata
(Horsha Formation, also termed Maresha). Therefore, both the cross-fold joints and
the fault must have evolved before the Middle Eocene, under a tectonic regime that
prevailed in the Lower Eocene. The strike-parallel s.l. joints, oriented about 059°, were
formed partly intermittently with the cross-fold joints (Bahat 1991a, Fig. 4.8) and partly
at a later stage (Gross et al. 1997).
The Horsha Formation, which is exposed at the center of the syncline (Fig. 3.34),
differs from the Mor Formation by the absence of chert beds. Single-layer joints in
both formations show a general similarity, with several exceptions. Joints of the Horsha
Formation vary considerably in orientation, plumes on the joint surface are generally
more delicate (Bahat 1987a) and en echelon fringes are common (Bahat 1986a). In
addition, joints of the Horsha Formation show indications of “delayed fracture” (Bahat
1991a, p. 245, 1997).
The patterns of fold-joint relationships of the Beer Sheva syncline are consider-
ably more diversified than those described in the literature, although they doubtlessly
are to be found in other locations also. The detailed planning of environmental engi-
252 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.33a,b. a Location map. The N-S sigmoid trace delineates the Syrian Arc, and the rhomb marks
the investigated area in Fig. 3.33b–d. b Regional map, showing Eocene outcrops (gray) and major
structures of the Syrian Arc system in Israel and adjacent countries. The city of Beer Sheva is marked
BS (modified after Buchbinder et al. 1988)
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 253

Fig. 3.33c,d. c The city of Beer Sheva (BS) is located at the center, between the southern tip of the Shephela
syncline in the north and the northern part of the Beer Sheva syncline in the south, axes symbolized by
straight lines. Curved solid lines are boundaries of exposed Lower Eocene rocks (Mor Formation) after
Bentor et al. (1970) and dashed lines are approximate boundaries between the Lower and Middle Eocene
(Horsha Formation). Barbed lines are inferred faults south of Beer Sheva, after Gvirtzman (1969). See
references to stations X, Y, R and S in the text. d Station-map. Stations designated “+” are in the Mor For-
mation and those marked by “∆” are in the Horsha Formation. The ∆ stations from 70 to 95 are located
along Wadi Secher (Fig. 3.37). Stations 37–40 are on outcrops along Wadi Naim. Maps are marked by the
Israeli grid (modified after Bahat and Grossmann 1988) (modified after Bahat and Grossmann 1988)
254 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.34. Lithostratigraphy, biostratigraphy, time scale and eustatic sea-level curve of the Early (ap-
proximately Lower) to Middle Eocene with reference to the Shephela area. Legend: 1: Chalk; 2: Marl;
3: Chert beds or nodules; 4: Mass-transported units of chalk and chert (from Buchbinder et al. 1988)

neering projects requires basic data on the whole spectrum of structural relationships.
The objective of this section is to synthesize the new results with previous findings
and to report on the diversity of fracture in the Beer Sheva syncline. Also, the prob-
lems of joint set classification will be considered in some cases, and the various fac-
tors that could have contributed to the fracture complexity in this fold will be discussed.
The methods of investigation that were applied in the investigation of the syncline
are presented in Sect. 6.2, in connection with additional aspects of this topic.
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 255

3.4.2
Fractures in the Beer Sheva Syncline

3.4.2.1
General

Folding of the syncline is slight, and joints are mostly vertical or near-vertical. Twenty
three fracture sets were distinguished in the Beer Sheva syncline, which are arranged
in sixteen groups (Table 3.3). Each group commonly represents a set, except groups 1,
4, 7 and 11 that consist of two or more sets, which have specific genetic affinities. The
groups are assembled in four rock units (I–IV in Table 3.3) on the basis of stratigraphy,
lithology (Bentor and Vroman 1960; Braun et al. 1977; Buchbinder et al. 1988) and frac-
ture appearance (Fig. 3.35a). The joints in the Mor Formation are represented by unit I,
while units II, III and IV represent the joints of the Horsha Formation. Unit II is com-
posed of thick layers (≥5 m), while units I, III and IV are composed of thin layers (≤1 m).

Fig. 3.35. a Diagrammatic (not to scale) illustration of the various fracture elements which are distin-
guishable in the Mor and Horsha Formations in the Beer Sheva syncline. The diagram consists of four
stacked rock units (I–IV), see text for explanation. b Three sections from outcrops in unit I showing
fracture relationships: Multilayer joints (dashed lines), vertical faults (solid vertical lines) and normal
faults (inclined lines); the latter are emphasized where displaced by vertical faults. Numbers at the right
indicate strike of outcrops. Stations 20, 63 and 58 (Bahat and Grossmann 1988) are marked by A, B and C,
respectively. c Diagram of the conjugate three sets from the Mor Formation (see text for explanation)
256 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks
3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 257
258 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fractures in the syncline include s.l. joints, m.l. joints and normal faults. Each of these
features occurs in several variations with different stratigraphic-structural associations.
Some of the s.l. joints are associated with the burial stage, while the m.l. joints are consid-
ered to be uplift joints (Bahat 1991a). Other s.l. joints were clearly formed during uplift
(Bahat 1999a). Vertical faults occasionally occur in association with m.l. joints, and there
are some small strike-slip faults. Wherever two types of fracture strike in different direc-
tions, they are assigned to different sets, since their propagation (which is normal to the
least principal stress SHmin) could not be the same for both structures at the same time.

3.4.2.2
Fractures in Unit I (Mor Formation)

Nine joint sets are identified in the Mor Formation (Table 3.3). Those that are associ-
ated with the same stress system are grouped together and marked by letters within
the group (sets 1a–1d and 4a–4b). Three typical outcrops are shown in Fig. 3.35b.

Group 1. These include s.l. joint sets that are associated with burial. Four sets are iden-
tified according to characteristic orientation and appearance (Bahat and Grossmann
1988). Sets 1a and 1b consist of a cross-fold set striking 328°, and approximately or-
thogonal to it a set striking 059°. Set 328° generally predates set 059°, but there are also
indications of reverse order (Bahat 1988a, Fig. 5). These sets are observed particularly
in the northeastern distal part of the syncline. In some outcrops, both orthogonal sets
are exposed abutting each other, in others only one of the sets is exposed. The orthogo-
nal sets are also concentrated in the southwestern distal part of the Shephela syncline
and along the southeast flank of this fold (Bahat and Grossmann 1988). As far as can be
seen from exposures, cross-fold joints are rare in the northwestern flank of the Beer
Sheva syncline.
Sets 1c and 1d are hybrid s.l. joint sets (Hancock 1985; Bahat 1987a): 1c strikes 309°,
and 1d strikes 344°. Both sets occasionally occur in association with set 328° (Fig. 3.35c
and 3.36a), but they appear mostly in separate outcrops. The age relation between
sets 1a–1b vs. 1c–1d is not clear, although in one outcrop there is evidence that set 328°
preceded set 344° (Bahat 1991a, p. 242).
At rare localities, the time sequence of adjacent joints of a given set can be deter-
mined based on the curving of joint terminations, where late joints curve toward early
joints and arrest at approximately right angles (Bahat 1983; Dyer 1988). In the expo-
sure shown in Fig. 3.36b, joint C postdates joint B, which is younger than joint A. In other
exposures, the sequence is more difficult to determine (e.g., Bahat 1988b, Fig. 5; Bahat
1991a, Fig. 3.34).

Group 2. These include joints that are concentrated on the southeastern flank of the Beer
Sheva syncline, trending predominantly 055° (Bahat and Grossmann 1988, stations 26, 27,
48, 49 and 50). Cross-fold joints are very rare in these stations. Morphologically, they do
not resemble joints from set 1b, but look more like joints of group 7 (Sect. 3.4.2.4). There is
a slight, but perhaps significant age difference between the layers that host fracture groups 1
and 2. Both groups belong to the Mor Formation, but in the Beer Sheva syncline the layers
that host group 2 occur in an area of a gradual transition to the Horsha Formation along
Wadi Secher (Fig. 3.33c) (Bahat and Grossmann 1988, Fig. 3).
Fig. 3.36a–d. Photographs of outcrops. a Joints of sets 328° (marked by horizontal arrow) and 309° (marked by vertical arrow), cutting the Mor Formation (unit I).
Scale bar is 25 cm. b Joints of set 328° cutting the Mor Formation (unit I), showing the sequence of formation: A postdates joint B, and C is younger than B. c Multilayer
joints cut thick layers in the Horsha Formation (unit II). Scale is 90 cm. d Cross joint cutting the Horsha Formation (unit III). Scale bar is 5 cm
3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel
259
260 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.36e–g. Photographs of outcrops. e A syntectonic joint set curves sympathetically with a primary
A and secondary faults B and C in the Horsha Formation (unit IV). f Joints from set 062° dissect unit IV
along Wadi Naim. They cut the wadi floor anastamotically and become planar m.l. joints as they cut the
cliff (unit IV), layer marked by arrows is 90 cm thick. g Joints of group 12 are displaced by a normal
fault from group 15 (unit IV)

Group 3. These include low-angle (dip ≤ 45°) normal faults that strike about 294° and
are closely associated with s.l. joint sets 328° and 059° (Bahat 1985; Gross et al. 1997).
These faults are thought to be a result of intra-synclinal seismic tremors (Bahat 1985;
Buchbinder et al. 1988) (Fig. 3.35bA).

Group 4. This group, trending about 335°, includes vertical m.l. fractures that cut many
layers (Bahat 1988a; Bahat 1991a, p. 306). The m.l. fractures are divided into two sets
(Fig. 3.35bC). Set 4a consists of non-displacing m.l. joints, occasionally adjacent to
vertical faults of set 4b, which developed from the former as a result of changing stress.
Set 4b consists of vertical faults, one of which displaces a low-angle normal fault of
group 3 (Bahat 1991a, p. 283), suggesting a sequence of (1) early, low-angle normal
faults from group 3, (2) set 4a, and (3) set 4b.
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 261

Group 5. These include post-uplift joints that occur in the Shephela syncline (Bahat
1991a, Fig. 5.24). They cross the local Lower Eocene chalks at 55° to 75° dip angles and
show a relationship to the local topography. The surface morphology of these joints is
quite different from burial, syntectonic and uplift joints. They are composite radial
fractures, interacting with each other, resembling other post-uplift composite frac-
tures in sandstone and in granite (Bahat et al. 1995, 1999) (Fig. 3.23c and 4.27).

3.4.2.3
Fractures in Unit II (Horsha Formation)

Rocks of this unit are exposed in the northeastern part of the syncline, particularly
along the Beer Sheva-Nitsana road (Bahat and Grossmann 1988).

Group 6. Included here are joints oriented 336°, cutting thick chalk layers of about 6–8 m.
These joints are flat and smooth, not marked by plumes or en echelon fringes, or any
other fracture markings. Joint spacing is irregular, varying from 50 cm to 20 m
(Fig. 3.36c) (Bahat 1991b). The relation between fracture spacing and bed thickness
(e.g., Ladeira and Price 1981) suggest that this set formed after, rather than in the
course of sedimentation (Hodgson 1961b). By way of comparison, periodic arrest marks
on syntectonic joints in strongly deformed, thick Senonian chalk layers (Bahat 1991a,
Fig. 3.22) imply rhythmic fracture propagation, most likely induced by pore pressure
(Secor 1969) under intense tectonic strain of the kind that never affected the almost
horizontal sediments of the Eocene. These layers never attained over-pressure of flu-
ids during the burial stage, and the m.l. joints of group 6 could develop only after the
chalk was sufficiently dewatered and lithified, possibly not before uplift.

3.4.2.4
Fractures in Unit III (Horsha Formation)

Rocks of this unit are exposed along Wadi Secher, which runs down the southeastern
flank towards the synclinal center.

Group 7. This group consists of s.l. joints that display strike rotation along Wadi Secher
(Fig. 3.33d and 3.37), from about 090° to 020° along its southeastern part, through the
300–360° range along the sector of the wadi that crosses the central part of the syn-
cline further west. Fracture patterns were categorized along Wadi Secher (Fig. 3.37)
(with the exception of station 83), according to the “selection method” (Sect. 6.2.1.2).
The following features stand out:

1. In the eastern area (stations 71 to 80), most outcrops are dominated by ENE strik-
ing joints (050–083°). Further west, the joints strike gradually more to NNE, and
there is an increase in the number of N-S and NNW trending joints.
2. Adjacent stations usually have similar joint sets, with subtle transitions. For in-
stance, the major set in the eastern part of station 86B is 060°, and the minor set
is 330°; a little further west set 040° becomes dominant, and more to the west set 330°
becomes the more important set. In station 86C, set 025° is more important on the
eastern side and set 320° is dominant on the western part of the outcrop. However,
262 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.37. Map-diagram summarizing joint orientations at stations 71–95 along Wadi Secher. Circles show
the azimuths of sets at each station (with the exception of station 83) (after Bahat and Shavit 1997)

set 060° unexpectedly returns as the dominant set in station 86D, where 020° and
330° are the two minor sets.
3. In spite of the many “unexpected” returns of ENE trending joints, the overall emerg-
ing pattern is the increase in importance of NNE and NNW trending joints to-
wards the west.
4. A summary of joint orientation in stations 71–95 (not including station 83) shows
the general increase in importance of NNE and N-NNW trending joints relative to
ENE trending joints, when moving from east to west along Wadi Secher (Fig. 3.38a–c).
Hence, there is a subtle counter clockwise joint rotation in Wadi Secher.

Station 83 (in Fig. 3.37) was studied by the “comprehensive method” (Sect. 6.2.1.2)
in greater detail, along a 700 m exposure in Wadi Secher. Joint strikes were measured
separately for five “layer groups” (Fig. 3.39a). Average joint orientations in the
various layer-groups show distinctive traits (Table 3.4). The combined strikes of
s.l. joints from all layer groups of the Horsha Formation at station 83 shows a single
maximum at 48 ±16° (F in Table 3.4). Average trends rotate gradually counter clock-
wise from the older group-layer B through C and D to the younger group-layer E
(from 63° through 55° and 51° to 42°, respectively). An exception is the average 42°
in layer-group A, which contradicts the trend that was through layer-groups B–E. Joints
fringed with en echelon segmentation occasionally occur in upper parts of the west-
ern cliff of the wadi resemble the en echelon fringes that occur in Wadi Naim (see
group 12 below).
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 263

Fig. 3.38. Histograms summarizing data on s.l. joints from all stations in Fig. 3.37, not including sta-
tion 83. a Stations 71 to 86A. b Stations 86B to 95. c Stations 71 to 95. d Multilayer joints in station 83
(after Bahat and Shavit 1997)

Table 3.4.
Strikes and properties of joints
trending NE and NW in station
83, Wadi Secher
264 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.39.
a Section across Wadi Secher
at station 83, showing layer-
groups A to E on the cliffs (note
vertical scale) (after Bahat and
Shavit 1997). b Diagram of out-
crops X (set 16, Table 3.3) and Y
(set 12), including floors N–R
between them, showing rota-
tion in mean azimuths of joints
from 357° to 034°. Beds of soft
chalk (s.c.) alternate with hard
chalk beds designated by num-
bers. The two small vertical
lines between floors N and Y
represent talus (from Bahat
1986a)

Group 8. Twenty five m.l. joints constitute a set along a 600 m exposure at station 83,
with an average trend of 055.0 ±13.1° (Fig. 3.38d). Their average differs by 7.5° from
the s.l. average (F in Table 3.4). This relative orientation matching is noticeable, com-
pared with the more pronounced deviations of m.l. azimuths from those of s.l. in the
Mor Formation (Bahat 1991a, p. 306). Like m.l. joints in units I and II (Bahat 1991a,
Fig. 5.16), group 8 often occurs in fracture zones with closely spaced joints.
The discrimination between s.l. burial joints and m.l. uplift joint in unit I is quite
simple due to considerable morphological differences between the two fracture types.
On the other hand, the distinction between s.l. and m.l. uplift joints in unit III is not
always clear, because in some outcrops there are “transitional” fractures between the
s.l. joints of small widths (heights, Fig. 2.8a) and m.l. joints of great widths. Single-
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 265

layer joints and “transitional” joints are offset by normal faults, but no contacts be-
tween m.l. joints and normal faults were found in unit III. Therefore, it appears that
these faults are younger than the s.l. and the “transitional” joints. It is however not clear
that the normal faults are younger than the m.l. joints.

Group 9. This group consists of four normal faults, dipping above and below 45°, and
striking 052° and 130°, south of station 83. A study of the largest and best exposed
fault of this series is discussed in Sect. 6.7).

Group 10. This is a distinct s.l. group which consists of “cross-joints” (as distinguished
from “cross-fold joints”) that are straight along their central parts, oriented in the N-
NW direction, but their tips gradually curve orthogonally towards joints of an earlier
set (Fig. 3.36d). Their shape resembles the geometry of curving cross joints whose
central straight part parallel to the SHmax of the neotectonic stress field or its Tertiary
predecessor in eastern North America (Engelder and Gross 1993). These cross-joints
presumably represent the latest s.l. joints in unit III, in association with uplift and
erosion (Hancock and Engelder 1989). There is no indication, however, that these joints
are younger than joints of unit IV.

3.4.2.5
Fractures in Unit IV (Horsha Formation)

This unit is exposed in Wadi Naim, along the synclinal axis (Fig. 3.33c,d).

Group 11. Fractures of this group, which cut unit IV, exhibit a unique fault termina-
tion zone that consists of three major elements:

1. A primary fault.
2. Secondary faults.
3. A set of fault-associated joints (Fig. 3.36e). A description of group 11 is presented
in Sect. 6.6.

Group 12. This group consists of s.l. joints oriented in the 012–035° range, rotating
counter clockwise from younger to older layers upstream Wadi Naim (Bahat 1986a,
1999a) (Fig. 3.39b). En echelon segmentation occurs intensively along the lower fringe
of these joints, and rotates covariantly with the parent joints. Note that en echelon
fringes are almost exclusive to groups 12 and 16 (see below), and occasionally in group 7
(see above).

Group 13. Joints of set 062° occur in various parts of Wadi Naim, along tens and even
hundreds of meters. Unlike the straight m.l. joints of groups 5, 8 and 14, these joints
undulate and “branch” from each other anastomotically (viewed in horizontal sec-
tions on a platform) (Fig. 3.36f), but on the average strike 062°. They display irregular
spacing, and their openings vary from about 1 mm to several centimeters. Intensive
irregular cracking occurs at and adjacent to these fractures (as revealed by opening
with a chisel, and on vertical exposures). This set was apparently formed by breakage
266 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

during an extended buckling-rebound when the compression normal to the syncline

axis relaxed (Bahat 1987b), implying an important difference between the genesis of
set 062° as compared with the other joint sets (e.g., group 12). Unlike those other sets,
which were formed by remote compression co-axial to their strike, set 062° was formed
by remote tension. The multilayer joints in general and set 062° in particular are the
most important vertical conduits of ground water in the syncline (Sect. 6.9).

Group 14. These m.l. joints strike in the 040–060° range and form remarkably planar
cliffs in Wadi Naim, and they are straighter than m.l. joints that cut other units in the
syncline. These m.l. joints overprint s.l. joints of group 12 or cut them at small angles
(less than 30°). The wadi exposures provide a perfect view of fracture, both at the cliff
and in the adjacent floor, revealing all details of the transition from group 13 to
group 14. Joints of set 062° significantly change their appearance from an anastomose
style when dissecting layers 1–2 in the wadi floor (Fig. 3.39b) to planar joints of
group 14 that cut the upper layers of the cliff (Fig. 3.36f) (Bahat 1987b). Hence, fol-
lowing the buckling-rebound that created set 062°, the m.l. joints of group 14 were
formed during an advanced uplift process, preserving the same approximate strike
for both joint groups. Compared with the profusion of fracture markings on s.l. burial
joints cutting Lower Eocene chalks and syntectonic joints cutting Santonian chalks,
fracture markings are rare on all m.l. joints. Perhaps it reflects fracture by fluid pres-
sure in the former two cases and tension unrelated to fluid pressure in the latter.

Group 15. These are three neighboring conjugate normal faults (dipping both above
and below 45°), striking 358 ±10°. These faults offset s.l. joints from group 12 along
Wadi Naim (Fig. 3.36g). There are no contacts between normal faults and m.l. joints
from group 14, leaving no clue as to their age relationship (see also Sect. 6.7).

Group 16. This is a 350–014° striking joint set, with en echelon fringes both at the top
and bottom of the parent joint (Fig. 2.35a) (see group 12 for comparison, and elabo-
ration in Bahat 1986a, 1997).

3.4.3
Fracture Diversity in the Beer Sheva Syncline

The high fracture diversity, allowing the distinction of no less than sixteen groups,
may appear bewildering to those not familiar with the Beer Sheva syncline, making
one wonder if all the above described twenty three proposed fracture sets in Table 3.3
are really distinct sets, particularly since there are certain proposed sets from differ-
ent fracture groups that have similar orientations. This issue is addressed below.

3.4.3.1
Strike Superposition of Different Joint Sets

Seven strike superpositions of suspected different joint sets (Table 3.3) are examined
for the possibility of genetic linkages, which if found, would reduce the actual num-
ber of sets.
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 267

Superposition 1. The distinction between s.l. sets 1b and 2 (striking 059° and 055°,
respectively) is mostly based on considerable morphological differences. Set 1b char-
acteristically occurs in orthogonal relationship with a cross-fold set 1a, has very small
apertures (<0.1 mm) and is confined to the distal end of the syncline. Set 2, on the
other hand, does not have an orthogonal counterpart set, has variable openings (ap-
ertures up to 10 mm and more) and occurs in the southeastern flank of the syncline.
These differences imply different genetic origins for the two sets despite their almost
identical orientations: Set 1b is probably an early burial set, whereas set 2 was possi-
bly formed later (Sect. 3.4.2.2).

Superposition 2. Sets 8 and 14 strike about 055°. These m.l. joints are similar in some
respects, and they may be genetically related. It is possible that fracturing in these sets
took place in different times, or perhaps, fracture occurred in units III and IV at about
the same time. There is no compelling evidence either way.

Superposition 3. There is a genetic linkage between joints of sets 4a and 4b trending

335° (Fig. 3.35bC). They are geographically adjacent and possibly the 4b vertical faults
developed from the 4a vertical multilayer joints when local stress conditions changed.
Thus, they formed in two distinct stress fields (SHmax parallel and inclined to sets 4a
and 4b, respectively) into two fracture sets. Early joints that developed into faults are
well known from other fracture provinces (e.g., Segall and Pollard 1983).

Superposition 4. There is no relative age indication for (group) set 6 m.l. joints. Their
orientation (336°) suggests genetic association with set 4 m.l. joints that trend in the
same direction. This possibility is however challenged by the presence of Upper
Pleistocene gypsum mineralization along joints of set 6 (Issar et al. 1988). Some of
these gypsum crystals are orthogonal to the jointing, which is not characteristic
of set 4. This orthogonality suggests that opening of set 6 joints might be related to
the age of mineralization (Ramsay 1980). Thus, in spite of uncertainties of relative
age, it appears that set 6 is associated with processes younger than those associated
with set 4.

Superposition 5. The strike slip fault (set 11a) preceded the adjacent set 028° that de-
veloped during an early stage of the uplift (Bahat 1987b). This fault was formed by
horizontal compression along 344°, as evidenced by the orientation of joint set 11c
(Fig. 3.36e). The direction of compression is at an approximate 30° angle to the strike
of the strike-slip fault. Group 11 is found in a small area only about 16 m long and 7 m
wide in Wadi Naim. Thus, a synclinal stress is more likely to be the cause of group 11
formation rather than a regional one.
However, compression from 344° coincides with the strike of set 1d, which oc-
curs in various areas of the syncline and is considered to be a regional direction
(Eyal and Reches 1983; Bahat 1999b). Although there is probably no genetic connec-
tion between groups 1d and 11, such a connection may still be proven by new find-
ings. A regional compression from 344° could be tectonically effective in the syn-
cline, forming group 11, a short time before the synclinal rebound and stress rotation
(Bahat 1999a).
268 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Superposition 6. Joints of set 10 are oriented N-NW and may be suspected to have
been produced by the stress field that formed some of the joints of set 7 of a similar
trend. However, their distinctive morphologic features, the rotational style of set 7
and the characteristic curvature of set 10 (Fig. 3.36d) all point to different genetic
groups.

Superposition 7. Both groups 15 and 16 strike approximately N-S, yet they represent
two different mechanisms: One that produced normal faults, and one that that pro-
duced joints with two en echelon fringes. Nevertheless, their genetic linkage cannot be
excluded (Sect. 3.4.6.2).

In summary, we tried to show in the above review how fracturing episodes may be
fitted to the documented sets. There is no compelling evidence to rule out genetic link-
age between sets 4 and 6, 8 and 14, 1d and 11c, as well as 15 and 16. Combining repre-
sentative sets from these four couples would reduce the total number of sets in Table 3.3
from 23 to about 19. Yet it is equally likely that superpositions 1, 3 and 6 combine sets
of different genetic origin, and mere similarities in strike directions are not a suffi-
cient reason for grouping them together. The great structural diversity in the Beer
Sheva syncline may therefore be real.

3.4.3.2
Single-Layer Burial Joints versus Single-Layer Uplift Joints

Burial s.l. joints form early in the history of the rock and typically arrest at layer bound-
aries. Uplift m.l. joints, on the other hand, develop late in the history of the rock and
cross layer boundaries. The discrimination between these two groups is relatively
simple, because the m.l. joints, cutting many layers (or thick strata), have wider open-

Table 3.5. Fracture properties of single-layer burial and single-layer uplift joints
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 269

ings and their spacing is irregular and considerably larger than of the s.l. joints
(Fig. 3.36c) (Bahat 1991a; Ghosh 1993, p. 492). Much more difficult is the distinc-
tion between burial s.l. joints and uplift s.l. joints, because very often they display
similar outcrop features. The distinction is important because it can be useful in deci-
phering geological processes. Table 3.5 lists nine diagnostic properties of s.l. burial
joints and uplift joints in the Mor and Horsha formations (see also Bahat 1999a). One
of these properties, the rotation of joint strikes under regional stresses, is treated in
Sect. 6.3.

3.4.4
The Structural Interpretation of En Echelon Fringes in the Beer Sheva Syncline

En echelon fringes in the Beer Sheva syncline are characterized in terms of style and
in terms of paleostress directions (see Sects. 2.2.5 and 2.2.7).

3.4.4.1
Three Styles of En Echelon Segmentation

Three styles of en echelon segmentation can be distinguished in the Beer Sheva syn-
cline. They are represented by set X (unit IV), set O-Y3 (unit IV) and joint S (unit III)
(Table 2.1). Set X has en echelon fringes both below and above the parent joint
(Fig. 2.35a), and in both, segment rotation is counter clockwise, ranging in strike be-
tween 338° and 342°. Quite distinct from set X is set O-Y3 in station Y. In this set, a
fringe occurs only below the parent joint and segment rotation is clockwise, ranging
in strike between 026° and 053° (corresponding to the clockwise rotation of the par-
ent joints). The curved joint in station Y (Fig. 2.35c,d) undulates considerably along
the strike but the segmentation is confined to a small azimuth range (043–047°), which
is only a limited spread within the range 026° and 053° mentioned above. The concen-
tric joint (Fig. 3.40b) is a subvariation in set O-Y3. Joint S is best represented by part D
that strikes 073°, whose general clockwise segment rotation is in the 073–083° range
(Fig. 2.38 and 3.40a). However, some counter clockwise segment rotation also is ob-
served on this joint. These three styles differ from the Mt. Carmel style (Fig. 3.40c)
(Bahat 1997).

3.4.4.2
Three Paleostress Directions Set

Set X consists of joints that strike between 350° and 014° in outcrop X, and seems to
reflect a regional N-S post Middle Eocene paleostress. This stress corresponds to the
“last phase” N-S regional compression reported by Letouzey and Tremolieres (1980)
from southern Israel. Mart and Horowitz (1981, Table I) also described a regional N-S
trending joint system along the western shoulder of the Dead Sea rift. Joints from set O-
Y3 on the other hand strike over a wide range, from 012° to 035° and are apparently
associated with synclinal stresses (Bahat 1997). The joint direction (073°) and en ech-
elon characteristics found in part D from station W.S. (Fig. 3.40a) are quite different
from X and O-Y3 sets, suggesting three different stress fields.
270 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Fig. 3.40.
a Morphological sketch of a
composite joint in Wadi Secher
(station S in Fig. 3.33c, see
Fig. 2.38). The fracture surface
is divided into five parts, A to E.
Lower edges of tear surfaces
and arrest marks are marked t
and a, respectively. Numbers on
segments are azimuths (°). Two
late fractures are marked by
thick, sub-vertical lines. Scale
bar is 15 cm. b The “concentric
joint”. A discontinuous break-
down occurs below the outer
arrest mark a. The sense of
rotation of overlapping barbs
is counter clockwise in zones A
and B. Segments at the right
side of fringe rotate clockwise
with respect to the parent joint
(triangles along heavy lines
point to direction of overlap-
ping segments). This rotation
conforms to the other rotations
in outcrop Y. However, seg-
ments at the left side of fringe
show counter clockwise rota-
tion, which is in conformity
with the rotation of overlap-
ping barbs on the joint surface.
Scale bar is 20 cm. c A discoid
joint on Cenomanian chalk
Mount Carmel, northern Israel
cutting. The discoid displays
several concentric fringes with
a non-uniform breakdown.
Segment rotation is counter
clockwise on the left side of
the discoid and is clockwise
on the right side of the joint.
There is no overlapping in the
upper and lower parts of the
discoid (see coin for scale)

3.4.5
Fold-Crossing Joints

3.4.5.1
General

The geometric relationships between joints and asymmetric folds are discussed in
Sect. 3.2. Quite different are the geometries of joints that develop across symmetric folds
(Fig. 3.20b). These have been studied by many investigators (e.g., Price 1966, Fig. 43;
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 271

Stearns 1968). De Sitter (1964) demonstrated a 90° rotation of the SHmax between sym-
metrical syncline and anticline folds. The rotation occurred due to local tension at the
crest of the anticlines and compression at the trough of the synclines, like in a bent beam
(outer-arc, Fig. 3.19a) (e.g., Hobbs et al. 1976, Fig. 4.33a; Burger and Hamill (1976); Ghosh
1993, p. 254). All these studies show relatively simple fold-fracture structures, with only
a few fracture sets (Fig. 3.41). For instance, Stearns and Friedman (1972) found only
five major fracture patterns associated with folds. Ramsay and Huber (1987, p. 655)
consider the combination of four sets, two extensile sets (in cross-fold and in strike
orientations) and two conjugate sets to be “classical” (which approximates F1 and F2 in
Fig. 3.20b). When this classical view is “maximized” by additional sets, one obtains all
the sets in Fig. 3.20b and 3.41 that still maintain a “simple” fold-joint geometry.
This simplicity often reflects a compromise between deviations in fracture direc-
tion that are dictated by local stress field changes around singularities (e.g., Rawnsley
et al. 1992) and by the ongoing process of folding (Dieterich and Carter 1969; Burger
and Hamill 1976), acting against the tendency of the remote stress to avoid changes
by “self-correction”. Self-correction, also known as “local symmetry”, is a term adapted
here from small-scale cracking mechanisms. It is a property that distinguishes between
fracture behavior in isotropic and anisotropic materials (Ravi-Chandar and Knauss
1999). While arbitrary loading on a structure could result in a mixed mode loading, a
growing crack quickly turns so as to reorient itself into mode I symmetry in isotropic
materials. On the other hand, anisotropy forces the crack to grow under mixed mode
loading along specific planes in crystals and along faults in earthquakes. Occasion-
ally, however, self-correction into mode I may manifest itself along specific planes in
crystals (Cramer et al. 2000 and Fig. 2.23d) as well as along layer boundaries in strati-
fied formations (e.g., Bahat 1991a, Fig. 5.22) and along faults in earthquakes (Bahat
1982). Thus, fold-crossing joints tend toward simplified patterns, due to control by some
self-correction mechanism.

Fig. 3.41. Simple fold-fracture relationship. a, b and c are three tectonic axes. The orientation of b is
constant, parallel to bedding, while a and c change their attitudes in different parts of the fold. The
stereographic projections show orientations of the coordinate system, of the bedding where it is not
horizontal (dotted great circles) and of the fractures (solid great circles) (from Twiss and Moores 1992)
272 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

3.4.5.2
Fold-Joint Relationships in the Beer Sheva Syncline
versus the Simple Fold-Fracture Concept

Structural features of the Mor and Horsha Formations. We compare the “simple”
fold-fracture pattern in Fig. 3.41 to the fracture pattern in the Beer Sheva syncline
(Table 3.3, Fig. 3.35a).

1. Outstanding in unit I are two orthogonal sets of s.l. burial joints, trending NW and
NE. These are only found in the distal parts of the Beer Sheva and the Shephela
synclines. Less common are non-orthogonal s.l. joints, striking NE-ENE. In some
outcrops, NNW or WNW joints are added to or occur instead of the more common
NW joints. Uplift m.l. joints are often oriented NW-WNW. In one station, a normal
fault is displaced along a sheared, vertical m.l. fault (set 4b).
2. NW trending uplift joints cut massive chalks of unit II in the Horsha Formation,
above the Mor Formation.
3. In the eastern flank of the Beer Sheva syncline, the most common fractures in unit III
are NNE-NE-ENE s.l. joints and m.l. uplift joints trending NE. Further west towards
the synclinal center, NW trending s.l. and m.l. joints occur in increasing abundance
along Wadi Secher. A series of conjugate, normal faults also occurs along the wadi.
4. An early feature in unit IV is a right-lateral strike slip fault system, striking NNW.
A syntectonic, anastomosing joint set sub-parallel to the synclinal axis occurs along
Wadi Naim. Almost exclusive to this unit is the abundance of en echelon segmenta-
tion in NNE-N trending s.l. uplift joints. Single-layer joints reveal a strike rotation.
Multilayer joints overprint some of the s.l. joints. A normal fault from a NS striking
conjugate set initiates on top of the earlier right-lateral strike-slip fault.

Similarities between the above two patterns. Single-layer joints in unit I and
m.l. joints in units II and III generally correspond to the pattern in Fig. 3.41, in the
sense that they follow the orientations of elements from groups A and B. Sets 328°,
309° and 344° correspond to the pattern of conjugate hybrid joint structure (Fig. 3.35a),
and m.l. joints in unit I partly follow elements from both groups A and B. Similarity
also applies to the syntectonic 062° and other m.l. sets in unit IV that fit the extensile
direction of group B.

Dissimilarities between the above two patterns. Multilayer joints in unit I occasion-
ally develop in other orientations that differ from the “simple” view (Fig. 3.41) (see
Bahat 1991a, p. 306). Neither the post-uplift joints in unit I nor the rotational pattern
of s.l. joints in units III and IV have parallels in Fig. 3.41. Groups C and E from Fig. 3.41
are not found in the Beer Sheva syncline. A remarkable difference between the two
systems is the considerable vertical continuity of joints as is visible in photographs
and diagrams (Price 1966; Stearns 1968; Stearns and Friedman 1972), compared to
the short vertical continuity of fractures in the Beer Sheva syncline. With the excep-
tion of set 062°, the vertical extent of most fracture groups, including m.l. sets from
Table 3.3, is limited to one thick layer (Fig. 3.36c) or to a few tens of thin layers (Bahat
1991a, Fig. 5.16). No m.l. joints have been observed to cross the boundaries between
units I–IV, or to reach rocks below the Eocene chalks.
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 273

3.4.6
Factors Influencing Fracture Diversity in the Beer Sheva Syncline

The great diversity of fracture structures in the chalks of the Beer Sheva syncline
(Table 3.3) points to a considerable diversity of fracture mechanisms. The role of sev-
eral factors that influenced this diversity is discussed below.

3.4.6.1
The Influence of Paleostress on Fracture Diversity

General. Paleostress is a prime factor in the fracture evolution of the Beer Sheva syn-
cline. Several scales of stress are distinguished. First are the regional stresses of the
Mediterranean continental collision area (e.g., Letouzey and Tremolieres 1980) which
prevailed from the Upper Cretaceous to the Quaternary. Second are synclinal stresses
that originate within the fold and cause structural changes in the syncline (Bahat
1986a). Third, also important, are local stresses that are confined to a particular out-
crop (Bahat 1991a, p. 301) or to a particular lithology at a given location (Gross et al.
1997) or to layers of particular thickness (Bahat 1997). In the following discussion,
these scales are referred to as regional, synclinal and local.

Regional influences in unit I. The orthogonal sets 328° and 059° (1a and 1b in Table 3.3)
cut Lower Eocene chalks and maintain these exact strikes and fracture characteristics
in outcrops that are 17 km apart in two distinct folds, the Beer Sheva and the Shephela
synclines (Fig. 3.33b,c). The two synclines differ considerably in their orientation (strik-
ing about 050° and 028°, respectively). The implication is that sets 328° and 059° were
formed in response to large-scale (inter-continental) SHmax stresses and stress relax-
ation (Bahat 1989), which were “indifferent” to the already existing shape of the Syrian
Arc (e.g., Letouzey and Tremolieres 1980; Zoback 1992; Buchbinder and Zilberman 1997).
In places, the extensile set 328° is associated with conjugate hybrid joint sets (1c and 1d)
(Fig. 3.35c and 3.36a), reflecting mixed modes I + II (Sect. 2.2.13). Therefore, sets 1a–d
are genetically related to a particular regional stress. Bahat (1999b) suggests an alter-
native possibility that the three sets 1a, 1c and 1d are genetically unrelated, and that they
were formed at different times under different regional stress conditions (Sect. 6.3.9.1,
see also Goodwin 1995). According to both models, however, sets 1a–d are products of
regional stresses.

Synclinal influences in unit IV. This unit is exposed along the Beer Sheva synclinal axis
and consists of Middle Eocene rocks. Sets 12–16 that succeeded group 11 in time, display
rock responses to synclinal rebound and uplift. The s.l. joints of set 12 partly interact with
joints of set 062° from group 13 that were associated with the synclinal rebound, exhibiting
a concurrent relationship (Bahat 1987b). Those s.l. joints of set 12 that show no interaction
with set 13 possibly suggest that some of them had developed before set 062°, i.e., before
synclinal rebound, and that they continued a prolonged process (Sect. 3.4.6.2 and 3.4.6.3).

Local influences in unit I. The normal faults cutting the Mor Formation (set 3) show
evidence of repeated slip, mostly during the Lower Eocene. However, some fracturing
can be assigned to local stresses responding to specific lithological conditions (Gross
274 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

et al. 1997). Unlike post-uplift joints in sandstone, which tend to follow the orienta-
tion of earlier joints (Sect. 3.3), the post-uplift joints of set 5 (Table 3.3) developed in
response to local stresses, which show affinity to the erosional morphology of the
rock surface (Bahat 1991a, p. 301).

3.4.6.2
Rotations of Synclinal Stresses in Units III and IV

The systematic azimuth rotation of the s.l. joints in Wadi Secher (Fig. 3.37–3.39a and
Table 3.4) and in Wadi Naim (Fig. 3.39b) describes fracture sequences, which strongly
suggest that these joints were formed by linked mechanism(s), rather than erratically.
A mechanism of clockwise stress field rotation was suggested for the formation of the
s.l. uplift joints in Wadi Naim (Bahat 1986a, 1999a, present Fig. 3.42a–c). The joints ori-
ented 360° and 012° occur in upper layers, upstream along Wadi Naim, compared to the
joints oriented 019° and 028° (wadi slope is exaggerated in Fig. 3.42d). Since the layers
are approximately horizontal, the joints oriented 012° fractured earlier than the joints
oriented 028°. This sequence is consistent with the propagation of uplift joints. During
uplift the stress gradient is such that the least principal stress decreases downward,
causing the greatest strain at the surface. Figure 3.42d shows that jointing along Wadi
Naim that straddles sub-parallel to the fold large axis occurred in older layers, closer to

Fig. 3.42.
Proposed evolution of fracture
in the Beer Sheva syncline un-
der changing synclinal stresses.
a During buckling, maximum
horizontal stress SHmax parallels
the short axis x of the syncline,
and the minimum horizontal
stress SHmin parallels the long
axis y. b Buckling rebound oc-
curred when the compression
along x relaxed and became
SHmin, and the compression
along y became SHmax. c Clock-
wise paleostress rotation is
shown sequentially in Wadi
Naim (from Bahat 1999a).
d–f Diagrammatic wadi section
(angle of slope is exaggerated)
showing different joint rotation
directions (opened arrows); see
text for explanation
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 275

the synclinal periphery, before taking place in younger layers closer to the synclinal
center. The detailed results from station 83 in Wadi Secher (Fig. 3.39a and Table 3.4)
show that the joints oriented 042° fractured in a younger group-layer (E) before the
joints oriented 063° in the older group-layer (B) (Fig. 3.42e), revealing a clockwise joint
rotation. Thus, a clockwise joint rotation occurred both in Wadi Secher and in Wadi
Naim, implying that jointing along the two wadies probably was associated with the
same process of stress field rotation at about the same time. This interpretation does
not fit, however, the general fracture pattern from the twenty-eight stations along Wadi
Secher (Fig. 3.42f). Here, joints oriented ENE formed earlier in older layers, closer to
the synclinal periphery, and a shift towards NNW joint orientation took place later, in
younger layers closer to the synclinal center (Fig. 3.37). This shows a counter clock-
wise rotation (Fig. 3.38a,b and 3.42f), which differs from the clockwise rotation dis-
played in Fig. 3.42d,e. How can this difference be explained? One possible interpreta-
tion is that the s.l. joints in Wadi Secher did not form in a single continuous process.
Perhaps, in the southeastern part of the wadi, which is dominated by Lower Eocene
chalks (see description of groups 2 and 7 above, Table 3.3), jointing occurred earlier,
by a mechanism that differed from the one causing jointing in a later stage, in the
Middle Eocene chalks, towards the synclinal center, as shown in station 83. Hence, a
unique model of joint rotation in Wadi Secher is uncertain, but this scenario seems to
fit the data best.
There are deviations from simple stress rotations, particularly, the “unexpected”
departures from the rotation pattern in two scales along Wadi Secher. First, there are
repeated azimuth fluctuations in joint rotation along the wadi (Fig. 3.37). Second, the
average azimuth of one particular joint set in group-layer A at station 83 is reverse to
that of the general rotation in station 83 (Table 3.4). There is also some fluctuation in
joint rotation along Wadi Naim: Joints cutting floor R diverge from the general rota-
tion pattern (Fig. 3.39b). These deviations, however, do not invalidate the joint rota-
tion model suggested here; they merely demonstrate the versatility of these processes.
The exclusion of groups 16 and 15, both striking about N-S (Table 3.3) seems in-
evitable, not least because group 16 is linked with regional stresses (Sect. 3.4.4.1 and
6.3) and group 15 is associated with probably the youngest fracturing along Wadi Naim
(Sect. 6.3). The sum of the evidence suggests that sets 15 and 16 were formed after the
synclinal stresses had completed their rotation.

3.4.6.3
Additional Factors of Fracture Diversity in the Syncline

The observed fracture diversity in the Beer Sheva syncline is linked to several addi-
tional factors:

1. The Beer Sheva fold is a synform.

2. The low strength of chalk.
3. Developments during various stages of burial, unfolding and uplifting.
4. Different lateral position in the fold.

It is often difficult to separate the effect of single factors from the influences of other
factors. Accordingly, combined factors are discussed below.
276 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Synforms vs. antiforms and rock strength. Interestingly, systematic joints are rare in
Cenomanian chalks, anticlinally folded and uplifted in northern Israel – Mount Carmel
and the cliffs of Rosh Hanikra. Particularly intriguing is the absence of m.l. uplift
joints comparable to groups 6 or 8 (Table 3.3) from the cliffs of Rosh Hanikra, which
extend along hundreds of meters. On the other hand, systematic joints are quite abun-
dant in the synclinal early Senonian (Santonian) and Eocene chalks in Israel. This
comparison suggests that joints are abundant in chalks when they are part of synforms
but not when they are part of antiforms.
The influence of lithology on jointing is well demonstrated by the profusion of joints
in Eocene chalk formations in synclines of south and central Israel, whereas in syn-
clines in northern Israel where the Eocene is dominated by limestone (the Gilboa’ and
Rosh Pinna synclines, Flexer et al. 1984), jointing is much less developed. It seems that
the combination of synform structures with weak rock is particularly conducive to
intense and diversified fracturing. The dramatic reduction of strength of wet chalk
(Hayati 1975; Bahat et al. 2001a) probably also contributed to differential fracturing,
because various fracture episodes are expected to have developed under different de-
grees of wetness.

The influences of burial, folding, unfolding and uplifting on fracturing in the Beer
Sheva syncline. Buchbinder and Zilberman (1997) present a detailed account of the
uplifts that affected the eastern Mediterranean margin during the Miocene-Pleistocene.
Since Oligocene times, the eastern Mediterranean paleo-shelf (today coinciding with
the Judean foothills and extending north-south through the Beer Sheva syncline) has
been serially uplifted. Early Miocene, Middle Miocene and Pleistocene uplifts are rec-
ognized by sedimentary hiatuses during the periods 25–17 m.y., 14–9 m.y., and 1.5 m.y.
to Recent.
The overall character of jointing makes it fairly clear that joints of groups 1 and 3
(Table 3.3) were formed during burial stages (Bahat 1985) and groups 4 and 5 were cre-
ated during the uplift and post-uplift stages, respectively (Bahat 1991a). Joint rotation
in units III and IV is best correlated with unfolding and uplifting that were more influ-
ential in the central parts of the syncline, and in various cases specific fracture episodes
can be related to particular tectonic phases. Some fracture groups, however, such as 2, 8
and 14, cannot be fixed with certainty in the frame of events (Sect. 3.4.3.1).

Limited sedimentary cover. The young age of the Eocene sediments and the fact that
they never were covered by a thick overburden was possibly significant in controlling
fracturing processes in the syncline.

3.4.7
Summary

The Beer Sheva syncline is an asymmetric basin in which the sediments are only
slightly folded. Nevertheless, the chalky Eocene strata of this syncline display a great
diversity of fractures, not visibly related to the basin’s asymmetry. The Lower and
Middle Eocene chalk formations of the Beer Sheva syncline are divided into four rock
units (I–IV), which display considerable differences in fracturing mechanisms.
Table 3.3 presents a summary of twenty-three sets in the syncline.
 3.4 · Jointing and Faulting in Eocene Chalk Formations in the Beer Sheva Syncline, Israel 277

No diversity of joints as rich as in the Beer Sheva syncline has been reported from
other folds in rocks of comparable strength.
Different fracture episodes relating to joint sets of similar strikes are presented in
“strike-superposition”, a method that helps to verify the genetics of sets.
Three scales of paleostresses regimes can be distinguished: Regional (extra-syn-
clinal), synclinal (intra-synclinal) and local (confined to a particular site). The differ-
ent paleostress scales are most clearly identified in different parts of unit I, particu-
larly in the distal part of the fold, and to a lesser extent in the other units.
Orthogonal sets of single-layer burial joints (328° and 059°) occur in unit I at the
northeastern distal part of the syncline which trends about 050°, while the predomi-
nant set in the southeastern flank is oriented NE.
Nine criteria of distinction were established between s.l. burial joints in unit I and
s.l. uplift joints in unit IV.
Single-layer joints cut units III and IV. Two cases of s.l. joint strike rotation are mea-
sured in unit III and one in unit IV. Two measurements, in Wadi Naim and Wadi Secher,
show clockwise rotation, and one in Wadi Secher shows a counter clockwise one. These,
are best explained by intra-synclinal paleostress rotations: A simultaneous clockwise
rotation along Wadi Naim and the younger layers of the Horsha Formation along Wadi
Secher. On the other hand, a different, counter clockwise paleostress rotation took place
in the older layers of the Horsha Formation and younger layers of the Mor Formation.
Single-layer joints of unit IV, which are the youngest in the syncline, exhibit en ech-
elon segmentation along layer boundaries. Some en echelon segmentation occurs spo-
radically along boundaries of s.l. joints in unit III, and in one outcrop, in connection
with composite fracturing. Styles of en echelon segmentation and particularly their
limited spreads of azimuth helped to identify three sets of joints in unit IV. These in-
clude, a set produced by a regional N-S post Middle Eocene paleostress, a set with wider
azimuth range (012° to 035°) which is connected to stress rotation, and a third set ori-
ented 073°, whose origin is not clear.
Multi-layer uplift joints overprint syntectonic joints (the 062°) in unit IV.
All multi-layer joints cutting units I–IV appear to be related to uplift processes. Their
“fine tuning” however is not clear. Strike patterns vary in the four units: A consider-
able range of strikes in unit I, NNW jointing in unit II, NE jointing in unit III, and joint
spreading around NE in unit IV. Joint morphologies vary as well, particularly in unit IV.
The multi-layer joints are the most important vertical conduits of ground water in
the syncline (Sect. 6.9).
Compared with the profusion of fracture markings on s.l. burial joints cutting Lower
Eocene chalks, and syntectonic joints cutting Santonian chalks, fracture markings are
rare on all m.l. joints. Perhaps it reflects fracture by fluid pressure in the former two
cases, and tension unrelated to fluid pressure in the latter.
A vertical fault, genetically linked to multi-layer uplift joints, offsets a normal fault
in unit I.
In unit I, low angle (dip ≤ 45°) normal faults interacted with s.l. burial joints. Gen-
erally, the interaction took place during the sedimentation stage. High angle (dip ≥ 45°)
normal faults interacted with uplift single-layer joints in unit IV during advanced uplift
stages of the syncline.
Unit IV exhibits a unique fault termination zone of a right-lateral strike slip fault.
This zone displays primary and secondary faults, and a joint set exclusive to the zone.
278 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

Additional factors that contribute to fracture diversity in the Beer Sheva syncline,
besides paleostress regimes, include: The synform structure, the low strength of the
chalk (especially when wet), and differential developments during stages of burial, un-
folding and uplifting.

3.4.8
Open Questions and Future Studies

In spite of detailed studies, many problems relating to fracturing in the Beer Sheva
syncline are still unresolved. Following is a brief review of these open questions.
Sets 1a–d are possibly conjugates that formed in one stress field (Bahat 1987a). Al-
ternatively, the three sets 1a, 1c and 1d (Table 3.3) evolved in a process of clockwise
stress rotation (Bahat 1999b). Both processes, however, would be caused and sustained
by regional stresses (Sect. 6.3.9.1).
If both sets 1d and 11c were formed by regional compression from 344°, future field-
work may yet discover a genetic relationship between them.
An intriguing issue is fracture by stress relaxation. Currently, it is considered that
strike-parallel joints (set 1b) are more widely spaced because they were formed in re-
sponse to smaller stresses relieved by recurring phases of relaxation, as compared to
the cross-fold joints (set 1a) (Bahat 1991a, Table 5.4). The latter were formed by greater,
periodically recurring increases in fluid pressures associated with regional stresses.
According to Rawnsley et al. (1998), the joints in the Bristol Channel basin were
formed in five main phases, at least partly during relaxation of the Alpine stresses
(Sect. 3.2.1). Engelder and Peacock (2001) do not argue against relaxation but assert
that the J3 joints in the south-dipping beds of the anticline at Lilstock Beach propa-
gated during folding (before relaxation). Although jointing during stress relaxation
has been proposed by various authors (e.g., Price 1974, Bahat 1989) the available ar-
guments are circumstantial and require more compelling geological evidence, though
there is experimental evidence of fracturing under this condition (e.g., Sect. 2.2.3.7).
To date, no study has dealt with the connection between “transitional” joints from
unit III (group 8) with layer boundaries (Underwood and Cooke 2001 and Sect. 2.2.3.15).
Such investigations may help to understand what prevents such joints from becoming
m.l. joints.
The unclear genetic relationship between groups 2 and 7 needs to be investigated.
There is currently no proof that s.l. joints from group 10 in unit III are younger than
s.l. joints from unit IV.
The structural and genetic relationship of the four adjacent normal faults in Wadi
Secher (group 9) needs to be established.
Three adjacent normal faults strike 358 ±10° (group 15). These faults offset s.l. joints
from set 12 along Wadi Naim. There are however no contacts between the normal faults
and m.l. joints from group 14, leaving their age relationship open.
Field data enabling the timing of groups 8 and 14 are still lacking: Were groups 8
and 14 formed simultaneously, or separately? Were they formed earlier in unit III and
later in unit IV, or vise versa?
The concept of regional stress meets a difficulty: If the stresses that induced joints
were not controlled by the early folding pattern of the Syrian Arc, as suggested above,
why are their products (group 1, sets 1a–1d) clearly associated exclusively with the
 3.5 · Comparison of Unrelated Fracture Provinces 279

distal parts of both the Beer Sheva and the Shephela synclines? Should not these sets
be uniformly dispersed throughout the two synclines during the Lower Eocene?
More studies are required to explain how lithological differences between chalk and
limestone, other than strength (and possibly related to their history of lithification)
may influence their modes of fracturing.
The young age of the Eocene sediments and the fact that they never were covered
by a thick overburden is most probably significant for the development of fracturing
in the syncline, but this topic needs be further investigated.
Finally, a micropaleontological study along Wadi Secher might confirm that unit II
is lacking there and establish the rate of joint rotation in station 83.

3.5
Comparison of Unrelated Fracture Provinces
The four different fracture provinces that were selected for presentation in this chap-
ter, the Appalachian Plateau, Bristol Channel, Zion National Park in the Colorado Pla-
teau and the Beer Sheva syncline, do not represent a comprehensive review of the sub-
ject. Certainly there are additional such provinces that are worth concentrating on.
Nevertheless, even in the present scope it may be instructive to bring up the similari-
ties and differences between these provinces. Soon it will be seen that in spite of be-
ing universally subject to the physical laws of rock mechanics and fracture mechan-
ics, the fracture products differ considerably in the four provinces due to the diversity
of geological conditions in which they evolved and the unique properties and mate-
rial individuality of any rock package.
The following topics might be a good basis for inter-regional comparison of joint-
ing.

1. Rock age, lithology and age of fracturing.

2. Genetic classification.
3. Changes in stress fields, structural complexity and overburden.

3.5.1
Rock Age, Lithology and Age of Fracturing

Fracturing took place on the Appalachian Plateau, northeastern United States, in De-
vonian multi-rock formations of siltstone, shales and limestone. Most foreland frac-
turing occurred in the Upper Paleozoic during four phases of the Alleghanian orog-
eny. Later, post-Alleghanian joint sets developed. In the Bristol Channel province,
southwestern Britain, Lower Jurassic limestone underwent fracturing during several
tectonic stages. The important joint set J3 was formed during the compressional Al-
pine orogeny that started in the Eocene to Oligocene, and continued through the Late
Oligocene to Miocene. In the Beer Sheva province, south Israel, fracturing started
during sedimentary burial in the Lower Eocene and continued throughout the Oli-
gocene/Miocene, or later uplift periods, cutting Lower Eocene and Middle Eocene
chalks. In the Jurassic sandstone of the Zion province, northwestern U.S., fracturing
probably started in the Mesozoic, but main geo-morphological developments occurred
at younger Pliocene-Pleistocene ages. In all these events, in unconnected provinces,
280 Chapter 3 · Four Fracture Provinces in Sedimentary Rocks

different time intervals elapsed between sedimentation and rock fracture, and in each
province fracturing took place in different multi-stages, lithologies and stratigraphies.
Apparently, the numbers of fracturing styles and packages of fracture-sediment type
are limited, but their combinations are numerous.

3.5.2
Genetic Classification

The attempts at genetic classification of joints in the various provinces (i.e., distin-
guishing between burial, syntectonic, uplift and post-uplift joints) are not simple.
Notwithstanding, efforts for obtaining such a classification are justified because they
enable one to discriminate between tectonic phases and to establish the actual circum-
stances of fracturing, such as depths, and sequence of paleostresses.
Fracturing in the Appalachian Plateau probably started in the burial joints of
set 320–330° (Younes and Engelder 1999), and these can be linked quite likely to the
Ib (325–337°) burial set (Bahat 1991a, p. 251). Fracturing continued with three to four
syntectonic stages during the Alleghanian orogeny. Young, post-Alleghanian joints are
possibly related to post-uplift events (Hancock and Engelder 1989). In the Bristol
Channel province, all jointing seems to be syntectonic (Engelder and Peacock 2001).
The Beer Sheva province presents joints of all the four genetic groups, while the mag-
nificent fractures at Zion National Park are mostly post-uplift joints.

3.5.3
Changes in Stress Fields, Structural Complexity and Overburden

Fracture-producing stresses were notably dissimilar in the various provinces. Post-

uplift jointing at Zion National Park takes the form of parallel, simple longitudinal
splitting (Bahat et al. 1995, 1999), mostly preserving the orientation of older fractures.
In this case, maximum compression was vertical, induced by overburden weighing
down on present-day plateau rocks consisting of strata that were not folded.
Horizontal compression of the Alleghanian orogeny induced two sequences of ro-
tation (Sect. 3.1) in the Appalachian Plateau, and these were characterized by small
differential stresses (6 MPa, Younes and Engelder 1999) and only slight folding. In the
Bristol Channel province, joints propagated in response to Cenozoic Alpine compres-
sion. This compression also generated the growth of buttress anticlines through the
reactivation and reversed pre-Alpine normal faults. The Alpine inversion, which was
accompanied by intense deformation, reflects large regional, differential stresses
(27.8 MPa in limestone and 32.2 MPa in shale, Engelder and Peacock 2001, Fig. 9). On
these buttress anticlines, local stresses were superimposed, forming joints along faults,
and other singularities.
In the Beer Sheva syncline, two stress types are distinguished: A regional stress that
activated various synclines in the Syrian Arc and affected mainly burial joints, and an
intra-synclinal stress, confined to the Beer Sheva syncline that was connected with the
formation of syntectonic and uplift joints in the syncline. Young uplift joints super-
imposed on burial or syntectonic joints are common. Single layer fractures in the
slightly folded Eocene chalks were created by very low differential stresses (–0.5 MPa
extensile effective stress, Bahat 1989).
 3.5 · Comparison of Unrelated Fracture Provinces 281

When comparing the four provinces, the degree of structural complexity increases
from Zion through the Appalachian Plateau, to the Bristol Channel buttress anticline,
and finally, the most complex pattern is found in the Beer Sheva syncline. Overbur-
den was thin and fracturing took place close to the ground surface in the canyon at
Zion National Park. In the Beer Sheva syncline, overburden was also thin, particularly
when fractures developed during and after uplift (possibly tens of meters, probably
not more than several hundred meters). The question at what depths fracturing of the
various sets occurred, remains partly enigmatic and justifies further research.
Chapter 4

Jointing in Granites

Whereas Chap. 3 concerns fracture provinces in sedimentary rocks, Chap. 4 relates to

fracture provinces and domains in granites. While a fracture province may concern
many plutons that maintain geographic and genetic affinities, a fracture domain often
relates to a particular pluton or a pluton-complex in a province. There are many publi-
cations on granite that relate to different fracture aspects of the rock (e.g., Johnson 1970;
Segall and Pollard 1983). This chapter does not provide a review of the subject; rather
it focuses on certain topics. We divide Chap. 4 into two parts. Part 1 (Sect. 4.1–4.6) re-
lates to geological and fractographic aspects of joints; it presents a characterization
of joints cutting granites in central Europe and in California, showing the diversity of
joint formation in different fracture provinces and domains. The descriptions relate
to joints throughout various stages in the granite histories, from fracture during the
cooling history of the pluton to fractures that formed after the pluton had arrived at
the ground surface. Part 2 (Sect. 4.7–4.10) concerns fracture mechanic aspects of joints,
particularly problems of dynamic fracturing by drawing on fractographic properties
of joints as follows:

1. Plotting joints on the velocity v vs. stress intensity K curve.

2. Presenting a general discussion of various manifestations of v and K in granites.
3. Calculating the new fracture areas on en echelon vs. hackle fringes.
4. Estimating the “index of hackle raggedness” of joints.

We present at the end of this chapter (Sect. 4.11) a description of a method how to
drill in situ parts of fractures from the rock.
284 Chapter 4 · Jointing in Granites – Part 1

Part 1
4.1
Background
Bankwitz and Bankwitz (1984, Fig. 8 and 13) describe joints about 40 m long in granite
from the South Bohemian Pluton near Mrákotin, west of Telc in the Czech Republic.
These joints were considered to be representatives of the uplift stage (Bahat 1991a)
because they resembled uplift joints cutting granite in Sinai (Bahat and Rabinovitch
1988). According to the latter authors, the presence of markings on the fractures in those
granites was promoted by the rapidity of “late fracture” propagation during the uplift
process, despite the coarse textures of the rocks. Recent studies of fracture in the South

Fig. 4.1.
The Cloos fracture model. Ori-
entation of primary, near verti-
cal Q- and S-joints and oblique
faults D, as well as sub-horizon-
tal L-joints with respect to the
regional maximum horizontal
principal stress σ1, which is per-
pendicular to flow lines in the
country rock (modified after
Müller et al. 2001)

Fig. 4.2a. Distribution of magmatic rocks in the Northern Bohemian Massif including the Lusatian
Granodiorite Complex (upper right) and the Erzgebirge (center) (from Müller et al. 2001)
 4.1 · Background 285

Fig. 4.2b–d. b A simplified geological map of the South Bohemian Pluton (marked +), the axis of major
fold the orientation of foliation of gneiss county rock on various sides of the pluton (modified after
Benes 1971). c Late to post-Variscan granites of the Bohemian Massif: A: South Bohemian Pluton; B: Ober-
pfalz plutons; C: Erzgebirge plutons; D: Lusatian Granite Massif. Dots mark localities of investigation by
Bankwitz et al. 2004 (from Bankwitz et al. 2004). d The Sierra Nevada Batholith from central California
(modified after Segall and Pollard 1983)

Bohemian Pluton reveal however that many vertical joints are in fact pre-uplift as will
be demonstrated in the following sections. These studies suggest that rapid fracture can
occur by “early fracture” in granite while this body still exists at great depths.
We begin by presenting the Cloos (1921, 1922) pioneering model of fracture in gran-
ites (Fig. 4.1). The Cloos model provides a sound basis for the classification of fracture
in granites and for distinguishing between “early fracture” and “late fracture”. A series
of recent investigations on fracture in granites from three different provinces, the North
286 Chapter 4 · Jointing in Granites – Part 1

Bohemian Pluton in Germany (Fig. 4.2a), the South Bohemian Pluton in the Czech Re-
public (Fig. 4.2b), both in central Europe (Fig. 4.2c), and the Sierra Nevada Batholith in
central California, USA (Fig. 4.2d) are described. The North Bohemian Pluton is repre-
sented by two domains, the Lusatian Granodiorite Complex and the Erzgebirge. The
South Bohemian Pluton is portrayed by data gathered particularly in two distinct quar-
ries (Mrákotin and Borsov). The two domains representing the Sierra Nevada Batholith
are the Knowles Granodiorite and the pluton complex in Yosemite National Park. Syn-
thesized into these descriptions are several investigations that have not been published
before. A section on comparative jointing ends the first part of this chapter.

4.2
The Cloos Model
According to Müller et al. (2001), Cloos (1921, 1922) and his associates investigated
the development of fracture in igneous rocks in central Europe, in relation to the mode
of emplacement of an intrusive mass. Cooling and hence crystallization, shrinking and
degassing of the igneous melt first takes place at the walls and roof of the intrusive
mass. Continued movement and intrusion of the still liquid core give rise to stresses
in the outer shell of the intrusion.
A system of three main types of “primary fractures” that are orthogonal to each other
is recognized and defined by Cloos with respect to flow-lines of oriented crystals, flow-
planes and schlieren, which develop during the movement of the viscous liquid melt
through the process of intrusion. These are Q-joints, (“cross”, or “dip fractures”), S-joints
(“longitudina1 fractures”), and L-joints (“flat-lying fractures”). A fourth type that forms
later than the “primary fractures” is termed “diagonal fractures”. The orientation of these
fractures with respect to flow-lines is shown in Fig. 4.1.
Q-joints are among the earliest of fractures to develop in the cooling mass. Typi-
cally, they form parallel to the main stress direction (σ1) and perpendicular to the flow
lines, i.e., stretching lineation. These fractures are sometimes opened and mineralized
(aplite and hydrothermal minerals), and correspond to ac-joints (Fig. 4.1). Q-fractures
are less jointed (in terms of divisibility, Müller et al. 2001) than S-joints.
S-joints are oriented perpendicular to the main stress direction (σ1) and parallel
to the flow-lines. The orientation of these fractures is little affected by variations in
pitch of the flow-lines. However, variations in trend of the flow-lines are faithfully fol-
lowed by changes in the strike of the S-joints. S-joints are closed, rarely filled with aplite
or “dike” material, and the minerals are usually different from those found in the other
kinds of primary fractures.
L-joints tend to develop sub-horizontal near the apex, or dome, of an intrusion or
in flat sheets and laccoliths. It is difficult to see how these structures can he interpreted
on dynamic grounds. It has, however, been suggested that they form when the center
of an intrusion shrinks due to cooling. These structures are also filled with hydrother-
mal minerals, and have been referred to as primary flat-lying fracture so that they may
be distinguished from exfoliation joints that may be much younger.
Diagonal fractures form at 45° or more, to the trend of the flow-lines, resulting from
compression normal to, and extension in the azimuthal direction of flow-lines. These
fractures are also commonly filled with aplite or hydrothermal minerals. Displacements
along the diagonal fractures are clear, indicating that they are shear phenomena.
 4.3 · Fracture in Granites from the North Bohemian Massif 287

4.3
Fracture in Granites from the North Bohemian Massif
in the Lusatian Granodiorite Complex and the Erzgebirge
Bankwitz and Bankwitz (2004) describe various European granites (mostly Variscan),
with a special reference to their age and some fracture properties. This section is mostly
cited from Müller et al. (2001). The study areas are the Lusatian Granodiorite Complex
(LGC) and the Erzgebirge in SE Germany (Fig. 4.2a), with the main emphasis of the study
on the LGC. Kossmat (1927) assigned both the LGC and the neighboring Erzgebirge to
the Saxothuringian Zone of the Variscan orogen (Fig. 4.2c).

4.3.1
The Lusatian Granodiorite Complex

4.3.1.1
Geological Background

Based on geological considerations, Watznauer (1954) wrote of a “granite flood”, which

was assumed to extend from the LGC across the Erzgebirge into Thuringia. In contrast,
Suess (1926) proposed to separate the LGC from the Saxothuringian Zone due to the
different structure and development of the basement in both domains. Since then, a
number of differences and similarities of the crustal structures and their development
have been found. The principal difference between the provinces is the main age of intru-
sion, which is Cadomian (540 to 530 Ma) in the LGC, and late-Variscan (325–318 Ma) in
the Erzgebirge (Bankwitz and Bankwitz 2004).
The LGC is separated from the Erzgebirge by the Elbe lineament and by the rock
collage of the Elbe Zone (Fig. 4.2a). The LGC is bordered by major faults and over
thrusts. The virtually undeformed Cadomian granitoids cover an area of about
2 500 km2 and constitute a genetically related, mostly peraluminous magmatic suite
(Krauss et al. 1992). Most of the contacts of the granodiorites and wall rocks are de-
fined by faults resulting from uplift since the Late Proterozoic (Eidam et al. 1992). The
late-Variscan granitoid activity in the LGC is represented by the Stolpen and Konigs-
hain monzogranites, which intruded as stock-like bodies along large deep-seated re-
gional faults (e.g., Hirschmann 1966).

4.3.1.2
The Gravity High of Lusatia

Müller et al. (2001) investigated the gravity field of Lusatia and of the Erzgebirge, using
the tomographic method of Linsser-filtering (Linsser 1967; Conrad et al. 1996; Trzebski 1997).
The Linsser-filtering provides an alternative technique to the conventional selective gravity
field methods since it allows depth selective gravity gradient analyses down to 20 km depth,
based on the Bouguer gravity field. Thus revealing information about the three-dimensional
extent of steep-dipping (<45°) rock density contrasts in selective depth levels. The LGC and
the Erzgebirge are distinguished by the most significant gravity anomalies of the Bohemian
Massif: The gravity high of Lusatia and the gravity low of the Erzgebirge. The widespread
occurrence of high density rock beneath the LGC is interpreted in terms of the presence
288 Chapter 4 · Jointing in Granites – Part 1

of a large-scale layered gabbroic intrusion (“gabbroic layer”, Lindner 1972; Behr et al. 1994)
of unknown age associated with a mantle upwelling. This poses the question of whether a
gabbro intrusion beneath the LGC could generate around 25 000 km3 (2 500 km2 × 10 km
thickness) of magma from metagraywacke-metapelitic turbidites (anatexis of late Prot-
erozoic, e.g., Eidam et al. 1992), which are exposed at the NW margin of the LGC.

4.3.1.3
Results and Discussion

According to Müller et al. (2001) the strike of the interpolated Q-jointing is harmonic over
the whole area of the LGC and forms a sigmoidal pattern (Fig. 4.3a). The general parallel-
ism of the strike of the Q-jointing and the axis of the gravity high is notable. Subsequent to
the formation of the LGC, gabbroic magmas mostly used these Q-joints as ascent paths in
the western and central part of the LGC (Fig. 4.3b). In the east of the LGC the mafic magmas
intruded diagonal joints. The sigmoidal striking of the Q-joints is parallel to the indicators
of 5, 7.5 and 10 km depths, which mark the southern boundary of the Lusatian gravity high.
The second vertical joint direction, the S-jointing, forms angles between 90° and 60° with
the Q-jointing (Fig. 4.3c). Angles of more than 80° between the S- and Q-joints are in accord
with Cloos’s joint model. The primary orthogonal joint system is overprinted by sets of diago-
nal joints. This occurs in association with large-scale fault zones at the margin of the LGC.
Generally, the primary joint sets Q and S are best preserved in the core zone of the south-
ern part of the gravity high in the area of Bischofswerda and Bautzen (Fig. 4.2a). Thus, it is
interpreted by Müller et al. (2001) that since the high density rock beneath the LGC is con-
sidered to be a large-scale layered gabbroic intrusion and the pattern of Q- and S-joints
fit the high gravity pattern quite well, the updoming of the postulated gabbroic intrusion
could have induced these fractures during the solidification of the LGC between 520 and
590 Ma. Lobst (2001) examined field relationships of fractures in the Klosterberg quarry,
some 10 km SW of Bautzen, and found that steep E-W oriented joints were older than N-S
joints. Apparently, the Klosterberg quarry is located geographically, such that the E-W and
N-S joints fit the directions of the Q-joints and S-joints, respectively, in the maps drawn by
Müller et al. (2001) (Fig. 4.3). This suggests that the Q-joints were formed before S-joints.
Müller et al. (2001) could not confirm the existence of late-magmatic, flat-lying L-joint-
ing (Fig. 4.1) in the Lusatian granite. On the other hand, they identified a joint set sub-
parallel to the surface topography that was related to late processes of surface exposure of
the granite body, which they consider to be an exfoliation set. They observe that the exfo-
liation is best developed in granites and granodiorites with large joint spacing (>5 m). Such
conditions are given in the top area of the Lusatian gravity high between Bautzen, Bischofs-
werda and Kamenz (Fig. 4.2a). The exfoliation is poor and irregular in regions where joint
spacing is <1 m. The distance between the surface parallel joints increases with increas-
ing depth, as observed in other granite quarries (e.g., Johnson 1970, p. 358).
In summary, Müller et al. (2001) show that the primary joint system of the Cadomian
Lusatian granodiorite fits the model of updoming by an underneath intrusion: In this
case, the presumed gabbroic intrusion at depth. The fractures under investigation
formed during the solidification of the LGC between 520 and 590 Ma, such that the
Q-joints were formed before S-joints. However, in contrast to the Cloos model it has
been found that the formation of sub-horizontal joints in the granite is related to the
external process of exfoliation and has no relationship to the primary joint set.
Fig. 4.3a–c. Structural maps of the Lusatian Granodiorite Complex. The maps are underlain by isoanomals of the gravity high of Lusatia with the local maximum
NW Kamenz. a Main faults (thick lines) and interpolated striking Q-joints (thin lines) (from Müller et al. 2001)
4.3 · Fracture in Granites from the North Bohemian Massif
289
 290
Chapter 4 · Jointing in Granites – Part 1

Fig. 4.3b. Structural map of the Lusatian Granodiorite Complex. The map is underlain by isoanomals of the gravity high of Lusatia with the local maximum NW Kamenz.
Devonian gabroic dikes used predominantly Q-joints as ascent paths, whereas Tertiary basaltic magmas intruded along N-S to NE-SW trending faults (from Müller et al. 2001)
Fig. 4.3c. Structural map of the Lusatian Granodiorite Complex. The map is underlain by isoanomals of the gravity high of Lusatia with the local maximum NW
Kamenz. Main faults (thick lines) and interpolated striking S-joints (thin lines) and additional steep-dipping joints (from Müller et al. 2001)
4.3 · Fracture in Granites from the North Bohemian Massif
291
292 Chapter 4 · Jointing in Granites – Part 1

4.3.2
The Erzgebirge

We consider previous investigations on two plutons in the Erzgebirge, the Niederbo-

britzsch granite (northeastern Erzgebirge, Fig. 4.2a), studied by Müller et al. (2001) and
the Eibenstock granite (southwest Erzgebirge), studied by Bankwitz and Bankwitz 2004).

4.3.2.1
The Niederbobritzsch Granite

According to Müller et al. (2001), the Erzgebirge covers an area of about 6 000 km2. This
area is characterized by Late Carboniferous plutons that intruded Upper Proterozoic
to Lower Palaeozoic metamorrphic sequences. The plutons are composed of spatially
separated multiphase intrusions with complex textural and field relationships. The gran-
ites are traditionally divided into the older (330–320 Ma), and the younger (300–290 Ma)
intrusive complexes (Laube 1876; Breiter and Seltmann 1995).
The Bouguer gravity low of the Erzgebirge is the most significant negative gravity
anomaly in Central Europe with an extreme of –76 mgal near Karlovy Vary. It extends
along the northern margin of the Bohemian terrain and is limited by the Oberpfalz
block in the west. The 250-km-long and 40-60-km-wide NNE-SSW striking gravity low
continues south of the gravity high of Lusatia to the Jizerské Hory and Krokonose
Mountains, where the minimum turns to an E-W direction.
Müller et al. (2001) point out that the NNE-SSW trending joints of the Niederbobritzsch
pluton (Fig. 4.4a) are concordant to the ac-plane of the Proterozoic gneisses. These
joints are interpreted as Q-joints, because they are opened joints partly filled with
quartz. The S-joints strike in the direction of the main stretching axis of the wall rock.
Thus, the development of the Q- and S-jointing of the Erzgebirge granites is adapted
to the syn-intrusive stress field of the intrusion frame (wall rock), which corresponds
to Cloos’s joint model. On the other hand, the primary joint sets of the Cadomian gra-
nodiorites in the Lusatian Granodiorite Complex reflect the updoming of a postulated
gabbroic intrusion, which underlies the granodiorites (as mentioned above).

Fig. 4.4a.
The orthogonal Q- and S-joint-
ing of the late-Variscan Nieder-
bobritzsch granite in the east-
ern Erzgebirge (from Müller
et al. 2001)
 4.3 · Fracture in Granites from the North Bohemian Massif 293

Fig. 4.4b,c. b Synoptic diagram of lower hemisphere stereographic projection of joints in the Eibenstock
granite. Numbers 1–8 of maxima indicate the relative sequence of joint formation (from Bankwitz and
Bankwitz 1995, 2004). c Schematic illustration of the mapped joint sequence in the Eibenstock pluton,
see text for explanation (from Bankwitz and Bankwitz 1995, 2004)

4.3.2.2
The Eibenstock Granite

The Eibenstock granite is a large laccolithic (exposure of 20 × 40 km) with complex

fracturing. Bankwitz and Bankwitz (1995, 2004) distinguish a series of joint sets in
this granite, which they describe in great detail. Jointing started with flat to oblique
dipping (18–35°) “lager” joints (Cloos 1921) (Fig. 4.4b,c). The other sets were steeply
dipping with the exception of set 9. The first subvertical joints (phase 2) followed the
shape of the exposed pluton. The relative timing of the various other sets suggests frac-
ture migration with time, e.g., from phase 3 to phases 4 and 5, and from phase 6 to
phases 7 and 8. Sheet fractures formed in stage 3, and finally, a second generation of
smaller, N-S and E-W-trending joints was created, terminating at the sheet fractures.
A dense deposition of greenish Uranium-micas on joints of sets 2 and 4 helps to dis-
tinguish the two generations of orthogonal sets.
294 Chapter 4 · Jointing in Granites – Part 1

4.4
Joints in Granites from the South Bohemian Pluton
This section concentrates on investigations that were carried out in the Borsov and
the Mrákotin quarries, citing primarily Bahat et al. (2001a, 2003), Bankwitz and Bank-
witz (2004) and Bankwitz et al. (2004).

4.4.1
Geological Setting of the South Bohemian Pluton

The study areas under consideration here are located in the South Bohemian Pluton
(SBP) (Fig. 4.2b), which is the northeastern part of the large South Moldanubian Mas-
sif that belongs to the internal zone of the Variscan belt of Europe (Fig. 4.2c). The country
rock consists predominantly of kyanite-sillimanite bearing gneisses and schists of Late
Proterozoic to Early Paleozoic age (Suk 1984; Matte et al. 1990). The metamorphism
occurred at 370–350 Ma. The stretching lineation trends NW-SE with indications of a
south-eastward shearing (Matte et al. 1990). The SBP intruded the core of an antiform
composed of gneisses and schists (Benes 1971). The SBP trends NNE-SSW through the
Czech part of the Bohemian Massif. According to Bankwitz et al. (2004),the SBP extends
over a distance of 150 km between Havlíckuv Brod in the north and the Danube River,

Fig. 4.5.
Geological sketch map of the
South Bohemian Pluton, show-
ing the dominant first vertical
fracture orientation, set NNE.
Dots: Locations of investigated
fluid inclusions. Cross hatched
areas: Granite bodies in the
axial part of the SBP (modified
Breiter 2001 by Bankwitz and
Bankwitz 2004)
 4.4 · Joints in Granites from the South Bohemian Pluton 295

and further south, recognized from drillings below the Molasse Zone of the Alps
(Scharbert 1998). The root zone of the SBP occurs in its southern part between the
Danube and the Pfahl faults. The granitoids were emplaced at mid- to upper-crustal
levels into hot country rocks (~15–18 to 7–10 km, ~450–650 °C) shortly after the ther-
mal peak of regional metamorphism (Büttner 1997). P-T-paths in the country rocks
suggest that granite formation, low-P/high-T metamorphism and extensional thinning
were preceded by a phase of intense crustal thickening that occurred during late
Palaeozoic continent-continent collision (Gerdes et al. 1998; Matte et al. 1990).
The SBP is a large late-orogenic batholith, which consists of several phases of in-
trusion with isolated sub-types (e.g., Cimer, Mrákotin, Eisgarn, Weinsberg, Freistadt,
Mauthausen types) (Breiter and Koller 1999; Breiter and Sokol 1997; Bankwitz et al.
2004), one of which is the Borsov granite that has a close affinity to granodiorite. Its
geochemistry is strongly influenced by the metamorphic country rock. The granitic
rocks have developed from a high-K peraluminous melt (Breiter and Koller 1999;
Matejka and Janousek 1998) with the presence of biotite and biotite-muscovite petro-
graphic manifestations in the rock. According to Matte et al. (1990), the intrusion age
is about to 330 Ma. Field contacts in the SBP at the Borsov quarry show that the main
fracture systems occur in the following sequence:

1. Early sub-horizontal tension fractures.

2. Vertical or sub-vertical E-W and NNE trending joints.
3. Late, sub-horizontal joints that are represented by the uplift sheeting (e.g., Dennis
1987), cutting the other joints without influencing them, i.e., the joints of the sec-
ond group had ended their formation before the late joints started.

Both early horizontal and vertical tensile fractures are expected to form in asso-
ciation with the increasing magma pressure in the pluton (Knapp and Norton 1981).
This section focuses on the NNE vertical set (Fig. 4.5) due to its particular regional
significance, minimizing the treatment of other sets in the pluton.
The South Bohemian Pluton was chosen for investigation for several reasons (Bahat
et al. 2003). First, these granites are cut by joints that exhibit a great variety of fracture
surface morphologies (displaying most of the features in Fig. 2.1a,b). Second, the frac-
ture morphologies of the joints are usually well preserved, because most of these gran-
ites were not deformed in post-Variscan (Late Paleozoic) times. Therefore, the frac-
tures maintained their original shapes from the time before the uplift. Third, the bulk
of the existing knowledge on jointing is related to sedimentary rocks, and field obser-
vations are normally made along road cuts or natural exposures, which are shallow
and seldom allow observations of large joints to their natural ends. These granite rocks
have been quarried for more than a hundred years for construction purposes through-
out Europe. The large quarries expose large walls required for obtaining good per-
spective views of large joint fractographic features that granites often exhibit. Fourth,
and perhaps most important, the unique exposure of the Borsov granite allows the
study of paleo-jointing at the intrusion depth before uplift and erosion. This provides
an insight into fracture processes at great depths in the crust. Finally, the SBP is an
important part of the Variscan orogen in central Europe and our study may add im-
portant information for use in future investigations of the orogen.
296 Chapter 4 · Jointing in Granites – Part 1

It is convenient to distinguish two classes of rock fracture: Those formed by shear

stress and those by tensile stress. The vast majority of experimental work in fracture
mechanics has concentrated on mode I, tensile fracture propagation (Atkinson 1987).
The results of these studies, quite apart from their importance in understanding joint
formation, may be applied to a wide spectrum of other geological problems. These
include hydrofracturing, microcracks associated with larger structures, tensile frac-
tures of macro and mega-scales (or extensile at great depth), dike formation, magmatic
intrusion, boudinage and crevasse penetration (Atkinson 1987). Earthquake studies
relate more to mode II shear operations (Scholz 1990), but may have mode I manifes-
tations as well (e.g., Bahat 1982).

4.4.2
Fractographic Versatility on Pre-Uplift Joints in the Borsov Quarry

Existing fractographic studies on fracture in granites generally concern late histories

of the granites, following their cooling and uplift (e.g., Holzhausen and Johnson 1979;
Bahat et al. 1999), whereas investigations of fractures that formed during the cooling
stages (e.g., Knapp and Norton 1981; Lee and Diehl 1989) commonly have not used
fractographic techniques. The following sections concern fractures in two outcrops,
the Borsov and the Mrákotin quarries from the South Bohemian Pluton in the Czech
Republic (Fig. 4.2b), which reveal joints that formed during the early history of the
granite.

4.4.2.1
Fractographic Characterization of Ten Joints from the Borsov Quarry

The ten joints with highly diversified fracture surface morphologies (Bahat et al. 2003)
are briefly described photographically (Fig. 2.54–2.56), and in their corresponding draw-
ings (Fig. 4.6–4.8). Their properties are summarized in Tables 4.1 and 4.2. The geologi-
cal interpretation of the fractographies of these joints is presented in Sect. 4.4.5.1, and
the fracture mechanic analysis is given in Sect. 4.8.
▲

Fig. 4.6. Drawings of joints from the Borsov quarry (see photographs in Fig. 2.54–2.56, modified from
Bahat et al. 2003). Mirror planes that do not contain fringes. a The “multi-arrest marks joint”, J1. Some
fifteen arrest marks superposed by a faint plume. The fracture markings are confined in a “wedge” be-
tween two sub-horizontal boundaries, above and below the fracture pattern. Thickness of wedge at the left
side is about 90 cm. b The “radial plume joint”, J2, showing the fracture origin at the center and a radial
plume representing the propagation mode. The joint is bounded by two horizontal partings. c The “quasi-
core joint”, joint J3. A mirror plane is decorated by elliptical arrest mark which is partly surrounded by
additional partial arrest marks. The origin is situated in an eccentric position with respect to the ellipse,
and a continuous plume that starts at the origin propagates in a radial manner to all directions, with
various intensities. A symmetric axis passes through regions of greatest intensity plumes. Sub-horizontal
fracture p, cuts the mirror. d The “multi-joint exposure”, J4. The plume m1, of the “semi-circular joint”
reveals a three-stage concentric growth into a semi-ellipse.The long axis of this semi-ellipse approxi-
mately coincides with an horizontal plane p1, that forms its upper boundary. The “circular joint” is charac-
terized by a circular single arrest mark and a faint, single continuous plume, m2, that starts at an eccentric
position with respect to the arrest mark and spreads approximately radially. Two upward propagating
plumes occur on distinct parallel planes, m3 and m4 at the left side of the “circular joint”. Note also the
plume m5. Sub horizontal fractures p2 and p3 cut m3 and m2, respectively. However, m4 and m5 seem to
truncate against the continuation of p2, suggesting that p2 predates m4 and m5
4.4 · Joints in Granites from the South Bohemian Pluton 297
298 Chapter 4 · Jointing in Granites – Part 1

Fig. 4.7a,b. Drawings of joints from the Borsov quarry (see photographs in Fig. 2.54–2.56, modified
from Bahat et al. 2003). Mirror planes that contain small en echelon fringes. a The “divided arrest mark
joint”, J5 starting at o. A major horizontal parting, p divides the lower semi-circular series of arrest
marks from the upper ones of the “divided arrest mark joint”. Note a rhythmic increase and decrease in
barb-coarseness in certain spacings between the arrest marks, including an inner spacing, i which has
a greater barb-coarseness than its adjacent outer one, t. Another plume started at o1 on parting p1 and
was cut by later q. Two plumes terminated in small en echelon fringes F. b The “asymmetric bilateral
joint”, J6 displays some ten concentric arrest marks (E1–E10), that propagated from the origin at an
eccentric position in ellipse E1 (inset at right). The arrest marks are superposed by a continuous plume
divided by a symmetry axis. At the left side of the origin there is a large wavy plume, w almost without
any arrest marks. A “triangle” fringe of en echelon cracks, F1 is separated from w by a “girdle” F2 of
short en echelon segments. Inset at left shows en echelon segmentation, s that initiates from arrest mark u
that corresponds to E9. F1 and F2 at the right side are interpreted to be the continuations of F1 and F2
at left. E11 and F3 are additional hypothetical arrest marks and fringe, respectively. J6 is limited by
partings p and is cut by a series of sub-horizontal fractures (like p1). Vertical scale at the center is 2 m
 4.4 · Joints in Granites from the South Bohemian Pluton 299

Fig. 4.7c,d. Drawings of joints from the Borsov quarry (see photographs in Fig. 2.54–2.56, modified
from Bahat et al. 2003). Mirror planes that contain en echelon fringes. c The “uniform fringe joint”, J7.
The boundary between the mirror that is marked by a faint radial plume and the fringe of en echelon
segments of approximately equal lengths is well defined. d The “partly erased joint” J8 shows a mir-
ror that is surrounded by a convex downward circular boundary and a fringe of en echelon segments
of non uniform sizes. Diameter of the mirror is approximately 10 m
▲

Fig. 4.8. (see next page) Drawings of joints from the Borsov quarry (see photographs in Fig. 2.54–2.56,
modified from Bahat et al. 2003). Mirror planes with diversified fringes. a The “trefoil joint” J9, that
forms the shape of a three leaflets flower, showing two types of ripple marks, delicate concentric
undulations u (hardly visible), at the mirror center, and two lateral arrest marks a (showing pro-
nounced morphologies), at the left side of mirror. Plumes with a general clockwise, cw stepping are
superposed on the latter. The two sides of the mirror contain intensely fractured fringes. The fringe
at the left side of the mirror is divided in correspondence with the zigzag of the mirror boundary,
into five sectors by the designations i, n, t, r, and m, where segmentation is counter clockwise, ccw
(for details see Bahat et al. 2001). Frame shows location of b. b The fringe at the right side of the
mirror has the shape of a fan which starts at a low side next to the mirror, and opens upward away
from the mirror, divided to five distinct sub-fringes. The sense of stepping is not uniform along the
various sub-fringes. c The “hackled fringe joint” J10, showing a joint on which the fringe is particu-
larly large compared to the mirror size (partly removed), and is heterogeneous with respect to hav-
ing adjacent hackles and en echelon segments. d The large diameter in the simplified diagram of (c)
is about 2 m
300 Chapter 4 · Jointing in Granites – Part 1
4.4 · Joints in Granites from the South Bohemian Pluton 301
302 Chapter 4 · Jointing in Granites – Part 1
4.4 · Joints in Granites from the South Bohemian Pluton 303
304 Chapter 4 · Jointing in Granites – Part 1
 4.4 · Joints in Granites from the South Bohemian Pluton 305

4.4.2.2
The Unique Features of the Mirror Plane, Undulations, Arrest Marks and Fringes
of the Trefoil Joint

General. The Trefoil joint is one of the ten Borsov joints described in Tables 4.1
and 4.2 and it displays a unique fracture surface morphology. The mirror
plane of this joint deviates considerably from the common circular shape. Part
of the mirror boundary extends downward as a “big tongue” beyond the com-
mon circular perimeter, and it forms the shape of a three-leaflet flower (Fig. 2.56a
and 4.8a). The mirror is marked by two types of ripple marks (Sect. 2.2.4.2). The
concentric set of delicate undulations (hardly noticeable on Fig. 2.56a) that im-
ply relatively rapid fracture propagation, differs considerably from the set of
lateral arrest marks, which reflect slow fracture propagation (u and a, respec-
tively, in Fig. 4.8a). The arrest lines occur on the left side of the mirror but not at
its right side. A faint set of radial striae (a plume) superposes the undulations
on the mirror, indicating downward and radial joint propagation. A set of intense
striae, emphasize the “semi-box” superpose orthogonally the lateral arrest marks
(Sect. 2.2.4.2).
The two sides of the “tongue” contain large fringes that are intensely fractured,
and their fracture styles differ considerably from each other. There is a resemblance
in shape between the curvatures of the arrest lines and the series of sub-fringes at
the right side of the mirror. These sub-fringes show contrasting senses of en ech-
elon stepping (see Fig. 2.35e), implying a change of fracture mode under the influ-
ences of modified stress fields. The fringe on the left side of the mirror shows a com-
plex fracture (Bahat et al. 2001b) on which we elaborate below. Finally, the mirror is
jointed by sub-horizontal parting fractures that generally slope gently from left to
right; several of them cut the mirror and the two fringes.

The left fringe. Here we focus on a joint with a particular relationship between
a “primary mirror” (the mirror) and a “secondary mirror” that occurs on the
fringe. The boundary between the mirror and the fringe is not a continuous curve
but a zigzag one (Fig. 4.8a). The fringe is divided into four sectors that gener-
ally correspond to the zigzag boundary. These are from left to right, i–n, n–t,
t–r, and r–m. We concentrate on the structure in the n–r sector. Two growth
styles of en echelon segments and steps can be distinguished: First, vertical
downward growth of alternating segments and steps from the boundary of
the mirror (Fig. 4.9a,b) can be discerned; second, some six segments merge
into one step that displays a tensile, secondary mirror plane with semi-circu-
lar boundary and a fringe that convex downward in continuation with the lengths
of the segments that form the merger (Fig. 4.9c,d). This morphology of a ten-
sile-step differs known fractographies in sedimentary rocks, where mirrors and
other tensile fracture markings occur on en echelon segments but are absent
on steps. Hence, the mirror of the trefoil joint was fractured into a segmented
fringe that then continued to fracture into a secondary mirror, which also was
fractured into a fringe. We discuss the conditions under which this joint formed
in Sect. 4.9.3.1.
306 Chapter 4 · Jointing in Granites – Part 1

Fig. 4.9. a–d The left fringe of the “trefoil joint”, where e and “s” are en echelon segments and steps,
respectively (see text for explanation) (after Bahat et al. 2001b)
 4.4 · Joints in Granites from the South Bohemian Pluton 307

4.4.3
“Dynamic Joints” in a Fracture Zone from the Mrákotin Quarry

4.4.3.1
Characterization of Mirror Planes and Fringes in the Mrákotin Quarry

According to Bahat et al. (2001c; Bankwitz and Bankwitz 2004), two joint types are
distinguished in the Mrákotin quarry (Fig. 4.2b), both of them trend NNE-SSW. First,
vertical individual joints that are spaced several meters apart, and second, fracture
zones that contain several parallel joints spaced about 5–10 cm (Fig. 4.10a). We con-
centrate on a set, which is exposed in a fracture zone about 30 m long and about 18 m
high. These joints parallel each other on a single wall. The “Mrákotin set” consists of
nine distinct neighbor joints, all striking about 025° (Fig. 4.10b,c). Joints A and B oc-
cur on one plane, joints C–H form a composite fracture on another plane that is fur-
ther inside the rock body, and joint I forms a third plane still deeper in the rock. Spac-
ing between the three planes varies from 5 to 20 cm, while joint fringes partly touch
neighbor planes. Joints A–H with the exception of C contain mirrors that have one or
two hackle-fringes above and below them (Fig. 2.1b). Joint C on the other hand ex-
poses only a fringe at its lower part, which consists of en echelon segments (Fig. 2.1a).
The mirrors of joints A–H deviate from the conventional penny shape into various
ellipticities. The hackles of all fringes have the same sense of stepping in that they show
clockwise rotation both above and below the mirrors; that is, they all are arranged such
that walking across the hackles is downstairs from left to right (from south to north).
Joint I exposes a large faint plume, and its mirror boundaries are not revealed. The
nine individual joints are described below from south to north (Tables 4.3 and 4.4).

Joint A. Two large hackle fringes occur above and below the mirror of joint A. Whereas
the upper joint boundary is somewhat curved, the lower boundary is fairly straight
(Fig. 4.11a–c). Quarrying has removed the southern (left) mirror half and a larger
part of the upper fringe so that its length parallel to the joint length is not fully re-
vealed. The length of the lower fringe is 9 m. It becomes gradually wider (the lengths
of individual hackles increase) from south (left side of Fig. 4.11a,b) to north (right
side of figure) towards the center of the fringe. Hence, maximum width of the fringe
occurs at about the middle of its length. There is correspondingly an increase in hackle
coarseness in this direction.
In profile, the hackles at the lower fringe show strong deviations from the common
uniform “shingle-structure” of en echelon fringes. Hackles vary considerably in lengths
and show extreme variations in thickness among individuals and along the length of
an individual hackle, appearing as wedges (Fig. 4.11c). They are not tabular, but some
of them are rather distorted as if responding to a twist mechanism. They have similar
strikes, but they differ in their dips. Note that whereas most of the lower hackles dip
away from the mirror (towards the left, Fig. 4.11c), some hackles close to the mirror
boundary of A dip towards the parent joint (to the right in this figure). Generally, the
fringe forms an angle (φ) of about 10° with the parent joint. An intriguing feature in
this fringe is one hackle, A* that strikes 290° and dips 65° towards NE, i.e., this joint
rotates 85° with respect to the strike of A. This hackle has its own mirror and is fringed
with a secondary hackle-fringe (Fig. 4.11a–c, Tables 4.3 and 4.4).
308 Chapter 4 · Jointing in Granites – Part 1

Fig. 4.10. Joints in the Mrákotin quarry. a A large joint (18 m high), roughly trending E–W, with vertical
traces of approximately NNE trending individual joints and fracture zones (arrowed). The latter corre-
sponds to the discussed Mrákotin set in the next figures. (from Bahat et al. 2001c). Photograph (b) and draw-
ing (c) of the Mrákotin set of nine joints (A–H and the large plume at the upper right). The sub-vertical
fracture V1 separates joints A and B and V2 cuts joint D. The fringe of jont C is somewhat erroded. Vertical
scale bar at c is 2 m (from Bahat et al. 2002b)
4.4 · Joints in Granites from the South Bohemian Pluton 309
310 Chapter 4 · Jointing in Granites – Part 1

Table 4.4. Fracture parameters of the Mrákotin 025°-joint set

Fig. 4.11a.
Joints in the Mrákotin quarry.
Photograph of the 025° trend-
ing joint A. Vertical ruler is 2 m
(from Bahat et al. 2001c)
 4.4 · Joints in Granites from the South Bohemian Pluton 311

Fig. 4.11b,c. Joints in the Mrákotin quarry. b Drawing of the 025° trending joint A. Vertical ruler is
2 m. c Profile drawing of joint A, where A* is possibly an equivalent of M2 in Fig. 2.30e (after Bahat
et al. 2001c)

Joints B–H. Joint B reveals only the upper part of the mirror and the upper fringe.
Mirrors A and B are separated by a vertical fracture (V1 in Fig. 4.10c), which had pre-
ceded the mirrors because the mirrors and fringes of these joints look as if their shapes
were modified along fracture V1 (Bankwitz 1966). This fracture seems to belong to
the set that includes the large joint (Fig. 4.10a). The upper part of the parent joint C is
partially covered by joint B, but the lower boundary of the mirror of joint C is ex-
posed, and it is approximately straight (Fig. 4.12a). The visible part of joint C has only
one fringe, below the mirror, consisting of en echelon segmentation. The small mirror D
has a fringe only at its upper side. In its lower part, mirror D interacts with mirrors C
and F (which are much larger) (Fig. 4.12a–c). Joint E occurs above joint D and has
fringes both above and below the mirror. The upper fringe of joint D is relatively large,
and it appears that it interacted with the lower fringe of mirror E, penetrating some-
what into the E mirror.
By far the longest mirror in the composite is F (Fig. 4.12b,c). The F-joint seems to
have propagated from north to south. There are large fringes above and below the
parent joint and both boundaries are fairly straight. The lower fringe is partly erased
by the quarrying process. The upper fringe, on the other hand, is well preserved and it
has its maximum width at about its mid-length. Joint G is the smallest one; it occurs
north of joint E and above the upper fringe of joint F. Joint H is located above joint G
and is a little smaller than that joint.
312 Chapter 4 · Jointing in Granites – Part 1

Fig. 4.12. Joints in the Mrákotin quarry. a Joints A–E. b Photograph and c drawing of a part of the
“composite fracture” wall with joints C–H. Ruler is 2 m (from Bahat et al. 2001c)

Ellipticity of mirrors and joints. The mirrors of joints D and E are surrounded by
curved boundaries. On the other hand, the boundaries of joint A and the lower bound-
ary of joint F approach straightness, and the upper boundary of joint F is straight.
 4.4 · Joints in Granites from the South Bohemian Pluton 313

Our measurements of the Mrákotin set (Table 4.4) reveal two features. With the ex-
ception of C, for all other joints the ratio of long to short axes is similar for the various
elliptical mirrors, ranging from 3.3 to 4.5, independent of elliptical size. However, this
ratio is reduced to the range from 1.3 to 1.4 for the full joints (combined mirrors and
maximum widths of the fringes for every joint).

Shape of fringes. Joints C and F have only lower fringes, whereas joints D, E, G and H
have only upper fringes. The increase in fringe width to a maximum (towards a shape
of quasi-triangle) seems to repeat itself in the upper fringes of joints D, E and F, and
possibly also in the lower fringes of joints A and F, and the upper fringe of joint B. This
characteristic is different from the conventional fringes of uniform width in synthetic
materials (e.g., Mecholsky 1991, Fig. 1) and many rocks (e.g., Fig. 2.34b and 2.35a,b).

4.4.4
Twist and Tilt Angles on Joints Cutting Granites in Europe

Two angular relations that fringes form with mirror planes of joints are of particular inter-
est: The twist angle ω and the tilt angle φ (Fig. 2.53). The characterization of ω and φ may
be used in the interpretation of fracture conditions (Cooke and Pollard 1996 and Bahat and
Rabinovitch 2000, respectively). Bankwitz and Bankwitz (2004) measured ω and φ on joint
surfaces that cut various granites in Europe. Their results are summarized in Fig. 4.5, 4.13a,b
and 4.14a, providing the first comprehensive account of the tilt angle φ in a series of rocks.
They illustrate the steady NNE orientation of seventeen pre-uplift joints throughout the
SBP and the consistent sense of en echelon segmentation with respect to the parent joints,
and a large, 16° spread of ω values (between 22° and 38° (Fig. 4.5). Larger joint populations
show even greater spreads of ω value, i.e., 40° in the Borsov quarry, and some 35° in the Mrá-
kotin quarry (Fig. 4.13b).Values of ω from the Erzgebirge vary in the 9–22° range (Fig. 4.14a).
The spreads of the φ values vary considerably from pluton to pluton and the spreads are
particularly large in the Borsov quarry (Fig. 4.13a) and in the Erzgebirge (Fig. 4.14a).
Bankwitz and Bankwitz (2004) plotted the logarithmic tilt/twist ratio vs. palaeo-depth
of pluton emplacement (Fig. 4.14b). They suggest a correlation in which the shallowest
intruded pluton (Eibenstock, Germany) represents the largest ratio range and the deepest
pluton (now exposed) is characterized by the smallest ratio range. Apparently, the depth
of joints from Borsov is 7.4 km, and the ratio range is very large, while the depth of joints
from Weinsberg is 15 km and the ratio range is much smaller, raising the suspicion that in
shallow depths fracture conditions are variable, while great depths unify them, and this
occurs between 7.4 and 15 km depths. This however does not coincide with the great spread
of φ values observed in the shallow Erzgebirge (Fig. 4.14a). Quite intriguing is also the low
spread of ω on joint surfaces in the Erzgebirge. Perhaps it suggests that fracture took place
under relatively uniform stress conditions, which maintained only small variations in the
local KIII / KI ratio (Fig. 2.21a). Hence, it appears that generally the spreads of φ and ω are
quite large (Fig. 4.13a,b) and for no clear reasons. Also noticeable are the φ values of some
joints that vary very considerably in a given fringe (see gray markings in Fig. 4.13a). This
suggests that an interpretation of the petrological conditions (e.g., depth) of these gran-
ites during jointing should be carried out with great caution. The great angular ranges found
at the Borsov quarry is not surprising in view of the extreme diversified fractography found
in this quarry (Fig. 4.6–4.9).
314 Chapter 4 · Jointing in Granites – Part 1

Fig. 4.13. a Tilt, φ· and b twist, ω· angles on joint surfaces from various plutons in Europe: SBP: South
Bohemian Pluton; CBP: Central Bohemian Pluton. KTB: Near the KTB (Continental Super-Deep Bore-
hole); Flossen: Flossenbürg Granite, Bavaria (from Bankwitz and Bankwitz 2004)

4.4.5
Geological Interpretation of the Joint Fractographies
in the Borsov and Mrákotin Quarries

While the set described here from Borsov consists of joints that occur in various out-
crops within the large quarry, the investigated set from the Mrákotin quarry consists
of joints that are displayed on a single wall. The sets from the two quarries display
different fractographies, and their respective interpretations focus on different issues
complementing each other.
 4.4 · Joints in Granites from the South Bohemian Pluton 315

Fig. 4.14.
a Relationship between twist,
ω tilt, φ angles of early gran-
ite joints in different plutons
in Europe. b The logarithmic
ratio of φ / ω vs. palaeo-depth
of emplacement of the plutons
plotted in Fig. 4.14a. The most
shallowly intruded pluton (Ei-
benstock) shows the widest
range of ratios; the deepest
pluton (Weinsberg, now ex-
posed) is characterized by the
smallest range of ratios, be-
tween ~0.5 and 1.0. Symbols
as in Fig. 4.14a (from Bank-
witz and Bankwitz 2004)
316 Chapter 4 · Jointing in Granites – Part 1

4.4.5.1
Geological Interpretation of Joint Fractography in the Borsov Quarry

In this section, we limit our discussion to three aspects of jointing: Fluid-driven frac-
turing, mirror ellipticities and their unilateral growth, and plutonic vs. sedimentary
rock’s fractographies.

Fluid-driven fracturing. The Borsov joints cover very wide v and KI ranges (Fig. 4.35).
Particularly intriguing is the question of what could cause these fractures to grow
under such extreme different conditions, from sub-critical to post-critical. These con-
ditions could be changes in local stresses due to changes in the magnitude of remote
stresses that would alter the component of local normal stress. It appears however
that drastic changes in fluid pressures in the granite history could have been the main
mechanism leading to joint fracture surface diversification at Borsov. The rhythmic
arrest marks (particularly in Fig. 4.6a–c and 4.7a) are linked to the fundamental con-
cept in the Secor theory (1969) of fracturing by pore pressure (Bahat and Engelder
1984). This process involves “fluid-driven fracturing” (Engelder et al. 1993, p. 44) and
“rhythms of fluid-pressure build-up and fracture-induced release” Davis and Reynolds
1996, p. 248). Thus, the frequent appearance of numerous arrest marks (Sect. 2.2.4.2)
probably indicates fluid-driven fracture processes, at least during the early stages of
joint growth at Borsov. This interpretation fits well the presence of biotite and biotite-
muscovite petrographic manifestations in the rock.
The elongated axis of the South Bohemian Pluton coincides with the dominant strike
direction of the flat dipping main foliation in the neighboring metamorphic country,
and structurally coincides with the crest line of the pluton (Benes 1971). This suggests
that the shape of the pluton could have controlled significant thermal and fluid-pres-
sure gradients, which in turn influenced the orientation of the joints. It could also in-
fluenced one of the regional, horizontal principal stresses in the NNE direction (see
below). Such relationships most likely have existed during the late stage of the granite
solidification before its uplift. Field contacts at the Borsov quarry support this inter-
pretation: Aplite thin dikes that are associated with two joints seem to register melt fill-
ings along the existing joints, i.e., the NNE joints were available when this late fluid of
the granite injected. If aplite represents the fluid that created the joints, this rock should
have been associated with all (or most) joints, which is not the case. See an alternative
interpretation in Sect. 4.4.6. These joints are less likely to be the result of mechanical stresses
induced by thermal gradients during early stages of the cooling process. Such gradients
are likely to be influenced by local conditions, causing considerable strike deviations from
regional uniform joint patterns. This could not be recognized. It is also unlikely that the
NNE fractures are late (Tertiary) uplift joints, mainly because uplift horizontal sheets cut
the earlier vertical joints, and the latter do not show fracture morphologies indicating
downwards propagation, which characterize uplift joints.

Mirror ellipticities and their unilateral growth. Practically all the joints at Borsov
that reveal their boundaries show that their mirrors deviate from circularity into vari-
ous ellipticities. The incremental crack growth exhibited by J6 shows a pronounced
unilateral growth of the radii of arrest marks (Fig. 4.7b), suggesting a general slow
fracture growth but a relative faster fracture propagation along the 8° dip of the “mir-
 4.4 · Joints in Granites from the South Bohemian Pluton 317

ror symmetry axis” than normal to this direction. Note that the dip of the symmetry
axis of J3 is even greater (24°) (Fig. 4.6c); both deviate from either horizontal or verti-
cal directions. Apparently, the same joints that display eccentric origins also show
small angle dips of their symmetry axes on the mirror planes. Hence, some earlier
fractures that constrain the Borsov joints are not quite horizontal, leading to occa-
sional inclined mirror symmetry axes. The constraints that caused this unilateral
growth could be sub-horizontal unisotropies imposed by either petrographic bound-
aries (Bankwitz and Bankwitz, unpublished), differences between horizontal and ver-
tical compression, or earlier fractures. It appears that constraints by earlier fracture
influenced the growth of most Borsov joints as seen on joints J1, J2, J4, J5, J6 and J7.

Plutonic vs. sedimentary rocks. The wide spectrum of fracture surface morphologies
displayed by the Borsov joints that reflect different fracture processes in a single set
seem to be exclusively associated with jointing in granite during its late solidification
stage and do not occur in sedimentary rocks. Fracture markings from a given joint set
in sedimentary rocks are commonly limited to simple combinations of one to three
features shown in Fig. 2.1a, because they form by a restricted number of fracture mecha-
nisms. Rare complex combinations often reflect special processes. A case in point is
displayed by “mark overprinting”, where a superposition of a late uplift joint on an
early burial (diagenetic) joint shows different fracture surface markings, but these dif-
ferent markings occur along two different sets of joints (Bahat 1991a, p. 288). Unfortu-
nately, only partial mirror boundaries are known from sedimentary rocks, and these
are not sufficient for comparison with the present results. The eccentric origins and
small dips of the symmetry axes on the mirror planes are rare in sedimentary rocks. In
horizontal sedimentary layers, plumes generally are parallel to layer boundaries.

4.4.5.2
Geological Interpretation of Jointing in the Mrákotin Quarry

The joints of the Mrákotin set display composite fractures on three adjacent closely
spaced parallel planes in a fracture zone. The joints on each plane interacted inten-
sively with each other such that fringes between mirrors are either missing or increase
to large sizes. This interaction resembles the interacting early joints cutting the
El Capitan granite at Yosemite National Park (Bahat et al. 1999) (Sect. 4.5.3.3). There
are however several important differences between the two interactions.

Dynamic growth. Compared to the El Capitan set that did not develop full mirrors
and fringes, the Mrákotin set formed full mirrors and hackle fringes. The hackle fringes
imply dynamic growth, i.e., rapid growth of most joints under high stress intensity
conditions (Sect. 4.8.4), with the consequence of fracture interaction. Apparently, joints
of strongly reduced spacing within fracture zones can be distinctly parametrized from
joints in sets that do not occur in fracture zones (Rabinovitch and Bahat 1999). The
series of distinct, adjacent joints implies independent fracture nucleation from many
sites. The considerable size difference of the mirrors (Table 4.4) suggests that the growth
of these joints was strongly affected by local fluid pressure conditions. While most of
the Borsov joints (J1–J7) propagated sub-critically during most of their fracture his-
tories (Sect. 4.8.5), most of the Mrákotin joints essentially formed dynamically.
318 Chapter 4 · Jointing in Granites – Part 1

Ratios of long to short axes of joints. The reduction in the ratio of long to short axes of
the mirrors from the 3.3 to 4.5 range to the ratio of almost 1 (Table 4.4) for the full joints
including fringes is intriguing. The growth towards the ratio 1 is expected from the theory
that an ultimate growth of a fracture should follow the trend of decrease in ellipticity (e.g.,
Irwin 1962; Roy and Saha 1995). The long axes of the elliptical mirrors are horizontal.
What stopped the mirrors from growing to a circular (penny) shape? We did not identify
early sub-horizontal fractures comparable to those found to have influenced the growth
of some of the Borsov joints. The depth of emplacement of the Mrákotin granite was
determined by Büttner and Kruhl (1997) to be ~15–18 to 7–10 km (compare to Sect. 4.4.6).
Great overburden pressure at these depths may explain what prevented the mirrors from
propagating vertically, upward and downward and induced lateral propagation of the
mirrors. The question still remains of what has changed when growth occurred beyond
the mirror boundaries, and if it was gradual or abrupt. It appears that the change was
violent and abrupt, as suggested by the hackles and the quasi-triangle fringe of joint F
(Fig. 4.7b). Possibly, the change was caused by a sudden rise in fluid pressure that increased
the extensile stress on the joint mirrors, such that the overburden pressure was overcome.

Joints A and C. Joint C differs from all the other fractures in the Mrákotin set in hav-
ing a fringe consisting of en echelon segments, compared to the hackle fringes that
characterize the others. Furthermore, joint C is surrounded upward and sidewise by
the other joints (Fig. 4.10c and 4.12a). This again implies that the joints did not form
by a uniform stress field. It is suggested that fluid pressures were smaller or less effec-
tive at the center on joint C and greater around it.

4.4.5.3
Summary

This section focuses on the Mrákotin quarry, where a single “vertical wall” exhibits a set of
nine joints in a fracture zone. These joints nucleated independently from many sites.
The considerable size difference of the mirrors probably suggests that the growth of
these joints was strongly affected by local fluid pressure conditions.
The hackle fringes that characterize these joints imply dynamic growth, i.e., a rapid
growth of most joints under high stress intensity conditions, with the consequence of frac-
ture interaction.
For most joints, the ratio of long, sub-horizontal to short axes of the elliptical mir-
rors ranges from 3.3 to 4.5. However, this ratio is reduced to the range from 1.3 to 1.4
for the full joints (combined mirrors and maximum widths of the fringes for every joint).
The sub-horizontal mirror boundaries of the joints, particularly joint F, imply growth
constraints above and below the mirror possibly due to great overburden pressure, which
prevented the propagation of the mirror to a circular shape. At some stage the growth
was resumed, as demonstrated by the fringes, possibly by a rise in fluid pressure.
It is expected that the fluid pressures that induced fracture in the Mrákotin quarry had
to surpass the overburden pressure, and they reached maxima that were greater than the
pressures calculated by the fluid inclusion analysis (see below). Possibly, the fluid pressure
was smaller or less effective on joint C at the center and greater on the joints around it.
There is a genetic linkage between the joints investigated in the Mrákotin and Borsov
quarries. The strike of all mirrors at the Mrákotin quarry is roughly uniformly directed
 4.4 · Joints in Granites from the South Bohemian Pluton 319

at NNE, characteristic to the SBP, with the implication that this direction coincides with
one of the regional horizontal principal paleostresses (Sect. 4.7, below).

4.4.6
Timing, Depth and Temperature of Joint Formation

This section is cited from Bankwitz et al. 2004, who combined several techniques in
studying the timing, temperature and depth of NNE trending joints in the SBP (Fig. 4.5).

Methods. Samples of the Mrákotin granite were taken from fringes of the “Trefoil
joint” and the “Bilateral asymmetric joint” that outcrop in the Borsov quarry (Fig. 4.8a
and 4.7b, respectively) and from the the Weinsberg granite (Fig. 4.5) at the Friepeß
quarry in Austria. The K-Ar method was used on these samples for age determination
and for fluid inclusion analysis (applied on muscovites) of the temperature and depth
of intrusion of the granites (Scharbert 1998). The latter analysis was also applied on
Erzgebirge plutons (Fig. 4.2a).

Temperature and depth of intrusion of the Mrákotin granite and a late dike. Ac-
cording to Bankwitz et al. 2004, the minimum intrusion depth was estimated from the
P-T conditions of the trapping data of fluid inclusions (Table 4.5), using a lithostatic
model with a rock density of 2 600 kg m–3. In the Mrákotin granite, the average tem-
perature of homogenization in the fluid phase of seventeen fluid inclusions is T = 620 °C,
i.e., the trapping temperature of the inclusions that are under pressure conditions of
P = 1.89 kbar. Under the assumption of a lithostatic pressure at this crustal level, the
depth at the time of inclusion formation of these samples is estimated to be 7.4 km. For
the inclusions in the Weinsberg granite, the level of generation was much deeper, at
14.3 km with associated higher temperature and trapping pressure of 3.65 kbar.

Temperature and depth of intrusion of plutons in the Erzgebirge. The heating-

quenching microthermometric method was applied by Bankwitz et al. 2004 in the
analysis of 1 386 melt inclusions in quartz, apatite and feldspar of several plutons (e.g.,
Eibenstock, Kirchberg, Bergen and Ehrenfriedersdorf granites) of the Erzgebirge
(Fig. 4.2a). According to the deep seismic reflection profile, most of these plutons be-
long to a shallow granite “sheet” within the upper crust without deep roots (Behr et al.
1994). The estimated trapping pressures of melt inclusions in the various plutons range
between 1.1 ±0.1 kbar and 1.0 ±0.1 kbar, and trapping temperatures of 562 ±2 °C to
577 ±7 °C correspond to intrusion levels of about 3 km. Geological evidence for shal-
low intrusion levels comes from vitreous glass inclusions that occur together with

Table 4.5. Trapping data of fluid inclusions in quartz of a late-granite dike, intersecting the Mrákotin
granite (FK-1), and in quartz of the Weinsberg granite (FK-11). Both granites are granite-types of the
South Bohemian Pluton; n: Number of determinations
320 Chapter 4 · Jointing in Granites – Part 1

normal crystallized melt inclusions in granitic quartz of several Erzgebirge plutons..

Glass inclusions demonstrate rapid cooling or quenching that is not compatible with
a deep intrusion level (Thomas and Klemm 1997). Therefore, the first fractures could
be initiated during granite cooling below 560 °C at a depth of about 3 km, where the
trapping of the investigated inclusions of the Erzgebirge samples occurred.

Interpretation of timing, depth and temperature of joint formation. Bankwitz et al. 2004
arrived at the following sequence of events. The intrusion age of the Mrákotin granite is
327 ±4 Ma, and cooling took place during the 327–325 Ma interval. According to the cool-
ing age of muscovite extracted from a late-granite dike associated with jointing, fractur-
ing started at 324.9 Ma. After joint initiation cooling of the SBP continued to 320 Ma though,
312 Ma was measured in the southern part of the Cimer granite (Mrákotin granite is the
local name of the Cimer granite in the more northern SBP intrusion area). The younger
granite Homolka intruded at 320–314 Ma, and then the deep-rooting granite bodies in-
truded in the axial part of the SBP (the Zvule granite, Breiter 2001). The Mrákotin granite
intruded in about 800 °C at a deeper level during 327–328 Ma and exhumed to 7.4 km at
325 Ma. The NNE oriented joints in the Borsov quarry formed at 7.4 km depth in 324.9 Ma
by the late-granite dikes that induced natural hydraulic fracturing. The fluid inclusions of
the dikes were trapped in 620 °C. The dike then cooled to about 350 °C, given by the cool-

Fig. 4.15.
Model of early deep-seated
fracturing in solidifying gran-
ite magma of the South Bohe-
mian Pluton, summarized from
fractographic investigations on
fluid and melt inclusions (from
Bankwitz et al. 2004)
 4.4 · Joints in Granites from the South Bohemian Pluton 321

ing temperature of the muscovite in the dike in 325 Ma (Scharbert 1998). Finally, most
intriguing is the occurrence of NNE oriented joints, presumably of the same set, in plutons
that originated in extremely different depths: Joints from Borsov at 7.4 km and joints from
Weinsberg (Fig. 4.5) at 15 km.
According to Bankwitz et al. (2004), the sequence of intrusions in the SBP (Table 4.6)
combined with the above set of depths and temperature conditions, provide evidence only
for weak exhumation during the cooling stage between 327 Ma and 325 Ma. A recogniz-
able exhumation that started at only 310 Ma is known from the Bohemian Massif (west of
the SBP where analogous intrusions exist), based on results derived from apatite-fission
tracks. This process lasted during the 310–280 Ma period in association with the late-
Variscan exhumation (Wagner and Van Den Haute 1992; Menzel and Schröder 1994). The
exhumation was not a continuous process. Exhumation of 3.3 km is estimated to have taken
place during the first stage (Coyle and Wagner 1995). The second period of exhumation
occurred in Lower Triassic and the third in Upper Cretaceous/Tertiary times. Finally,
Bankwitz et al. (2004) conclude that the sequence of major events was:

1. Cooling of the Mrákotin granite melt down to the solidus.

2. Injection of residual melt into the solid phase of the Mrákotin granite, thus form-
ing late-granite dikes and creating the NNE-trending hydraulic fractures.

They hypothesize a multi-stage process of repeated cycles of melt generation and

fracture (Fig. 4.15), a model that may guide a future stimulating research.

Table 4.6. Succession of early main joints in granites of the SBP

322 Chapter 4 · Jointing in Granites – Part 1

4.5
Joints in Granites from the Sierra Nevada Batholith in California
4.5.1
Introduction

Bateman and Wahrhaftig (1966) discriminate between several types of joints in the Si-
erra Nevada (Fig. 4.16a). These include early “cooling”, later “steep”, presumably near
vertical (including vertical), and younger “gently dipping sheeting” as well as “exfolia-
tion joints” that are oriented like the steep joints. They found that generally it is pos-
sible to identify two principal groups of steep joints almost at right angles, one group
striking NE and the other NW. Bateman and Wahrhaftig (1966) did not distinguish dif-
ferent sets within the groups of steep joints. However, in some places nearly parallel
joints cross one another, and where a significant change takes place in the strike of one
joint set, the joints of one strike generally interfinger with the joints of another strike;
a description that raises suspicion about more than one set (Younes and Engelder 1999
and present Sect. 3.1.2 and 3.1.6). Furthermore, Bateman and Wahrhaftig (1966) note
that spacing between joints ranges from less than an inch to hundreds or even thou-
sands of feet. Such changes in spacing often provide guidelines in separating joints into
distinct categories and possibly distinct sets (Bahat 1991a, p. 308). Thus, it seems quite
possible that the near-vertical joints belong to more than two orthogonal sets.
Bateman and Wahrhaftig (1966) consider that the continuity of joints across bound-
aries between individual plutons indicates that the joints formed after the consolida-
tion of the entire Sierra Nevada batholith, before the Eocene. The abundance of regional
near-vertical joint sets striking N-NW and E-NE that are approximately perpendicular
to each other in the Sierra Nevada Batholith was noted by additional others (e.g., Becker
1891; Huber 1987, Fig. 33). Lockwood and Moore (1979) observe that strike-slip
microfaults (original joints on which some displacement is measurable) are pervasive
throughout the granitic rocks of the eastern Sierra Nevada. Offsets typically range from
less than a millimeter to several tens of centimeters, but exceed 100 m in some places.
The microfaults are oriented in two nearly vertical conjugate sets: A N-NE striking set
showing right lateral offset and an E-NE striking set showing left lateral offset, the lat-
ter is more common. They also plotted and measured lineaments (Fig. 4.16b,c) and
found that most lineaments visible on aerial photographs are microfaults. The age of
this microfaulting is not precisely known. It developed after consolidation of the young-
est granitic plutons in the Sierra (79 Ma) and is known to cut a late Miocene volcanic
dike in one area. Slickensides along the microfaults are sub-horizontal but show an av-
erage plunge of about 3° westward, suggesting that much of the deformation occurred
prior to the westward tilting of the Sierran block in late Tertiary time. The direction of
maximum horizontal extension strain (determined as the bisector of average microfault
trends) changes (rotates) systematically from north to south (WNW at 38.5° N; NW
at 36.5° N). A detailed study of microfaults at 37°20' N indicates a maximum extension
of 2.3% in a N 61° W direction. These extension directions are remarkably parallel to
late Mesozoic to present-day tectonic extension directions in the Basin and Range prov-
ince. Thus, Lockwood and Moore (1979) differ in the description of fracture from
Bateman and Wahrhaftig (1966); they found two major conjugate sets of microfaults,
(N-NE and E-NE) rather than two major groups almost at right angles, one set striking NE
 4.5 · Joints in Granites from the Sierra Nevada Batholith in California 323

Fig. 4.16.
a Map of California showing
the Sierra Nevada Batholith.
Area of Fig. 4.16b is shaded and
locations of Shaw quarry (S)
and Yosemite National Park (Y)
are marked (modified after
Lockwood and Moore 1979).
b Average trends of most promi-
nent lineaments in thirteen
15-min quadrangles (modified
after Lockwood and Moore
1979). c Lineaments mapped
from aerial photographs in four
15-min quadrants. Heavy lines
are normal faults (from Lock-
wood and Moore 1979)
324 Chapter 4 · Jointing in Granites – Part 1

and the other NW. However, Lockwood and Moore (1979) agree with the interpreta-
tion of Bateman and Wahrhaftig (1966) regarding the “post consolidation” timing of
fracture. They conclude that the pattern of microfaulting demonstrates that the (sup-
posedly monolithic) Sierran terrane was affected by the late Cenozoic and possibly ear-
lier regional extension of western North America. Huber (1987) considers that jointing
could have occurred during late tectonic events, such as tilting of the Sierra region, 25
to 15 million years ago. Accordingly, these fractures reflect syntectonic (possibly uplift)
processes rather than cooling ones.
In contrast to Bateman and Wahrhaftig (1966) and Lockwood and Moore (1979),
Balk (1937) and Mayo (1941) consider that the steep orthogonal fracture systems in
the Sierra Nevada developed during the emplacement of the plutons, and Segall et al.
(1990) suggest that they formed soon after the host pluton was emplaced, between 85
and 79 Ma. Zawislanski (1994) suggests that in the Knowles granodiorite that has been
dated as 111 Ma (see below), pegmatites, representing late granite differentiates, may
have filled early tension-fractures related to the cooling of the pluton. Thus, it seems
justified that the fractures described by Balk (1937), Mayo (1941), Segall et al. (1990)
and Zawislanski (1994) be termed cooling joints.
Restating the problem, it is possible that different authors may have assigned steep
joints to distinct times in relation to the petrological history of the pluton (during
emplacement vs. post consolidation of the pluton), and that they consider different
fracture systems to represent three or four regional sets rather than two that devel-
oped during different geological periods. Alternatively, as inferred from Lockwood and
Moore (1979), there was a systematic change in stress direction that resulted in the
formation of a series of orthogonal sets of joints and faults (Segall and Pollard 1983),
which underwent rotation with time. This problem that confronts the concepts of pro-
longed paleostress directions vs. rotating paleostresses is analogous to a similar one
that relates to the distinction of joint sets in sedimentary rocks (Sect. 6.3).
There are in addition, very young joints that seem to fall into the two categories,
termed, “gently dipping sheeting” and “exfoliation joints” (Bateman and Wahrhaftig
1966). Becker (1891) suggests that (presumably the same fractures) are Pliocene in age
and Lockwood and Moore (1979) propose that they are post-Pliocene. In summary,
there are still many uncertainties about the multi-fracture behavior in the Sierra Ne-
vada Batholith that await further investigation. The present section narrows its inter-
est into only two fracture domains in the Sierra Nevada Batholith, the oblique joints
in the Knowles Granodiorite and the exfoliation joints that cut plutons in Yosemite
National Park.

4.5.2
Petrographic Study of Two Oblique Joints from the Knowles Granodiorite
in Western Sierra Nevada

4.5.2.1
Previous Studies on the Knowles Granodiorite

This section relates to the surface properties of two joints from the Knowles grano-
diorite (Bateman and Sawka 1981). These joints are exposed in the Shaw quarry, next
to the Raymond field site (Karasaki et al. 2000), which is located near the edge of
 4.5 · Joints in Granites from the Sierra Nevada Batholith in California 325

Rte 411, 5 km east of the town of Raymond and 70 km north of Fresno, at the western
foothills of the Sierra Nevada Mountains (Fig. 4.16a). The Knowles granodiorite plu-
ton is 15 km by 5 km, elongating in the NS direction and is fine to medium grained,
generally not deformed, and has an isotopic age of 111 Ma (Stern et al. 1981; Bateman
and Sawka 1981). It intruded the tonalite of Blue Canyon that has been dated at 114 Ma.
Zawislanski (1994) measured 438 fractures in the area around the Shaw quarry and
grouped them into four: Two orthogonal, near vertical sets striking at N 30° W and
N 60° E and dipping between 75° and 90°, a sub-horizontal set that dips to the west at
about 5°, and a less prominent set, termed “oblique”, striking at about N 45° E and dip-
ping between 30° to 60° to the NW. He observed that pegmatite dikes are pervasive in
the granodiorite. The dikes are oriented in two sets roughly 90° apart, one striking
N 35° E, the other striking N 70° W. Dikes in both sets dip between 20° and 60° to the
NW and NE. Thus, the orientations of the pegmatite dikes do not coincide with either
the near vertical or sub-horizontal fractures, but are similar in orientation to some of
the oblique joints. Based on the lack of well-developed chilled margins in the pegma-
tite dikes, Zawislanski (1994) considers that the dikes were emplaced shortly after the
granodiorite emplacement. The dikes are routinely cut by the two near vertical sets,
generally with very little to no displacement. Also, the rough coincidence of the
NW dipping dikes with oblique joints suggests that the pegmatites may be fillings of
early cooling joints. Hence, the sequence that followed the rock emplacement seems to
have been oblique joints, pegmatite dikes and near vertical fractures. Zawislanski (1994)
also found that sub-horizontal fractures cut the sub-vertical ones. Since the dip of the
sub-horizontal fracture roughly coincides with that of the regional slope, he concludes
that they are young, unloading fractures.
Zawislanski (1994) also observed that most fractures are very tight and have no sig-
nificant secondary mineralization, although apertures are often ambiguous in surface
exposures due to the tendency of weathering processes to erode and “open” fractures.
Displacements on the order of 1 to 2 cm were observed on some sub-horizontal and
oblique fractures. However, no measurable displacement was observed on near verti-
cal fractures. Spacing is irregular, with some fractures as closely spaced as a few centi-
meters (in fracture zones), and some as widely spaced as tens of meters. The average
spacing of the near vertical fractures is 0.95 m (standard deviation = 0.85; n = 85).

4.5.2.2
Joints C and D in the Knowles Granodiorite from the Shaw Quarry

The fracture classification in granites by Cloos (Fig. 4.1) that divides fractures into
near vertical, sub-horizontal and diagonal (oblique) categories provides a good basis
for comparison of fractures from different domains and provinces. Most previous ef-
forts of fracture investigation concentrated on the first two categories, almost ignor-
ing the oblique fractures. However, some intriguing observations that were made on
diagonal fractures in plutons (e.g., Cloos 1921, 1922; Zawislanski 1994) (Sect. 4.5.2.1)
seem to justify their detailed study. Furthermore, near vertical fractures seem to be
rare in the Shaw quarry (Fig. 4.17). Accordingly, the present section focuses on two
diagonal joints that cut the Knowles granodiorite in the Shaw quarry. Joints C and D
that are described in this section differ in their orientations. Joint C strikes 078°
(N 78° E) and dips at 42° to the SE and joint D strikes 045° and dips at 36° to the NW
 326
Chapter 4 · Jointing in Granites – Part 1

Fig. 4.17. Photographs of a part of the Shaw quarry, showing joint D in three scales. a The joint dips from left to right at the lower center of picture and appears to
be sub-parallel to several joints above it, that belong to the same set (D is about 3 m long). The joints partly show concentric arrangement. An oblique plume
(horizontal arrow) is seen on a joint from set C. b Three drills along joint D from which rock cylinders have been removed. c Part of the joint in a close up (note two
syringe needles). d A large plume (delicate lines) is examined by students on a “oblique-concentric joint” in a granite outcrop from the Bohemian Massif
 4.5 · Joints in Granites from the Sierra Nevada Batholith in California 327

(Fig. 4.17a). The detailed petrography of the fracture apertures of joints C and D is the
main subject matter here. In studying the joints, the “epoxy procedure” was applied in
order to obtain in-situ samples for preparing thin sections of the fractures (Sect. 4.12).

Microphotography. A series of photo micrographs each covering a square area of

4.5 × 4.5 cm were taken by both plane light and plane polarized transmitted light from
the examined thin sections. These micrographs, partly overlapping each other in or-
der to minimize photograph distortion, were assembled into mosaics covering sec-
tions along the joint apertures. A comparison between images obtained by the two
lights enables one to distinguish the boundaries of the primary fractures C and D
(Fig. 4.18–4.20) and branching of a secondary fracture from joint C (Fig. 4.18a
and 4.19c), as well as secondary precipitates, additional very thin cracks, and other
common petrographic features.

Distinction between transgranular and intergranular fracture. Both transgranular

and intergranular fractures were observed on joints C and D. Several criteria for
transgranularity were defined. First, a mineral that shows a uniform extinction at both
margins of the joint is more likely to represent a transgranular fracture and is less
likely to represent late crystallization of two independent minerals in precisely the
same orientation. A further confirmation of transgranularity is obtained when corre-
sponding locations of a certain boundary and/or polysynthetic twins of the mineral

Fig. 4.18. Thin sections of joints C and D constructed by mosaics of micro-graphs. Photographed
sections are 4.5 cm long. a Joint C, showing straightness of the primary fracture (sub-horizontal)
that branches at a large embayment (dark) to the secondary fracture, which inclines downward, along
a transverse section (parallel to the face of the original rock). b Joint D, showing curvatures along a
longitudinal section (parallel to the hole formed by drilling). Quartz appears in white and gray, feld-
spars and muscovite in gray and biotite in black colors
328 Chapter 4 · Jointing in Granites – Part 1

Fig. 4.19.
Profiles of fracture C.
a Waviness in double
refraction occurs on a
quartz crystal that is
partly cut radially by
microcracks. Aperture
at center is 0.5 mm
(longitudinal section).
b Inclined crystals of
micas along the upper
margin of the primary
fracture. An angular
embayment at lower
right of fracture
(arrow) is filled with
unoriented, altered
micro muscovite grains.
A possible third gen-
eration of micro micas,
quartz and sericite are
scattered along the
fracture. Aperture at
right is 0.5 mm (trans-
verse section). c Trans-
granular fracture by
mode I (without shear
displacement) along
the branching second-
ary fracture. Matching
is very good in quartz
two grains and biotites
(scale of several mm)
but is poor in the sub-
millimeter scale. Aper-
ture at center, next to
the biotites (arrow) is
0.1 mm (transverse
section)
 4.5 · Joints in Granites from the Sierra Nevada Batholith in California 329

Fig. 4.20.
Profiles of fracture D.
a A curved transgranu-
lar fracture that cuts a
K feldspar 5.5 mm long,
but propagates around
the boundary of a bi-
otite, at lower right
(longitudinal section).
b Two adjacent large
quartz crystals at dif-
ferent orientations, ap-
pearing as dark and
light patches. The dark
one is cut by a trans-
granular fracture. An
intergranular fracture
propagated along the
grain boundaries of me-
dium size plagioclase
and biotite at left of cen-
ter. The length of biotite
at center (arrow) is
0.8 mm (transverse
section). c A transgranu-
lar fracture through a
twinned plagioclase
(arrow) and K feldspars
(light and dark colors),
showing considerable
variation in aperture
and size of mineral
fragments along the
fracture. The length of
twinned plagioclase is
1.6 mm (traverse sec-
tion). Matching is good
in the feldspar scale
(several mm) but is
poor in the sub-milli-
meter scale where frag-
ments required to re-
construct the original
mineral are missing
330 Chapter 4 · Jointing in Granites – Part 1

can be traced on both sides of fracture (Fig. 4.20b,c). A morphological matching on

the sub mm scale, additional to above criteria, provides the highest degree of certainty
for transgranularity. Such matchings, however, were rare (see below). Transgranular
fracture is a useful tool in determining the extents of opening by mode I vs. shear
offsets along the joint walls. When above transgranular criteria are established on a
series of different adjacent minerals, their reconstruction to “zero displacement” re-
veals the extent of opening. A reconstruction that shows deviations from “mode I clos-
ing” reveals the extent of shear offset. In contrast to transgranular fracture that cuts
through the grain along curved (e.g., Fig. 4.18b and 4.20a) or more often, along straight
(e.g., Fig. 4.18a and 4.19c) trajectories, intergranular fracture curves along grain bound-
aries (Fig. 4.20a). Accordingly, different extinction angles of two grains of the same
species on both sides of fracture imply intergranular fracture, particularly when the
grain boundaries are not straight.

4.5.2.3
Results

Fracture C. The primary fracture C branches out into a secondary one so that it is rep-
resented by two fractures (Fig. 4.18a). Both primary and secondary fractures are straight
and maintain approximately uniform apertures, ranging from about 0.1 mm to about
0.5 mm, but the joint walls recede considerably from each other into a large embayment
at the branching site and in several other locations along the fractures. Fracture is char-
acteristically transgranular, mostly cutting through the larger feldspar and quarts crys-
tals. Occasional intergranular fracture cuts around medium and small size minerals
(feldspars, quartz and biotite). There are mineralogical and morphological mismatches
along the section shown in Fig. 4.19a, raising the suspicion of a lateral offset along the
longitudinal section of primary C. One or two transgranular contacts of quartz grains
(arrows in Fig. 4.18a) suggest left lateral offset of about 4 mm. Wave extinction in a
quartz crystal, accompanied by local radial microcracks reveal strain in the quartz,
possibly caused by an inclined impingement on the primary fracture wall.
The space along the primary fracture is empty in certain locations, but two or pos-
sibly three generations of secondary muscovite are aligned in others (Fig. 4.19b). The
first generation consists of 0.7 mm long fresh muscovite grains that display high or-
der double refraction colors. Some of them form small angles (0–20°) with respect to
the fracture walls. These micas are kinked, reflecting shear strain during their history.
The second generation occurs as small altered patches 1–2 mm in size that often fill
small embayments along the fracture walls (making it look straighter than it origi-
nally was). These muscovites are often altered to sericite, and their double refraction
colors are hardly recognizable. The joint space is filled in some locations by late, mi-
cro crystals (<< 1 mm) of mica/sericite and xenomorphic quartz.
Uniform extinction and grain-boundary fit of several minerals along parts of
transgranular fracture surfaces of the branching joint testify to an exclusive mode I frac-
ture without any discernable lateral offset (Fig. 4.18a and 4.19c). Matching of the fracture
walls along a transgranular fracture in quartz occurs in a scale of several millimeters.
However, characteristically, some quartz is missing, thereby negating an unequivocal match
at the 0.5 mm scale (left center of figure). A considerably altered grain of biotite 3 mm long
is cut by the fracture without lateral displacement (at right side of fracture) (Fig. 4.19c).
 4.5 · Joints in Granites from the Sierra Nevada Batholith in California 331

Fracture D. Fracture D occurs along a single structure (Fig. 4.18b). The transverse and
longitudinal sections reveal curvy fractures to various extents. The longitudinal section
exhibits curvatures at two wave lengths. The larger one is about 5 mm (Fig. 4.18b) and the
smaller one is about 1 mm (Fig. 4.20a), although there are some straight parts as well. The
aperture size deviates considerably from uniformity, between 1.2 mm (Fig. 4.20a,c) down
to 0.2 mm (Fig. 4.20b). Fracture is transgranular through the larger crystals of K feldspar
(Fig. 4.20a), quartz (Fig. 4.20b) and plagioclase (Fig. 4.20c), along both straight and curved
traces. But it is occasionally intergranular around medium- and small-sized minerals (feld-
spars and biotite) (Fig. 4.20a,b). The transverse section shows a very slight left lateral off-
set (Fig. 4.20c). Secondary micas are rare along this section, but fragments from broken
quartz and feldspars are common along trans-granular parts of the fracture. There is a
mismatch on the scale comparable to the joint aperture. No shear offset can be identi-
fied in the longitudinal section, and matching improves considerably at 5 mm and above
(see good matching in the large K feldspar, in Fig. 4.20a). Thus, the longitudinal and
transverse sections both show evidence for just mode I operation, or nearly so.

4.5.2.4
Fracture Behavior in the Two Joints from the Knowles Granodiorite

Comparison with a previous study. Moore and Lockner (1995) investigated the
microcrack density (crack length per unit area) in a 19.5 cm-long cylinder of Westerly
granite, after inducing a shear fracture by an axial load and 50 MPa confining pressure,
at rates less than 10 mm s–1. Their results show that three-quarters of the examined fault
length follows intragranular cracks, with only one fourth along grain boundaries. Also,
microcrack densities are consistently higher on the dilational side of the shear than on
the compressional side. The uneven distribution of microcracks along the “fault” tends
to pull the propagating fracture tip towards the dilational side, even though the trend
is away from the overall fault strike. As a result, the propagating shear follows the
microcrack trend for some distance and then changes direction in order to maintain
an overall in-plane propagation path, resulting in a zigzag or saw tooth fracture pat-
tern. Although no systematic investigation of microcrack density was carried out on
fractures C and D, an overall impression is that also in the latter fractures, intragranular
cracking is more common than intergranular (Fig. 4.18–4.20). However, no distinctions
can be made between zones of higher and lower microcrack densities, i.e., no separa-
tion between “dilational sides” and “compressional sides” is possible. The waviness along
fracture D (Fig. 4.18b) does not resemble the angularity of the zigzag fracture pattern
reported by Moore and Lockner (1995, Fig. 14), and fracture C is straight. This com-
parison possibly suggests that the fracture mechanisms in these two studies were dif-
ferent, shear in the Westerly granite and essentially tensile in the Knowles granodior-
ite. Nevertheless, in both cases intragranular cracking was dominant.

Fracture sequence. While there seems to have been some shear offset along the primary
fracture, no offset is identified in the secondary fracture of joint C. The differences in the
modes of displacement between the primary and secondary parts of the joint and lack of
muscovite along the secondary fracture suggest that fracture branching occurred after
displacement(s) along the primary had taken place. The small angle between the two frac-
tures suggests that a reverse sequence would have caused offsets along the secondary frac-
332 Chapter 4 · Jointing in Granites – Part 1

ture as well. This implies that the geometric splay (Fig. 4.18a) does not reflect a dynamic,
single branching of the secondary from the primary in the strict mechanical sense of the
term, i.e., the two fractures did not propagate simultaneously under conditions of KI > KIC
(Sect. 4.8.1 and Fig. 2.30c). Thus, a slight opening of the joint enabled the introduction of
K-silica rich solutions that produced muscovite, sericite and quartz along the primary
joint. Next, a minor shear offset took place along the primary, and then, slightly tilted
tensile stress resulted in the formation of the branching fracture.
Straightness characterizes both the primary fracture and the secondary branch of
joint C, regardless of the fracture mode, i.e., the offset (s) along the primary that
amounted to about 4 mm did not influence the fracture straightness. Quite possibly
during the early history of the rock while at depth, lithostatic stresses could have closed
the fracture, inducing an impingement of the two fracture walls (Fig. 4.19a), which pro-
duced the fragments along the joint, even without lateral offset. This could be an im-
portant mechanism contributing to the formation of embayments along the fractures.
The curved part of fracture D reveals a typical transgranular mode I loading
(Fig. 4.18b and 4.20a). This precludes the possibility that fracture occurred by curving
due to local shears, e.g., due to two approaching non axial fractures from opposite di-
rections (e.g., Yokobori et al. 1971; Acocella et al. 2000). The lack of muscovite flakes along
joint D implies that joint C is younger or older than joint D, i.e., joint D formed before
the initiation of the hydrothermal activities or after their termination, respectively.
However, if analogy is made to the lack of muscovite in the secondary joint C compared
to the older, primary C, it appears that joint D is younger than joint C.
It is possible that the relative multitude of joints deviating considerably from vertical-
ity in the Shaw quarry may have been the reason for the rarity of vertical or near-vertical
fractures in the quarry. A rational for this would be that early oblique fractures released a
major amount of stored energy in the rock, minimizing the driving force of subsequent,
vertical fracture. This would imply that joints C and D are older than the vertical joints in
the area. Such a sequence is supported by the observations made by Zawislanski (1994) in
the nearby Raymond field site (Karasaki et al. 2000) that the formation of oblique (diago-
nal) joints and intrusion of pegmatite dikes preceded the formation of vertical dikes. Thus,
further studies are needed to establish whether systematic diagonal fractures have pre-
ceded vertical fracture sets in the Sierra Nevada. Finally, some of the oblique straight joints
in the Shaw quarry appear to be oblique-concentric joints (Fig. 4.17a), and there is no clear
categorization of these fractures. Similar dilemmas occur in other granite provinces as well
(Fig. 4.17d), possibly suggesting a distinct class of oblique-concentric joints.

4.5.3
Joint Exfoliation and Arch Formation at Yosemite National Park, California

4.5.3.1
Introduction

The granitic rocks at Yosemite National Park constitute part of the Sierra Nevada
batholith (Fig. 4.16a). The “monoliths” El Capitan and Royal Arches are part of the “in-
trusive suite of Yosemite Valley” (Huber 1987). The El Capitan granite intruded older
plutons about 108 million years ago and now makes up the bulk of the western half of
the valley area (Fig. 4.21). Huntington (1966) and Huber (1987) found that fracture in
Fig. 4.21. Bird’s-eye view of Yosemite Valley, with selected landforms. Note particularly locations of El Capitan, Royal Arches and Half Dome (from Huber 1987, Fig. 18)
4.5 · Joints in Granites from the Sierra Nevada Batholith in California
333
334 Chapter 4 · Jointing in Granites – Part 1

the Yosemite Valley varies with the rock. Both composition and texture influence the
spacing of joints, such that the finer-grained diorite is the most closely jointed rock and
the El Capitan granite is the least fractured (Huber 1987). Huber (1987) suggests that
the vertical cliffs of Yosemite may have hidden vertical fractures behind and parallel to
the cliff face, i.e., vertical exfoliation possibly takes advantage of free surfaces provided
by previous jointing. Important studies on exfoliation in granitic rocks from Yosemite
National Park have also been carried out by Cadman (1969) and Holman (1976).
Arches also occur in granitic rocks. However, arches in granites are far less abun-
dant than in sandstone at Zion National Park (Sect. 3.3). Moreover, their occurrence
is selective. We divide arches for convenience in this section into “small” and “large”
ones. Small arches exclusively develop at the foot of the cliff of El Capitan. Large arches
“decorate” selectively one monolith at Yosemite National Park with the majestic “Royal
Arches” (Fig. 4.21) but almost no others. Hence, there is some intriguing variability in
exfoliation and arch formation in granitic rocks.
Qualitative and quantitative fractographic and electronic surveying methods have
been applied in studying exfoliation joints in middle elevations of the cliff of El Capitan
and on the cliff of Royal Arches (Bahat et al. 1995, 1999; Grossenbacher et al. 1996),
and these are cited here. Characterization of arches at the foot of El Capitan and a
comparison of their fracture mechanisms to those that formed arches at Zion National
Park are included as well.

4.5.3.2
Joint Exfoliation and Arch Formation at the Foot of El Capitan Cliff

We start in characterizing the exfoliation joints at the foot of El Capitan cliff, by

fractographically describing two types of plumes, the “curved-concentric plume” that
exhibits some unique features and the “sub vertical” one, which displays rather com-
mon features. We then describe the “full arch” and the “series of partial-arches”.

a. The curved-concentric plume. This plume occurs on a curved section of the cliff
along which there is a gradual strike change of approximately 2–4° (a precise mea-
surement in this location is technically very difficult). The strike change is greater in
the upper part of the plume than in the lower one (Fig. 4.22a,b). The curved-concen-
tric plume is divided into three areas (Fig. 4.22b):

1. The plume at the center, p.

2. A series of sub-concentric fractures, cf, at its southwestern side (at left upper part
of figure).
3. En echelon cracks in a fringe, e, at its northeastern side (at the right side of figure),
as itemized below.
▲

Fig. 4.22. a A photograph consisting of a mosaic of five photos and b drawing, showing the curved
concentric plume on a curved surface (resulting in some optical distortion), at the foot of El Capitan
cliff. In the drawing, p, cf, e and n, designate the curved-concentric plume, sub-concentric fractures, en
echelon cracks and niches, respectively. Also, small B designates fracture boundaries of the fringe,
f, and z marks merge zones along the boundary fracture, BF. Numbers 1–3 designate the three distinct
areas. Arrows show directions of fracture propagation. The line AB marks a section, which is rotated at
90° in c, to show (exaggerated) third dimension of the joint surface
4.5 · Joints in Granites from the Sierra Nevada Batholith in California 335
336 Chapter 4 · Jointing in Granites – Part 1

Area 1. The most striking feature in this fracture morphology is the curvature of the
plume that propagates upward, and after reaching a peak it bends and the barbs (indi-
vidual segments of the plume) turn sidewise and then downward, revealing a rotation
of the fracture front. The barbs are not uniform in shape along their lengths. As the
plume curves, the barbs bend slightly away from the cliff, forming small niches in
certain intervals (n in Fig. 4.22b,c), as arches of a centimeters or tens of centimeters
in size. The bends become particularly prominent at locations of strong surface cur-
vatures, where barbs at higher elevations partly superpose those below them, such
that they form a “staircase” climbing up the cliff. There is a gradual increase of the
radius of curvature of the plume in higher elevations.

Area 2. The curved-concentric plume becomes coarser, transforming into a series of

large niches and concentric fractures at the meter scale (n and cf at the left sides of
Fig. 4.22b,c). These fractures cut laterally (asymptotically) into the rock, partially as-
suming the shape of partial-arches at meter or tens of meter scales. The upper end of
this structure consists of three lateral boundary fractures, BF that meet at two “merge
zones” (z in Fig. 4.22b) that appear as triangle-like discontinuities along the other-
wise continuous curved boundary (Fig. 4.22a).

Area 3. The termination of the concentric plume at its right side is not abrupt; it oc-
curs along two, parallel, straight boundaries (B in Fig. 4.22b) that separate the parent
joint from the fringe (Hodgson 1961a). The plumes approach these boundaries sub-
orthogonally, and beyond, en echelon cracks, e, maintain the same sense of stepping

Fig. 4.23. Growth of cracks round by point after bends: a Straight-fronted crack moving on a pris-
matic surface with crack tip parallel to axis of prism. b Principal frame fields on a curved crack front.
c Continued growth of a crack, with a curved front, by twisting into numerous small en echelon cracks.
d Continued growth of a crack with a curved front, by bulging (from Hull 1999)
 4.5 · Joints in Granites from the Sierra Nevada Batholith in California 337

shown in the plume (the same “staircase” as in area 1). Many en echelon cracks are
straight and are slightly open. The propagation directions (arrows) are marked in the
various transitional parts from area 1 to areas 2 and 3 (Fig. 4.22b).
The fractography displayed by the curved-concentric plume reflects the mechanism
of exfoliation around the curved foot of El Capitan (Fig. 4.22a). The mechanism of crack
growth into round bends has been investigated by Hull (1999, p. 88) and is cited below.
According to Hull (1999, Fig. 3.25), a curved surface is formed by crack tilting (mode II),
provided the crack front remains straight and normal to the plane of the specimen. For
a variety of reasons, real cracks do not have straight fronts, even in flat plate specimens
tested in uniaxial tension. When changes in the local externally applied stress fields cause
a crack with a curved front to bend, more complicated crack movements are required.
Figure 4.23 is a 3-D representation of the simple case for a straight-fronted crack. The
crack front AB is always parallel to the x-axis (Fig. 4.23a). A curved surface is formed
by tilting about this axis. At any position PQ, the crack tip is always straight and paral-
lel to the x-axis. For the realistic case (Fig. 4.23b), the crack front is curved (see ripple
marks in Fig. 3.24). Suppose that initially, the crack is in a planar surface and that as it
grows the stress fields caused it to bend on a path described by the broken lines. The
principal frame fields marked at the tip of the crack show that the crack can continue
to propagate in the planar surface without any geometrical constraint. Any stresses that
cause the crack to bend, as illustrated, result in local stress fields at the crack tip with a
mode III component. This applies to all elements of the crack tip not parallel to the x-
axis. The crack tip experiences twist forces. However, to maintain a smoothly curving
surface, the crack front cannot move by twisting.
There are four possible ways for a crack that is forced to bend to follow the general
path marked by the broken lines in Fig. 4.23b. First, the curved crack front straightens
and becomes parallel to the x-axis. The crack then follows a prismatic surface for con-
tinuous tilting at the straight crack tip (Fig. 4.23a). Second, the crack front remains
curved and the crack is driven by the stress fields to bend on a path equivalent to the
prismatic surface. This inevitably means that the surface cannot remain smooth and
continuous. The local mixed I / III stress fields at the crack tip nucleate an array of en
echelon segments, separated by steps on the fracture surface, parallel to the direction
of crack growth (Fig. 4.23c). Third, the crack front remains curved and follows a
smooth 3-D double curvature path. This is achieved by the crack front bulging out of
the prismatic surface Fig. 4.23d). This diagram shows striae and undulations on the
surface with the undulations mapping out the successive positions of the crack tip for
a non-twisting crack. Fourth, the crack front moves by a combination of the first three
mechanisms, such that the relative contributions depend on the ease of initiating ar-
rays of secondary cracks at the tip of the parent crack.
The propagation of the curved-concentric plume (Fig. 4.22) approximates Hull’s
(1999) 3-D model (Fig. 4.23). The curved front of the exfoliation joint propagates
around the curved surface of the cliff (see the curvilinear shapes of the long arrows in
Fig. 4.22b). Breakdowns due to local mixed I / III stress fields occur at the joint tip,
initially into delicate plumes, then to coarse plumes and finally, to en echelon segmen-
tation. The breakdown of the joint front (Fig. 4.23c) and the bulge (Fig. 4.23d) reflect
3-D dimensions of Hull’s model, while the niches of plumes and steps of the en ech-
elon segmentation represent the actual 3-D on the cliff (Fig. 4.22a,b).
338 Chapter 4 · Jointing in Granites – Part 1

b. The sub-vertical plume. This plume occurs on a planar section of the cliff on a
back wall of a full arch (Fig. 4.24). The convex side of an undulation faces downward,
indicating that the plume propagated downward. The plume produces an asymptotic
fracture at an angle ≥ 1° with the cliff. This undercutting reveals its tensile (Bahat
1979) role in the formation of joint exfoliation.

c. The full arch. A few tens of meters from the curved-concentric plume southwest-
wards along the cliff, there is a fully developed arch that consists of a ceiling and of
two sidewalls (Fig. 3.28 and 4.25). The height of the arch from base to ceiling is about
7 m. At the top of the upper part of the arch is a merge zone of two boundary frac-
tures, which assume the shape of partial-arches. Each partial-arch consists of parts of
the ceiling and of a sidewall, such that the concave side of the partial-arch on the right
faces towards the left, and on the left the concave side of the partial-arch faces towards
the right. Hence, the partial-arches behave independently of each other when form-
ing a full arch.

Fig. 4.24.
A coarse, vertical plume un-
dercuts thin slices from the
back wall of an arch (at both
sides of the plume) at the foot
of El Capitan cliff; the sidewall
of the arch is shaded at right.
The undulation (u) maintains
orthogonal relation with the
plume. Note hammer for scale
at lower left
 4.5 · Joints in Granites from the Sierra Nevada Batholith in California 339

Fig. 4.25.
A full-shape arch at the foot of
El Capitan cliff. A little off the
top of the arch is a merge zone
(arrowed) of two partial-arches.
Each partial-arch consists of
part of the ceiling and a side-
wall. Picture is taken looking
upward the cliff (note an opti-
cal distortion). Size of arch ap-
proximates those of the side-
walls in Fig. 4.24 and 4.26

d. The series of partial-arches. There are many partial-arches that do not match
with other partial-arches to form full arches. A series of some five individual left
partial-arches that are convex in the same direction is shown in Fig. 4.26. Some
partial-arches are subdivided into parallel thin slices, about 1 to 10 cm in thickness
that form a small angle ≥ 1° with the cliff and laterally undercut into the rock.
The sidewalls of the various partial-arches resemble the sub-concentric fracture cf
(Fig. 4.22a–c).

Discussion and conclusion. The gradual increase in the topographic coarseness from
delicate barbs, through small en echelon segments into large ones, can be correlated
with the local increase of the KIII / KI ratio on the curved section of the cliff. Particu-
larly, large segments occur at the upper part of the fringe (around e in Fig. 4.22a,b).
These are also the most twisted and opened cracks, implying the greatest ratio of KIII / KI
(Cooke and Pollard 1996). Sommer (1969) observed segmentation in AR glass at as
little as 3.3° rotation of the parent fracture from normality to the tensile stress, and
this is approximately the change in strike occurring across the areas 1 and 3 (Fig. 4.22b).
The joint growth upward and downward (as demonstrated by the curved-concentric
plume) along a curved front resulted in plume coarsening and en echelon segmenta-
tion, following Hull’s model (Fig. 4.23c).
340 Chapter 4 · Jointing in Granites – Part 1

Fig. 4.26.
A series of partial-arches that
concave towards right at the
foot of El Capitan. Thickness
of sidewall (between arrows)
is about 80 cm

The plume trajectories mark the circular path of the joint front (arrows in Fig. 4.22b).
This path differs radically from the unilateral propagation of the vertical plume on
the back wall of an arch (Fig. 4.24); both are at approximately the same height on the
cliff, only tens of meters away, implying different fracture mechanisms, possibly in
different times. It demonstrates the small changes in local stress fields (across tens of
meters) within larger stress fields (across hundreds of meters) dominated by remote
stresses. There seems to be a gradual structural transition from curved-concentric
plumes through niches and sub-concentric fractures into partial-arches. They form
sets of various sizes that undercut laterally into the rock, forming small angles of sev-
eral degrees with the cliff. When matching (left and right) partial-arches merge, full
arches develop. A merge often occurs at one (Fig. 4.25) or two zones along the ceiling.
Thus, the various fracture features shown in Fig. 4.22, 4.25 and 4.26 provide a com-
plete evolutionary sequence of exfoliation and arch formation at the foot of El Capitan.
Note that the mechanisms leading to the formation of the latter arches differ from those
that created arches in sandstones (Fig. 3.31).
 4.5 · Joints in Granites from the Sierra Nevada Batholith in California 341

4.5.3.3
Joint Exfoliation at the Middle Height of El Capitan Cliff

Methods of investigation. Exfoliation surfaces were examined in the middle height

of the approximately 1 000 m high granitic cliff of El Capitan (Fig. 4.21), within the
frame shown in Fig. 4.27a (this section is cited from Bahat et al. 1999). An electronic
total station, which incorporates a laser range-finder, provided the means to survey
points and measure angular and size parameters on the cliffs of El Capitan, the Royal
Arches and the southwestern side of Half Dome (Sect. 6.5) at Yosemite National Park
(see also Grossenbacher et al. 1996). A principal assumption in this technique is that
these cliffs are planar, which applies to the investigated exfoliation joints. For the ge-
ometries used, a deviation of 3 m from a true planar surface yields an error of at most
1 m in a calculated position on a cliff. With distances measured of about 50–100 m and
a distance from base station to cliff of over 300 m, this does not appear to involve sig-
nificant error, as the relief on the cliff is less than 5 m (as verified by examination of
topographic base maps). Taken together, these yield an error of less than 3%, which
was deemed acceptable, considering the large distances in the park.

Fractography. Vertical fracture surfaces on the N 32° E oriented exfoliations are marked
by tens of early cracks, which are shaped as fans of radial plumes. Each fan is com-
prised of an origin and radial plumes. In some cases, a single or several concentric
ripple marks (Fig. 4.27a,b) form orthogonal relationships with the plumes at their
intersections on the fans. Distorted fans manifested by plumes deviating from radial
structures are common as well. The size of a fan varies from several meters to several
tens of meters. Occasionally, only partially developed radial plumes appear.
These fans cut country-rock and dikes of the Taft Granite (Fig. 4.27b) without ap-
parent change in direction. Early circular cracks resembling in shape and size similar
fractures from Zion National Park (Bahat et al. 1995 and present Fig. 3.24b) also oc-
cur sporadically on the cliff of El Capitan. The fans are more common than the early
circular cracks. Many neighboring fans interacting with each other merge into verti-
cal composite joints, which are several hundreds of meters in size.
The fan population produces a specific fracture pattern at the middle height of the
cliff. This pattern is not so clear for fractures A and B (Fig. 4.27c). However, a particu-
lar preferred orientation appears for fractures C and D, where fans indicate upward
propagation above the approximate height of D (where ripple marks in the fans con-
vex upward), and fans below this height show downward propagation (where ripple
marks in the fans convex downward). Hence, fans cutting the granite at El Capitan
propagated upward and downward from initiation zones on individual exfoliations.
These zones occupy similar heights on the cliff (Fig. 4.27d).
A few exfoliation joints occur parallel to each other on the cliff, such that each two
adjacent joints contain a rock “slice” between them. Some of the slices do not maintain
a constant thickness, causing deviations from parallelism of the exfoliation joints. Let-
ters A–D represent surviving remnants of distinct exfoliation surfaces, such that D reveals
that the previous more external slices (A–C) have been removed at this location (Fig. 4.27c).
Exfoliated slices apparently detach selectively from the cliff in a gravity driven process,
as revealed by certain undetached parts from the cliff (dash lines in Fig. 4.27c).
342 Chapter 4 · Jointing in Granites – Part 1

Fig. 4.27a,b.
Exfoliating joints in the El
Capitan granite at Yosemite
National Park (after Bahat et al.
1999). a Photograph of a cliff
on El Capitan striking N 32° E,
showing the frame which is
detailed in b. b Photograph of
the exfoliated cliff within the
frame, showing multiple groups
of radial striae. Some groups
occur in association with con-
centric ripple mark, together
forming fan-shapes. The cliff is
divided by sub-vertical frac-
tures into distinct exfoliations.
Fractures cut sub-horizontal
dikes of the Taft granite (which
have a light color)
 4.5 · Joints in Granites from the Sierra Nevada Batholith in California 343

Fig. 4.27c,d. Exfoliating joints in the El Capitan granite at Yosemite National Park (after Bahat et al.
1999). c Scaled diagram of b, showing exfoliation boundaries between slices belonging to distinct
joints A–D. Rectangles designate initiation zones at which the propagations depart to opposite direc-
tions, as shown by traces of convex ripple mark or radial plumes. d Traverses x–x' and y–y' on joints C
and D from c, depicting fans. The fans are either radiating barbs or convex ripple marks pointing to the
direction of fracture propagation upwards in the upper part of diagram and downwards in the lower
one. Rectangles are as in c

Mechanism of exfoliation. According to Holzhausen and Johnson (1979) longitudinal

splitting occurs parallel or sub-parallel to the direction of loading in uniaxially com-
pressed rock cylinders. Although longitudinal fracture often originates at the ends of
the specimens (due to sample end-effects) and propagates toward the center, uniform
loading also results in fracture origins within the specimens (Peng 1970). Holz-hausen
and Johnson (1979) postulate that an “internal fracture” is formed by the coalescence
of earlier axial cracks (analogous to the term composite fracture used in this section),
which have created stably under high loads (local compression in Fig. 4.28a,b). An ex-
tension of the internal fracture to longitudinal splitting may be caused by buckling of
the rock away from this surface. This would happen when the two separated parts of
the cylinder bend outward (local tension in Fig. 4.28c). Tension at the tips of the longi-
tudinal split may cause unstable growth to the ends of the specimen. Holzhausen and
Johnson (1979) consider several tensile and shear criteria but refrain from advocating
one mechanism leading to splitting (Sect. 2.2.3.11 and 2.2.13, see also Fig. 2.65).
344 Chapter 4 · Jointing in Granites – Part 1

Fig. 4.28.
Schematic representation of how buckling causes lon-
gitudinal splitting of cylindrical specimens. a Axial
cracks coalesce under applied axial stress σa to an
“internal fracture” of length 2c at a distance d from
the free surface. b The internal fracture shown at 90°
position to a, exhibiting early cracks in shapes of fans
pointing upward and downward, before coalescence.
c Axial propagation of the internal fracture (in the
a position) occurs when it buckles outward (from
Bahat et al. 1999, modified after Holzhausen and
Johnson 1979)

Fig. 4.29.
a A section model showing the
profile of joint exfoliation by
longitudinal splitting and buck-
ling of the “internal fracture”
from Fig. 4.28a. Vertical arrows
indicate overburden load. Frac-
ture (longitudinal splitting) oc-
curs in association with buck-
ling (an exaggerated buckling is
shown by the dash line), which
develops sub-normal to σ3 (hori-
zontal arrow). b An early slice
collapsed (at 0) and washed away,
and new exfoliations develop
(earlier ones closer to the free
surface at arrow numbered 1 to
arrow numbered 10). c A profile
of an arch at the lower part of the
cliff, which forms when part of
the slice collapses (white arrow)

The findings in the field are in harmony with various experimental observations
made by Holzhausen and Johnson (1979). Coalescence of coplanar fans (Fig. 4.28b–
d) that are close to and parallel to free surfaces (the outer surface of the cliff) is the
main fracture process in the middle height of El Capitan cliff that leads to exfoliation.
Exfoliation developed along principal planes populated by multiple early cracks by a
mechanism of longitudinal splitting, which was caused by the overburden load. Split-
ting occurred upward and downward from an initiation zone at an elevation where
the normal resistance to buckling was minimal (Fig. 4.28) next to a free surface in
analogy to Fig. 4.29. Fracture propagation on El Capitan cliff was polarized upwards
and downwards from this zone. It followed the creation of multi-nucleation sites of
independent cracks that had developed into fans and then merged with each other
into composite joints (Fig. 4.27c,d).
Considering tension vs. shear mechanisms causing splitting (Holzhausen and
Johnson (1979), the fractographic observations suggest the following. First, the ripple
marks and plumes on the surfaces of the early joints reflect mode I opening as the
predominant mode (Bahat 1979), i.e., on principal planes normal to the minimum
principal stress. Second, minor local shear was superposing the tensile mode (Bahat
1987b). Third, significant shear strain can be ruled out because such a process would
 4.5 · Joints in Granites from the Sierra Nevada Batholith in California 345

have erased the fracture markings, and fourth, the lack of hackle or exfoliation-branch-
ing (Fig. 2.1) raises questions about an unstable fracture analogous to that observed
experimentally (Holzhausen and Johnson 1979). These observations imply stable joint
propagation at the middle height of El Capitan cliff that may be correlated with the
compressive state in the uppermost lithosphere. If however buckling is promoted close
to the ground surface of the granite, unstable exfoliation may occur (Fig. 2.65).
A formula used by Holzhausen and Johnson (1979) is adapted for the analysis of
the critical conditions for buckling:

(4.1)

where σa,critical is the critical axial stress for buckling of a beam with fixed ends, E is
Young’s modulus, d is the distance between the internal surface (nucleation of the lon-
gitudinal split) and the free surface, and 2c the internal fracture length (Fig. 4.28). This
formula shows that σa,critical increases with d 2. It explains the observations both in gran-
ite and in sandstone (Fig. 4.30 and Fig. 3.25c) that vertical dark streaks produced by
dripping solutions mark external exfoliations and not internal ones. That is, the ear-
lier longitudinal splits develop close to the free surface at small d values (Fig. 4.29),
and then new ones gradually form further inside the rock. Also, Eq. 4.1 implies that
rock slices produced between successive exfoliations should be much thinner than the
lengths of the associated “internal fractures”. This may explain the great ranges of
exfoliation spacing depicted in various publications (e.g., Holman 1976, p. 2; Twiss and
Moores 1992, Fig. 3.5).
Joints are classified into four fracture categories: Burial, syntectonic, uplift and post-
uplift groups. These four groups differ in their various properties, including their
fractographies. It turns out that the merger of early circular cracks (Fig. 3.23b,c) and
early fans (Fig. 4.27b–d) is common to exfoliation joints in both sandstone and gran-
ite. These are characteristic fractographic features exclusive to post-uplift joints, which
develop close to a “free boundary” above the ground surface. A similar merger of fans
into a larger post-uplift joint has been observed in chalk (Bahat 1991a, p. 301). Mergers
of early circular cracks or early fans that occur on post-uplift joints are not known from
surfaces of burial and uplift joints, and they are rare on syntectonic joints. As mentioned
above, Huber (1987) suggested that the vertical exfoliations possibly took advantage of
free surfaces provided by previous jointing. These observations support his suggestion.

4.5.3.4
The Royal Arches

The series of Royal Arches is located between El Capitan and Half Dome in Yosemite
Valley (Fig. 4.21) and provides an exceptional occurrence of very large arches
(Fig. 4.42a,b). They are considerably larger than arches in Zion National Park (e.g.,
Fig. 3.24a,b) and are by far larger than arches at the foot of El Capitan. These arches
are confined to a single cliff, cutting the entire monolith, in contrast to the division
into distinct structures at different heights on El Capitan and the limitation of arches
to its lower elevations.
346 Chapter 4 · Jointing in Granites – Part 1

Fig. 4.30.
a Photograph of the Royal
Arches. b Scale diagram of the
Royal Arches showing mea-
sured dimensions. There are
angular merges of ceilings with
the sidewalls at points 5 and 9
and in the upper left arch. Note
the division of parallel thick
slabs along side-walls and ceil-
ings into series of thin slabs
(particularly along arch 7–8–9),
often combined with sharp con-
tours that outer surfaces make
with sidewalls and ceilings.
Note vertical, dark streaks of
dripping solutions between 8
and 9

The Royal Arches occur in a concentric structure and show both symmetric and
asymmetric features (Fig. 3.28a). The lower and smaller arches tend to be symmetric,
whereas the upper and larger arches are asymmetric. The latter arches display angu-
lar merges of the ceiling with the sidewalls at one side of the arches (their western
 4.5 · Joints in Granites from the Sierra Nevada Batholith in California 347

side) and an arcuate continuity of the ceiling with the sidewalls of the arches on the
other side (the eastern one). The close association of exfoliation with arch formation
observed at the foot of El Capitan is repeated in the Royal Arches. However, the ceil-
ings and sidewalls in the Royal Arches are subdivided by more undercut exfoliation
joints (e.g., between points 8 and 9 in Fig. 4.30b).

4.5.3.5
Arch Formation in Granitic Rocks and Sandstones

Structural features common to the various arches. (a) Intimate associations of arches
with exfoliation joints are almost invariable, suggesting that joint exfoliation is a pre-
requisite for arch formation. (b) Many contacts between the outer surface of cliffs and
the ceilings and sidewalls are angular (in some cases close to an angle of 90°), imply-
ing a development by fracture, rather than by erosion. (c) There are both symmetric
and asymmetric arches. (d) On many arches, particularly asymmetric ones, there are
merge zones of sidewalls and ceilings or matching partial arches, implying that these
arches generally form by the connection of distinct structural components. (e) Niches
occur in multiple sizes at bending barbs on centimeter scale and at sub-concentric
fractures on meter scales at the foot of El Capitan (Fig. 4.22). They also appear on large
scales of tens of meters on partial-arches and arches in Zion National Park. Niches
appear to be initiation sites of many symmetric and asymmetric arches. The shape
and symmetry of an arch depends on the number of initial niches and their location
along the arch perimeter. A single niche at the highest position of the arch is likely to
result in a symmetric arch. On the other hand, niches at positions below the arch sum-
mit are likely to produce asymmetric arches.

Structural features that differ in the various arches. (a) There are many more arches
in sandstone at Zion National Park (Sect. 3.3) than in granitic rocks at Yosemite Na-
tional Park. (b) Arches at the foot of El Capitan differ significantly from the Royal Arches.
(c) The mechanisms of partial-arch formation at Zion National Park and at the foot of
El Capitan are different, as follows: In the sandstone, the ceilings and sidewalls often
start tangent or vertical to a niche and propagate downward to yield a curved surface
normal to the trajectories of least compressive stress induced by the gravitational load
of the overlying rock. This propagating curviplanar fracture often cuts normally into
the rock body across a series of rock “slices” sandwiched between exfoliation joints
that develop parallel to the cliff and to each other. On the other hand, partial arches at
the foot of El Capitan are formed by curved-concentric plumes (Fig. 4.22) that under-
cut laterally and form small angles (1–4°) with the cliff. Correspondingly, arches at
Zion National Park are commonly impressively thick, since they consist of many slices
of rock, whereas arches at the foot of El Capitan are considerably thinner.

Open questions. The differences in fracture mechanisms suggest that additional to

the driving exfoliation, a certain set of conditions needs to be provided for arch for-
mation. The available information does not warrant a speculation on what these com-
bined conditions at each location should be. This leaves several open questions as
summarized below.
348 Chapter 4 · Jointing in Granites – Part 1

1. Huntington (1966) observed that for unknown reasons, sheeting developed most
abundantly in quartz monzonite and granite, “spottily” in granodiorite, and never
in quartz diorite or gabro. That is, arch formation is dependent on the petrography
of the rock, which in turn should significantly influence the strength and resis-
tance properties of the rock to fracture. Thus, it is pertinent to enquire how such
differences resulted in distinct exfoliation and arch formation on the cliff of
El Capitan vs. the Royal Arches.
2. Moreover, there are significant differences in fracture behavior at different heights
of El Capitan (Fig. 4.22, 4.24–4.26 vs. Fig. 4.27). These differences cannot merely be
assigned to differences in rock properties. We need to determine what the distinct
fracture stress conditions were in such cases.
3. Dry conditions probably prevailed during fracture at the middle heights of El Capitan,
whereas pore pressure activities quite likely, influenced exfoliation in the sandstone
at Zion National Park (Bahat et al. 1995). In view of the multitudes of arches both at
Zion National Park and at the Royal Arches, may one assume that fracture condi-
tions (dry vs. wet) have been similar at the two sites?
4. The exfoliations investigated at El Capitan and at Half Dome generally follow the
NNE directions of earlier regional fractures in Yosemite National Park (Bahat et al.
1999). The implication is that previous fractures provided free surfaces parallel to which
exfoliation occurred in these two sites. This fracture overprinting pattern is repeated at
Zion National Park (Bahat et al. 1995). The Royal Arches are approximately E-W and
deviate from the regional NNE direction in Yosemite National Park. This could be
related to an exfoliation that relates to a different regional-structural setup.

4.6
Fractographies that Pertain to both Quasi-Static
and Dynamic Fractures
We start by raising the question of how en echelon segments (generally reflecting quasi-
static propagation) and hackles (generally reflecting dynamic propagation (Sect. 2.2.5) can
occur adjacently on a single fringe (e.g., Fig. 2.1c). The answer is based on the assump-
tion that only “slight deviations” of the direction of the minimum principal compres-
sion (or maximum principal tension) from orthogonality to the fracture surface would
induce en echelon segmentation (Sommer 1969). Accordingly, fringes that form by both
quasi-static and dynamic propagations may be influenced by such “slight deviations”.
Fractures formed quasi-statically when the ratio KIII / KI is high will be dominated by
en echelon fringes (Fig. 2.1a and 2.35), and these fringes will have the tendency to as-
sume simple, uniform widths (Fig. 2.34b). On the other hand, fractures formed dynami-
cally when KIII / KI is low will be dominated by hackle fringes, on which, however, en
echelon segmentation in a shingle-like arrangement may be superimposed due to this
ratio. Such superposition can explain the frequent occurrence of fringes that show hack-
les in a shingle-like arrangement (Fig. 4.12b). The latter fringes often exhibit non-uniform
widths (Fig. 2.34a and the center of Fig. 4.12b). Hence, it is quite possible that on the
fracture surface obtained by De Freminville (1914) (Fig. 2.1c), the two fringe types re-
flect local slight changes in the extent of normality of the maximum principal tension:
The fringe area that displays hackles reflects the “pure” mode I, whereas the fringe area
that displays hackles in a shingle-like arrangement reflects a mixed mode I + III loading.
 4.7 · Comparative Jointing 349

In summary, most fringes from sedimentary rocks suggest quasi-static growth, while
rare fringes display cuspate hackles (Bahat 1991a, Fig. 3.41a) or hackles in shingle-like
arrangements (e.g., Fig. 2.33e). Note the resemblance between the latter fracture mor-
phology and the fracture surface of a turbine that is interpreted to have initiated in fa-
tigue growth and possibly approached catastrophic fracture when KI ~ KIc conditions
were attained (Fig. 2.44a) (Bahat 1991a, p. 132). In many outcrops, the distinction be-
tween these fringe types would be relatively easy, whereas in others that exhibit tran-
sitional fractographies, this distinction is expected to be more problematic. Hackle
fringes are far more abundant in granitic rocks than in sedimentary ones, but even in
granites hackles often show shingling. Note that in Fig. 4.11 in the lower fringe shin-
gling is almost absent, whereas in the upper fringe shingling is more pronounced.

4.7
Comparative Jointing
4.7.1
Cooling Joints in Granites

4.7.1.1
Regional Axial Fracture in Two Provinces

Maps of the two fracture provinces, the Sierra Nevada Batholith (Fig. 4.31) and the
South Bohemian Pluton (Fig. 4.32), both show that these are regional, elongated in-
trusion-complexes consisting of series of large and small discrete plutons that are
aligned along the complex axes (Bateman and Wahrhaftig 1966 and Breiter 2001, re-
spectively). According to Breiter (2001), the evolution of the South Bohemian Pluton
started with the intrusion of the peraluminous Lasenice granite. Next was the intru-
sion of high-K peraluminous melt that crystallized as porphyritic Cimer granite in
the south and non-porphyritic Mrákotin granite in the north. Further fractionation
of the “Cimer melt” produced the Eisgarn granite in the central part of the intrusion-
complex. Additional small bodies were then produced by intense fractionation of the
“Eisgarn melt” and a still younger magmatic input formed the Zvule, Melechov and
Cerinek deep-seated stocks.
A detailed geochemical investigation of the Eisgarn, Zvule and Melechov granites
enabled Breiter (2001) to construct the inner structures of these plutons. Breiter (2001)
shows that the Eisgarn (Fig. 4.33a), and Melechov (Fig. 4.33b) granites are elongated
bodies and their long axes are oriented NNE, whereas the Zvule granite (Fig. 4.33c)
has a slight elliptical shape with the longer axis slightly oriented E-W. It has been found
that NNE is an important direction of a set of cooling joints along the South Bohe-
mian Pluton (Fig. 4.5). The possible implication of the corresponding direction of the in-
ner structures of the Eisgarn and Melechov granites to the regional direction of the
NNE joint set is that there is a genetic connection between the two. Possibly, the cooling
joints are aligned in the South Bohemian Pluton along certain discrete plutons as long as
they also maintain the regional orientation, i.e., the regional influence is more pronounced
than the one exerted by the individual plutons on joint orientation. This interpretation
would be challenged if future studies will prove the non-existence of NNE oriented joints
in the Zvule granite, which is approximately oriented E-W (Fig. 4.33c).
350 Chapter 4 · Jointing in Granites – Part 1

Fig 4.31. The general distribution of Mesozoic plutonic rocks in California and Lower California.
The Peninsula Ranges are located in Southern and Lower California. The Sierra Nevada Mountains
are in east-central California (modified from Ehlers and Blatt 1982)

To the best of our knowledge, no special treatment of cooling joints exists for the Sierra
Nevada Batholith. Furthermore, the classification into distinct genetic groups/stages of
joints in granites is challenged by existing investigations. For instance, Davis and Reynolds
(1996, p. 265) cite works by Rehrig and Heidrick (1972, 1976) regarding jointing in Late
Cretaceous to Eocene plutons in southern Arizona. The latter authors concluded that these
plutons were intruded into the upper levels of the crust and as they cooled and congealed,
 4.7 · Comparative Jointing 351

Fig. 4.32. Simplified geological map of the South Bohemian Pluton (after Breiter 2001)

and cracked in directions systematically related to the regional tectonic stress field that
existed at the time. Thus, the interpretation seems to be that these joints formed on cool-
ing (like cooling joints) by regional tectonic stress (syntectonic joints) at the upper levels
of the crust (as uplift joints?), i.e., this explanation combines three distinct stages and mecha-
nisms in one process. This interpretation may well be legitimate for south Arizona, but it may
or may not be applicable to other fracture provinces, like the Sierra Nevada Batholith.
352 Chapter 4 · Jointing in Granites – Part 1

We now return to the debate about the age of the joints in the Sierra Nevada
Batholith (Sect. 4.5.1). Suppose some of the joints formed during the emplacement of
the plutons (e.g., Balk 1937; Mayo 1941) or soon after it (Segall et al. 1990), so that they
might be considered to be cooling joints. According to (Bateman and Wahrhaftig 1966)
the granitic rocks of the Sierra Nevada Batholith occupy the axial part of the N 40° W
trending, faulted synclinorium, which began to form in Permian or Triassic time in
central Sierra Nevada. Intense deformation is manifested by fold axes, steep or verti-
cal beds, cleavage, and lineations that took place, both before and during the magma
emplacements. Folds, at least, in the western part of the Ritter Range pendant may be
related to the axial region of the synclinorium. Thus, it appears that an early NNW-
NW trending structure may have significantly influenced the emplacement of the Ju-

Fig. 4.33a. A detailed geochemical investigation of the Eisgarn, Zvule and Melechov granites by Breiter
(2001). An idealized arial distribution of selected elements within the Eisgarn body. The isolines of
element contents or ratios were constructed by methods of data interpolation by polynomial regres-
sion of third order. A: Simplified geological sketch of the interpreted body. The interpreted area is
dotted; localization of used samples is expressed by small squares. B: Idealized distribution of P205
in wt.%; C: Idealized distribution of Na2O / K2O-ratio in wt.%; D: Idealized distribution of Th in ppm;
E: Idealized distribution of Zr in ppm; F: Idealized distribution of Rb in ppm
 4.7 · Comparative Jointing 353

Fig. 4.33b.
Idealized arial distribution of
selected elements within the
Melechov body. A: Simplified
geological sketch of the inter-
preted area; B: Idealized distri-
bution of Zr in ppm; C: Ideal-
ized distribution of Rb / Sr
in ppm; D: Idealized distri-
bution of Th in ppm (from
Breiter 2001)

Fig. 4.33c. Idealized distribution of selected elements within the Zvule body. A: Simplified geological sketch
of the interpreted area. The interpreted area is dotted; localization of used samples is expressed by small
squares. B: Idealized distribution of Rb in ppm; C: Idealized distribution of Sr in ppm (from Breiter 2001)
354 Chapter 4 · Jointing in Granites – Part 1

rassic granites in central Sierra Nevada. This structure may also have controlled frac-
turing during cooling of the plutons and possibly also subsequent fractures along this
direction. In support of this suggestion, Tobisch et al. (1989, Fig. 7) show a definite
NNW-NW trend of a continuous, planar cleavage, predominant throughout the foot-
hills terrain in central Sierra Nevada, where it is heterogeneously developed, occasion-
ally in shear and mylonitic zones in the Upper Jurassic granitoids. Hence, there are
similarities in the European and Californian fracture provinces, where in both there
are indications that emplacement of granites was influenced by earlier, regional elon-
gated structures, that also controlled later regional jointing.

4.7.2
A Comparison to the Cloos Model

In view of the general NNW-NW orientation of the elongated Sierra Nevada Batholith
(Fig. 4.31), the NNW-NW cooling joints in the batholith (Sect. 4.5.1 and 4.5.2.1) seem
to have significant similarities to the NNE cooling joints that are oriented along the elon-
gated South Bohemian Pluton which is oriented NNE (Fig. 4.5 and 4.32). It appears that
common to these two fracture groups is their fit to the S-joint category in the Cloos
model (Fig. 4.1). The NNW-NW fractures developed perpendicular to the main hori-
zontal compression that created the synclinorium and axial structures that subsequently
controlled the formation of these fractures in the Sierra Nevada plutons. Furthermore,
Lockwood and Moore (1979) found that the direction of maximum horizontal exten-
sion strain in the Sierra Nevada Batholith varies between WNW and NW, precisely fit-
ting Cloos’s criterion for S-joints, if applied to the fractures in this direction.
Sub-parallelism between fractures and isoanomals of the gravity high of Lusania ex-
pressing structural contours of the country rock has convinced Müller et al. (2001) that
these fractures are S-joints (Fig. 4.3). The elongated South Bohemian Pluton sub-parallels
the earlier structural contours of the country rock (Fig. 4.2b). The similarities in these two
sub-parallelisms support the suggestion that the NNE set represents S-joints. It has been
suggested that the shape of the South Bohemian Pluton could have controlled significant
thermal and fluid-pressure gradients that in turn influenced the orientation of the NNE
cooling joints (Bahat et al. 2003). This interpretation would coincide with the Cloos model,
if proven in the future that the fluid-pressure gradients were not merely local on the sur-
faces of the individual joints (Sect. 4.4.5.2), but rather regional, i.e., perpendicular to the
whole elongated pluton. The present interpretation relating the two groups of joints to the
S-joint category would be challenged if proven that the main horizontal compression was
axial during the cooling of the respective plutons, i.e., possibly implying that these frac-
ture groups are Q-joints, which does not seem a likely possibility.
The descriptions of joints C and D (Sect. 4.5.2.2) partly resemble Cloos’s “diagonal
fractures” (Fig. 4.1). They dip at about 40° and they are partly filled with hydrother-
mal minerals. However, their strikes are more compatible with that of the Q-joints than
with the strike of S-joints (Sect. 4.5.2.2 and Fig. 4.1). Moreover, the present observa-
tions suggest that they predate the formation of steep joints. If so, they differ from the
Cloos model, in which the diagonal fractures are younger than the vertical joints. Fur-
thermore, Cloos’s diagonal fractures are clearly faults, whereas fractures C and D (D in
particular) seem to fit better the term joints. Perhaps C and D represent a new frac-
ture type that does not appear in the Cloos model?
 4.7 · Comparative Jointing 355

4.7.3
Joint Genetic Grouping in Sedimentary Rocks and in Granites

The genetic division of systematic joints that occur in sedimentary rocks is into four
groups: The burial, syntectonic uplift and post-uplift (Bahat 1991a, pp. 239–322). There
are certain indications that the above classification of joints in sedimentary rocks may
be partly applicable to joints cutting granite. Bankwitz and Bankwitz (1994, 1997) char-
acterized syntectonic joints in two granite massifs from Europe. Bahat and Rabinovitch
(1988) described uplift joints in granite from Sinai, Egypt. Holzhausen (1989) observed
fractographic features in sheet joints cutting granite in Massachusetts, and Bahat et al.
(1999) described post-uplift joints in granites from California. Thus, there seem to be
resemblances in tectonic setting for the last three joint groups (syntectonic, uplift and
post-uplift) between granites and sedimentary rocks. On the other hand, there are no
corresponding similarities between early joints. The earliest systematic joint sets in
sedimentary rocks are burial ones that maintain parallel orientation and approximate
uniform dimensions and spacing. Probably the earliest joints in granites form during
the cooling process as soon as some critical range of viscosity develops. The cooling
of a pluton may be a long, multistage process, consisting of different thermo-mechani-
cal modes. Hence, a correlation between “cooling joints” in granites to burial joints in
sedimentary rocks is not possible, and a special genetic classification of cooling joints
in plutonic rocks is required.
356 Chapter 4 · Jointing in Granites – Part 2

Part 2
Part 2 concerns fracture mechanic aspects of joints by drawing on fractographic prop-
erties of joints as follows:

1. Plotting joints on the velocity v vs. stress intensity K curve (Sect. 1.3–1.5).
2. Presenting a general discussion of various manifestations of v and K in granites.
3. Calculating the new fracture areas on en echelon and hackle fringes.
4. Estimating the “index of hackle raggedness” of joints.

4.8
Analysis of Fracture Velocity versus Stress Intensity Factor
in the Borsov Joints Based on Fractographic Criteria
This section presents six aspects of the v vs. KI diagram:

1. Physical meaning.
2. Justification for application to geological fracture.
3. Construction for geological application.
4. General criteria and method of placing the joints on the diagram.
5. Plotting of the Borsov joints on the v vs. KI diagram.
6. Present limitations of the method.

The basic ideas of this scheme were outlined in Bahat (1991a, p. 234), and these were
elaborated in Bahat et al. (2003).

4.8.1
The Velocity versus Stress Intensity Factor Curve

Wiederhorn (1967) and Wiederhorn and Bolz (1970) experimented on glass and showed
the dependence of crack velocity v on the applied force and the tensile stress intensity

Fig. 4.34.
Schematic variation of stress in-
tensity factor K with crack ve-
locity v (modified from Evans
1974). The subcritical curve left
of KIc is divided into three re-
gions I, II and III. K0 and v0 are
the stress corrosion limit and
the crack velocity at this limit
(Wiederhorn and Bolz 1970).
Supercritical growth occurs
mostly along a terminal velocity
plateau following a rapid in-
crease in crack velocity at KIc.
A velocity overshoot o, immedi-
ately after KIc and before the
plateau has been predicted by
Rabinovitch and Bahat (1979)
 4.8 · Analysis of Fracture Velocity versus Stress Intensity Factor in the Borsov Joints 357

factor KI in the sub-critical range (below KIc, the critical stress intensity factor). Wiederhorn
and Bolz (1970) obtained a curve that was characterized by three distinct regions (Fig. 4.34).
The slow crack growth in region I is manifested by a linear logarithmic plot. The behavior
of region I is attributed to stress corrosion. The rate of crack growth is controlled by the
chemical reaction between the reactant (solution) and glass at the crack tip when the sys-
tem is essentially at equilibrium. Region II is controlled by the rate of reactant transport
to the crack tip. In this region V almost does not change and is independent of KI. In
region III crack velocity is quite high; it is insensitive to the reactant and to stress-corro-
sion. It was also established that regions I–III were sensitive to water environments (vs.
vacuum conditions) and were influenced by chemical conditions such as pH. The KIc line
shows that at a critical stage in region III, the crack undergoes a catastrophic growth asso-
ciated with a sudden jump in V, i.e., at KI ≥ KIc post-critical crack propagation occurs.

4.8.2
Application of Fractography and Fracture Mechanics to the Study of Joints

Fractographic investigation of joints rests on analogies between the fracture surface

morphologies of joints in geological exposures to fractures in glass, ceramics, metals
and polymeric materials (e.g., Woodworth 1896; Bankwitz 1965, 1966; Kulander et al.
1979; Bahat 1979a, 1991; Cooke and Pollard 1996). It has been shown that KIc estimates
derived from geological measurements are compatible with existing data from labora-
tory fracture experiments (Segall and Pollard 1983). Quantitative tectonofractography
is a method that was developed to estimate paleostresses from strain measurements
on joint surfaces, and this method stems from principles of fracture mechanics (e.g.,
Bahat 1979a, 1991; Bahat and Rabinovitch 1988; Weinberger 2001a). A strong correla-
tion between the behavior of en echelon segmentation in polymethyl methacrylate
(PMMA) (by mixed modes I + III) and en echelon segmentation in field outcrops was
demonstrated by Cooke and Pollard (1996) and Bahat (1997). Sophisticated experimen-
tation by Müller and Dahm (2000) further developed this concept and showed that
knowledge of rupture velocities in sedimentary rocks can be gained from experimen-
tation on starch-water mixtures. Particularly important were a series of demonstrations
by various investigators (e.g., Atkinson and Meredith 1987a,b) that plots of minerals
and rocks on V vs. KI diagrams greatly resembled analogous plots of glasses. All these
(as well as other sources) justify the suggestion by Pollard (2000) that the mode is emi-
nently suitable for the taxonomic classification and description of fractures in the field.

4.8.3
The Velocity versus Stress Intensity Curve for Granite

The curve in Fig. 4.35a is a synthesis of nine elements:

1. The basic concept stems from the sub-critical V vs. KI curve determined on glasses
(Fig. 4.34).
2. Section A–B on the curve corresponds to the sub-critical V vs. KI curve determined
for Westerly granite, under 15 kPa p(H2O), at 300 °C by Atkinson and Meredith (1987b).
3. From the two hypothetical curves, B–C and B–D, the former assumes the existence
of regions I–III and the latter favors the existence of only region III (Fig. 4.35a).
358 Chapter 4 · Jointing in Granites – Part 2

Fig. 4.35. a The V vs. KI curve for joints in granite from the South Bohemian Pluton, synthesized from
laboratory experiments. b The estimate of a range of crack velocities for ten joints in granite from
the South Bohemian Pluton. For each joint the range of fracture velocity is given in a rectangular
frame constructed on the curve from Fig. 4.35a, where the vertical right side of each frame gives the
maximum fracture velocity that was attained by the joint

4. The position of D approximates KIc = 1.09 MPa m1/2 (the KIc value of Westerly granite
at 300 °C, Atkinson and Meredith 1987a).
5. The position of D independently fits well the definition of KIc at crack velocity of
≥ 10–1 m s–1 (Wiederhorn et al. 1974).
6. A velocity overshoot immediately beyond KIc at E was mathematically predicted
by Rabinovitch and Bahat (1979) and was recently verified experimentally (Cramer
et al. 2000, Fig. 1A).
7. Kerkhof (1975) proposed a unified curve for both sub-critical and post-critical
conditions (at both sides of KIc) and suggested that the positions of mist and bifur-
cation initiations correspond to m and b, respectively. These positions are approxi-
mated in the post-critical part of the present curve.
8. We add h (Beauchamp 1996) representing an approximate location for hackle for-
mation between m and b.
9. Finally, the two ends of the curve are at K0, signifying zero slow crack growth be-
low point A, and at b where the fracture bifurcates along the terminal velocity pla-
teau (Evans and Wiederhorn 1974).

Hence, Fig. 4.35a is essentially based on experimental data from glass but it is con-
strained by results from granite and the definition of KIc (elements 2–5).
 4.8 · Analysis of Fracture Velocity versus Stress Intensity Factor in the Borsov Joints 359

In further justification of these elements for the construction of Fig. 4.35a, we need
to add that the Westerly granite was chosen because of its textural similarity to that of
the granite under investigation (fine to slightly medium grain size). Curve A–B was
selected because we consider joint fracture under considerable pore pressures at depths
of about 10 km, so that assuming 300 °C due to the thermal-gradient seems reason-
able. There is no compelling data that indicates which curve, B–C or B–D is more real-
istic, so we use both. A velocity overshoot immediately beyond KIc is anticipated to be
an inherent property of fracturing in elastic materials following great velocity “jumps”
such as the one from D to E. Although a realistic velocity jump is expected to occur
along a slope (through a certain range in KI), a conventional vertical C–D is adapted
(Atkinson and Meredith 1987b, Fig. 4,15).

4.8.4
Methodology for Plotting Joints onto Velocity versus Stress Intensity Factor Curve

Various sub critical and post critical growth conditions of joints may be inferred on the
v vs. K diagram (Fig. 4.35a) by two fractographic distinctions: First, plumes and ripple
marks (see Sect. 2.2.4.2 for definitions) vs. hackle cracks (or hackles) and second, en ech-
elon cracks vs. hackles. The first distinction is between characteristic fractographic fea-
tures below and above KIc (Fig. 4.34 and 4.35a). Both plumes (striae) on the mirror within
the sub critical regime and hackles on the fringe in the post critical regime often occur
radially, such that the latter often appear to be in continuation with the former (Fig. 2.30a).
However, they can be distinguished quite readily. Whereas striae are delicate and approxi-
mate unidimensional textures by their barb lines (Fig. 2.55b and 4.8a), hackles are three-
dimensional, coarse complexes consisting of superimposed flakes of material that are sepa-
rated by microcracks in all orientations. In fringes formed in glass by extremely intense
fracture (high KI values), each flake may display a miniature assemblage of fractographic
features shown in Fig. 2.1a. On surfaces cutting granite that represent moderately intense
fracture conditions. the differences between plumes and hackles are less discernable than
in glass, but they are quite pronounced.
Kerkhof (1975) observed that arrest marks (ripple marks) were induced in plate
glass at propagation rates below v = 4 × 10–5 m s–1 and KI < 0.73 MPa m1/2 that corre-
spond to the range of regions I and II in Fig. 4.35a. Accordingly, we propose that frac-
ture markings in rocks that display arrest marks reflect the range of regions I and II.
The frequent occurrence of ripple marks superposed by radial plumes on joint sur-
faces suggest that the plumes are formed under fracture conditions similar to those
that produce ripple marks but under slightly more intense ones, i.e., closer to region II.
Fracture experiments on PMMA have shown that circular ripple marks develop before
radial plumes on the fracture surface (Bahat 1991a, Fig. 2.50c), i.e., these ripple marks
developed at lower KI values than the plumes. Striae developed on a smooth fracture
surface of soda-lime silica glass in water when the velocity of fracture propagation
reached about 10–2 m s–1 at around KI = 0.7–0.8 MPa m1/2 (Michalske 1984, Fig. 1). This
was above the mid range between the stress corrosion limit, KI = 0.3 MPa m1/2, and the
fracture toughness KIc = 0.9 ±0.1, i.e., striae formed in the range of regions II to III
(Fig. 4.35a). We propose that markings like the short rhythmic plumes from the Appa-
lachian Plateau (Fig. 4.36) that show periodic increase of barb intensities reflect the
range of regions II to III, whereas long uniform plumes free of arrest lines relate more
360 Chapter 4 · Jointing in Granites – Part 2

Fig. 4.36. Rhythmic plume growth. Fan perimeters designate loci of arrest marks where barbs reach
their maximum intensity and vanish into smooth areas. Barbs then resume their growth as fine lines
within the plume that mark the local direction of fracture propagation. The perimeters convex to-
ward the direction of fracture propagation (from right to left) and the barbs may be partly “branched”
due to lagging bays (modified after Bahat and Engelder 1984)

to region III (Engelder et al. 1987 and Bahat 1994a, Fig. 3.18). On the other hand, the
periodic smooth areas where a new cycle of fracture begins (Fig. 4.36) represent V and
KI conditions within region I and possibly below it.
The second distinction relates to the range of KI conditions under which en echelon
cracks develop. This range is unclear because it varies considerably. En echelon cracks
have been observed in borosilicate glass at very low KI (Wiederhorn and Johnson 1973
Fig. 5). There are some indications that en echelon cracks formed in rocks and rock-
like material under low KI values (Kulander et al. 1979; Bahat 1997; Müller and Dahm
2000). On the other hand, Hertzberg (1976, Fig. 14.9) analyzed the fracture conditions
of a turbine rotor made of steel and found that the en echelon cracks developed at close
to KIc conditions. This problem is further complicated by the occasional presence of
hackle cracks (Fig. 2.1a) and en echelon cracks (Fig. 2.1b) at close proximities, suggest-
ing some transitional relationship at the fringe (Bahat 1991a, p. 68). All these support
early observations that en echelon segmentation depends on the KI / KIII ratio (Sommer
1969). Thus, plotting en echelon segmentation (Fig. 2.1b) on the V vs. K curve is still
problematic, and a transitional range of some uncertainties may exist up to and around
KIc conditions for en echelon cracks (Hertzberg 1976). On the other hand, the situa-
tion that concerns hackle cracks (Fig. 2.1a), which are expected to occur above KIc up
to branching, is clearer (e.g., Kerkhof 1975; Rabinovitch et al. 2000a).

4.8.5
Plotting Joints of the Borsov Pluton

It is unrealistic to determine the points of joints J1–J10 (Fig. 4.6–4.8, respectively) along
the curve of Fig. 4.35a. A more reasonable approach is to estimate the range of crack
velocities for each joint (Bahat et al. 1999), given in rectangular frames, where the ver-
tical right side of each frame gives the maximum fracture velocity that was attained
by the respective joint (Fig. 4.35b). It is assumed that joints generally start in low ve-
locities before reaching fast propagation (e.g., Bahat 1991a, p. 58).
 4.8 · Analysis of Fracture Velocity versus Stress Intensity Factor in the Borsov Joints 361

We first consider the four joints of group 1 (Fig. 4.6). Based on the criteria in the
previous section, the arrest marks superposed by radial plumes on joints J1 and J2
(Fig. 4.6a,b) signify maximum velocities close to region II and region III. There is a
considerable variability in the plume intensity of the quasi-core joint, J3 (Fig. 4.6c),
implying that fracture stresses, as well as V and KI were not uniform on the joint sur-
face. The long plumes not associated with arrest marks signify monotonous propaga-
tion (approximately constant velocity) that could reach region III. However, the plumes
of J3 that are associated with arrest mark seem to signify vacillation between region I
and part of region III. These plumes did not develop into fringes. The joints from the
multi-joint exposure, J4 (Fig. 4.6d), mostly exhibit fractographies that occur on
joints J1–J4, placing them between regions I and III.
The fringes in group 2 display a series of fringes between the two end members shown
in Fig. 2.1a,b). The divided arrest mark joint, J5 (Fig. 4.7a) exhibits a small en echelon
fringe, which may indicate that KI temporarily reached KIc, because these fringes re-
semble the features of the fringe studied by Hertzberg (1976) that formed close to critical
conditions (at KIc). The extended fringe F1 of J6 (Fig. 4.7b) relates to the entire struc-
ture consisting of the ten (or eleven?) concentric arrest marks. Accordingly, it appears
that both the right side of the joint and the triangle were formed before the propaga-
tion of the large wavy plume w from the triangle towards the left. Therefore, the large
wavy plume represents an intriguing example of transition from en echelon fringe in F1
and F2 to a low intensity plume. We assign to J6 a frame similar to that of J5 (Fig. 4.35a,b).
The en echelon segmentation is clearly defined on the fringe of J7 (Fig. 4.7c) and
plotting this joint is somewhat problematic, as mentioned above. Prolonged growths
under sub-critical conditions, i.e., through regions I–III, may have transformed to the
mode of post-critical growth, i.e., possibly beyond KIc (dashed line in Fig. 4.35b). The
extent of this frame is still a topic for future research.
The partly erased joint, J8, shows a well-defined circular mirror boundary, but the
fringe is very irregular (Fig. 4.7d). A delicate radial plume reaches the boundary from
an inside location of the joint and continues radially downward and sidewise beyond
the boundary as irregular en echelon segments in diverse morphologies. The en ech-
elon segments are delicate in certain areas, while in other locations they assume coarse
relieves of hackles. This diversity fits between the two end members shown in
Fig. 2.1a,b, suggesting a range in KI between m and h (Fig. 4.35a,b).
The trefoil joint, J9, has the two most complex fringes on the right and on the left
sides of the mirror (Fig. 4.8 and 4.9). However, these fringes fit the pattern of Fig. 2a and
resemble the fringes of J7. Therefore their position along the V vs. KI curve is further left
from that of J8 (dash line in Fig. 4.35b). The joint with the hackled-fringe, J10 (Fig. 4.8c),
shows features resembling both Fig. 2a and Fig. 2b, but closer to Fig. 2b than to Fig. 2a.
We therefore place the right side of the frame further right from h (Fig. 4.35a,b).

4.8.6
Limitations of this Methodology

The curve shape in Fig. 4.34 depends on various parameters, including material prop-
erties as well as environmental conditions such as temperature, fluids and pH (e.g.,
Wiederhorn and Bolz 1970; Guin and Wiederhorn 2003); it is particularly sensitive to
stress and stress rate. Therefore, Fig. 4.35a,b may be modified in future research along
362 Chapter 4 · Jointing in Granites – Part 2

the various segments of the curve. Changes may occur along the sub-critical range due
to different water/vapor contents, such that various rocks would fit different region II–
III relationship. This will come with a deeper understanding of the properties of undu-
lations vs. arrest marks and plumes (striae), i.e., when criteria of distinguishing them on
mirrors that form sub-critically from mirrors that form post critically are established
(Sect. 2.2.4). Also, plotting of joints in the sub-critical range would be influenced by the
dependence of KI on v that may change along different v ranges (Guin and Wiederhorn
2003). Additionally, the KIc for granite may change, as KIc values are refined.
The shape of the post-critical “terminal velocity” segment may vary to fit different
dynamic fracture data. Results generally show a strong increase in fracture velocities
at the transition from the mirror to the hackle and branching zones (in glass) (Cramer
et al. 1999). But velocity fluctuations can also be expected (Sharon and Fineberg 1999).
The plot of hackles within a limited range close to h (Fig. 4.35b) would be quite reli-
able. However, the plot of en echelon fringes remains problematic until further infor-
mation on the dependence of these fringes on KI (rather than KI / KIII) will become
available. Meanwhile, a useful criterion would be (Sect. 2.2.4.2): Faint plumes and lack
of arrest marks imply relatively high KI (Fig. 2.56a), whereas pronounced plumes and
arrest marks suggest relatively low KI (Fig. 2.54a,b). All these may somewhat shift v and
KI values for particular joints, but they would not challenge the principles of the method.
Finally, note that while in Fig. 4.34 the curve of terminal velocity is horizontal, it has a
slight slope in Fig. 4.35. The actual shape of this curve is still unknown, and it may vary
in various materials. Moreover, the exact conditions of tension (or extension) are im-
portant. While under pure tension the terminal velocity may be attained rapidly (per-
haps on the mirror plane), deviations to bending by slight lateral compression may delay
the attainment of terminal velocity (Shand 1954, also Bahat 1991a, Fig. 2.19).

4.9
Velocity and Stress Intensity Manifestations of Fracture Propagation
in Granites
The plot in Fig. 4.35b offers a generalized quantitative frame within which various
other natural and experimental fractures can be examined; it also demonstrates some
of the limitations of this methodology. When using V and K in analyzing the fracture
history of joints, one needs to bear in mind several points:

1. Fracture velocity changes all the time, generally by growing up to the terminal
velocity, but also up and down.
2. En echelon segmentation is a complicated indicator of crack velocity.
3. When quantitative estimate of fracture behavior is impossible (Fig. 4.35b), it is
convenient to describe qualitatively the relative increase or reduction in either v or
K of specific fracture exposures.
4. There are significant differences in the fitting of curves that correlate v and K with
fractographic features (Arakawa and Takahashi 1991; Sect. 2.2.3.12).
5. There are theoretical disagreements regarding the issue of which parameter actu-
ally controls dynamic fracture (aspects of fracture velocity in Sect. 1.4, and aspects
of stress intensity (Sects. 1.4.3.1 and 1.4.3.2).
6. Both remote and local forces that act on the fracture, influence v and K.
 4.9 · Velocity and Stress Intensity Manifestations of Fracture Propagation in Granites 363

We present below two examples of fracture in granites that touch on these six points.
The first relates to fractographic features on the mirror plane of a post-uplift joint from
California (after Bahat et al. 1999), while the second concerns the fringes of cooling
joints from the Czech Republic (after Bahat et al. 2001a,b, 2003).

4.9.1
Fracture Velocities of Exfoliation Joints Cutting the El Capitan Granite
and Comparable Fractures

Fan-like markings on exfoliation surfaces from the middle height of El Capitan

are made of radial plumes and concentric ripple marks orthogonal to each other at
their intersections, which imply simultaneity (Fig. 4.27). This fracture morphology
suggests a sub-critical propagation rate below 4 × 10–5 m s–1, assuming dry conditions,
and below 10–2 m s–1, if the fracture were immersed in water (Bahat 1991a, p. 234 and
present Fig. 4.35). Accordingly, it would take longer than fourteen days to fracture a
fan having radial plumes 50 m long in dry conditions, and a fan immersed in water,
whose plume is 36 m long would fracture in more than an hour. Due to the uplift of
the El Capitan block, the middle height of the cliff is expected to be above the water
table. In addition, due to the low porosity of the granite, the above dry conditions are
preferred. If however buckling and splitting (Fig. 4.28) would enable water flow
into the fracture zone, a more rapid fracture would occur, requiring a much stron-
ger preference of downward fracture propagation compared with an upward propa-
gation, due to the flow of more water towards lower elevations. This is not sup-
ported by the fan pattern on the cliff (Fig. 4.27) so that propagation under dry con-
ditions is favored. Olson’s (1993) calculations for joints from Sierra Nevada granites
(generated by a mechanism other than exfoliation) suggest that fractures of some 10 m
long were formed in about three minutes or faster, possibly under conditions approach-
ing critical.

4.9.2
Fracture Velocities Relative to Estimated Paleostress Magnitudes

The calculated tensile paleostresses that initiated joints in granitic rocks under three
different geological conditions are presented in Table 4.7. The results compared here
were obtained by different procedures. The method applied to the analysis of a frac-
ture on Half Dome (Fig. 6.18) and granite from East Sinai (Bahat and Rabinovitch 1988)
is based on measuring fracture markers on surfaces of individual joints. The method
used by Segall and Pollard (1983) relies on measurements of extension strain accom-
modated by the dilation of many joints (and was used for the analysis of fracture in
granodiorites from Mt. Givens). Accordingly, these two methods pertain to local and
remote stresses, respectively.
Segall and Pollard (1983) assumed quasi-static crack propagation conditions in their
analysis of the initial tensile stresses that induced regional joints in the Mt. Givens
Granodiorite from the Sierra Nevada. The relatively high fracture stress values result-
ing in jointing at Mt. Givens (ranging from 1.2 ±1.0 MPa to 26 ±11 MPa) seem to sug-
gest syntectonic fracture (Bahat 1991a). Fracture was enhanced by pore pressure, as
testified by mineral fillings (Segall and Pollard 1983), but under considerable depths,
364 Chapter 4 · Jointing in Granites – Part 2

Table 4.7. Tensile paleostress conditions for jointing in granodiorite and in granite

that prevented fracture branching (Segall and Pollard 1983). Such conditions would
fit sub-critical fracture velocities (Table 4.7). Greater fracture stress results than those
mentioned above for the granodiorites were obtained for an uplift joint in granite from
Sinai. The latter paleostress results are less, but comparable to those obtained by ten-
sile experiments on granite (from 7 MPa to 18 MPa, Segall and Pollard 1983). Quite
possibly, the Sinai joint started to propagate from certain depths below the water table
while being exposed to excessive pore pressures that resulted in high fracture stress
values.
In summary, fracture velocities may change considerably during the fracture pro-
cess. A two stage process is assumed for the three granitic rocks:

1. Sub-critical fracture growth, which had occurred before critical conditions were
attained.
2. Post-critical fracture velocities that could have initiated under reduced overbur-
den pressures. The difference is in the amount of time needed for fracture growth
during the first stage.

The results suggest that the first stage was relatively long for the exfoliation in the
cliff of Half Dome, whereas for the joint in East Sinai this stage was relatively short.
However, many joints, possibly including the joints from Mt. Givens (Table 4.7) never
reach post-critical fracture velocities along their history, mostly because of overbur-
den constraints.
 4.9 · Velocity and Stress Intensity Manifestations of Fracture Propagation in Granites 365

4.9.3
Cycling Fracture

The previous sections relate to the general property of fracture to increase its frac-
ture velocities up to the terminal velocity (Fig. 4.34 or 4.35). Quite common however
are alternating changes in fracture velocity (and stress intensity) that are reflected from
fractographies, which are displayed on mirror planes (e.g., Fig. 4.7b, 4.36). Rarer and
more intriguing are cycles that are exhibited by both mirrors and fringes (e.g.,
Fig. 4.8a,b). We now interpret the observations from the mirror-left fringe relation-
ship of the trefoil joint (Fig. 4.9c,d) in light of some experimental results.

4.9.3.1
Fracture Velocities During Growth of the Trefoil Joint

Müller and Dahm (2000) formed plumes and fringe zones on surfaces of desiccation
tensile cracks that were created by drying starch-water mixtures (Sect. 2.2.3.9). Particu-
larly relevant to the present section were their observations that at transitions from the
parent joints to fringes, topographic amplitudes occurred, i.e., enlargements in the crack
area, were associated with a reduction in fracture velocities. The latter findings resemble
the experimental results obtained by Sharon et al. (1995). They applied tensile stresses
perpendicular to a thin plate (0.8 or 3 mm in thickness) made of cast PMMA. These
authors found that cracks discontinue their straight propagation and become “unstable”
beyond a critical velocity, vc, of 0.36vR, where vR is the Rayleigh wave speed in the ma-
terial (the velocity at which a wave moves along a free surface). This transition from
quasi-static velocities to dynamic ones involves two coupled phenomena, velocity os-
cillations and secondary micro-branching from the primary parent crack (Fig. 1.33).
Sharon et al. (1995) observed that the increase in new crack area by the formation of
the micro-branches was associated with reduction in fracture velocities.
The fractography of the trefoil joint (Fig. 4.9c,d) seems to fit the fracture principles
derived by Sharon et al. (1995). The trefoil primary mirror plane and en echelon seg-
ments (first fringe) are analogous to the primary and secondary (branch) cracks
(Sharon et al. 1995), respectively. The trefoil secondary mirror plane and secondary
en echelon segments reflect the stages of resumption of rapid propagation of the pri-
mary crack and subsequent appearance of secondary cracks (Sharon et al. 1995), re-
spectively. The implication is that the remote stress operating on the trefoil joint did
not subside (or perhaps was resumed after a quiescent period due to local conditions).
Therefore, exerted tension resulted in the creation of a new, secondary mirror that
propagated in the same downward direction, as shown by its downward facing con-
vex boundary. The increase in length of the secondary mirror was associated with
increasing its fracture velocity in analogy to the above mentioned fracture principles
(Sharon et al. 1995). This fracture velocity was reduced again, resulting in the second-
ary fringe, i.e., the energy dissipation that occurred by creating the new segment sur-
faces (tensile and shear) was associated with a reduction in fracture velocity, as also
observed by Müller and Dahm (2000). Thus, the primary mirror of the trefoil joint
was segmented into a fringe that then continued to fracture into a secondary mirror,
which also was segmented into a secondary fringe.
366 Chapter 4 · Jointing in Granites – Part 2

4.9.3.2
The Formation of the Trefoil Joint under Mixed Modes Loading

Couplings of velocity decrease with the increase in new fracture surface area were
observed both under quasi-static conditions (Müller and Dahm 2000) and through
dynamic ranges (Sharon et al. 1995). What are the differences between the two?
Sommer (1967, 1969) simulated en echelon segmentation on the fracture surfaces of
circular glass rods loaded in tension by the superposition of a small amount of tor-
sion, creating conditions of increasing mode III / mode I ratios. On the other hand, high
KI / KIII ratio increased crack velocity, leading to crack branching. Bank-Sills and Schur
(1989) showed the inherent tendency of increasing mode III / mode I ratios from the
center of metallic plates towards their boundaries, and Bahat (1997) reported that the
latter tendency occurred also in sub-horizontal chalk beds.
All these experimental and field observations match into the following scenario of
the trefoil joint growth. The primary joint propagated down under conditions of in-
creasing KI magnitudes and very low KIII / KI ratios to the mirror boundary at which
an abrupt increase in the mode III / mode I ratio occurred, leading to the breakdown
into en echelon cracks (Cooke and Pollard 1996). This was linked with energy dissipa-
tion on many new crack surfaces and a reduction in crack velocity. The crack speed
reduction was probably associated with the “mode III crack closure” effect, which was
linked with friction and interlocking interferences along the fracture surfaces (Tschegg
1983). Crack speed reduction can be very substantial under these conditions: En ech-
elon cracks have been observed in borosilicate glass after application of stress with
an intensity factor less than the fatigue limit at very low crack velocities (Wiederhorn
and Johnson 1973, Fig. 5). As mentioned above, recent experiments on rupture veloci-
ties in starch show similar results (Müller and Dahm 2000). The second mirror reflects
resumed conditions of increasing KI magnitudes. Hence, the micro-branching pro-
cesses described by Sharon et al. (1995) relate to crack velocity oscillations under high
KI / KIII ratios, whereas the experiments conducted by Müller and Dahm (2000) took
place under increasing KIII / KI ratios.

4.9.3.3
Fractographies that Pertain to both Quasi-Static and Dynamic Fractures

How can en echelon segments (generally reflecting quasi-static propagation Sect. 4.7.3.1)
and hackles (generally reflecting dynamic propagation) occur adjacently on a single
fringe? (e.g., Fig. 2.1c). The answer is based on the assumption that even “slight devia-
tions” of the direction of the minimum principal compression (or maximum principal
tension) from orthogonality to the fracture surface would induce en echelon segmen-
tation (Sommer 1969). Accordingly, fringes formed by both quasi-static and dynamic
propagation may be influenced by such “slight deviations”. Fractures formed quasi-stati-
cally when KIII / KI are high will be dominated by en echelon fringes (Fig. 2.1a), and these
fringes will have the tendency to assume simple, uniform widths (Fig. 2.34b). On the
other hand, fractures formed dynamically when KIII / KI are low will be dominated by
hackle fringes, on which however en echelon segmentation in a shingle-like arrange-
ment may be superimposed in accordance with this ratio. Such superposition can ex-
 4.10 · New Fracture Area in the En Echelon and Hackle Fringes on Joint Surfaces 367

plain the frequent occurrence of fringes that show hackles in a shingle-like arrange-
ment (center of Fig. 4.12b). The latter fringes often exhibit non-uniform widths
(Fig. 2.34a). Hence, it is quite possible that on the fracture surface obtained by De
Freminville (1914) (Fig. 2.1c) the two fringe types reflect local slight changes in the
extent of orthogonality of the maximum principal tension: The fringe area that displays
hackles reflects “pure” mode I, whereas the fringe area that displays hackles in a shingle-
like arrangement reflects mixed mode I + III loading.
In summary, most fringes from sedimentary rocks suggest quasi-static growth, while
rare fringes display cuspate hackles (Fig. 2.30a and 2.42) or hackles in shingle-like
arrangements (e.g., Fig. 2.33e). Note the resemblance between the latter fracture mor-
phology and the fracture surface of a turbine that is interpreted to have initiated in
fatigue growth and possibly approached catastrophic conditions, i.e., under KI ~ KIc
conditions (Fig. 2.44a) (Bahat 1991a, p. 133). In many outcrops the distinction between
these fringe types would be relatively easy, whereas in others that exhibit transitional
fractographies, this distinction would be more problematic. Hackle fringes are far more
abundant in granites than in sedimentary ones, but even in these rocks hackles often
show shingling. Note that in Fig. 4.11a in the lower fringe shingling is almost absent,
whereas in the upper fringe shingling is more pronounced.

4.10
New Fracture Area in the En Echelon and Hackle Fringes
on Joint Surfaces from the Mrákotin Quarry
4.10.1
Fracture Area Increase in the Transition from En echelon to Hackle Fringe

A basic physical tendency of a cracked body is to minimize its stored strain energy
(Eq. 1.19), which is done by the creation of new crack area (Eq. 1.20). Therefore, mea-
suring changes in crack area is important for the characterization of the fracture pro-
cess. A comparison of the relative flat areas of the “segments” on the fringe of joint C
(Fig. 4.12a and 4.37a) to the extremely wrinkled “flake” surfaces of the hackles on the
fringe of joint A (Fig. 4.11a–c and 4.37a,b) intuitively suggests that the overall frac-
ture area of the latter fringe is significantly larger than the former fringe. This intui-
tive observation regarding the difference in new area in the two fringe types needs to
be proven, and meeting this requirement is the main objective of this section, citing
Bahat et al. (2002b).
The extent of superposition of two adjacent segments (or flakes) in a fringe is of-
ten an enigma. The best way of resolving this question is by removing the upper seg-
ment and examining the area underneath the upper segment (Bahat 1986a, Fig. 7). This
provides two measurable parts: The revealed area of the upper segment before its re-
moval and the hidden area that belongs to the lower segments. The ratio between the
two can then be determined. This determination is however difficult to accomplish
because in many outcrops the segments are inaccessible, requiring the pursuit of esti-
mation methods that rely on certain fracture mechanic considerations.
The en echelon segments are approximately flat on the fringe of joint C and are not
divided into multiple smaller cracks. Therefore, the actual fracture area is given by
368 Chapter 4 · Jointing in Granites – Part 2

Fig. 4.37. Joints A and C from the Mrákotin quarry (Fig. 4.10). a The “triangle” within the hackle fringe
of joint A (left), where cM is on the mirror boundary, and X is the most distant point on the lower fringe,
at a continuation of the triangle, and CB is remote from x. The “rectangle” within the en echelon fringe
of joint C (right). b An expansion of the triangle, showing the drawn traces of flake boundaries

the sum of the measured, revealed segment areas plus the sum of the estimated un-
derlying, hidden surfaces due to shingling. The number of the hidden parts equals that
of the segments, and they are approximately the same for all the observed segments
of similar size, which maintain the same twist angle.
The fracture surface is however considerably more complex in the hackle-fringe of
joint A. A superposition in an en echelon fringe involves the shingling of two adjacent
segments, whereas in a hackle-fringe, several (from two to five) small hackle-like flakes
often superpose a single “large hackle” (Fig. 4.11a). Furthermore, in this flaky texture,
the multi-surfaces that often deviate from planarity overlap each other on a given
nominal area such that the actual crack area in the hackle-fringe is a greater sum than
the shingling surfaces.
The estimation of the difference between the ratios of the hidden surface area (HA)
divided by the revealed surface area (RA) in the two fringes is presented below. We
recognize that the difference in the nature of the steps between adjacent segments is
different in the two fringes. The step areas are similar for all the adjacent segments in
the en echelon fringe, whereas in the hackle-fringe, some steps are much larger than
others. However, for the sake of simplicity we concentrate on the fracture areas of seg-
ments and hackles and do not enter the contribution of the steps and secondary seg-
mentation into our calculation, which are far smaller than the segments. This would
reduce the accuracy of our measurements by about 10%.

4.10.2
Estimation of the Fracture Areas in the Fringes of Joints A and C

The en echelon fringe on joint C. The exposure of the fringe of joint C is not of a uni-
form quality throughout its entirety. Therefore we selected a representative area of a
rectangle shape for the detailed study. The “rectangle” is 8 m2 on which 19 ±1 m frac-
ture length of en echelon segments were measured (Fig. 4.37a). The rectangle was se-
lected as a representative of the fringe, because:
 4.10 · New Fracture Area in the En Echelon and Hackle Fringes on Joint Surfaces 369

1. This is a nominal fringe area on which the distribution of fracture length per area
is relatively large compared to other fringe areas of this joint.
2. The appearance of fractures is clear, rendering them to be measurable with a rea-
sonable confidence.
3. The rectangle is located at a far end of the fringe, at a relatively long distance from
the mirror boundary, such that it fits our “assumption” (see below).

The total fracture area in the rectangle consists of the revealed (nominal) area plus
the hidden area. The hidden area between two neighboring segments is not known. In
two locations, the maximum superposition at the bottom of the fringe is measured to
be (0.2 ±0.1 m) (looking upward towards the mirror). However, this superposition does
not necessarily represent the hidden area throughout the various segments, because
there is no certainty that the same relative superposition is maintained along the seg-
ments. En echelon segmentation forms by mixed mode I + III (Sect. 1.1.3). Mode III
contributes to tearing the rock in a scissors-like operation. This requires that shingling
is zero at the segmentation initiation and the hidden area is increased gradually, stepwise
or otherwise. Our “assumption” is that there is a linear increase in the hidden area with
distance from the location of superposition initiation (O in Fig. 4.38a).
At the end of the en echelon fringe, the average overlap of two adjacent segments (AB)
is 0.2 ±0.1 m. Therefore, the angle ϕ = 1.2 ±0.6° (Fig. 4.38a, see also difference between
l and li). The hidden area (HAE) is produced by two adjacent segments along the width
of the rectangle, and is given by

(4.2)

Fig. 4.38. Diagrams for calculating crack areas in fringes. a A schematic picture of the overlap develop-
ment from the location of superposition initiation O on the rectangle in Fig. 4.37a. AB: The amount of
overlap at the end of the rectangle. 2ϕ is the opening angle of overlap (exaggerated); l = OB; l1 = OC;
dashed area: Overlap inside rectangle; ϕ = 1.2°; therefore, sin ϕ / 2 = 1 / 2 × AB / l = 1 / 2 × 0.2 / 4.8 = 0.0208.
b Schematic picture of overlap of a single flake. li: Length of flake (sub orthogonal to mirror boundary)
inside the hackle triangle. Dashed area designates overlap = HPHi; 2ϕ is the opening angle (exagger-
ated). HPHi = 1 / 2li sin 2ϕ [area of triangle = 1 / 2base × height = 1 / 2 (2li sin ϕ (li cos ϕ))]
370 Chapter 4 · Jointing in Granites – Part 2

Accordingly, HAE = 1 / 2 (4.82 – 2.82) sin 2.4° = 0.32 ±0.16 m2. The hidden area per
1 meter segment boundary length is therefore 0.16 ±0.08 m2 m–1. The revealed area
(RAE) of the rectangle is 8 m2, and the total segment boundary length in the rectangle
is 19 ±1 m. Hence, the total hidden area is 2 × 0.16 × 19 = 6.08 m2 ±3.00 m2 within the
nominal area (the factor of 2 comes from the mirror image to the fracture on its other
side that was lost by quarrying) and the ratios of the hidden surface area (HAE) di-
vided by the revealed surface area (RAE) is: uE = 0.76 ±0.4.

The Hackle fringe on joint A. The “triangle” was selected as a representative of the
fringe, because it contains a fairly uniform and clear fracture distribution, rendering
them to be measurable with a reasonable confidence. Note that the base of the tri-
angle is on the mirror boundary (Fig. 4.37a). In analyzing the “triangle” we use a simi-
lar approach to that applied for the “rectangle” in that we assume that each flake bound-
ary (equivalent to segment boundary) overlaps its neighbor by the same angle as that
of the en echelon segments, i.e., ϕ = 1.2 ±0.6° (Fig. 4.38b). Table 4.8 summarizes the
measurements and calculations of the flake boundaries across the triangle (Fig. 4.37b).
Fractures (flake boundaries) li are measured along their trends sub orthogonal to the
mirror boundary. These measurements do not include flake boundaries that curve
into orientations that form angles greater than 45° with these trends. Accordingly, the
total hidden area is

(4.3)

Accordingly, (see Table 4.7 for the calculation of Σ Si l i2)

i

The revealed area (RAH) of the triangle is ~1.86 m2; hence the ratio of hidden area
to the revealed area in the triangle, uH = 2HAH / RAH = 0.38 ±0.2.
For a fracture growing in tension under a constant load, the energy release rate G
is defined as the elastic energy (per unit width) released from the field per unit crack

Table 4.8. Flake boundaries and hidden area in the triangle of joint A
 4.10 · New Fracture Area in the En Echelon and Hackle Fringes on Joint Surfaces 371

length (Sect. 1.3.3.5). G is usually proportional to σ2c where σ is the stress field value
and c is the crack length: G = nσ2c. On the other hand, by the Griffith criterion, when
the crack becomes critical, i.e., for c = ccr (at the critical flaw, Fig. 2.1a), the energy re-
lease rate is equal to 2Γ , where Γ is the energy per unit area, 2Γ = nσ2ccr. Beyond ccr,
the excess energy released per unit crack length is therefore nσ 2(c – ccr). This energy is
manifested either as new created area (as in the hackle and bifurcation zones, and to a
lesser extent in the mist zone), or as kinetic energy, heat, deformation, radiation, etc.
We assume that the linear relationship between G and c continues throughout the
fracture length. Therefore, if the fracture had bifurcated, at say, cB (an imaginary point
that should be further away, beneath point X (Fig. 4.37a), G there would have to be close
to twice that of G at cM (Beauchamp 1995) (to create twice the fracture area, provided
that the kinetic energy per unit crack length changes very little at the bifurcation);
then we can estimate the maximal additional (hidden) area ratio as a function of crack
length, c, to be ∆ = (c – cM) / (cB – cM), where cM is on the mirror boundary (Fig. 4.37a).
In the fringe of joint A, the most distant point, which can be observed on the triangle
(point X) is at a distance of 5.20 m from the origin of the crack, and it is clear that no
bifurcation has occurred down to this point. Therefore, ∆ ≤ (c – cM) / (X – cM). At the
hackle region at the center of the triangle c ~ 3.06 m, which limits the hidden area
ratio uH (see below) at this point to be less than (3.06 – 1.65) / (5.20 – 1.65) = 0.4.

4.10.3
A Comparison between the Fringes of Joint C and Joint A

The part of the energy release rate, which is used to create the overlapping area is
G – 2Γ – qc – p, where qc is the kinetic energy per unit crack length (q is proportional
to v2, where v is the crack speed). For simplicity, we neglect the plastic energy, p. At cM,
G(cM) = 2γ + qcM. Therefore, for cM < c < cB we get that the ratio of the additional area
to nominal area is proportional to c / cM – 1 = (η).
In order to compare hidden area ratios between the en echelon and the hackle
fringes, they should be both at the same scaled distance from the respective crack ori-
gins. Since they are situated at different distances, the following procedure is conducted.
For the en echelon segments in the rectangle, its average distance from the origin is
5.88 m and the distance from the origin to the mirror boundary is cM = 0.94 m; there-
fore, ηE = 5.88 / 0.94 – 1 = 5.26. For the hackles in the triangle, a similar calculation
yields ηH = 3.06 / 1.65 – 1 = 0.85. If it is assumed that the hidden area is linear with crack
length, the hidden area should be normalized by division by η for an appropriate com-
parison. We therefore compare the ratio of hidden area to revealed area at the “same
distance” for both fringes. For the hackle fringe we have rH = uH / ηH = 0.38 / 0.85 = 0.45,
while for the en echelon fringe it is rE = uE / ηE = 0.76 / 5.26 = 0.14, which is 3.14 times
smaller. In summary, the normalized (real) ratios of hidden area divided by revealed
area on the fringes of joints A and C are respectively, rH / rE = 0.44 / 0.14 = 3.14.

4.10.4
How was Energy Released Differently in the Two Fringes?

It turned out that energy release in the actual formation of new fracture area (per a given
nominal fringe area) is substantially greater in the hackle-fringe than in an en echelon one.
372 Chapter 4 · Jointing in Granites – Part 2

What is the reason for this? We follow the principle that G = GI + GII + GIII (Lawn 1993,
Eq. 2.18) and we neglect GII because of its minor importance in relation to fracture condi-
tions in the two fringes under consideration. The G relationship then becomes

(4.4)

That is, the energy release rate by mode I loading in the hackle fringe equals the
energy release rate by modes I + III loading in the en echelon fringe. How do we ex-
plain the reduction of the “effective” G to produce new fracture area in the transition
from the hackle fringe to the en echelon one?
The state of stress influences the size of the plastic zone, and shear stresses pro-
mote the enlargement of this zone (Hahn and Rosenfield 1965). More specifically, the
plastic zone associated with the mode III loading is several times larger than that for
mode I for similar materials and at the same stress intensity values (Tschegg 1983).
Plastic deformation is associated with the loss of considerable energy into heat. The
latter energy transition may reach extreme values at rapid fracture velocities (up to
60% or even 90% of G in PMMA and some metals) (Döll 1976), but even at low crack
velocities heat losses may be significant under induced plastic conditions. Hence, the
transformation of some GI into GIII in the en echelon fringe involves an increase in
plastic deformation and subsequent heat loss, which is not used in the creation of new
fracture surfaces.

4.10.5
Summary

The distinction between fringes of en echelon segmentation and hackles on joint

surfaces is important because they reflect different fracture conditions. It was found
that the normalized (real) ratios of hidden area divided by revealed area on the hackle
and en echelon fringes of joints A and C are respectively, rH = 0.44 and rE = 0.14, hence,
rH / rE = 0.314. This shows that considerably more new area is formed in hackle zones
than in en echelon zones per any given nominal area of the fringe. Possibly, the trans-
formation of some GI to GIII in the transition from hackle to en echelon fringe involves
an increase in heat loss, which is the portion of energy not used in the creation of new
fracture surfaces.

4.11
The Index of Hackle Raggedness (IHR) on Joint-Fringes
of the Mrákotin Joint Set
This section concentrates quantitatively on the fractographic differences between the
fringes of joints A, C and F (Sect. 4.4.3), citing Bahat et al. (2004). It has been shown
that considerably more new area is formed in hackle sections than in en echelon sec-
tions per given nominal area of fringe (Sect. 4.9). Since the crack area is a fundamen-
tal parameter in fracture physics, we seek to devise a method of correlating changes
in crack area by tectonofractographic techniques, i.e., by the characterization of the
secondary fracture on joint fringes. The two distinct morphologies manifested by the
en echelon and hackle fringes (Fig. 2.1) represent contrasted fracture mechanisms,
 4.11 · The Index of Hackle Raggedness (IHR) on Joint-Fringes of the Mrákotin Joint Set 373

reflecting different local stress conditions (Sect. 4.8.4). We start dealing with this re-
lationship by formulating a series of criteria for estimating the value of the Index of
Hackle Raggedness (IHR) on a given fringe, based on geometric distinction of crack
features between the hackle and en echelon end members in the fringe. We then apply
it to joints A, C and F from the Mrákotin set, summarizing the results in Table 4.9.

4.11.1
Criteria of IHR and Application to the Mrákotin Joint Set

Before getting into the actual definition and procedure, one has to consider three limi-
tations that qualify the rules. First, there is a need to establish which of the seven key
criteria presented below is applicable to a given outcrop. Nature is quite selective in
revealing good fracture markings and very often only small parts of the ideal
fractography (Fig. 2.1a,b) is exposed. Occasionally, only one fringe is well displayed
(the upper or lower). Therefore, in many outcrops only a certain number of these cri-
teria would be useful, according to the quality of the crack exposure. Second, it is re-
quired to apply independent estimations of the IHR, ranging from 1 to 10 for each of
the above criteria. Third, since the IHR is new (even for the engineering literature),
lack of previous experience justifies at this stage the assumption of equal weight for
the various criteria. Therefore, the IHR for a given joint will be the mean of IHR val-
ues from all measurable parameters (criterion 3 is an exception, where the angle of
“triangularity” sinα is converted to IHR by multiplication by 10). The total IHR error
per joint, consisting mainly of measurement errors is thought to be ±0.7.

4.11.1.1
Length Variability

In many fringes en echelon segments maintain uniform lengths, if microcracks in

root zones (Bankwitz 1966) and elsewhere on the segments are excluded. Hackles on
the other hand show great length variability even among neighbor cracks. Hence, in-
creasing deviation of length ratios (of maximum to minimum length in a given fringe)
from length uniformity (Fig. 2.34) produces an enlargement of IHR. The length ra-
tios of fringe-cracks for joints A, C and F are 6, 1 and 3, corresponding to the respec-
tive IHR.

4.11.1.2
Rectangularity

En echelon segments approximate rectangular shapes whose aspect ratios can be

readily measured. Hackles on the other hand deviate considerably from such shapes.
Measurement of the deviations from straight lines (DSL) along the rectangle length
and width can provide information regarding the departure of the fringe crack from
its rectangular shape. We use a set of profiles that resembles the scale of fracture-sur-
face roughness constructed by Barton and Choubey (1977), which shows a gradual DSL
ranging from 0 to 10 (Fig. 4.39). The DSL value is the mean of combined deviations
along the length and width of three fringe-cracks, assessed manually. Thus, the DSL
values of A, C and F are 6.5, 2.5 and 3.5, respectively.
374 Chapter 4 · Jointing in Granites – Part 2

Fig. 4.39.
A set of profiles that show
gradual deviations from straight
lines on a scale ranging from
0 to 10 (modified from Barton
and Choubey 1977)

4.11.1.3
Fringe Width

The tendency of en echelon fringes (to be distinguished from individual segments) to

maintain uniform widths, compared to non-uniform widths of hackle fringes
(Fig. 2.34) has been repeatedly observed (Bahat 1998b). This is also demonstrated by
the “quasi-triangular” shapes of the two fringes of joint F (Fig. 4.12c) compared to the
approximate parallelism of the mirror and fringe boundaries at the lower side of joint C
(Fig. 4.10c). A measure of deviation from a uniform fringe width is given by the angle α
that increases with the “triangularity” of the fringe. Unfortunately, this parameter can-
not be measured with high confidence for joint A because only certain parts of the
fringe lengths of joint A are revealed (sufficient for the study of criteria 1–5 but not
for this parameter) so that an appropriate index cannot be established for the latter
joint. Accordingly, the sin α values for joints C and F were estimated to be 0.17 and 0.34
(α = 10 ±2° and 20 ±2°), respectively, and the corresponding IHR for these joints are 1.7
and 3.4. When clear triangularity is recognized this method would be more reliable
than the one given in Sect. 4.11.1.1.

4.11.1.4
Fringe Thickness

En echelon segments have approximately (quasi two-dimensional) planar surfaces and

display uniform thickness along their lengths. On the other hand, hackles consist of
multi-fracture surfaces superposing each other not uniformly along the crack lengths,
often leading to the increase in thickness of individual cracks and in the fringe thick-
ness (FT) with distance from the mirror-fringe boundary. Therefore, FT is normally
maximum at the far end of the fringe and minimum close to the mirror boundary.
FT is abruptly reduced when fracture branching occurs, but there is not clear evidence
that such a process actually took place in this outcrop. The change in FT often renders
 4.11 · The Index of Hackle Raggedness (IHR) on Joint-Fringes of the Mrákotin Joint Set 375

the hackle-fringe to look quasi three-dimensional (Fig. 4.11c), such that IHR increases
with the ratio of maximum / minimum FT. Changes in thickness were measured on
an enlarged profile of a photograph of joint A. These ratios for A and C were found to
be 6.5 and 1 ±1, respectively (no measurements for joint F are available).

4.11.1.5
Fringe Angularity

There are indications that the angle φ that the fringe (as a set of cracks, rather than
an individual one) forms with the mirror, increases with the “fracture-dynamics” of
the process, i.e., roughly corresponding to the increase in fracture velocity. Bahat and
Rabinovitch (2000) found that in the same rock (chalk) the φ varies from 0° to 30 ±5°
on a natural joint and from 30 ±8° to 50 ±8° on an artificial fracture along a road cut
(that was formed by explosion). It is assumed that this relationship may change sig-
nificantly from rock to rock (see more in Bankwitz and Bankwitz 2004 and Sect. 4.4.4).
It was found that φ = 5 ±2° for joint A, while φ = 0–1° for joint C (it could not be mea-
sured for joint F). We therefore assign IHR values of 5 and 1 ±1 to joints A and C, re-
spectively.

4.11.1.6
Crack Dipping

Deviation of individual fringe-cracks from a uniform dip also increases the IHR. In-
dividual en echelon segments in a fringe commonly dip in the same direction of the
mirror, i.e., the “normal direction”, forming zero to small +φ angles. However, occa-
sionally, different hackle flakes may also dip into the mirror plane, i.e., in the “oppo-
site direction”, forming –φ . For determining the IHR value we combine the maxima
of +φ with those of –φ for different cracks on the same fringe to a single angle φ f rep-
resenting a fringe. Our results show that the φ f values for joints A, C and F are 10 ±5°
(+φ = 7 ±3° plus –φ = 3 ±2°), ~0.5°and 2–3°, respectively. Accordingly, the correspond-
ing IHR are 10, 1 and 2–3.

Table 4.9.
Summary of IHR indices for
joints A, C and F
376 Chapter 4 · Jointing in Granites – Part 2

Fig. 4.40. A diagrammatic summary of six criteria for estimating the value of the index of hackle
raggedness on joint fringes (see text for explanation)
 4.12 · A Method of Drilling in situ Fracture from the Rock 377

4.11.1.7
Secondary Mirror

Gross deviations from “normal fringe morphologies” require special attention. They are
very rare on en echelon fringes and are rare on hackle fringes. Quite intriguing is the
rarity of “secondary mirrors” in natural exposures (Fig. 4.11a–c), compared to their
frequent appearance in dynamically fractured glass (e.g., Fig. 2.30c). It is presently im-
possible to construct a scale for secondary mirrors. Quantification of this parameter
remains a challenge for future study. Qualitatively, however, this distinction supports
the relatively high IHR value assigned to joint A that exhibits a secondary mirror.

4.11.2
Discussion

The seven criteria. Criteria 1–6 are summarized diagrammatically in Fig. 4.40. Note that
criteria (1–3) relate to changes observed parallel to the mirror, while criteria (4–6) con-
cern parameters perpendicular to the mirror. Criterion 7 requires measurements both
parallel and perpendicular to the mirror. We did not use deviations of fringe cracks
from a uniform strike as an IHR criterion, because we unexpectedly found it not to be
diagnostic. The gap between the top value of the scale 10 and the value 6.8 found for
joint A probably suggests that joint A was not formed by the most dynamic conditions,
and the 10 IHR value should be obtained under more extreme dynamic conditions,
such as intense fracture of glass would occur in the laboratory (e.g., Fig. 2.30, see also
Bahat et al. 1982, Fig. 2).

4.11.3
Summary

A method of quantifying the “spectrum” of secondary fractures on joint fringes rang-

ing from en echelon segments to hackle cracks is presented. The method is based on a
set of seven criteria that enable one to place every fringe on an “index of hackle rag-
gedness” (IHR) along a scale from 0 to 10. This method is applied to the “Mrákotin
joint set” from a granite quarry in the Czech Republic part of the South Bohemian
Pluton. The results show an IHR range between 1.2 ±0.7 and 6.8 ±0.7. The gap that exists
between 10, the top value of the scale and 6.8 the maximum obtained value, suggests
that the joint for which IHR = 6.8 was not formed by the most dynamic conditions,
and the 10 IHR value should be obtained under more extreme dynamic conditions.

4.12
A Method of Drilling in situ Fracture from the Rock
An important structural problem is the determination of shear offsets along joints,
which often are concealed in an outcrop scale, particularly when offsets are limited to
short distances (millimeter scale). Thus, transitions from joints to faults are often
obscured. This enigmatic situation may be fully or partly resolved by resorting to pet-
rographic methods. A detailed study of millimeter and sub-millimeter scales of past
378 Chapter 4 · Jointing in Granites – Part 2

natural offsets along the fracture requires the assurance that in situ conditions are being
maintained in the laboratory.
The objectives in sample preparation are to keep intact both margins of the exam-
ined joints following their removal from the outcrop so that the original aperture (or
thickness, Fig. 2.8a) would be examined. The sample preparation process consists of
several stages:

1. Coat the trace of the fracture with gray epoxy.

2. Inject red epoxy into the aperture to fill the space and stick to the two margins of
the fracture.
3. Core drill samples.
4. Vacuum imbed selected samples with blue epoxy.
5. Prepare thin sections.

The “epoxy procedure” consisting of different epoxy colors helps to follow the tim-
ing of shear offset in the thin sections.
A 25 gauge 5/8" syringe needle is inserted into the selected fracture (Fig. 4.17c).
Then, a Hardman brand double-bubble epoxy is smeared on the front of the fracture.
This secures the needle in place, and prevents the forthcoming red epoxy from flow-
ing out of the front of the fracture upon injection. Also, it affixes the two margins of
the fracture at the rock surface together to maintain it in the original juxtaposition
after sample removal. This epoxy sets in about 5 min.
The next step is to prepare the red epoxy mixture for injection into the fracture.
Twenty drops of toluene are mixed into 4 pouches (totaling 8 g) of Hardman epoxy, to
which fifteen drops of Morefast brand red epoxy die are added. All of this mixture is
stirred for 1.5 min. The epoxy mixture is poured into a 5 cc Luerloc syringe or a
1 cc slip-tip syringe with their plunger removed. The plunger is put back in, and the
syringe is then attached to the needle that had previously been imbedded in the rock.
The plunger is then depressed, causing the red epoxy to flow into the fracture. Because
of the tightness of the aperture, it is impossible in some fractures to obtain penetra-
tion of large quantities of injected epoxy. However, even little penetration may occa-
sionally be enough to secure original juxtaposition in a few cylinders. Handling time
for the red epoxy mixture is 10–15 minutes, and the setting time, 1 hour.
Chapter 5

Electromagnetic Radiation Induced in Fractured Materials

The key to a successful application of electromagnetic radiation (EMR) to rock frac-

ture investigation is a good understanding of the basic physics of EMR emission. Such
an understanding is expected to provide useful universal tools or “fingerprints” of the
process involved. In this chapter we describe our efforts in the last decade investigat-
ing EMR in the radio frequency range (1 kHz to 50 MHz, Fig. 5.1) emitted by fracture.
These investigations were conducted in three consecutive principal stages:

1. “Small-scale” laboratory investigations, which consisted of two sub-stages, i.e., quasi-

static and dynamic. The aim of this stage was to derive the basic physical prin-
ciples operating behind EMR emission from microcracking excited by stiff com-
pression and tension and to find the similarities and dissimilarities between EMR
emissions from different dynamical and quasi-static cracking.
2. “Medium-scale” pilot investigations carried out both in open quarries during blast-
ing and in underground mines prior to rock bursts and rock and gas outburst. The
goal of this stage was to understand the differences in EMR during “low scale” and

Fig. 5.1.
Electromagnetic wave
spectrum (E (Exa): 1018,
P (Peta): 1015, T (Tera): 1012,
G (Giga): 109, M (Mega): 106,
k (kilo): 103, c (centi): 10–2,
m (milli): 10–3, µ (micro): 10–6,
n (nano): 10–9, p (pico): 10–12
(Rabinovitch et al. 2000c)
380 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

“medium scale” cracking and thereby to pave the way for planning “large-scale”
investigations.
3. The “large-scale” stage is meant to study EMR behavior during earthquake nucleation.
Since earthquake monitoring and prediction by seismological techniques is “still in its
infancy” (Wyss 2001), the possible addition of more sophisticated EMR methods should
be welcome. In particular, the EMR method measures micro-cracking, i.e., the initial
process in an earthquake, a process occurring much earlier than the actual earth move-
ments; the study of which might thus be beneficial for earthquake forecasting.

Accordingly, this chapter (a) summarizes our investigations in the small and me-
dium scales, (b) reports some improvements suggested thereby for monitoring and
prediction in the mining industry and (c) points out guidelines for large-scale experi-
ments (earthquake forecasts) based on our acquired theoretical understanding and
the “fingerprints” accumulated from tensile and shear experiments.
We hope that this chapter, by describing the current state of the art, could be inter-
esting not only to EMR specialists, but also to rock physicists, geoscientists, and min-
ing engineers who always seek new methods for multi-scale failure investigation.
We begin the chapter with a review of previous EMR investigations (Sect. 5.1), then
present our new EMR model (Sect. 5.2) and describe in detail its experimental verifi-
cation (Sect. 5.3). A comparison of this model with previous ones is presented in
Sect. 5.4. Individual EMR pulses and their classifications are given in Sect. 5.5 followed
by Sect. 5.6, where influence of the elastic properties of rocks on EMR activity is ana-
lyzed. A novel EMR phenomenon, polarization induced in materials during percus-
sion drilling, is examined in Sect. 5.7. EMR effects induced by blasting in open quar-
ries and appearing prior to volcanic eruptions and earthquakes are considered in
Sect. 5.8, while universal EMR features are discussed in Sect. 5.9. In Sect. 5.10, which
we believe would be interesting mostly for mining engineers, we focus on underground
EMR observations prior to rock burst and gas outburst deriving criteria of their fore-
casting. Finally, a brief summary of the whole chapter is presented.

5.1
History of EMR Research

Emitted EMR from fractures was firstly observed by Stepanov (Urusovskaja 1969). He
loaded samples of KCl and detected an appearance of electromagnetic pulses during
specimen failure. At that period, the EMR did not gather any implementations, and in-
vestigations of this phenomenon were continued basically on samples of alkaline ha-
lide crystals (Nabaro 1967; Kornfel’d 1971, 1977; Gol’d et al. 1975; Finkel et al. 1975; Golovin
and Shibkov 1986a,b). It was observed that the creation of new microcracks stimulated
excitation of EMR (Gol’d et al. 1975; Golovin and Shibkov 1986a,b). These studies were
followed by numerous ones that investigated various EMR aspects. The EMR phenom-
enon was measured for fracture in different materials, including metals and alloys (Misra
1975; Jagasivamani and Iyer 1988), single crystals (Gol’d et al. 1975; Khatiashvili 1984),
rocks (Nitsan 1977; Warwick et al. 1982; Ogawa et al. 1985; O’Keefe and Thiel 1995;
Rabinovitch et al. 1995, 1996, 1998) and ice (Fifolt et al. 1993; Petrenko 1993).
EMR investigations during rock specimen compression, which in the majority of
cases were combined by acoustic emission (AE) measurements (Nitsan 1977; Goncharov
 5.2 · Foundations of EMR Model 381

et al. 1980; Warwick et al. 1982; Sobolev et al. 1982; Ogawa et al. 1985; Cress et al. 1987;
Yamada et al. 1989; Frid 1990), showed that EMR was indeed detected at the same mo-
ment that microcracks were formed. EMR features on different scale levels (earthquake
(EQ), rock burst (RB) and rock sample fracture) were studied by Khatiashvili (1984).
In all of these cases, EMR anomalies were observed preceding the final rock failure.
According to the theory of material destruction (Zurkov et al. 1969; Regel’ et al. 1972;
Kuksenko and Mansurov (1986); Kuksenko et al. 1985, 1987; Petrov and Gorobetz 1987;
Gueguen and Palciauskas 1994), the process of rock fracturing consists of two stages: The
first one consists of an accumulation of microcracks, and the second consists of microcrack
clustering and final failure (Sect. 1.2.3.3). Rock destruction processes on all scale levels (from
rock samples up to crust fracture) adhere to this “scheme” (Kuksenko et al. 1987; Mansurov
1994). As hypothesized by Gershenzon et al. (1987), EMR is initiated in the first stage, ap-
pears before final failure and could be an EQ forecast indicator.
Large-scale EMR studies prior to rock burst (RB) in mines are also registered.
Nesbitt and Austin (1988) registered EMR in a gold mine (2.5 km depth). Frid (1997a,b,
2000, 2001) observed EMR anomalies before RBs and gas outbursts. Vozoff and Frid
(2001) measured EMR anomalies prior to a roof fall in a coal mine. Sakai et al. (1992);
Tomizawa et al. (1994); Tomizawa and Yamada (1995); Rabinovitch et al. (2002a) in-
vestigated EMR induced by explosions.
During the seventies and eighties, the interest in EMR emanating from failure of rocks
was renewed in connection with the problem of EQ prediction. Numerous investigations
(King 1983; Sadovskii et al. 1979; Gokhberg et al. 1979, 1982, 1985; Morgunov 1985;
Gokhberg et al. 1986; Gershenzon et al. 1987; Yoshino et al. 1993; Yoshino and Tomizawa
1989), a significant part of which was carried out in Japan and in the former USSR, showed
that the magnitude of EMR sharply increased several hours or even days before EQ and
quickly decreased at the moment of EQ or directly before it. The magnitude of the detected
EMR anomalies was of up to 10–15 dB over the usual noise level. Thus, EMR is associated
with fractures on multi-scale lengths and is connected with various applications.
Several atomic-scale models have hitherto been suggested to explain the EMR phenom-
enon. All of these models were qualitative, and efforts to use EMR quantitatively were un-
successful, due to the lack of a detailed understanding of the EMR phenomenon (King
1983; Rabinovitch et al. 1998). Until quite recently, the physical mechanism of EMR was
unknown (Freund 2000, 2002). An essential requirement for this type of knowledge is a
careful laboratory investigation of rock failure. Such an investigation has been carried out
by the group of Rabinovitch et al. (1995, 1996, 1998, 1999b, 2000a,b, 2002a,b, 2003a,b).

5.2
Foundations of EMR Model

A general shape of an EMR pulse induced during fracture experiments is shown in Fig. 5.2.
It can be characterized by the following general relationship (Rabinovitch et al. 1998):

(5.1)
382 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

where t is the time, t0 is the time from the origin up to the pulse beginning and T is
the time from the origin up to the EMR pulse envelope maximum. Thus, T' = T – t0 is
the time interval to reach pulse maximum; τ is the rise time and the fall time (RFT),
which turn out to be the same; A0 is the pulse amplitude and ω is the frequency. The
signal’s general shape including its envelope, frequency and duration can be obtained
by our basic (see below) theory, while the parameters themselves can be derived by
least squares fitting of the experimental results (Rabinovitch et al. 1998, see Fig. 5.16).
Four basic elements of a physical model, which would result in such simple rela-
tionships, are (Frid et al. 2003):

1. Time of growth (Rabinovitch et al. 1998): The pulse amplitude, A, increases as long
as the crack continues to grow, since new atomic bonds are severed and their con-
tribution is added to the EMR. When the crack halts, as it does if created under a
compressive load, the pulse amplitude starts to decay. The time from the start of
the pulse up to its maximum, T', is proportional to the number of severed atomic
bonds and thus to the crack length l (the crack velocity vcr is almost constant).

(5.2)

2. Dynamics of growth and decay of the amplitude: In the time increment, dt, ‘A’ de-
creases by dissipation (to phonons, see Sect. 5.3.7) to

but it is “replenished” by a term proportional to the number of severed bonds, αdx,

where dx is the length increment of the crack in the interval dt (dx = vcrdt) and α is a

Fig. 5.2.
An experimental pulse shape
(output voltage against time)
and a numerical fit according
to Eq. 5.1 (dashed line) (Rabi-
novitch et al. 1998)
 5.2 · Foundations of EMR Model 383

coefficient relating the antenna output to dx. α is assumed proportional to the width
of the crack, b, (the following calculations assume a constant crack width), to the num-
ber of bonds per unit area, to the emittence of a bond, to the solid angle of the crack
seen by the antenna and to the latter’s conversion factor. It follows that:

(5.3)

Hence,

(5.4)

For t > T the crack stops and the amplitude thereafter diminishes resulting in

(5.5)

3. Regarding the oscillatory parts of Eq. 5.1, we assume that these EMR waves are
created by charge oscillations at both sides of the propagating crack (Fig. 5.3).
Consider the line of bonds located at the front (tip) of the propagating crack
(Fig. 5.3a). These bonds break when the front moves to the next line. Following this

Fig. 5.3. a A schematic picture of crack propagation; b, c schematic “optical surface wave” at crack
surface (a similar wave propagates on the other surface) at a specific time. Crack surface is in the
x-z-plane and the crack moves in the x-direction. Note that charge separation can either be
longitudinal (b) or transverse (c) with respect to the surface, with appropriate EMR polarizations.
Note also the exponential decay of the wave into the material. Charge separation is oscillatory so that
at a later time the dipole directions are reversed (Frid et al. 2003)
384 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

break, the atoms on both sides of the bonds are moved to “non equilibrium” posi-
tions in relation to their (new) steady state ones and will perform oscillations around
them. If each atom (ion) vibrated individually, the situation would resemble the
Einstein model of lattice vibrations and the frequency of these oscillations would
approach 1015 Hz. Since, however, the lines of vibrating atoms move together and
are also connected to atoms around them (in the forward direction and also to
atoms on their side of the two surfaces newly created by the fracture), the ensuing
vibrations are similar to those obtained for the bulk by the Debye model, having
much lower frequencies. It is the oscillations stimulated by this process at the new
surfaces, which give rise to the EMR, as follows. The positive charges on these sur-
faces move together in one direction away from the equilibrium plane (one crack
side), while the negative charges move in unison in the other direction from the
equilibrium plane (the same crack side), and vice versa, retaining an overall charge
neutrality throughout. These surface oscillations, similar to “optical” Rayleigh waves,
decay exponentially into the bulk (Fig. 5.3b,c) – “surface vibrational optical waves
(SVOW)” (e.g., Srivastava 1990), much like “bulk optical (phonon) waves” (e.g., Kittel
1987) observed in vibrating crystals.
4. It has been shown elsewhere (e.g., Srivastava 1990) that surface waves decay in time
as a result of an interaction with bulk phonons. We therefore (Rabinovitch et al.
2003a) consider τ to be the relaxation time of such a (Rayleigh-like) surface wave,
which interacts with a bulk phonon, leading to the creation of another bulk phonon
(a three-phonon process), and use Eq. 13 of King and Sheard 1970) to characterize
the process. The rate of occurrence of the process per unit time is given by the
golden rule formula (Schiff 1986):

where the initial and the final states are

H3 is the time dependent anharmonic part of the crystal Hamiltonian, nR, nb1 and nb2
are the numbers of surface phonons, initial and final bulk phonons, respectively. Ei and
Ef are the initial and the final energies of the three-phonon system, so that

where ω denote the related frequencies. The relaxation time τ, which we identify
with the rise and fall time of the EMR pulse, is obtained from the golden rule for-
mula (1 / τ being proportional to the transition probability) using the explicit ex-
pression for H3 (Eq. 6.47 in Srivastava 1990) and the displacement field due to the
surface modes written in second quantized notation. Integrating over the states of
the initial and final bulk phonons b1 and b2, one obtains (Eq. 8.36, Srivastava 1990)

(5.6)
 5.3 · EMR Model Verification 385

where T is the temperature, ρ is the material density, and vR is the Rayleigh wave
velocity. The proportionality coefficient contains data of bulk phonons and of the
crystallographic orientation and is considered to be constant for the same mate-
rial. Note that T is the local temperature at the crack tip. It is much higher than
room temperature (Fuller et al. 1975).
In a developing crack, a surface waves propagate along the crack surfaces. Their
emitted EMR frequency is the same as that of the oscillating ions of the crack sides
(Rabinovitch et al. 1998), ω . Therefore we equate ω = ω R.

Several important results can be obtained even from these basic ideas.

1. Equations 5.4 and 5.5 together with oscillations part sin ω(t – t0) immediately lead
to the correct empirical shape (5.1).
2. If the half wavelength, λ / 2, of the surface waves creating the EMR is limited by the
crack width b (since at both sides of the crack, atomic movements are restricted),
then (Frid et al. 2000; Rabinovitch et al. 2000c)

(5.7)

3. From Eq. 5.2 and Eq. 5.7,

(5.8)

where S = l b is the crack area. Since T' and ω are measurable quantities (from the
pulse shape, using Eq. 5.1), the ratios of the crack surface areas, if not their exact
values, can be calculated using Eq. 5.8.
4. Since EMR pulse amplitude A is proportional to α (Eq. 5.4), which was assumed to
be proportional to crack width b, it is expected that EMR amplitude should be in-
versely proportional to the EMR signal frequency ω = 2πƒ.
5. The rise and fall time, τ, is expected to be inversely proportional to the pulse fre-
quency ω (Eq. 5.6).

5.3
EMR Model Verification

5.3.1
Experimental Arrangement

In order to investigate EMR properties through the range of the whole stress strain curve,
a triaxial load frame (TerraTek stiff press model FX-S-33090; axial pressure up to
450 MPa; confining pressure up to 70 MPa; stiffness 5 × 109 N m–1) was used (Fig. 5.4).
It is combined with a closed-loop servocontrol (linearity 0.05%), which is used to
maintain a constant axial piston rate of displacement. The load was measured with a
sensitive load cell (LC-222M, maximum capacity 220 kN, linearity 0.5% full scale). The
confining pressure was continually controlled by a clock-type sensor to preserve its
386 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.4.
Compression press: a the triaxi-
al load frame, b the sample stack,
c schematic diagram of the EMR
experimental arrangement

preset value through volumetric changes of the sample during the loading process.
The cantilever set (consisting of axial and lateral detectors; strain range about 10%;
linearity 1% full scale) enables us to measure sample strains in three orthogonal di-
rections parallel to the three principal stresses.
A magnetic one-loop antenna (EHFP-30 Near Field Probe set, Electro-Metrics Penril
Corporation) 3 cm in diameter was used for the detection of the EMR. It is wound
within a balanced Faraday shield, so that its response to external electric fields is neg-
ligibly small. A low-noise micro-signal amplifier (Mitek Corporation Ltd, frequency
range 10 kHz to 500 MHz, gain 60 ±0.5 dB, noise level 1.4 ±0.1 dB across the entire
frequency band) and a Tecktronix TDS 420 digital storage oscilloscope connected by
way of a General Purpose Interface bus to an IBM PC completes the detection equip-
ment.
 5.3 · EMR Model Verification 387

5.3.2
Materials

For our uniaxial and triaxial compression investigation we used the following materials:

1. Granite. A large Eilat granite “block” from the Nahal Shelomo area of Sourthen Is-
rael, nearly 3 km from Eilat was used. This granite is gray and consists of K-feld-
spars (about 40%), quartz (about 35%) albite-oligoclase (about 20%), and biotite
(about 3–4%). It is medium to coarse-grained (feldspars and quartz, 2–4 mm; bi-
otite 0.5–1 mm) and non-porphyritic (Bogosh et al. 1997).
2. Chalk from the Horsha Formation (Middle Eocene), sampled in the Beer Sheva
syncline (Ramat Hovav, about 9 km from Beer Sheva, Israel, Bahat 1991a). Chalk is
a pure-white, soft and friable rock consisting of calcitic (nannoplankton) micro-
fossils. The color, purity, texture, cementation and hardness of the rock may vary
due to additives such as water, oxides, marl, clay and flint, and by diagenetic and
tectonic processes (Mortimore 1990). The strength of chalk under compression
depends on the above properties, particularly on the dryness of the sample, and
varies from values of ~1 MPa when wet, to some 50 MPa when extremely dry. A
strict drying process was therefore applied to the samples, which involved a cycle
of heating to 110 °C for 24 hours, and then an immediate transfer to a desiccator to
avoid water absorption by the samples. All samples were cut from one chalk layer,
with unified co-orientation within the rock, into standard cylinders of 100 mm in
length and 53 mm in diameter. The density of all investigated samples was
2.16 ±0.01 × 103 kg m–3. The ends of the samples were scrupulously polished to mini-
mize heterogeneity of the stress field under compression (Jaeger and Cook 1979).
The endcups had a diameter identical to the samples. To prevent the hydrostatic
loading oil from penetrating into the sample pores, all samples were enclosed in
plastic jackets (ALPHA FIT-221-3), and the contacts of their ends with the endcups
were carefully closed. Each chalk sample was tested axially with a strain rate of
1 × 10–5 s–1 and laterally by a different hydrostatic oil pressure. The axial pressure
varied from 37 MPa up to 65 MPa and the confining pressure from 0 to 15 MPa.
3. β-quartz solid solution glass ceramic. The composition of this material (VISION by
Corning) in weight percent is SiO2 = 68.8, Al2O3 = 19.2, Li2O = 2.7, MgO = 1.8,
ZnO = 1.0, TiO2 = 2.7, ZnO2 = 1.8, plus additional traces (Beall 1989). The glass-ce-
ramic is close to monophase crystals of β -quartz solid solution about 600 Å (60 nm)
in diameter and a small amount of zirconium titanate crystals (<100 Å in diameter)
which allow efficient light transmission (Beall 1989). The resistance to failure under
uniaxial compression of samples, which comply with the standard specification di-
mensions (106 mm length and 52 mm in diameter), is very high (greater than
450 MPa) and was beyond the load capacities of our press machine. Since our pur-
pose was to overcome the resistance to fracture and to induce controlled cracking
(rather than to obtain accurate strength results), we selected a sample geometry that
deviates from standard specifications. The best results were obtained for a sample
with the following specifications: Length 104 mm and having a frustum shape (the
part of a cone left after cutting off a top portion by a plane parallel to the base) and
an elliptical cross section, such that the large and small diameters at the base end
388 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

are 33 mm and 21 mm, respectively, and the two diameters at the top end of the
frustum are respectively 33 mm and 10 mm. The abraded flexural strength of the
β-quartz solid solution glass-ceramic is 69.3 MPa (Beall et al. 1967). The axial pres-
sures varied from 0 up to 112 MPa (there was no confining pressure in this case).
4. Glass. Cylindrically shaped soda-lime glass samples 10 cm in length and 3 cm in
diameter, which were uniaxially loaded. Soda-lime glass is homogeneous and trans-
parent. Its structure consists of glassy matrix consisting of silicon, oxygen and cal-
cium atoms, which are strongly bound, with weakly bound sodium ions. The den-
sity of all investigated glass samples was 2.6 ±0.01 × 103 kg m–3.

5.3.3
Amplitude Investigation

The amplitude of the envelope of the EMR pulse (Eq. 5.4) is determined by the fol-
lowing relation:

(5.9)

As we noted above (Eq. 5.7), the frequency of the EMR pulses is inversely propor-
tional to the crack width. EMR pulse amplitude increases as long as new atomic bonds
are severed (Rabinovitch et al.
_ 1998, 1999b); hence, this amplitude is related to the crack
area, and it is expected that A ∝ 1 / ω (Rabinovitch et al. 1998, 1999b).
The voltages of the EMR pulses depend on the antenna reaction (antenna
_ efficiency),
which changes with frequency. Compensating for this factor, E = h(A) (E being the field
amplitude reaching the antenna), by the appropriate antenna efficiency chart (h-EHFP-

Fig. 5.5.
Amplitude-frequency relation
of EMR pulses associated with
chalk, granite and glass-ce-
ramic compression: Chalk (filled
circles); granite (diamonds);
glass-ceramic (plus signs); chalk
trend line (unbroken line); gran-
ite trend line (dashed line); glass-
ceramic trend line (dashed
dotted line) (Rabinovitch et al.
1999b)
 5.3 · EMR Model Verification 389

30 Near Field Probe set, Electro-Metrics Penril corporation), we were able to compare
the heights of the various EMR signals that have different frequencies.
Figure 5.5 (Rabinovitch et al. 1999b) shows the compensated amplitudes (E) of the
electromagnetic field signals induced by the fracture of chalk, granite and glass ceramic
samples. Analysis of about 160 pulses shows that the amplitude-frequency ratio of each
of the three materials can be fitted by a power-law type relation. Thus, for chalk, the
relation is E ≅ 6 × 108ƒ–0.91 ±0.04 (squared regression coefficient R2 = 0.87), for granite
E ≅ 3 × 10 9ƒ –0.99 ±0.04 (R 2 = 0.89), and for the glass ceramic E ≅ 2 × 10 11ƒ –1.32 ±0.11
(R2 = 0.82).
The exponent of these three relationships is very close to –1, although for the glass
ceramic the value is somewhat scattered. EMR pulses induced by chalk and granite frac-
ture could generally be fitted by a single frequency. In contrast, some EMR pulses asso-
ciated with glass ceramic fracture were “multi-frequent”. Therefore the amplitude-fre-
quency ratios of these pulses were corrected in the following way. The contribution of
a specific frequency peak of the “fast Fourier transformation” (FFT) to the compen-
sated amplitude was calculated by

where Ai is the area under the ith peak and E is the measured pulse amplitude. Such a
procedure evidently adds errors; it increases the spread of glass ceramic results and
could also be the reason for the deviant exponent. If we collect all EMR data on one
figure (Fig. 5.6, Rabinovitch et al. 1999b), a single combined relation can be obtained:
E ≅ 5 × 109ƒ–1.06 ±0.04 (R2 = 0.84).
These results imply that the amplitude of the EMR field is indeed inversely propor-
tional to the signal frequency and is hence proportional to the crack width (Rabino-

Fig. 5.6.
General amplitude-frequency
relation of chalk, granite and
glass-ceramic samples (Rabi-
novitch et al. 1999b)
390 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

vitch et al. 1998). The final measured amplitude obviously depends also on other pa-
rameters such as fracture length, geometry of antenna-fracture orientation, attenua-
tion along the path, etc. These factors constitute the source of the (rather large) verti-
cal spread in the results of Fig. 5.5 and 5.6.
A controversy exists in the literature as to the magnitude of EMR from tensile and
shear fracture. While Yamada et al. (1989) maintain that tensile cracking excites more
intensive EMR than shear cracking, Petrenko (1993) states that the EMR amplitude is
rather much greater for the shear fracture. Here we consider the influence of tensile
vs. shear fracture modes on EMR. Since in our model (Sect. 5.2) no distinction is drawn
between shear and tensile fracture modes, it is expected that magnitudes of EMR from
both would be comparable.
We selected chalk for our study due to its micro-texture and low strength leading
to a relatively small number of fragments (between 1 to 18) at failure, which can be
analyzed fractographically (Bahat 1991a; Bahat et al. 2001a).
As expected from our previous studies (Bahat 1988c), the fractured surfaces exhibited
a useful fractography. Tensile fractures (Fig. 5.7a) were planar and intensely marked by
“striae” (plumes) (Bahat 1979a). These surfaces reflected a light grayish color. The shear
fractures, on the other hand, had rough morphologies that were marked by white coarse
“sugary” textures (Fig. 5.7b). The discrimination between these very distinct surfaces
provided us with a fractographic criterion to distinguish between the two modes of

Fig. 5.7.
A photograph of a fractured
surface on which the dashed
line marks the boundary be-
tween the tensile part (a, show-
ing planarity, gray tint and
strong striae) and the shear
part (b, showing coarse topog-
raphy, white color and sugary
texture). The tensile part
(‘a’ arrow) is still covered by a
piece of the shear part at the
center of the sample. Scale in
cm (Frid et al. 2000)
 5.3 · EMR Model Verification 391

fracture and helped us to develop a method for the determination of tensile vs. shear
fracture zones.
The area of each fracture was manually measured as the sum of all crack surfaces
obtained for each sample for each type (Fig. 5.7, Frid et al. 2000). Measurement errors
varied from 3 to 8%. Twenty-three chalk samples were investigated under different
loads. The first eleven samples were loaded uniaxially, while the twelve following ones
were loaded triaxially. During the triaxial test, the lateral load was changed from 1
to 15 MPa, which caused a monotonous rise of sample strength from 37.7 MPa up to
80.1 MPa. Young moduli for all tested samples ranged from 6.2 up to 11.3 GPa, while
the Poisson ratio changed from 0.07 up to 0.37. Since brittle-ductile transformation
after the Coloumb-Mohr region diminishes EMR activity (Frid et al. 1999), this region
was not treated. Tensile fracture was the preferable mode of failure during uniaxial
compression, while the shear mode became significant only during triaxial loading:
The “(shear area) / (tension area) ratio” changed from 0.5–0.8 during uniaxial com-
pression to 2–35 during triaxial loading (Bahat et al. 2001a).
We compared the total amplitude (sum of compensated amplitudes) of the EMR
pulses from each sample (measured above the sample’s elastic limit) with the total
respective areas (sums of crack areas of a given sample) of tensile and of shear cracks
(Rabinovitch et al. 1999b). Figure 5.8 (Frid et al. 2000) shows a 3D graph of total
compensated EMR pulse amplitudes vs. total areas of tensile and shear cracks of
all investigated samples. The total EMR amplitude E was fitted by the linear equation:
E = –33.39 + 0.00655 St + 0.00596 Ss (with a squared regression coefficient R2 = 0.86),
where St and Ss are the total areas of tensile and of shear cracks of a given sample, re-
spectively. The difference between the two coefficients multiplying St and Ss is of the
order of ±5%, which is rather small and might be attributed to the error in area mea-
surements (3–8%). This result shows that it is only the size of the entire area of the
crack, irrespective of its mode, which governs the amplitude of the EMR.

Fig. 5.8.
A 3D experimental graph of
total compensated EMR pulse
amplitudes (V m–1) vs. total
areas (mm2) of tensile and
shear crack areas of all investi-
gated samples and its fit (Frid
et al. 2000)
392 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

5.3.4
Investigation of the Time up to Envelope Maximum and its Relation with Frequency

To check the validity of Eq. 5.2, two experimental tests were carried out:

1. Mutual measurement of T' and crack lengths (Bahat et al. 2002a) in transparent
glass ceramics under compression. A glass ceramic sample was gradually loaded
uniaxially. Loading was stopped as soon as EMR pulses were recorded. The sample
was removed for a check and then the procedure started afresh. The procedure was
repeated five times (five stages-Fig. 5.9). Characteristic pulses of different modes
observed during the entire process are presented in Fig. 5.10 and Table 5.1 (Bahat
et al. 2002a).
Figure 5.11 shows an example of an EMR pulse observed during the collapse of
a glass ceramics sample with its least square fit to Eq. 5.1. As can be seen, both the
accuracy of measurement and that of the parametrization procedure are adequate
(R2 = 0.96).
The results of parametrization of this EMR pulse are marked by arrows in
Fig. 5.12, where we summarize our results that pertain to the correlations between T'
and crack length. Note that T ' values varied between 0.8 and 1.5 µs under low
stresses (0.36–1.7 MPa) in association with microcracks of several millimeters in
length. T' values of 5–15 µs were associated with stresses of up to 65 MPa and were
correlated with cracks of up to several centimeters in length. A value of T' larger
than 20 µs occurred under greater stresses (112 MPa) and was correlated with the
longitudinal splitting at failure (~10 cm). The correlation between T' and crack length
shows a positive trend (Fig. 5.12). A straight line can be drawn through the points rep-
resenting the stages from 3 to 5 (Fig. 5.9). The dashed line indicates our assertion,

Fig. 5.9.
Stress vs. strain for the fifth
stage of the incremental stress
increase. Dashed arrows indi-
cate stress levels achieved dur-
ing stages 1–4 (Bahat et al.
2002a)
 5.3 · EMR Model Verification 393

T' = 8.15 × 10–7 L (vcr ≈ 1 200 m s–1), where L is the crack length in mm and T ' is in
seconds. The R2 coefficient obtained was 0.73. Considering the small number (five)
of points, this coefficient is considered reasonable.

Fig. 5.10 Characteristic pulses of the different modes (Table 5.1). a A short pulse (stage 1, pulse 1).
b A medium, individual pulse (stage 4, pulse 7). c A medium, uneven pulse (stage 4, pulse 10). d A
lengthy pulse (stage 5, pulse 1). e A medium, individual pulse (stage 5, pulse 2) (Bahat et al. 2002a)

Fig. 5.11.
An example of an EMR pulse
of the fifth stage with its least
square fit to Eq. 5.1 (dashed line).
Results of fit: T ' = 42 ±0.3 µs,
f = 18.7 ±0.012 kHz, R2 = 0.96
(Bahat et al. 2002a)
394 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Table 5.1. EMR pulse parameters vs. stress and strain in fractured glass-ceramic

Fig. 5.12.
Correlation between T ' and
crack length. Open squares show
EMR-fracture data during the
first two stages; closed squares:
Stages 3–5; closed triangles show
spread of EMR data during the
forth stage. Error bars here were
estimated by the spread of the
EMR data of the forth stage of
loading to be ~3.5% for lnT '
(Bahat et al. 2002a)

Note that the dimensions of the flaky cracks (stages 1 and 2, Fig. 5.9) calculated
by EMR parameters (Fig. 5.12) are much less than the measured ones (Table 5.1)
and hence deviated from the regression lines in Fig. 5.12. We assume that the ini-
tially created cracks during compression were in fact of small dimensions (as mea-
sured by the EMR). Eventually these fractures caused flaking off the external sur-
face of the sample of much larger dimensions. The reason that the initial cracks
were optically undetected stems from the fact that their apertures, were apparently
below the wavelength of light.
 5.3 · EMR Model Verification 395

Fig. 5.13.
Histogram of log T ' (where T '
is the time interval to reach
the pulse maximum) and its
Gaussian fit Rabinovitch et al.
1998)

2. T' values distribution (Rabinovitch et al. 1998). If we assume that the fracture pro-
cess develops incrementally and that each new increment is proportional to the
existing crack length (as previously observed by Gillespie et al. (1992) and by Cowie
and Scholtz (1992a,b), then a log-normal distribution should be expected (Aitchison
and Brown 1976). Results show that T ' indeed obeys a log-normal distribution
(Rabinovitch et al. 1998) (Fig. 5.13).
Another problem is the possibility of a scaling relation between crack length
and width. Walmann et al. (1996) showed that the aperture, u, and the length l of
the crack were related by a fractal relation, u ∼ lβ, where β ranges between 0.5 and 1.
Although the relation was observed for u and not for crack width, we tried to corre-
late between T' and ω : ω ∼ (T')–ν, which would have implied a scaling between crack
length and width. Results, however, show that the obtained ν is of the order of zero, i.e.,
crack length is independent of its width, implying that the width is constrained by an
auxiliary mechanism, such as grain boundary, intergranular spacing, etc. A similar con-
straint was already noted (Bahat 1988c) in geological fractures in chalks, where in a
given layer of a constant width, there is a wide range of fracture lengths.

5.3.5
Frequency Investigation

Equation 5.7 reveals that for the same crack width, the EMR frequency for stiffer ma-
terials would be higher than that for the weaker materials. Indeed, since the Rayleigh
wave velocity in a material with a Young modulus E, Poisson ratio µ and density ρ, is
given by
396 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

the ratio of EMR frequencies for two materials is given by

(5.7')

To check the validity of Eq. 5.7, we compared EMR frequencies and fracture widths
for four different materials, the elastic properties of which are shown in Table 5.2 (Frid
et al. 2003).
Only during glass uniaxial compression experiments was an exact one-to-one cor-
respondence between the EMR signals and the longitudinal splits causing them ob-
tained. Even during the transparent glass-ceramic experiments not all pairs of EMR
signal-fracture (S-F) were perfectly related. In our analysis, we included only those
pairs of S-F that could be accurately “linked”. During chalk and granite (that are non-
transparent materials) experiments, a one-to-one link between EMR signals and frac-
tures was not possible. Moreover, fracture widths and lengths were uncertain since the
fracture directions were usually undefined. Hence we included in our analysis only
“maximal” fractures, which are, of course, easily discerned, and of those only the ones
that had a “stretched shape”, where one dimension (considered as length) was signifi-
cantly larger than the other (considered to be the fracture width). Figure 5.14 (Frid
et al. 2003) shows the relation of πvR / ω (where vR is calculated by the relation leading
to Eq. 5.7') vs. crack widths, b. The slope of this relation is 0.93 (R2 = 0.82), being very
close to 1, in agreement with the theoretical prediction (Eq. 5.7 and 5.7').

Table 5.2. Experimental results of material properties

Fig. 5.14.
The relation between reciprocal
EMR frequency and fracture
width for chalk, granite, glass
and glass ceramics under com-
pression (• chalk, ◊ granite,
+ glass ceramic, × glass) (Frid
et al. 2003)
 5.3 · EMR Model Verification 397

5.3.6
T' / ω Investigation (Eq. 5.8)

The T ' / ω parameter is directly calculable from the parametrized data, and hence is
independent of different auxiliary factors, which can be sources of large fluctuations
(errors) if we would like to relate fracture areas to A0 (Eqs. 5.1 and 5.9).

5.3.6.1
Experimental Procedure and Results

Twenty-four chalk samples were investigated in the conventional compression test

(Rabinovitch et al. 2000c). Nine samples were loaded uniaxially and fifteen samples were
loaded triaxially. Twenty-one of the twenty-four samples showed a similar EMR behavior,
while three (not treated here) behaved differently. Table 5.3 presents rock-mechanics prop-
erties of all investigated samples. Figure 5.15 shows a representative normalized stress-
strain curve (For the sake of comparison, stress values were normalized by the peak stress
value, while the axial strain values were normalized by the axial strain at peak stress)
of a chalk sample together with its T' / ω data. The latter was calculated from the EMR
data by a least square method as shown in the caption of Fig. 5.16. T' / ω data are shown
in Fig. 5.15a in a semi-logarithmic scale and in Fig. 5.15b in a normal scale. As is seen
from Fig. 5.15, an increase of stress (strain) is generally followed by an increase of the
T' / ω factor, indicating an increase of the area of newly formed cracks. The range of

Table 5.3. Rock-mechanics parameters of investigated chalk samples

398 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.15.
A representative normalized
stress-strain curve of a chalk
sample together with its T ' / ω
data (T ' / ω data are shown in a
semi-logarithmic scale (a), and
in a normal scale (b)) (Rabino-
vitch et al. 2000c)

Fig. 5.16.
An experimental EMR pulse
and its numerical fit. For this
pulse, t0 is 9.63 µs, T is 10.07 µs.
Hence, T ' = T – t0 is 0.44 µs. The
frequency ω is 6.76 × 107 s–1.
Hence, T ' / ω is 6.51 × 10–15
(Rabinovitch et al. 2000c)

the value of T' / ω stretches from 10–15 to 10–10 (s2) (5 orders of magnitude!). The minimum
values of T' / ω were observed at the beginning of the stress-strain curve. Later, the elastic
and elastic-plastic zones before the peak of the stress-strain curve, T' / ω values of the or-
der of 10–14–10–11 (s2) were measured, and the T' / ω value jumped to a maximum (of the
order of 10–10) in the zone immediately preceding the peak stress. This jump is indicative
of the collapse failure. After the peak stress, T' / ω amplitudes (and the corresponding crack
areas) decrease back to values of 10–14–10–11 (s2) (“relaxation” region). A similar behavior
of T' / ω vs. stress-strain (namely, a gradual increase of T' / ω with stress increase followed
by an abrupt jump) was observed during uniaxial compression of glass ceramics (Bahat
et al. 2002a), suggesting that such behavior could be of general validity (see below). These
 5.3 · EMR Model Verification 399

Fig. 5.17.
Fractography: Various fracture
types and sizes are shown on
specimen No. 1 (Table 5.3). Two
large fractures (~4 × 4.5 cm
and ~5 × 1.5 cm) are separated
by a ridge at the lower part of
the picture, whereas the upper
part of the picture reveals
length variations from 0.1 mm
(magnified), through 2–4 mm
up to 1–4 cm (cm scale at top
of picture) (Rabinovitch et al.
2000c, Bahat et al. 2001a)

Fig. 5.18.
Experimental largest T ' / ω
values vs. maximal fracture
areas and a linear fit (squared
regression coefficient R2 = 0.9)
(Bahat et al. 2001a)

results imply that following the early pore-closure zone of the stress-strain curve,
microcracking and possibly coalescence occurred, while collapse happened just before peak
stress and was followed by smaller relaxation cracking in the post-peak zone.
Fractographic examination of the failed chalk samples revealed fractures ranging
in size from 0.1 × 0.1 mm up to about 40 × 50 mm. The failed surface (Fig. 5.17) con-
sists of composites of cracks that vary in size from several mm to several cm. This range
of crack areas (10–2 mm2 to ~2 000 mm2) is in accordance with the variation by 5 or-
ders of magnitude of the T' / ω values mentioned above.
Since a one-to-one correspondence between the EMR pulses and the causative frac-
tures is not possible in the chalk measurement, only the cracks possessing the largest
400 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

areas were measured and correlated with the largest T' / ω values (Fig. 5.18, and see
also Bahat et al. 2001a). As can be seen, very good agreement exists between Eq. 5.8
and the experimental results both in the range of values and in the correlation of the
largest values (R2 = 0.9). In a recent experiment on transparent glass ceramics (Bahat
et al. 2002a) and glass (Frid et al. 2003), even a one-to-one correlation was shown to
exist between T' / ω and crack area.
The order of magnitude of the velocities in Eq. 5.8 can also be estimated from
these measurements. Thus v ≈ [S / (πT ' / ω ]1/2 and, e.g., for the 0.1 × 0.1 mm case, we
have T ' / ω = 10–15 s2 whereupon v ≈ 1.4 × 103 ms–1. This is of the correct order of
magnitude (since the average Rayleigh velocity calculated for chalk (Bahat et al.
2002a), at the known values of its dry density, Young modulus, and Poisson ratio is
1.2 × 103 ms–1).

5.3.7
Decay Time Investigation (Rabinovitch et al. 2003a)

5.3.7.1
Experimental Results and Analysis

We have analyzed EMR pulses of chalk, granite, glass and glass ceramics by fitting them
to Eq. 5.1 and deriving their parameters (in particular τ and ω ). Figure 5.19a shows the
dependence of τ on ω for glass, glass ceramics and granite. The slopes on a logarith-
mic scale are within –1 ±0.1, with R2 = 0.96, 0.85 and 0.91, respectively, agreeing with
the theoretical prediction (Eq. 5.6). For chalk, however, an “effective” slope of –0.7 is
obtained (Fig. 5.19b). It cannot be attributed to the low accuracy of the fitting param-
eters: The error for τ ranges between 2% and (seldom) 30%, which is quite small in a
logarithmic scale, while the error for ω does not exceed a few percent, and is usually
only a fraction of a percent. We would like to attribute this “change of slope” to the spread
in temperatures of the small cracks there (see below).
The distribution of (τ ω )–1/4, which according to Eq. 5.6 should be proportional to
the absolute temperature for small and large crack widths in granite is shown in
Fig. 5.20a and b, respectively. We denote (τ ω )–1/4 values henceforth as “temperatures”.
The mean values of the crack “temperatures” in relative units are respectively 0.67
and 0.65 for small and large cracks (in granite), which are considered to be the same
within the accuracy for “temperature” values (10%). Similar distributions for chalk are
shown in Fig. 5.21a and b. Here, however, the mean values are different, being 0.47 for
small crack widths and 0.67 for large ones. As mentioned, the error for ω is not more
than a few percent, mostly a fraction of a percent, while for τ it ranges between 2%
and 30%. Thus the error of (τ ω )–1/4 may be estimated as ~10%, yielding “tempera-
tures” of 0.47 ±0.05 for small cracks and 0.67 ±0.07 for large cracks.
Figure 5.22 shows the distribution (histogram) of log (1 / ω ) (where 1 / ω should
be proportional to crack width, Eq. 5.7, for both small and large cracks). We de-
note by small (large) cracks those cracks that have ω values larger (smaller) than
2 × 107 s–1 (e.g., Fig. 5.19). If one assumes the Rayleigh velocity in chalk to be (see above)
1 200 m s–1, then the calculated small crack widths in chalk (by Eq. 5.7) range be-
tween 52.5 and 130 µm. For large cracks, the corresponding range is from 130 µm
to 1.8 cm.
 5.3 · EMR Model Verification 401

Fig. 5.19.
Rise and fall time τ as a func-
tion of frequency ω : The points
are the experimental results of
individual pulses, and the line
is the power law fitting result:
a for glass (triangles), fitting
τ = 1.87ω –0.95 (short-dashed
line); for glass ceramics
(squares), fitting τ = 28.1ω –1.1
(solid line); for granite (×), fit-
ting τ = 2.23ω –0.93 (long-dashed
line); b for chalk (circles), fit-
ting τ = 1.87ω –0.7 (Rabinovitch
et al. 2003a)

Fig. 5.20.
Frequency counts of “tempera-
tures” (ωτ)–1/4 for cracks in
granite: a short cracks; b long
cracks (Rabinovitch et al. 2003a)
402 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.21.
Frequency counts of “tempera-
tures” (ωτ )–1/4 for cracks in
chalk (columns) and calculated
(solid line) (Rabinovitch et al.
2003a): a for short cracks;
b for long cracks; crack aper-
ture is assumed to behave as
ε ∼ kxα, where x is the crack
half-width (Rabinovitch et al.
2003a)

Fig. 5.22.
Frequency counts for log (ω –1)
(ω –1 is proportional to crack
widths, b = πvR / ω) in chalk fit
(Rabinovitch et al. 2003a)

5.3.7.2
Discussion

It has been shown that the temperature of dynamically propagating cracks rises (Fuller
et al. 1975). The actual temperature rise depends on crack velocity and on material
properties (Yatomi 1981). The spread of experimental points in Fig. 5.19 can be attrib-
uted to a spread in crack velocities that causes a spread in temperatures. Since we as-
sume that 1 / τ ~ ω T 4, the dependence on temperature is very strong. However, cracks
 5.3 · EMR Model Verification 403

in glass, in glass ceramics (which are not porous) and in granite (the porosity of which
is about 5%) lead to almost uniform distribution of crack “temperatures” around their
mean (Fig. 5.20, resulting in the almost exact ωτ = const. relation of Fig. 5.19a). In
chalk, however, ωτ ≠ const. The “temperature distribution” (histograms) for small and
large cracks are given in Fig. 5.21a and b. It can be seen that “temperatures” of small
cracks are significantly lower than those of large cracks.
We assume that this temperature difference in chalk is due to the interaction be-
tween the propagating cracks and the existing material pores. Chalk is the only mate-
rial of the four analyzed that has large porosity (of about 40%). The pore radii in chalk
range (Gueguen and Palciauskas 1994) between 0.05 and 100 µm. The results of our
measurements (see Rabinovitch et al. 2003a), together with their exponential fitting,
are given in Fig. 2.28b. Note that although the range of observed pore sizes was
between 2 µm and 210 µm, their 3D fitting, being exponential, includes all pores, also
those of lower sizes. The cracks in chalk encounter one or more pores during their
development. Thus crack heat can be partially spent on raising the temperature of the
air inside the pores; in addition, air from the pore can enter the crack and expand
adiabatically; this process could also lower the latter’s temperature. Although all cracks
in chalk are cooled down in this way, the temperatures of small-sized cracks are more
strongly influenced by these processes than those of large cracks, since the final
temperature depends on the ratio between the pore’s volume and the volume of the
crack. The greater this ratio, the stronger is the temperature decrease. To show the self-
consistency of our assumptions, we have theoretically derived the probability distri-
bution of final temperatures (Rabinovitch et al. 2003a). The experimental and the cal-
culated distributions of final “temperatures” for small and large cracks are shown in
Fig. 5.21a and b. The agreement is adequate.

5.3.8
Summary of the EMR Model

It is seen that all experiments undertaken to check the validity of the model turned
out to yield positive results. Thus,
_
1. The amplitude A was shown to be indeed proportional to 1 / ω .
2. EMR amplitudes from shear fractures are similar (for similar areas) to those from
tensile fractures.
3. T' is proportional to crack length l.
4. The distribution of T' values follows a lognormal curve.
5. No correlation between length and width of cracks is obtained.
6. ω is proportional to 1 / b and scales with Young modulus.
7. T' / ω is indeed proportional to crack area with the right proportionality constant.
8. τ is proportional 1 / ω and reasonable “temperature” distribution can be deduced
for porous and non-porous materials.

All of these results support the contention that the model of EMR, although laid
out only in a simplistic manner, is the right one. In the next section, we analyze the
hitherto existing models of the origin of EMR and show that they all fail to stand ex-
perimental or self consistent tests.
404 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

5.4
Comparison of the New EMR Model with the Previous Ones
5.4.1
Dislocations and Charged Electrons

Misra (1977) and Misra and Ghosh (1980) suggested that “during non-uniform distri-
bution of dislocations, which occur at the transition stages of elastic-plastic deforma-
tions under tension, namely, yield point, end of Luders’ strain, and crack propagation,
mobile dislocations arrange themselves into some configuration, that is mechanically
stable”. If, at the halting position of a dislocation, its energy is reduced, the disloca-
tion can become self-trapped. “Conduction electrons (CEs) associated with such a dis-
location would be stopped and trapped relative to the positive ions”. This CE braking
process would be similar to a “Bremsstrahlung” mechanism and would result in the
emission of EMR. Misra (1977) further assumed that “during each transition stage (like
a yield point and/or fracturing) there must be a re-adjustment of the CE distribution
within the microscopic system, and the latter creates an oscillating Hertzian dipole.
Thus, the EMR must appear only at transition deformation stages”.
In a further analysis Molotskii (1980) pointed out that in the adjustment of the
CEs to the slowly moving dislocations, their acquired energy would be of the order of
10–11 eV per mean free path, so that the maximum frequency of emitted EMR would
be of the order of 103 Hz, i.e., significantly lower than the value predicted and mea-
sured by Misra. The acceleration of CEs by moving dislocations could not therefore
be the cause of the EMR from metals, and Molotskii suggested therefore that the EMR
was due to the increase of the total dislocation length and velocity, which occurred at
transient deformation stages. Since these dislocations acted as electric dipoles, this
mechanism would lead to an accelerated rise in the dipole moment of the material
with an accompanying emission of EMR.
Both explanations however relating EMR to dislocation phenomena seem question-
able. As is well known the motion of dislocations can be totally neglected in the crack-
ing of brittle materials, and this mechanism therefore cannot explain EMR from glass
and other brittle materials (e.g., most geological materials). A typical EMR pulse from
the failure of a glass cylinder under uniaxial compression is shown in Fig. 5.23. The
shape of EMR signals is found to be invariant for different brittle materials (glass, glass
ceramics, granite, rhyolite, limestone, and chalk; Rabinovitch et al. 1995, 1996, 1998,
1999b, 2000a,b, 2002a; Frid et al. 1999, 2000; Bahat et al. 2002a), and even to be inde-

Fig. 5.23.
A typical EMR pulse observed
during glass cylinder failure
under uniaxial compression
(Frid et al. 2003)
 5.4 · Comparison of the New EMR Model with the Previous Ones 405

pendent of the type of loading (compression, drilling and blasting (Goldbaum et al.
2001; Rabinovitch et al. 2002b) and mining (Vozoff and Frid 2001)).
The basic weakness of the “dislocation movement-hypothesis” was also pointed out
by Jagasivamani and Iyer (1988), who showed experimentally that the EMR amplitude
even increased with the brittleness of the investigated metals. Indeed, this latter re-
sult is in line with our measurements (Frid et al. 1999), which also showed that EMR
activity increased with the brittleness of materials and decreased in the case of a tran-
sition from a brittle to a ductile behavior.

5.4.2
Discharge

Finkel et al. (1975) demonstrated that the splitting of alkali-halide crystals creates a
mosaic of positive and negative charges, which appears on both sides of the fracture
as it is formed. Since such charge separation can create an electrostatic field of the
order of 107 V cm–1, an electric discharge may occur, which was suggested as the ori-
gin of EMR (Finkel et al. 1975). However, Miroshnichencko and Kuksenko (1980) and
Khatiashvili (1984) have already noted that the spectrum of discharge radiation is
known to be of a “white noise” type and to be independent of the mechanical proper-
ties of the materials; the observed EMR behaves in an entirely different manner (Miro-
shnichencko and Kuksenko 1980; Rabinovitch et al. 1998, 1999b; Frid et al. 2000). Our
results show that the EMR appears as individual pulses or as clusters of pulses caused
by the different (individual or group) fractures (Rabinovitch et al. 1999b, 2000c; Frid
et al. 2000). The properties of the pulse are influenced by the dimensions of the frac-
ture (Rabinovitch et al. 1998, 1999b, 2000c; Frid et al. 2000) and by the elastic proper-
ties of the materials (Khatiashvili 1984; Frid et al. 1999). Moreover, the high resolu-
tion of our experimental system (Rabinovitch et al. 1998) allows us to obtain the ex-
act shapes of the EMR pulses (Figs. 5.2, 5.10, 5.23, 5.24a). These definitely do not be-
have as “white noise” but rather exhibit a very distinct character (Eq. 5.1). These re-
sults confirm those observed by Miroshnichencko and Kuksenko (1980) and Khatia-
shvili (1984). Note in particular that the EMR spectrum (Fig. 5.24b) is highly local-
ized around a single frequency.

Fig. 5.24. An example of EMR signal a excited by fracture of granite in compression and its spectrum b
(Frid et al. 2003)
406 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

5.4.3
Movement of Fracture Tips

EMR is observed (Gershenzon et al. 1985) during the propagation of cracks introduced
by cleavage of LiF crystals with a knife. The developing crack (tip) propagates in the
direction of indentation. According to their assumption, negative electrical charge
moves with the crack tip while positive charge was assumed to accumulate at the re-
gion of the indenter-material contact. Dipole radiation should occur from these op-
posite charges, separated by the length of the crack, by the acceleration of the latter.
Based on these assumptions, the power of the emanating EMR was calculated
(Gershenzon et al. 1985), further assuming that the charging mechanism arises from
moving charged dislocations.
As already pointed out, however, charge transfer by dislocations cannot consti-
tute a general explanation of the origin of EMR. The “movement of fracture tips”
explanation is questionable even for a dielectric like LiF, since the velocity of dis-
locations in alkali halide crystals (Tetelman and McEvly 1967) is smaller than the
velocity of the crack, implying that charged dislocations cannot be the reason for crack
tip charging.
An even stronger adverse argument to this model is that there is no clear rea-
son for any “symmetry breaking”, i.e., there is no known mechanism that would
select a certain tip (side) to become negative while the other tip (side) becomes
positive and not vice versa. In particular for the case discussed by Gershenzon et al.
(1985), it is not clear why the crack tip should accumulate negative and not positive
charge.

5.4.4
Movements of Fracture Sides (the “Capacitor” Model)

In order to determine whether movement of the crack sides during cracking could
cause EMR with the appropriate characteristics, Miroshnichencko and Kuksenko
(1980) used the acoustic wave emitted by a fracturing material to drive the plates of a
specially built auxiliary-charged capacitor. The EMR from the capacitor was moni-
tored by an antenna. Since the measured signal “shapes” were comparable to those
obtained by the “directly” observed EMR, they concluded that the latter was caused by
similar movements of the charged sides of the crack.
A decade later, O’Keefe and Thiel (1995) suggested another version of a capacitor
model (for EMR from the cracking of compressed ice), in which a charged parallel plate
capacitor is created whose plates (crack sides) are being drawn apart. After an initial
charge is formed on the crack surfaces, further separation should result in a decrease
of capacitance and an increase of the voltage across the crack. A space/time analysis
of the emanating radiation can be carried out using the diffusion equation for the elec-
tric field in the crack. Although it proved possible to simulate the time decay of the
EMR, no oscillatory behavior was predicted. O’Keefe and Thiel (1995) also considered
the many routes by which the net charge could be created, i.e., pre-polarization of the
material or applied physical gradients due to piezoelectricity or pseudo-piezoelectric-
ity, temperature, deformation, impurity concentration gradient effects, etc. Petrenko
 5.4 · Comparison of the New EMR Model with the Previous Ones 407

(1993) claimed that the electrification of crack sides could be caused by the surround-
ing non-homogeneous elastic strains, and Ogawa et al. (1985) assumed that crack side
electrification was due to piezoelectrification and to contact (or separating) electrifi-
cation.
However there are problems with this scenario.

1. In this model the EMR is assumed to be caused by an accelerated dipole created by

charged crack sides. This assumption implies that the EMR can arise only from
tensile cracks and not from shear ones. In contrast, our measurements of chalk
under uniaxial and triaxial compression (Frid et al. 2000) showed that the EMR
amplitude is independent of crack mode, tensile or shear and is related only to the
entire area of the crack (see Sect. 5.3.3).
2. The capacitor model implies a correspondence between EMR and acoustic emis-
sion (AE) signal appearance. However, our own measurements and those of Yamada
et al. (1989) show that although there are failure events for which AE and EMR
signals are measured together (e.g., Rabinovitch et al. 1995), there do exist events
for which AE is detected while EMR is absent and events for which EMR is ob-
served while AE is completely missing. These results therefore disagree with the
capacitor model. Note that the last discord between AE and EMR signal appear-
ance is probably due to the difference of mechanisms leading to the two types of
radiation and is not yet completely understood.
In addition, the general no “symmetry breaking” argument mentioned above
evidently applies here as well.
3. Even under the assumptions that the two crack sides could be charged in a charge
mosaic manner (Finkel et al. 1975), thus retaining an overall charge neutrality and
that the EMR was induced by the dipoles consisting of pairs of oppositely charged
mosaic “elements” on the two crack sides, the EMR induced by the moving crack
would be expected to be very weak due to the cancellation of the radiation origi-
nating from mutually oscillating oppositely charged dipoles (random distribution
of the mosaic elements).

5.4.5
Summary of the New Model and its Properties

The suggested model of EMR origin can be described as follows: Following the break-
ing of bonds, the atoms on both sides of the severed bonds are moved to “non equilib-
rium” positions relative to their steady state ones, and oscillate around them. Lines of
oscillating atoms move together, and by being connected to atoms around them (in
the forward direction and also atoms on their side of the two surfaces newly created
by the fracture), the latter also participate in the movement. The larger the number of
cut bonds, the larger is the number of excited atoms, and hence the greater becomes
the EMR amplitude. These oscillations behave like surface vibrational optical waves,
where positive charges move together in a diametrically opposite phase to the nega-
tive ones and decay exponentially into the material like Rayleigh waves. The resulting
oscillating electric dipole is the source of the EMR. The pulse amplitude decays by an
interaction with bulk phonons.
408 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Our model does not suffer from the defects of the former ones since

1. No dislocations are included in the model, and therefore it can be applied also to
brittle and amorphic materials. EMR frequencies obtained experimentally do agree
with those of the surface vibrational optical wave (SVOW) model.
2. The spectrum of the SVOW is definitely not of a white noise type. Moreover, it is in
line with given in Eq. 5.1 and the measured ones (see, e.g., Fig. 5.24b). In particular
the peak frequency depends on crack width according to Eq. 5.7.
3. Note that no symmetry breaking mechanism appears in the SVOW model.
4. Our model is applicable to shear, tensile and even mixed modes of fracture.
5. The exact correlation between AE and EMR predicted by the “capacitor” model is
not always found experimentally. No such correlation arises in the SVOW model.
6. Charge neutrality is assured here via a different mechanism (“optical” oscillations).
Therefore no cancellation is expected.

5.5
EMR Pulses Induced by Rock Fracture
5.5.1
Previous EMR Pulse Investigations

All previous investigations considered EMR activity (number of EMR pulses) or EMR
amplitude (intensity). Several investigators tried to measure EMR signals in different
materials: Metals, glasses, ionic crystals and different rocks.
Misra (1975) investigated EMR induced in samples of carbon steel, aluminum, brass,
zinc, and lead under tensile loads. The scheme of Misra’s experimental arrangement
is shown in Fig. 5.25.
Samples were 9 mm in diameter and 160 mm in length. An EMR sensor (a co-axial semi-
cylindrically curved foil 27 mm in internal diameter, 50 mm in length and 0.3 mm in thick-

Fig. 5.25.
Schematic diagram of the
Misra (1975) experimental
arrangement
 5.5 · EMR Pulses Induced by Rock Fracture 409

ness) was placed near the fracture zone. EMR signals were passed through an amplifier
operating at a frequency of 30 MHz (with a frequency band of 0.7 MHz) and were picked
up by a storage oscilloscope. Figure 5.26 shows four examples of EMR signals observed
during fracture of samples of carbon steel (a), aluminium (b), brass (c), and zinc (d). EMR
signals from different materials were similar in shape but varied in their amplitude.
Jagasivamani and Iyer (1988) also noted that heat-treated high carbon spring steels
subjected to tensile loadings emitted EMR during fracture. Figure 5.27 shows an ex-
ample of EMR bursts measured by Jagasivamani and Iyer (1988).
It was found that the bulk of the EMR energy was in the range of 200 Hz to 10 kHz.
Nitsan (1977) investigated EMR signals induced by the fracture of quatrz-bearing
rocks (several different granites, granodiorites, sandstones and quartzites) and single
crystals of quartz and tourmaline. Materials were fractured under uniaxial loads be-
tween two parallel flat plates or by a small metal ball against a flat surface. The EMR
signals were measured by a radio-frequency coil, the effective bandwidth of which was
1–10 MHz. Figure 5.28 shows an example of such EMR pulses observed by Nitsan

Fig. 5.26.
Four examples of EMR signals
observed during fracture of
samples of a carbon steel,
b aluminium, c brass and
d zinc (Misra 1975)

Fig. 5.27.
An example of EMR signals
measured by Jagasivamani and
Iyer (1988)
410 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

(1977). He noted that small cracks induced in quartz and tourmaline crystals were
followed by short radio frequency signals with an approximately exponential decay
with a time constants of about 10–5 s for quartz (Fig. 5.28a) and about half of this time
for tourmaline (Fig. 5.28b). Figure 5.28c shows a radio-frequency signal with a more
complex shape. This signal was excited by the fracture of a sandstone sample.
Miroshnichencko and Kuksenko (1980) investigated EMR pulses irradiated by ionic
crystals, glasses and different rocks during fracture. They detected EMR by an induc-
tance coil of 100 kHz resonance frequency and low Q factor (Q = 2). An example of
EMR signals induced in diabase is shown in Fig. 5.29.
This signal was induced by a crack of 1 mm in length. Assuming that the creation of
the crack took time that was equal to the duration of the first “pulse” namely was 1 µs, Miro-
shnichencko and Kuksenko (1980) found that the average crack growth velocity was about
1 000 m s–1, which was about one-third of the velocity of sound in this material.
Yamada et al. (1989) observed EMR and acoustic emission during the entire defor-
mation of a rock using a standard uniaxial deformation tests at a constant strain rate
of 10–6 s–1on cylindrical cores of Indian granites 25 mm in diameter and 62.5 mm in
length. In the first experiment, they used five acoustic sensors to detect acoustic emis-
sion and one coil antenna to observe EMR. In the second experiment, one acoustic
transducer and five coil antennas were used. The EMR antennas consisted of a coil of
20–40 turns of enameled copper wire 0.2 mm in diameter. The coil antennas were tuned
in rather narrow bands. Figure 5.30 shows the overall frequency response of the ex-
perimental system (Yamada et al. 1989).
An example of Yamada et al. results is shown in Fig. 5.31. They claimed that their
EMR signals were associated with cracks of 1–100 µm length. Note that the shape of
Yamada et al. (1989) signals is similar to the ones observed by Nitsan (1977) for poly-
crystaline rocks (Fig. 5.28c).
Ivanov et al. (1990) investigated EMR pulses induced in rock samples from the
Konstatinov, Saralin and Tashtagol ore deposits. All samples were investigated under
uniaxial compression with different strain rates (10–4–10–7 m s–1). The initial signals
were detected by a capacitance antenna (copper foil placed at the sample faces), trans-

Fig. 5.28. Three examples of EMR pulses observed by the Nitsan (1977): a quartz, b tourmaline,
c sandstone

Fig. 5.29.
An example of EMR signals
induced in diabase (Miroshni-
chencko and Kuksenko (1980)
 5.5 · EMR Pulses Induced by Rock Fracture 411

Fig. 5.30.
The overall frequency response
of the experimental system
(Yamada et al. 1989)

Fig. 5.31.
An example of EMR and AE
signals observed by Yamada
et al. (1989)

formed by an amplifier and a filter block and stored by a storage oscilloscope, from which
screen they were photographed. The filter block consisted of a high frequency filter,
which had ten fixed cut-off frequencies: 300 Hz, 1, 3, 10, 30, 100 and 300 kHz, and 1 and
10 MHz. The low-pass filter had seven fixed frequencies: 300 Hz, 1, 3, 10, 30, 100
and 300 kHz. An optimal working frequency band was selected by this filter and hence-
forth enabled the improvement of the signal / noise ratio. A total of 140 rock samples were
studied. Figure 5.32 shows two examples of EMR signals observed by Ivanov et al. (1990).
These authors linked the rise-time of the pulse front with the crack propagation
time and the decay time with a charge relaxation process.
Frid (1990) investigated EMR signals during stiff uniaxial compression. He exam-
ined rocks from Khibin Deposit (Kola Peninsula), Ural bauxite mines, Shpitzbergen
and North Kuzbass coal deposits. EMR was detected by a narrow band resonance an-
tenna (resonance frequency of 100 kHz and a Q factor of 2), passed through a 30 dB
narrow band amplifier and stored by a storage oscilloscope, from the screen of which
signals were photographed. Figure 5.33 shows an example of an EMR pulse obtained
in this investigation.
412 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.32.
Two examples of EMR signals
observed by Ivanov et al. (1990)
induced by gabbrodiorite
fracture: a horizontal scale
20 µs / division, vertical scale
1 V / division, crack length
2.5 cm; b horizontal scale
2 µs / division, vertical scale
20 mV / division, crack length
3.5 mm

Fig. 5.33.
Typical EMR pulse measured by
Frid (1990) during rock fracture

The shapes of signals from all investigated rocks were shown to be similar, whereas
amplitudes and decay times were different. Note that the shapes of the EMR signals
here are similar to those observed by Nitsan (1977) for a mono-crystalline material
(Fig. 5.28a–c).
O’Keefe and Thiel (1995) performed EMR measurements on samples of dimensions
100 × 100 × 100 mm. A schematic illustration of their experimental arrangement is
shown in Fig. 5.34. The samples were stressed by uniaxial compression. The signals
were detected by a 100 mm dipole antenna placed 100 mm from the sample. The sig-
nal was passed through an amplifier (54 dB) and an anti-alias filter (cut-off 160 kHz)
and was connected to a PC, where signals were digitized and saved.
The digitization rate was 500 ks s–1. An example of the electric signals observed by
O’Keefe and Thiel (1995) is shown in Fig. 5.35.
It is seen that this signal is very similar to those obtained by Ivanov et al. (1990)
(Fig. 5.32).
An analysis of all of these results shows that the EMR pulses consist of a rising part and
a falling part. The majority of cases also show oscillations under the whole rising and fall-
 5.5 · EMR Pulses Induced by Rock Fracture 413

Fig. 5.34.
A schematic experimental
arrangement of O’Keefe and
Thiel (1995)

Fig. 5.35.
An example of electric signals
observed by O’Keefe and Thiel
(1995)

ing envelope. However, these investigations were conducted under different conditions
and most of them with the help of a narrow band antenna, which makes a quantitative
comparison between them very hard. Moreover, the shape of the signals depends to a
large extent on the digitization rate, and the latter must be significantly higher than the
500 ks s–1 used, e.g., by O’Keefe and Thiel (1995) in order to obtain an accurate picture.

5.5.2
Our EMR Pulse Classifications

Our investigations (Rabinovitch et al. 1996) showed that

1. EMR pulses appear in two main different forms: “Lengthy” (total duration 30–110 µs
and up to 400 µs, e.g., Fig. 5.11) and “short” (0.5–6 µs, e.g., Fig. 5.10a).
2. An analysis of the EMR pulses emanating during uniaxial failure showed that the
deformation process can be divided into three zones according to the EMR fea-
tures (Fig. 5.36, Rabinovitch et al. 1996):
Zone I: Of individual EMR pulses;
Zone II: Of EMR pulses with an uneven shape and/or groups of pulses;
Zone III: Of lengthy EMR pulses.
414 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.36.
Stress-strain curve of rhyolite
sample. Arrows on the graph
show moment of EMR pulse
excitations. Zone I is defined
by individual pulses; zone II is
characterized by pulses with
uneven shape and/or groups of
pulses; zone III shows particu-
larly lengthy pulses (see elabo-
ration of arrows (a–e) signifi-
cance in Rabinovitch et al. 1996)

3. In the beginning of the deformation individual pulses occurred, while at the upper
part of the elastic zone the shape of the EMR events became more complex indicat-
ing a superposition of individual pulses (sequences), with internal interpulse spac-
ing of less than 3 µs. The interval between sequences was 40–80 µs.

Lengthy EMR pulses were registered immediately before and at the peak zone.
These results indicate a correlation between the EMR shapes and the failure stages. Thus,
individual short pulses (zone I) seem to be correlated with the stage of individual micro-
crack formation, multi-pulse sequences (zone II) can be correlated with the crack coales-
cence stage, and the lengthy pulses (zone III) can be correlated with the sample faulting.

5.6
EMR and Material Elasticity
Gol’d et al. (1975) found differences in activity of EMR emanating from different ma-
terials in a row of materials investigated under uniaxial compression. Materials were
ordered (from “high” to “low”) according to the number of EMR signals registered
during cracking as follows: Quartz, orthoclase, plagioclase, granites, sandstones, and
metamorphosed aleurolites. They noted that a decrease of the number of EMR sig-
nals could be due to a decrease of rock/mineral hardness (i.e., elasticity) from quartz
via granites to metamorphosed aleurolites. Khatiashvili (1984) investigated EMR in-
duced by splitting of LiF, NaCl, KCL and CaCO3 crystals. He showed that EMR frequency
spectra are specific for each crystal (Fig. 5.37a). The spectra differ in the positions and
widths of the radiation bands. He also observed that the intensity (amplitude) of the
EMR induced by alkali halide crystal failure was proportional to the squares of the
elastic moduli and the crack length (Fig. 5.37b). These investigations (Gol’d et al. 1975;
Khatiashvili 1984) were carried out only under uniaxial loading.
 5.6 · EMR and Material Elasticity 415

Fig. 5.37.
a EMR spectra from splitting
1-LiF; 0.1I = 1 mV, 2-NaCl;
0.1I = 80 µV, 3-KCl; 0.1I = 15 V,
4-CaCO; 0.1I = 0.1 mV; b EMR
intensity as function of frac-
ture length (Khatiashvili 1984)

Fig. 5.38.
An example of granite triaxial
deformation (confining pres-
sure: 10 MPa); a the pore clo-
sure region (10 EMR pulses
were measured); b the elastic
region (6 EMR pulses were
measured); c the nonlinear
region before the peak stress
(8 EMR pulses were measured);
unbroken line: Axial strain,
dashed line: Volumetric strain
(Frid et al. 1999)

For our investigations, we used granite samples (see above). Each sample was tested
by an axial strain rate of 1 × 10–5 s–1 and laterally by different hydrostatic oil pressures.
The axial pressure varied from 110 MPa up to 284 MPa and the confining pressure
from 0 to 14 MPa.
Figure 5.38 shows an example of a stress-strain curve of a representative granite
sample (confining pressure 10 MPa). All rock samples went through the “normal” three
deformation stages (Jaeger and Cook 1979) during compression up to the peak stress:
a the nonlinear region of pore closure (axial strain curve here is slightly bent down,
b the elastic region (the stress-strain curves are quasi linear), and c the nonlinear re-
gion before the peak stress.
As seen from the Mohr-Coulomb diagram (Fig. 5.39), the increase of confining pres-
sure from 10 MPa (third circle) to 14 MPa (fifth circle) is not accompanied by an ap-
preciable increase of shear strength (the Mohr-Coulomb envelope becomes much less
steep than for smaller confining pressures).
Since, the confining pressure does not cover a wide range of pressures, the ductil-
ity stage has not yet been completely reached. However, the failure plane inclination
to the axial load axis was measured to be 41 ±1°. Thus, our results are reasonably close
to obeying the von Mises ductile failure criteria.
416 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.39. Two stages of the failure envelope (thick line) around Mohr circles which represent the
different confining pressures, ______ uniaxial test, __ __ __ 5 MPa confining pressure, _ _ _ 10 MPa con-
fining pressure, __ _ __ 12 MPa confining pressure, __ _ _ __ 14 MPa confining pressure (Frid et al. 1999)

Fig. 5.40.
Relationship of the following
characteristics with confining
pressure: a Poisson ratio;
b Young’s modulus; c EMR
activity. Error bars on these
figures are determined by the
accuracy limits of our rock
mechanics equipment; that
is ±2% for Poisson ratio and
Young’s modulus and by the
dead time of the storage oscil-
loscope-IBM PC system (the
period it takes for registration,
digitizing and memorizing an
EMR sequence, during which
time the system is shut down.
The error in EMR activity is of
the order of +2% (Frid et al.
1999)
 5.6 · EMR and Material Elasticity 417

We denote here the EMR activity as the number of individual EMR pulses emitted
during tracing of the entire stress-strain curve. Its value is seen to change with the
loading type (Fig. 5.40c), increasing first with the confining pressure and then decreas-
ing again.
We checked the following stress-strain characteristics to investigate possible cor-
relations with EMR activity: Axial strain and its increment; confining pressure; de-
formation modulus (ratio of axial stress increment to rise of axial strain – the secant
modulus) and its absolute value); differential stress; differential volumetric com-
pression work (product of differential stress increment and volumetric strain changes)
and its absolute value; elastic strain work (product of average differential stress and
elastic volumetric strain changes); elastic volumetric strain (part of volumetric strain
that returns to its original value after applied stresses are removed); lateral deforma-
tion modulus (ratio of axial deformation increment to rise of lateral deformation –
the secant lateral modulus) and its absolute value; maximal stress; Poisson ratio (ra-
tio of lateral elastic deformation to an axial elastic one, measured in the elastic zone
of deformation); residual axial strain (remaining axial deformation after removal of
external stress); residual strain work (strain work minus elastic work); residual volu-
metric strain (remaining value of the sum of the three principal strains after removal
of external stress) and its absolute value; strain work (product of average differential
stress and volumetric strain changes) and its absolute value; volumetric strain (sum
of the three principal strains) and its absolute value; Young’s modulus (ratio of axial
stress increment to rise of elastic axial strain, measured in the elastic zone of defor-
mation).
Note that these stress-strain characteristics can be classified as either “elastic” or
“general”. For example, the Poisson ratio and the Young modulus are defined here as
elastic characteristics and measured in the elastic zone only. Lateral deformation
modulus and deformation modulus, on the other hand, are similar in definition to the
Poisson ratio and Young modulus, respectively, but they are “general” characteristics:
They can change during a loading process.
Our analysis shows that out of all these stress-strain characteristics the Poisson ra-
tio is the only one that shows good correlation with EMR activity. The Poisson ratio
(Fig. 5.40a) changed from 0.36 (for uniaxial test) via 0.27 for 10 MPa confining pres-
sure to 0.33 for 14 MPa confining pressure. Its squared regression coefficient was –
0.93. An increase of the Poisson ratio (Fig. 5.40a) is accompanied by a decrease of EMR
activity (Fig. 5.40c) and vice versa. Note that since all samples here are made of the
same material (granite), the Poisson ratio varies only with loading conditions, and does
not vary with material properties, on the other hand e.g. the Young’s modulus (Fig. 5.40b)
increases monotonically from 27 GPa to 48.5 GPa with the confining pressure, and
therefore it is obvious that its correlation with EMR activity should be poor (the ob-
tained squared regression coefficient was actually 0.18).
The result of a poor correlation between Young’s modulus and EMR activity is dif-
ferent from that found previously (Gol’d et al. 1975; Khatiashvili 1984) in which an
increase of Young’s modulus of a material was associated with an increase in EMR
activity. In the previous experiments, however, the correlation between Young’s modu-
lus and EMR activity was investigated only under uniaxial compression. Our loading
conditions were different. We studied changes of EMR activity during the transition
418 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

from tensional fracture (uniaxial test) via brittle shear fracture according to the Cou-
lomb criterion, to shear fracture in the brittle-ductile region under an increase of con-
fining pressure (see, for example, Twiss and Moores 1992, Fig. 9.9). As noted by Jaeger
and Cook (1979), Young’s modulus increases with the confining pressure, while the
Poisson ratio could both decrease and increase. We suggest that our sample deformed
with the confining pressure as follows: An increase of the confining pressure from 0
to 10 MPa closed granite microcracks and hence the Poisson ratio was decreased to
its intrinsic value of a material devoid of openings. Under a greater confining pres-
sure (12–14 MPa), however, the deformation changed from a brittle to a ductile one
causing an increase of Poison ratio. The brittle-ductile transition is confirmed by the
large angle (41 ±1°) produced between the axial load and the failure plane and in the
insignificant increase of shear strength (Fig. 5.39). Correspondingly during the brittle
deformation, EMR activity increased with confining pressure while in the brittle-duc-
tile region EMR activity decreased.
The correlation (or rather anti-correlation) of EMR activity with the Poisson ratio
seems plausible. The Poisson ratio measures the compliance of the material in the
transverse direction when stressed axially. The lower the Poisson ratio, the harder it is
for the material to strain transversally, and hence the higher is the probability of new
fractures (especially parallel to the axis) and of the ensuing EMR. On the other hand,
the higher the Poisson ratio the easier it is for the material to strain transversally, and
accordingly, fewer fractures and lower EMR activity should be expected. The key elas-
tic parameter for EMR characterization during triaxial compression is therefore the
Poisson ratio and not the Young’s modulus. Note that our results are in line with those
of Gol’d et al. (1975) and Khatiashvili (1984) for uniaxial load. Our results, however,
being triaxial, do show the correct global parameter to choose. They also indicate that
the EMR activity correlates well with the lateral resistance to axial fracture, which is
compensated by an increase in the number of new cracks.

5.7
EMR during Percussion Drilling
Recently, Rossmanith et al. (1996, 1997) carried out experimental and numerical in-
vestigations in order to understand the mechanism of material damage under per-
cussion drilling. They found that the stress applied during drilling, besides inducing
elastic waves emission, also resulted in defect creation: Every time a percussion oc-
curs, existing defects grow, microcracks are formed and propagate, and a “damage zone”
around the drilling tip is thus created (Rossmanith et al. 1996, 1997), a zone consist-
ing of a network of many small and a few larger cracks. Another observed defect con-
sists of circumferential cracks around the hole and appears when weak regions exist
in the material (Ravishankar and Murthy 2000).
Since fracturing also emits EMR, we discuss here an investigation of the EMR emit-
ted from fractures that develop during percussion drilling.
We used cylindrical samples 100 mm in length and 30 mm in diameter (Fig. 5.41)
that were percussion drilled. Drilling was carried out at about the mid length of the
cylinder, perpendicular to the cylinder axis. EMR was measured in the frequency range
1 kHz to 50 MHz with 1 µV sensitivity throughout. Radiation was detected by a one-
 5.7 · EMR and Percussion Drilling 419

Fig. 5.41.
Schematic diagram of experi-
mental arrangement during
drilling (Goldbaum et al. 2001)

loop magnetic antenna 3 cm in diameter, which is electrically “small” and exhibits neg-
ligible response to foreign electric fields. The antenna was placed at a distance of 3 cm
from the drilling hole. External EMR disturbances were further decreased by the two
following methods:

1. The sample together with the antenna were placed in a grounded Faraday cage.
2. The plane of the antenna was aligned perpendicular to the drilling direction, so
that the influence of electromagnetic disturbances emitted by the motor of the
drilling machine was minimized.

All EMR signals were electrically amplified by 60 dB, digitized and collected at a
PC hard disk. The data was analyzed after test completion.
The materials used in the study were chalk from Middle Eocene layers in the Beer
Sheva syncline, Eilat granite from the Nahal Shelomo area of southern Israel (Rabino-
vitch et al. 1999b), Solenhofen limestone from Germany, PMMA and soda-lime glass.
420 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

The 700-odd EMR signals measured during percussion drilling of all five materi-
als were classified into four groups:

1. The first group: “Short single pulses” of 0.3–1.5 µs duration.

2. The second group: “Short chain of single pulses” of 2–15 µs duration.
3. The third group: “Extended chain of pulses” of 15–60 µs duration.
4. The fourth group: “Pulses on top of base-line voltage changes” whose duration var-
ies between 10 and 800 µs.

Pulses of the first group (Fig. 5.42a) probably correspond to single cracks and are char-
acterized by a single main frequency ranging between 10 and 25 MHz. It means (Eq. 5.7)
that the corresponding cracks are very small, and indeed their T' / ω values (Eq. 5.8) indi-
cate that they range in area from 0.005 mm2 up to about 0.05 mm2. Pulses of the second
group (Fig. 5.42b) consist of several pulses of the first group type and probably correspond
to several overlapping small cracks. Their frequency spectrum is thus a bit more compli-
cated. Pulses of the third group consist of numerous pulses of the first two groups
(Fig. 5.42c). Hence, their frequency spectrum shows a broad band (11–23 MHz). These
pulses are probably emitted by tiny “powdered” particles fragmented from the walls of
the drilled hole, where each particle has its own size and emits its own frequency. Note
that like that of the first group, the area of the “powder” particles here ranges from
0.005 mm2 up to about 0.05 mm2.
Signals of the fourth group (Fig. 5.42d) (never seen in the compression experiments)
show baseline voltage changes with very low frequencies of several kilohertz and in addi-
tion contain lengthy groups of the third type. Cracks emitting such radiation should range
in area from 1 cm2 up to 16 cm2. The additional presence of the third group pulses indi-
cates that “powder” is simultaneously created.

Fig. 5.42. Experimental EMR pulses induced by percussion drilling of the first group (a), second
group (b), third group (c) and forth group (d) (Goldbaum et al. 2001)
 5.7 · EMR and Percussion Drilling 421

The first three group types have been observed in our compression experiments,
where however single pulse frequencies never exceed 10 MHz (Rabinovitch et al. 2000c;
Frid et al. 2000).
They provide a new method for a general identification by EMR profile of the wide
population of “powdered” and “chipped” particles of various sizes that form by per-
cussion drilling and a one to one identification of all the defects involved in the pro-
cess. The fourth group is discussed next.

5.7.1
The 4th Group and Polarization

The fourth group includes two different subgroups (4a): “Lengthy forms” (e.g., Fig. 5.42d),
whose duration varies between 10 and 800 µs and which show low frequencies (sev-
eral kilohertz) baseline voltage changes and also contain high frequency third type
pulses and (4b): A different type of “Lengthy forms” again decorated by high frequency
strings (Fig. 5.42, 5.43).
Here we present the results of investigation of the 4a and 4b signal types (Goldbaum
et al. 2001), where the “4” is dropped in following. The samples tested were cylinders
of soda-lime glass, 10 cm in length and 3 cm in diameter that were percussion drilled.
The experimental arrangement was described in Sect. 5.7.
Signals of groups a and b occurred in 5 glass samples (2, 3, 4, 5, 6). Group ‘a’ sig-
nals are similar to individual EMR pulses (Rabinovitch et al. 1998), but their frequen-
cies are smaller (several kHz). Group ‘b’ signals (Fig. 5.43) consist of a sharp voltage
“jump” (of a duration of ~3% of the whole pulse duration) and range in size between 6
and 12 µV; the high voltage remains more or less constant for a period that varies be-
tween 10 and 1 000 µs and is terminated by a rapid falloff (of about the same duration
as that of the rise time). Group ‘a’ signals are frequently followed by group ‘b’ signals
after a time interval of several hundreds of microseconds. This occurrence is denoted
by “(a/b) pair”. If a ‘b’ signal has not been detected within this time interval, then an
‘a’ or a ‘b’ signal appears after a time period ranging between 0.2 s and 2 s. Sometimes
a string of several ‘a’ signals appears and is followed by a ‘b’ signal.

Fig. 5.43.
Typical pulse of group ‘b’
(Rabinovitch et al. 2003b)
422 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

5.7.2
Fractographic Examination

Fractographic examination showed that in addition to glass powder and a network of

small cracks (Rossmanith et al. 1997), some large “chips” were formed. These are large
cracks, which originated at some depth and propagated until reaching the surface of
the sample. Their observed fracture areas vary between 1 and 30 cm2.
As shown in Goldbaum et al. (2001), the ‘a’ type pulses are excited by the creation
of these flakes and the ‘b’ type pulses are derived from a polarization process, as shown
presently.

5.7.3
Brief Theoretical Consideration

The measured quantity in our experiments is the electromotive force in the loop an-
tenna, which is given by

the time derivative of the magnetic flux

Here B is the magnetic field through the antenna and S is its (constant) area. Since
r / c = 10–10 s it can be neglected with regard to the apparatus sensitivity (2 × 10–8 s) in
comparison and we have (Bleaney and Bleaney 1965):

(5.10)

where µ0 is the magnetic permeability, c is the speed of light, l is the length of the ele-
ment in which current is present, r is the distance between the antenna and the cur-
rent, I = I(t) is the magnitude of the current, the angle θ is that created by the current
direction and the radius-vector to the antenna relative to the current, and a dot de-
notes time differentiation. For EMR, the current is produced by the oscillating surface
waves discussed in Sect. 5.2. Since our experiments were carried out in the so-called
“short” zone, we can neglect the second terms in (Eqs. 5.10) and write

(5.11)

Consider now a changing dipole p(t) with a constant length l and changing charge q(t),
p(t) = lq(t). The current (Bleaney and Bleaney 1965) is I = dq / dt, lI = dp / dt, and
 5.7 · EMR and Percussion Drilling 423

(5.12)

Thus, the electromotive force V in the antenna is proportional to the second de-
rivative of the dipole moment:

(5.13)

If instead of a dipole moment one considers the sample’s polarization P, which is

defined as its dipole moment per unit volume, the electromotive force would be pro-
portional to the second derivative of the polarization:

(5.14)

Here 1 / C > 0 is the proportionality constant. The polarization current is given by

the integral of (5.14):

(5.15)

·
where P0 is a constant, and the polarization P is given by the integral of (5.32):

(5.16)

5.7.4
Examination of ‘b’ Type Signals

We assume that the ‘b’ type signals (Fig. 5.43) originate by a change of material polar-
ization (Eq. 5.14). By Eq. 5.15, the first integral of the signal should be proportional to
·
P, while the second integral should be proportional to the polarization (P) itself. This
sequence is schematically shown in Fig. 5.44. A smoothed ‘b’-like signal is shown in
Fig. 5.44c, where units of amplitude and the time are evidently different from those of
Fig. 5.43, but signal shape is preserved. The first and second integrals appear in
Fig. 5.44b and 5.44a, respectively.
Ogawa et al. (1985) measured EMR from impacted materials using an electric sen-
sor. Hence the signals obtained by them (see Fig. 2 and 3 of Ogawa et al. (1985), re-
drawn here as Fig. 5.45a and b) should be proportional to the depolarization current
or to the first derivative of the polarization, and the time derivative of their signal
··
(Fig. 5.45c) should be proportional to P.
Comparing the pulse shown in Ogawa et al. (Fig. 3 of Ogawa et al. 1985, redrawn
here as Fig. 5.45b) with the integral of a ‘b’ pulse (Fig. 5.44b), it is clear that they are
similar, indicating that both stem from a similar process, namely polarization changes.
The shape of the polarization itself can be obtained by the appropriate integrals.
424 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.44.
Schematic polarization accu-
mulation and depolarization
process: a polarization: P.
Note: Only the final part of the
polarization increase is shown.
b First derivative of P, and
c second derivative of P, as
functions of time. Note the
negative tail appearing in c
(Rabinovitch et al. 2003b)
 5.7 · EMR and Percussion Drilling 425

Fig. 5.45. Pulses obtained by Ogawa et al. (1985): a their Fig. 2, b their Fig. 3, c smoothed time de-
rivative of 5.45a, and d the integral of 5.45a (Rabinovitch et al. 2003b)

5.7.5
The Polarization Model

The emerging picture of the process is as follows: During drilling a stress-strain field is
slowly accumulated in the sample by numerous accumulated impacts (percussions), fol-
lowed by an increase in polarization (Varotsos et al. 1999) (see Fig. 5.44a) for the final stage
of this process). This accumulation of stress and polarization can take place only suffi-
ciently ahead of the drill bit, since the region close to the latter is obviously reached by it,
releasing the stress, before the latter grows sufficiently. During polarization increase, p·· is
too small to be detected (Fig. 5.44c, the region up to t = 100). Note that electric polariza-
tion (i.e., the appearance of a dipole moment in a material) can be caused (Khesin et al.
1996, 1997; Rzhevskii and Novik 1978) by four basic processes: Electronic, ionic, dipolar
and space charge polarization. Since relaxation times of the first three processes are less
than 10–6 s and since measured relaxation times for glass polarization are of the order of 10 s
(Kuksenko et al. 1990), the process here can only be due to space charge polarization.
Crack development begins when the accumulated stress becomes sufficiently large.
Crack propagation (indicated in our measurement by an ‘a’ signal) is accompanied by
stress relaxation, and the latter causes depolarization and an appearance of a ‘b’ signal.
426 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.46.
Distribution histogram of time
lags between an ‘a’ and a fol-
lowing ‘b’ signal (Rabinovitch
et al. 2003b)

Natural polarization relaxation in glass is very slow (see above) and can therefore be
neglected. Here however polarization is relaxed by fracturing, and therefore the ‘a’ sig-
nals are followed by the ‘b’ signals, within a small time interval.
However, polarization does not fall off at once, because not all ions participating
in the depolarization current begin their movement at the same moment: An ion be-
gins to move when the acoustic waves excited by fracture reach it.
The time it takes a crack to travel through the entire sample whose size is 10 cm
(with a velocity of 300 m s–1, which is low but possible in glass) is 300 µs. We have regis-
tered thirty-three couples in which ‘a’ signals were followed by ‘b’ ones, and twelve of
them have a time lag of less than 300 µs (Fig. 5.46). There are seven couples with a
time difference between the ‘a’ and ‘b’ pulses of up to 600 µs. These longer times may
be due to the fact that not all ions are situated along the way of the propagating crack.
Some of them are placed at rather large distances from the crack, so stress relaxation
must first reach a definite point on the crack route, and then be transferred by acous-
tic waves to the distant ions. For larger measured differences, it is assumed that there
was an additional signal in between those observed that has not been detected by our
equipment possibly because of a very small solid angle at the antenna.
Note that the spread in the histogram (Fig. 5.46) can also be due to another reason.
To understand it, let us consider Fig. 5.47a and b that schematically show two examples
of possible depolarization events. The calculated ‘b’ signals (second derivatives)
(Fig. 5.47c,d) show that the second one (Fig. 5.47d) is shifted by ten time units with
respect to the first one (Fig. 5.47c). This means that a negligibly small change in the
depolarization shape may cause a significant shift in the ‘b’ signal origin, and hence
to a spread of delay times between the ‘a’ and the ‘b’ signals.
Flake cracking induced by the drill bit thus leads to the measurement of a group ‘a’
and group ‘b’ signal. When cracking stops, stress relaxation and depolarization current
increase also halt. Thus “immediate” depolarization can be incomplete, and some po-
larization could still remain in the material. The latter continues to decrease rather slowly
to zero (Fig. 5.44a). Depolarization current (Fig. 5.44b) generally changes too slowly to
 5.7 · EMR and Percussion Drilling 427

Fig. 5.47. (Schematic) Slight differences in depolarization shapes (a, b) can cause large shifts in the
time lags between the observed signal pair (c, d) (Rabinovitch et al. 2003b)

give a measurable second derivative (Fig. 5.44c), but in some pulses a “tail” appears
(Fig. 5.48a, arrow). Since measurements were carried out only up to 1.3 × 10–3 s, we ar-
tificially continued this “tail” (for the specific pulse of Fig. 5.48a) and evaluated its first
and second integrals (Fig. 5.48b,c). Note that the derivative of the pulse obtained by
Ogawa et al. (1985) (Fig. 5.45c) shows a similar “tail”.
The estimated decay (relaxation) time for such a “tail” is several milliseconds, which
is much shorter than the ~10 s natural relaxation times observed in glass (Kuksenko
et al. 1990). This decay time decrease can be due to material heating. It was shown
(Kuksenko et al. 1990) that the EMR pulse decay time decreases significantly with a
rise of temperature. Measurements of temperature rise during fracture have been car-
ried out for different materials yielding very high values. For instance, for PMMA the
temperature increase was about 500 °C (Swallowe et al. 1986), while for glass it was
~840 °C for a 200 m s–1 crack velocity (Li et al. 1988).
Comparison of the general shape and value of the rise-fall times of group ‘b’ pulses
enables us to assume that these pulses are excited by space-charge polarization
(Rzhevskii and Novik 1978; Gueguen and Palciauskas 1994). The charge carriers caus-
ing polarization are probably the weakly pinned Na+ ions (Kuksenko et al. 1990) of
the glass. Hovestadt (1902) described a change in the position of temperature mark-
ers on a thermometer over time. A structural change in a silicate mineral (feldspar) at
room temperature was ascribed to a Na+ displacement under strain (Bahat 1968). These
relatively heavy, weakly bound ions (Babcock 1977) can leave their equilibrium posi-
tions and move under stress, leading to polarization. When stress is removed or re-
laxed, these ions return to their equilibrium positions.
428 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.48.
A ‘b’ type pulse with a clearly
visible “tail” (arrow) (a), the
first integral (b) and the second
integral (c) of the extended pulse
(see text) (Rabinovitch et al.
2003b)

One remark is in order. The high frequency “decorations” of both the ‘a’ and the
‘b’ signals are evidently due to the powder and small cracks generated by the drill si-
multaneously with the flaking and depolarization process discussed here.
The flake creation process itself is accompanied by an ‘a’ pulse, while the depolar-
ization process is accompanied by a ‘b’ signal. The latter, being fast enough, is regis-
tered by our equipment. The flake/depolarization pairs should occur in a cyclic man-
ner, since following depolarization it takes some time for a repolarization process ahead
of the drill bit to reoccur. Hence also the emission of ‘a’/‘b’ signals should be cyclic
and not continuous. However, since following the first ‘a’/‘b’ pair creation the stress
does not fall off to zero, it should take less time for the stress to increase to the critical
value than the time to build up the first ‘b’ pulse. This phenomenon was actually ob-
served as a gradual decrease in the time periods between ‘a’/‘b’ pairs.
 5.8 · Intermediate- and Large-Scale EMR Detection 429

5.8
Intermediate- and Large-Scale EMR Detection
5.8.1
Blasting Experiments

5.8.1.1
Experimental Arrangement and Results

Blasting experiments were carried out in Turonian age limestone in an open quarry
near the village Omarim, located about 25 km north-east of Beer Sheva (Israel). Two
blasts of 5 and 1.4 t of explosive material respectively were performed. In each blast,
one slope of the quarry was exploded. The slopes are of a height of about 11 m and of
length of about 100 and 20 m, respectively. A magnetic one-loop antenna (Electro-
Metrics Penril Corporation) 0.5 m in diameter was used for the detection of the EMR.
A low-noise micro-signal amplifier (Mitek Corporation Ltd, frequency range 10 kHz
to 500 MHz, gain 60 ±0.5 dB, noise level 1.4 ±0.1 dB throughout) and an analog-to-
digital converter connected to a triggered PC completed the detection equipment. We
placed our magnetic antenna in front of the blasted slope at a distance of about 100 m
from it (Fig. 5.49).
Figure 5.50 shows ∼5% (in time) of the EMR-record that was observed during
blasting. A detailed analysis of the EMR record shows that the main frequency of
the observed pulses lies in the range 2–8 MHz, and that the intervals between
individual pulses are of about 1–5 µs (Fig. 5.51). The shapes of individual pulses
(Fig. 5.51) that were formed by a digital expansion of Fig. 5.50 are very similar to
those observed during compression and drilling (Fig. 5.52a,b). Our analysis shows
that it is only the entire durations of the EMR-records that were different. The dura-
tions of the EMR records from blasts were similar to those of the explosions them-
selves ∼0.5 s for the more powerful blast (5 t) and ∼0.3 s for the second one (1.4 t).

Fig. 5.49.
Schematic diagram of experi-
mental arrangement during
blasting
430 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.50. The EMR string induced by blasting (Rabinovitch et al. 2002a)

Fig. 5.51. An individual EMR pulse induced by blasting (Rabinovitch et al. 2002a)

5.8.1.2
Analysis of Results

As is known (Rossmanith et al. 1997), when stress waves propagate through a brittle
material, a population of micro-defects are generated and extended. In drilling and
blasting the rock is deformed under dynamic conditions, which create high intensity
stress waves. A stress wave of sufficient amplitude (like those created during blasting
or drilling) can initiate rock fragmentation at the expense of wave energy (Gueguen
and Palciauskas 1994). As is also known (Rabinovitch et al. 1998, 1999b, 2000a, 2000b),
 5.8 · Intermediate- and Large-Scale EMR Detection 431

Fig. 5.52.
a Three individual EMR sig-
nals from chalk drilling. b An
experimental EMR pulse in-
duced by chalk compression
(Rabinovitch et al. 2002a)

each crack is a source of an individual EMR pulse. Hence, a fragmentation process,

such as drilling or blasting, where numerous cracks are created, is an origin of strings
of EMR pulses. A larger stress wave energy leads to a larger fragmentation volume in
the rock and hence to a larger duration of these strings (and to a larger number of pulses
per strings). Therefore EMR pulse strings, which represent a small volume of rock in
drilling, are much shorter (2–60 µs) than those induced by blasting (∼0.3 and 0.5 s). In
comparison, during compression, where these strings stem merely from crack coales-
cence, the duration of each EMR string never exceeded 30 µs. A zooming in on the EMR
strings induced by blasting shows that both the amplitude and the frequency of indi-
vidual pulses are different from those obtained in drilling and in compression. Never-
theless an overall comparison of EMR signals excited by compression, drilling and blast-
ing shows their basic shape to be invariant to the dynamic/quasi-static loading mode.
Note that this basic shape was shown previously to be invariant to the fracture mode
432 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

(Rabinovitch et al. 2000c, Sect. 5.2.3). The similarity of shape of single EMR pulses in-
duced by these three processes therefore enables us to analyze the EMR signals induced
by drilling and blasting by the same technique we developed earlier for compression
(Rabinovitch et al. 2000c, Sect. 5.2). This technique allows us to calculate the dimen-
sions of the crack, which was the source of the measured EMR signal. Our analysis shows
that the time from EMR pulse origin up to its envelope maximum T' induced by blast-
ing ranges between 0.01–0.5 µs. The ω value of the signals ranges between 16 and
50 × 106 s–1. Hence their ratio T' / ω will be of about 0.06–1 × 10–14 s2. According to Bahat
et al. (2001a), the proportionality coefficient between T' / ω and crack area S is ∼3 × 107.
Hence, crack areas induced by blasting include a major fraction whose individual ar-
eas are estimated to be of about of 0.02–0.3 mm2. Note that cracks of similar areas are
created at the beginning of compression loading (Rabinovitch et al. 2000c, Sect. 5.3.4).
As noted by the Rossmanith et al. (1997), the set of fractures induced by blasting
can be divided into three zones: A crush zone (near the borehole) of short uniformly
distributed radial cracks; an intermediate zone, where short cracks from the crush zone
extend; and an outer zone with long radial cracks. Comparison of crack areas observed
during our blasting with this model shows that our EMR signals are induced by the
cracks from the two first zones (crush and intermediate) and can indicate the begin-
ning of a dynamic large-scale failure.

5.8.2
EMR Prior to EQ and Volcanic Eruptions

5.8.2.1
Current State of the Art

As noted above, during the nineteen seventies and eighties, interest in EMR increased
in connection with the problem of earthquake (EQ) prognosis. Several investigators ob-
served the appearance of a disruption of radio communications, atmospheric lumines-
cence and perturbations before EQs (Davis and Baker 1965; Finkelstein and Powell 1970;
Drobdev et al. 1978; Galperin et al. 1985; Chmyrev et al. 1986; Kustov and Liperovsky
1988; Bilichenko et al. 1990; Ouellet 1990; Johnston 1991; Derr 1986; Molchanov and
Hayakawa 1998a,b; Liu et al. 2000; Lee et al. 2000). Sadovskii et al. (1979) observed an
irregularity of EMR before some Karpatian EQs. Gokhberg et al. (1979) investigated
EMR prior to an Iranian EQ. This study consisted of EMR registration at frequencies
of 25, 385 and 1 630 kHz, for a period starting 55 min before the first EQ shock. EMR
perturbations were observed to gradually increase up to the first shock moment. Fol-
lowing this shock, the levels at frequencies of 27 and 1 630 kHz sharply decreased, while
those at 385 kHz remained high up to the last “aftershock”. These authors proposed
that if the EMR sources were deposited in the Earth’s crust at the EQ epicenter, the
original amplitude of the EMR signal would have to be of the order of hundreds of
Volts per meter. Similar works were also conducted in Japan (Gokhberg et al. 1982).
EMR was registered at a distance of 250 km from the epicenter of an EQ of magnitude
M = 7. Since an analysis of “industrial” noise showed its absence at a frequency of
81 kHz, that frequency was chosen for EMR studies. It was shown that 30 min before
the EQ, the EMR amplitude rose by up to 15 dB and then sharply decreased at the EQ
 5.8 · Intermediate- and Large-Scale EMR Detection 433

moment. EMR intensity (voltage) changes in a tunnel at a depth of about 60 m were

also checked. Analysis of diurnal cycles of EMR intensity variations showed that they
were significantly disturbed a few days before an EQ. Gokhberg et al. (1986) carried
out measurements of EMR features prior to an EQ in Gasli. They found that 2 h 40 min
before the EQ (M = 5), perturbations above the original noise level were increased by
up to 6 dB. Yoshino and Tomizawa (1989) measured EMR (at a frequency of 82 kHz)
before a volcanic eruption (VE). The first EMR anomaly was registered two weeks
before the VE, and another powerful EMR emission (20 dB higher than the usual noise
level) was observed one day prior to the VE. Rikitake (1997), analyzing 60 EQ events
measured in Japan, decided that EMR was a “short-term” precursor, with an estimated
mean time prior to an EQ of ∼6 hours. Nikiforova and Yadakhin (1989) investigated
EMR in the Issik-Kul seismic region at a frequency of 15 kHz. They obtained a sig-
nificant increase of EMR intensity in the immediate zone of active geological faults,
and claimed that the EMR sources were of ionosphere origin. On the other hand,
Morgunov and Matveyev (1991) conducted an EMR investigation during the Spitak
EQ and claimed that EMR could not be explained as an ionosphere disturbance; its
origin could only be ascribed to crust destruction. Gershenzon et al. (1987, 1989) and
Gokhberg et al. (1982, 1986) proposed that the anomalies of EMR prior to EQ were
originated by a mechanic-electric transformation. They noted that before an EQ, a
deformation of the Earth’s surface took place. Changes in the stress state of surfaces
resulted in local destruction, in the formation of micro-fractures and in friction of
the nearby rock blocks. Each of these processes could lead to EMR generation. Bella
et al. (1992), Morgunov (1985), Morgunov et al. (1988) and Morgunov and Matveyev
(1990) also favored the idea that EMR was excited by surface deformation in the vast
zone of the EQ vicinity. Gokhberg et al. (1985) noted the following principal features
of EMR occurrences:

1. Abnormally high EMR levels, which occur hours or even days before an EQ.
2. The EQ itself takes place during a decrease of this EMR anomaly or immediately
after it.
3. The EMR spectrum has a wide frequency range (from several Hz to hundreds kHz).

The third conclusion was also endorsed by Khatiashvili (1984) (who registered EMR
in the range of 1–1 000 kHz) and Bella et al. (1992) (who studied EMR in the range of
300 Hz to 300 kHz). It is clear from the above sections that there is no uniquely ac-
cepted mechanism regarding the mechanism of EMR prior to an EQ. Parrot et al.
(1993), after a detailed consideration of a large number of presently known EMR-EQ
investigations, remarked that although the existence of EMR in relation to seismic and/
or volcanic activities was clear, EMR selection out of a host of artificial signals remained
a significant problem. Nevertheless, investigations of EMR as a precursor to EQ are
still in progress (Drakopoulos et al. 1993; Hayakawa et al. 1993; Yepez et al. 1995;
Fujinawa and Takahashi 1998) and presently there is an agreement in the literature
that EMR might be a prospective forecaster for EQ (King 1983; Mogi 1985; Yoshino
et al. 1993). However, as we noted above, all efforts to use EMR for earthquake predic-
tion met with very meager success, due to the lack of a detailed quantitative under-
standing of the EMR mechanism (King 1983; Rabinovitch et al. 1996).
434 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

5.9
Global Features of EMR
5.9.1
EMR versus Gutenberg-Richter Relation

5.9.1.1
Fractal Nature of Earthquake

A fundamental observation in seismology is the Gutenberg-Richter law (Gutenberg

and Richter 1954):

(5.17)

where M is the earthquake magnitude (which is defined as the logarithm of the inte-
gral of slip along the fault during an earthquake), N is the number of earthquakes
having magnitudes greater than M, and a and b are constants. This power law or fractal
distribution is valid both for main events and for aftershocks (Nanjo et al. 1998). It is
presently thought that such a distribution can be a fundamental result of “multiple
fracturing” (Turcotte 1992), when spontaneously occurring microcracks tend to coa-
lesce, leading, by numerous upscalings, to a catastrophic failure. A similar relation is
also observed in laboratory studies of acoustic emission (Lockner et al. 1991; Lockner
1993). It is even true for energy distribution of neutron star-quakes (Jagasivamani and
Iyer 1988) (the source of a star-quake is a fracture in its neutron crust, which may re-
lease strain energies of up to 1046 erg! (Link and Epstein 1996)), which is many orders
of magnitude larger than that of an earthquake.
Recently, enough data has been collected to extract statistics on individual systems
of earthquake faults (Dahmen et al. 1998), and it was found that the distribution of
earthquake magnitudes may vary substantially from one fault system to another and
for different Earth regions.
Regarding the exponent ‘b’ of the power-law distribution, it was previously claimed to
be universal and close to 1 (Gutenberg and Richter 1954). In Molchan et al. (1997), the re-
lated b-values (for many earthquakes in Italy calculated for the period 1900–1993) are in
the range 0.7–1.35. Elgazzar (1998) estimated the b-value to be 0.85 ±0.2. Nanjo et al. (1998)
give b to be between 0.5 and 1.5. It has recently been claimed that ‘b’ fluctuates in time and
depends on the earthquake magnitude (Olami et al. 1992; Carlson et al. 1991).
In recent years, numerous investigators tried to explain b theoretically. For example,
“self-organized criticality” models were proposed (Bak and Tang 1989; Ito and
Matsuzaki 1990; Matsuzaki and Takayasu 1991; Olami et al. 1992) to explain this fractal
law as being a result of the extremal nature of the dynamic rules governing the sys-
tem. There are also other deterministic models (Carlson et al. 1991; Scott 1996) de-
scribing earthquake dynamics by the friction and elastic forces acting in the fault zone.
For example, the one-dimensional Burridge-Knopoff model (Carlson and Langer 1989)
leads to the Gutenberg-Richter law with b ≈ 1.
Anton (1999) proposed an analytic model based on a relation between stress release
rate and kinetic energy loss by seismic waves and noted that the relation b < 1 implied
that the stress release rate was greater than the loss of kinetic energy by seismic waves.
 5.9 · Global Features of EMR 435

Gabrielov et al. (2000a,b) developed a “colliding cascades model” consisting of a hi-

erarchical structure. They proposed that an external load applied to the largest block
is transferred hierarchically to the smallest components. Fracture processes expand
in an inverse manner. The two processes “collide” and interact. On the basis of this
model, they argued that the b-value should be 0.53 for main earthquakes and 0.69 for
aftershocks. Nanjo et al. (1998), Kossobokov et al. (2000), Elgazzar (1998), Anton (1999),
and Gabrielov et al. (2000a,b) for the distribution of EQ magnitudes in that the b value
is of the order of 2/3.

5.9.1.2
Comparison of Fractal Behavior of EMR and EQ

As we noted above, amplitude changes of acoustic emission from fractures during labora-
tory compression tests of brittle materials were considered by Lockner (1993) and Lockner
et al. (1991) to be similar to those occurring in EQ. This kind of similarity can also be gleaned
from a comparison between energy release during starquakes (Fig. 5.53a, Kossobokov et al.
2000) and our measurements of amplitude of EMR induced by chalk compression failure
(Fig. 5.53b). In this section this similarity is investigated, focusing on the possibility of a
power law for EMR induced during conventional uniaxial and triaxial fracture.
Figure 5.54 shows the cumulative number (the number of signals having amplitudes
larger than E) of EMR signals (measured additively from twenty-four chalk samples
during uniaxial and triaxial conventional tests) vs. _ EMR pulse amplitude.
Since the voltage output of the EMR pulses ‘A’ depends on the antenna reaction
(antenna efficiency), which changes with frequency, it was compensated (Sect. 5.3.3).
We were thus able to compare heights of EMR signals of different frequencies.
EMR pulse amplitudes E changed by 5 orders of magnitude (Fig. 5.54): From
0.001 V m–1 to 100 V m–1. The figure consists of four parts. Its main part is consistent
with a Gutenberg-Richter type law with a b-value of 0.62 (R2 = 0.95).

Fig. 5.53.
A qualitative comparison be-
tween energy changes during
starquakes (a after Kossobokov
et al. 2000) and b amplitude
changes of electromagnetic ra-
diation induced by rock com-
pression (Rabinovitch et al.
2002b)
436 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.54.
The number of signals having
amplitudes larger than E of
EMR signals (measured addi-
tively from twenty-four chalk
samples during uniaxial and
triaxial conventional tests)
vs. EMR pulse amplitude
(Rabinovitch et al. 2002b)

(5.18)

Note that this b-value is close to 2 / 3 (see above). On both ends of the graph there
is a deviation from the Gutenberg-Richter type law. The b-value for the small ampli-
tude range is very low (0.02, R2 = 0.76). This small value might be related to either an
incomplete sampling of small events, to noise or to a new physical effect governing
the process. In the range of large amplitudes 1 < A < 10 V m–1, the b-value is also
significantly lower (0.08, R2 = 0.76) than in the main part, which might possibly be
due to the finite size of the sample. Note that the EMR amplitude in the range of
1 < A < 10 V m–1 is measured when the external stress is higher than the sample’s elas-
tic limit (Fig. 5.53b). Figure 5.54 has also a fourth part (A > 10 V m–1) with b = 2.3
(R2 = 0.77). This is the range immediately before sample collapse (Fig. 5.53b).
Several investigators have noted this slope change effect for seismic data as values
in the smallest and largest ranges in their investigations also deviated from the
Gutenberg-Richter type law. For example, Lockner (1993) and Lockner et al. (1991)
showed a roll-off in the distribution of AE signals at low amplitudes and explained it
as incomplete sampling of small events (Lockner 1993; Lockner et al. 1991). Kossobokov
et al. (2000) noted this change of slope effect in the high range of earthquakes and
explained it as being due to the maximum energy release being limited by the size of
the crust and by the energy density. They also noted that for the largest earthquakes
the downward slope may altogether disappear or even turn into an upward slope.
A change of slope effect was also observed by Molchan et al. (1997) for induced
seismicity both in the range of small and of large earthquakes.
The decrease of the b-value with stress, as happens here between 1 and 10 V m–1
(Fig. 5.54), is also a known effect (e.g., Lockner et al. 1991).
Figure 5.53 exhibits the escalation of the fracture process before collapse.
Figure 5.53a (Kossobokov et al. 2000) shows the energy of starquakes vs. time, which
occurred at a distance of about 40 000 light years from Earth. Figure 5.53b gives an
EMR amplitude-stress-time graph of all chalk samples. Stress values (Fig. 5.53b) were
normalized by the peak stress value. Visual comparison indicates that the two graphs
are very similar.
 5.9 · Global Features of EMR 437

5.9.2
Benioff Strain Release

Fracturing processes can be measured either by intensity changes (see, e.g., Fig. 5.54)
or by their energy release. The latter is usually represented by the so-called “cumula-
tive Benioff strain release”(Newman et al. 1995; Kossobokov et al. 2000; Gabrielov et al.
2000a,b), which relates the total sum of the square root of the energy released for se-
quential fracture events to the time prior to the collapse failure. Hence, in addition to
the Gutenberg-Richter purely statistical law, the “cumulative Benioff strain release”
relation enables us to monitor the continuous development of the upscaling fracture
process through time.
The measured EMR amplitude is proportional to the magnetic field intensity (H)
reaching the antenna. Since the energy of the electromagnetic field is propor-
tional (Pointing vector) to ⏐H⏐2, the EMR amplitude is proportional to the square
root of the energy. Figure 5.55 shows a bilogarithmic graph of a cumulative ampli-
tude of all EMR signals registered during all samples’ compression vs. time before
their collapse failure. Since maximal values of EMR amplitude are excited by
the samples’ collapse failure, their occurrence time was taken as the zero time for
each sample.
The graph (Fig. 5.55) shows an almost constant slope in its middle part (about a
whole decade). It was fit to a power law (R2 = 0.987 and slope = 1.42) and is seen to be
accompanied by some logarithmic-periodic variations. This graph is very similar to
the usual “Benioff strain release” ones (e.g., Kossobokov et al. 2000) and the slope of
its main part is close to the usual Benioff slope of 1.35 (Newman et al. 1995).
Results obtained here show that both the b value of the Gutenberg-Richter type law
and the slope of the Benioff strain release relationship of EMR signals obtained dur-
ing chalk fracture in the laboratory are similar to those measured in earthquakes and
starquakes. The qualitative similarities and even more so the almost exact power laws
are striking. The fractal nature of the processes controlling earthquakes and starquakes
can therefore be extended to a microscale regime. This “global” nature of multiple frac-
ture effects evidently implies that a basic general process is “acting behind” all these
phenomena.

Fig. 5.55.
The “Benioff strain release”
relationship of a cumulative
amplitude of all EMR signals
registered during all samples’
compression vs. time before
their collapse failure (time goes
backwards from the moment
of collapse) (Rabinovitch et al.
2002b)
438 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

5.10
EMR Induced by Underground Rocks under Stress
5.10.1
General Background

Nesbitt and Austin (1988) registered EMR in a gold mine (2.5 km depth). An EMR sig-
nal (1.2 mA m–1 amplitude) was generated some seconds prior to the micro seismic event
(magnitude of –0.4). Registration of EMR activity in Ural bauxite mines showed
(Scitovich and Lazarevich 1985) that its value sharply increased with rock burst (RB)
hazard increase. Analogous works in Noril’sk polymetal deposit (Krasnoyarsk region)
revealed an increase of EMR amplitude (up to 150–200 mV m–1) and activity in the RB
hazardous zones (Red’kin et al. 1985). Markov and Ipatov (1986) investigated EMR activ-
ity changes in an apatite underground mine (Khibin deposit, Kola Peninsula) and found
out that EMR amplitude in RB hazardous zones was in the range of 8–25 mV m–1 and
EMR activity there was significantly higher than the regular noise level.
A survey of this general background shows that the EMR investigations in mines
are rare and non-systematic. We conducted careful investigations in underground
mines in different mining and geological situations associated with an increase of RB
and rock-gas outburst hazard. The main aim of these investigations was to correlate
between specific EMR patterns and an increase of RB and outburst hazard in under-
ground mines and hence to develop EMR criteria for rock bursts and rock-gas out-
burst. We hope that a good understanding of these relations would create new avenues
for successful “large-scale” EMR investigations of EMR pulses prior to ensuing earth-
quakes-enabling earthquake monitoring.

5.10.2
EMR in Mines

5.10.2.1
Criterion of Rock-Burst Hazard in Coal Mine and Short Description of the “Gum” Method

RB hazard increases if the load on a given part of a coal seam exceeds some critical
level, while the distance from this part to the stress maximum (in the zone of influ-
ence of a mine-working) is lower than a critical value (Petukhov and Lin’kov 1983).
Note that we use by the term “mine-working” for all underground openings irrespec-
tive of their size and purpose. If the potential energy accumulated in the zone of high
stress is enough to destroy rocks near a mine-working, RB takes place.
As shown by Petukhov and Lin’kov (1983), the RB safety condition is the following:

(5.19)

where Pn is the external applied load, Pm is the limiting load (strength), Mc is the coal drop
modulus (descending modulus of the stress-strain curve beyond the peak point under rigid
compression, Fig. 5.56), Er is the Young modulus of adjoining rocks, an is the distance from
 5.10 · EMR Induced by Underground Rocks under Stress 439

Fig. 5.56.
Typical stress-strain curve
during stiff sample compres-
sion (Frid 2000)

the mine-working face up to the position where there exists a peak stress, and h is the
half thickness of the coal stratum. If coal is under residual strength (non-hazardous
condition), both inequalities comply with the following: Load on the given zone is lower
than the limiting value (first inequality) and the left side of the second inequality is smaller
than 1 (the drop modulus in the residual strength zone is equal to 0).
The RB hazard is usually determined by some standard geomechanical method, e.g.,
volume measurement of gum (dry coal “powder”), measurement of the diameter of
the hole created or the number of disks that are created due to core fracturing as a
result of drilling in a highly stressed zone, etc. (Petukhov 1972).
Drilling of a highly stressed coal seam leads to an intensive fracturing process in the
zone influenced by the drill hole. The volume of this highly cracked zone depends on the
hole diameter, the drilling rate and especially, the stress level. Hence, if the first two pa-
rameters remain invariant for a given coal seam, the stress value in the coal seam is re-
sponsible for the volume of gum that is recovered from the hole (i.e., from the highly stressed
zone drilled by the hole). This method is widely used for RB prediction, although it is time
consuming and in some conditions even dangerous.
The normalized ratio, β, of the highly stressed zone (ratio of the non-elastic defor-
mation zone, diameter D, to the hole diameter d = 0.043 m, Fig. 5.57, Frid 2001) can
be calculated from the following formula (Petukhov 1972) is given by:

(5.20)

where nr is the coefficient of coal loosening on the borehole wall that is usually equal
to 1.3–1.4, M0 is the gum volume in the borehole (M0 = Λπd2 / 4, Λ is the borehole
440 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.57.
The zone of non-elastic defor-
mation excited by drilling in
the high stressed zone (β is the
non-dimensioned diameter of
this zone: The ratio of the non-
elastic deformation zone dia-
meter D to the hole diameter
d = 0.043 m) (Frid 2001)

Table 5.4.
Calculation of rock-burst
hazardous zone parameters

length), and Ms is the gum volume induced by drilling in a stressed zone) (Petukhov
et al. 1976).
The vertical stress in the coal seam can be determined as follows (Petukhov et al.
1976):

(5.21)

where k* is the coal shear strength. Forecast boreholes are usually drilled in succes-
sive depths (the length of each beeing 1 m). Hence, we can estimate the vertical stress
distribution in the coal seam near the mine-working face.
After the drilling of each interval, the gum volume is measured and if it exceeds a
definite limiting value (the experience at the mining works in North Kuzbass shows that
the limiting values are 5 to 8 liters per meter at the fourth and the seventh drilling meter
from the drift face, respectively, (Table 5.4, Frid 2001)), drilling is stopped, and that part
of the mine-working is considered RB hazardous.
As noted by Lockner (1993, 1996), there is a strong parallelism between the well-known
Gutenberg-Richter relation for seismic events (from macro (earthquake) to micro (RB))
and power law frequency-magnitude relationship for acoustic emission events. This anal-
ogy suggests that micro shocks (high frequency and small magnitude) are precursors of
macro failure (large magnitude and small frequency). This is the theoretical basis for RB
forecasting by the acoustic emission method (Kuksenko et al. 1982; Mansurov 1994). The
EMR frequency range is close enough to the acoustic emission band. Moreover, Gutenberg-
Richter type relationships similar to the seismic ones were also observed for EMR (Rabino-
vitch et al. 2002b, Sect. 5.9.12) Hence, it would be correct to assume that EMR could be useful
for RB hazard forecasting along with acoustic emission. Moreover, being non-contact, the
EMR method has some advantages over acoustic emission. For example, when a rapid and
comprehensive prognosis of a short-term mine-working region (for example, in a drift
face) is needed, the roughness of the mine walls becomes a marked problem for the acoustic
emission method for rapid data acquisition due to inferior contact between the acoustic
emission transducer and the mine wall.
 5.10 · EMR Induced by Underground Rocks under Stress 441

5.10.2.2
EMR Measurements in Coal Mines

EMR pulses excited by rock fracture show up as a “bell-shaped” oscillation charac-

terized by amplitude, a frequency and an individual rise and fall time (Rabinovitch et al.
1995, 1996, Sect. 5.2). As a rule, EMR is measured in mines within the narrow band in
the 30–150 kHz range. The middle part of this range (100 kHz) is the most convenient
one in coal underground mines (Frid et al. 1992). This range is conditioned both by the
industrial electromagnetic noise level in underground mines and the electromagnetic
pulses’ amplitude. The Electromagnetic pulses’ field amplitudes excited by coal failure
near the face of a mine-working range are between 5 and 50 µV m–1 (Frid et al. 1992)
that is at least one order of magnitude larger than mine industrial noise levels. An in-
dustrial noise level at the Earth’s surface is usually not lower than 5 mV m–1. Hence, an
industrial electromagnetic noise in coal mines springs from underground mining
machines, but not from the Earth’s surface. The daily work cycle of a mine consists of
four six-hour shifts: three extraction (mining) ones and one repair shift. The prob-
ability to pick up an electromagnetic transient process pulse due to the starting/stop-
ping of electrical mining machines during extraction shifts is large enough, while the
electrical noise level during the repair shift is as noted above. Thus, we carried out
our measurement during the repair shift. As shown by Frid et al. (1992), EMR regis-
tration near a face of coal drift enables us to obtain EMR signals excited by a coal fail-
ure up to a depth of 10 m from the drift face. The daily drift moving in North Kuzbass
conditions was usually of about 6 m. Hence, a measurement of EMR during the repair
shift can characterize at least the daily activity at a coal drift face.
The EMR was measured by a resonance 100 ±1 kHz antenna that was located 1 m
from the mine-working face and was connected to a counter. We defined the EMR
activity as the number of crossings of the EMR voltage signals (per unit time) through
a given voltage level of a special counter (Fig. 5.58, Frid 2001). At all studied points
30 EMR activity readings were taken, each with a duration of 10 s. For simplicity, we
named units of EMR activity “pulse per 10 seconds”, and they were symbolized by

Fig. 5.58. The EMR signal that intersects counter voltage level (Frid 2001)
442 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

“pulse/10 s”. A complete cycle of EMR measurements at a single drift face is of 5 min
duration. This method allowed us to determine a stable mean that characterized EMR
activity at a given point.
Our preliminary mine estimation of EMR compatibility conditions showed that
choosing the given resonance frequency would allow us to accurately extract the use-
ful signal from the industrial background noise. As is known, the use of the shortest
wavelength band of the measuring system is a necessary condition to increase the
system’s sensitivity. The Q factor in our case is 50, resulting in a system sensitivity not
lower than 0.1 mV m–1.

5.10.2.3
Mine Investigations prior to Rock Bursts

An increase of RB hazard is usually associated with the existence of pillars, geological

faults, sharp changes of inclination angle or thickness of coal seams (Petukhov 1972).
We conducted our investigations in coal mines of the North Kuzbass, which exhibited
all of these conditions.
An example of EMR registration at the zone of pillar influence is shown in Fig. 5.59
(Frid 1997a) The study was conducted in a mine-working where the influence of pil-
lars was found at a distance of 70 m from the conveyer drift 13. Profiling of the mine-
working (in spacing of 10 m) showed an anomalous EMR activity at stressed regions
under pillars.
A second example of EMR registration at a zone of pillar influence is shown in
Fig. 5.60 (Frid 2001). The EMR sequential measurements (profiling with 3 m spacing)
were carried out in an extraction gallery that partly drifted into the zone of pillar in-
fluence. As can be seen, EMR activity is anomalous under the pillar (120–140 pulse/10 s).
Gum value measurements showed the existence of a RB hazard in this region (gum
value was about 7–12 l m–1).
A third example of EMR registration is shown in Fig. 5.61 (Frid 2001). In this case,
drift was in a zone of an intensively variable thickness of coal seam and a sharp change-

Fig. 5.59.
EMR profiling under pillar
(Frid 1997a)
 5.10 · EMR Induced by Underground Rocks under Stress 443

Fig. 5.60. EMR profiling at the zone of pillar influence. The error in EMR activity is of the order
of ±7.5% (Frid 2001)

Fig. 5.61. EMR monitoring during drift excavation in an intensively changeable thickness of coal
seam. The error in EMR activity is of the order of ±7.5%, while in gum value is of the order of ±0.5 l m–1
(Frid 2001)
444 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.62. Anomalous EMR activity and gum value in the mine excavation near the uplift. The error
in EMR activity is of the order of ±7.5%, while in gum value is of the order of ±0.5 l m–1 (Frid 2001)

Fig. 5.63. EMR monitoring during pillar unloading. The error in EMR activity is of the order of ±7.5%,
while in gum value is of the order of ±0.5 l m–1 (Frid 2001)
 5.10 · EMR Induced by Underground Rocks under Stress 445

able inclination angle. The regular thickness of the coal stratum was 1.5–1.7 m. A sharp
decrease of the thickness of the coal stratum was associated with a RB hazard increase
(a decrease of coal stratum thickness induces a stress maximum nearer to the drift
face and hence a RB hazard increase). For example, on 7 and 26 January and on 4 March,
the coal stratum thickness decreased to 0.5–0.7 m, with a corresponding gum volume
and EMR activity increase of up to 7–10 l m–1 and 150–400 pulse/10 s respectively.
Figure 5.62 (Frid 2001) shows an example of an anomalous EMR activity and
gum value in a mine excavation that was opened at a distance of 8 m from an uplift
with a 2 m amplitude. As the excavation moved away from the uplift, the EMR and the
gum monitoring showed gradual decrease. Unloading of RB dangerous rocks near
mine-workings is usually performed by the following methods: Hydroprocessing,
drilling of boreholes of large diameter and blasting. We carried out EMR activity in-
vestigations both before and after these unloading operations. If these operations were
successful, i.e., mine-working became safe, EMR activity should have decreased to its
usual level.
Figure 5.63 (Frid 2001) shows an example of an EMR registration during pillar
unloading. The average level of EMR activity during the period of 21 October to 16 No-
vember was 130–150 pulse/10 s. Measurement of the gum volume showed that this pil-
lar was RB hazardous (gum volume was 8–10 l m–1). Unloading was performed by drill-
ing 100 mm diameter holes. The holes enabled us to decrease the pillar brittle response.
Drilling began at 17 November. This led to pillar unloading, and both gum volume and
EMR activity gradually decreased.
Figure 5.64 (Frid 1997a) shows an example of EMR profiling in a drift that was
initially RB dangerous along all of its length after it was unloaded. Unloading was
conducted by hydroprocessing that includes the drilling of a borehole of 6 m length
at intervals of 30–50 m followed by water pumping into the rock. Hydroprocessing
leads to controlled rock cracking and removes the zone of increased stress from
the mine-working wall deeper into the rock-mass. Figure 5.64 (Frid 1997a) shows an
example of EMR measurement when one part of drift (from 0 to 70 m) was previously
hydroprocessed and became safe (gum value was lower than 5.5 l m–1). The average
level of EMR activity at this region was not more than 40 pulse/10 s. However, the un-
processed drift zone was very hazardous (gum value of up to 8–9 l m–1), a fact that
was reflected by a high level of EMR activity (of up to 250 pulse/10 s).

Fig. 5.64.
EMR registration by drift pro-
filing (Frid 1997a)
446 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

We carried out EMR investigations in more than 400 different RB situations in

underground coal mines (North Kuzbass deposit) (Frid 1990; Frid et al. 1992; Frid
1997a,b). Our investigations showed that an increase of RB hazard, accompanied by
an increase of EMR activity, occurred when pillars were found above mine-working,
where geological faults existed in the area, and/or where seam thickness sharply di-
minished (Frid 1997b). These investigations also showed that the highly stressed zone
near mine-workings is the source of the EMR (Frid 1990, 1999). An analysis of all the
data enabled us to determine the following empirical criterion of RB hazard:

(5.22)

where N is the value of EMR activity registered at a mine-working, and Nemp is the lim-
iting value of EMR activity (115–120 pulse/10 s, Frid 1997b). If the value of the EMR
activity, N, exceeds Nemp, a RB hazard existed.
The amount of time needed to determine the degree of a RB hazard by the EMR
and the gum methods at a given point is about 5 and 30 min, respectively. The reduced
time is a significant advantage of EMR forecasting. However, the amount of time for
data collection to establish an empirical Nemp in a specific mine is much longer, even
up to several years. The only way to reduce this time is to develop a theoretical model
of this criterion. The following section establishes an empirical EMR criterion of RB
hazard and compares it with the theoretical criterion.

5.10.2.4
Theoretical EMR Criterion for Rock-Burst Hazard in a Coal Mine

Zone of EMR formation under external stress. Goncharov et al. (1980) carried out an
EMR study with a stiff press. They loaded concrete samples of 0.55 × 0.55 × 0.65 m
dimensions under compression. The EMR activity was measured with a magnetic
antenna that had a resonance frequency of 800 kHz (±1 kHz). The first significant
increase of EMR activity occurred at the post-peak zone under a load of 0.95 σc,
where σc is the limit of strength, and a second rise of EMR activity was noted for
a load of 0.9 σc (in the post-peak zone). Single pulses of EMR accompanied the sub-
sequent decrease of the load until 0.75 σc. In the zone of residual strength, EMR activ-
ity ceased.
EMR studies with a stiff press were also performed on ores and rocks from the
Khibin apatite deposit (Kola Peninsula, Ipatov 1989), on magnetic ores, granites, white
marble and coal from Shpitsbergen (Frid 1990). EMR activity was registered with a
100 kHz (±1 kHz) resonance magnetic antenna. The maximum EMR activity was
measured in the range of (1.0–0.8) σc in the post-peak zone, while in the zone of re-
sidual strength EMR activity stopped. Note that wide frequency band EMR measure-
ments (Rabinovitch et al. 2000c) confirmed these results. An analysis showed that the
following EMR features were common to all rocks:

1. A sharp increase of EMR activity near the peak stress.

2. An absence of EMR in the zone of residual strength.
 5.10 · EMR Induced by Underground Rocks under Stress 447

Fig. 5.65.
Typical stress distribution curve
in the zone of mine-working
influence (Frid 2000)

To check these conclusions in situ we investigated the EMR activity during bore-
hole drilling (hole diameter was 42 mm). Note that the residual strength zone (the last
zone during sample deformation on a stiff press, Fig. 5.56) is the nearest zone to the
mine-working face (Fig. 5.65).
All holes were drilled at 1 m separation intervals. During all interval drilling, a cu-
mulative value of EMR activity (summarized value of EMR activity per drilling inter-
val) was measured, and after drilling each interval the gum value was measured. Fore-
cast drilling in the mine does not generally achieve stress peak due to instrument “grip-
ping” in a borehole, and stops if gum volume (stress value) becomes higher than a
predetermined value. Forecast drilling (in North Kuzbass mines) is generally conducted
up to 6 m depth. Figure 5.66 shows two examples of EMR observation during bore-
hole drilling:

1. In the RB hazardous zone (gum value is higher than the limiting value, see below).
2. In the non-hazardous zone (gum value is lower than the limiting value, see below).

As is seen from Fig. 5.66a, gum value increases up to 12 l m–1 in the 5th drilling meter.
This value is significantly higher than the limiting value (5.5–6 l m–1), and hence the
forecast drilling was stopped. It evidently indicates the existence of a RB hazardous
zone at the 5th meter depth from the mine-working face. Figure 5.66a shows that when
a hole face approaches a stress peak (gum increase above the limiting value), EMR ac-
tivity sharply increases, and vice versa: If the gum value doesn’t change drastically (gum
value lower than 5.5–6 l m–1 at the 5–6 drilling meter non-hazardous zone, Fig. 5.66b),
EMR activity does not essentially increase.
448 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.66.
Two examples of EMR study
during forecast drilling: a in the
rock-burst hazardous zone, and
b in the non-hazardous zone.
Full-drawn curve is the Gum
value, while the hatched curve is
the EMR activity (Frid 2000)

The analysis of more than 150 drillings conducted on three different coal stratums
(Desyatiy, Koksoviy and Andreevskiy) enables us to summarize some general features
of EMR activity associated with borehole drilling:

1. The EMR activity sharply increases if the hole face approaches a zone of highly
stressed coal (that is near the stress peak, e.g., Fig. 5.66a).
2. Drilling in a non-hazardous zone does not tangibly affect EMR activity (e.g.,
Fig. 5.66b).
3. The EMR activity value induced by drilling is lowest nearest to the mine-working
face, i.e., in the zone of crushed coal or residual strength (see, e.g., first drilling
meter from the mine-working face, in Fig. 5.66).

Calculation of rock-burst hazard zone volume. Drilling for gum volume measure-
ments in a mine is usually conducted up to a depth of increased stress σkγ H, but stress
maximum is usually not measured due to instrument gripping (as a result of the hole
squeezing around the drill rods) in the borehole. However, for the theoretical EMR
analysis it is important to know the volume of the highly stressed zone, and hence the
distance from the mine-working face up to the stress maximum.
 5.10 · EMR Induced by Underground Rocks under Stress 449

Fig. 5.67.
The vertical stress distribution
in the zone of influence of the
mine-working (all parameters
are discussed in the text)
(Frid 2001)

The distance to the stress maximum (Fig. 5.67, Frid 2001) can be estimated as fol-
lows (Petukhov et al. 1976):

(5.23)

where X0 is the mine-working half width, a' is the distance from the mine-working face
to a boundary of the zone of increased stress (σkγ H, Fig. 5.67), h is the coal stratum half
thickness, and σc is the coal uniaxial compressive strength (11 MPa; Frid 1990). If we
assume that the drift width, 2X0 = 3 m (typical for vent and tramming drifts in North
Kuzbass mines) and the coal stratum thickness 2h = 2 m (also a representative value), we
can calculate the distance to the stress maximum by Eq. 523. Table 5.4 shows that for all
RB hazardous conditions (4–7 m from the mine-working face, Frid 1997a) the stress maxi-
mum is about ξ = 1 m from the commencement of the stressed zone (σkγ H zone, Fig. 5.67).
Let us assume that the RB hazard zone before coal drifting has a parallelepiped form
with a width equal to the drift width (2X0 = 3 m), a height equal to the coal stratum thick-
ness (2h = 2 m), and a length equal to the distance from the origin of the σkγ H stress zone
to the stress maximum (ξ = 1 m). Hence, the volume of the RB hazard zone would be 6 m3.
A formation of a 6 m3 stressed zone cannot be developed instantly. As noted by
Kuksenko et al. (1982, 1987), the time to form a highly stressed zone is related to its
dimensions. This relation is valid for laboratory fracture experiments, mine pillar load-
ing, RBs and even for earthquakes (Kuksenko et al. 1982; Mansurov 1994) and can be
approximated by the following formula:

(5.24)

where T is the time (s) to form the stressed zone, and L is an average dimension of
the stressed zone (m).
450 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

If the volume of the stressed zone is 6 m3, its average dimension, L, would be about 2.5 m,
and according to Eq. 5.24, the time to develop this zone will be approximately 4 hours.

EMR source size estimation. Assuming that the cracking rate n (number of cracks
per unit time) is nearly constant during the whole period of the highly stressed zone
formation, then

(5.25)

where N' is the total number of cracks created in the volume V.

The crack length is associated with the time from the EMR pulse origin up to its
envelope maximum (Rabinovitch et al. 1998, 1999b, Eq. 5.2). For 100 kHz pulses (Frid
1997a,b), this time is about T ' = 10–15 µs (Frid 1990; Rabinovitch et al. 1998).
As noted by Marder (1996a), crack velocity does not exceed the Rayleigh wave
velocity vR. For coal, vR is of the order of 1 100 m s–1 (Frid 1990). Hence, the length of
the crack, l, that radiates a 100 kHz EMR pulse would be

l = vRT' (5.26)

Estimation of EMR activity in the hazardous zone. As follows from the kinetic theory
of failure (Zurkov et al. 1969; Regel’ et al. 1972; Petrov and Gorobetz 1987), the maxi-
mum number of cracks of length l in a volume V is limited by the following relation:

(5.27)

where k is about 3, the concentration factor. N' ranges between 45 000 and 170 000 and
its average value is 107 500. Hence, the limiting cracking rate (according to Eq. 5.25)
will be nlim = 7.7 events per second. Hence, a RB hazard situation is predicted when
the cracking rate, n, is larger than this limit value:

(5.28)

The final step is to associate this criterion of cracking rate with a specific param-
eter of the EMR. There are numerous EMR parameters that could be measured in a
mine, for example, EMR activity (see Sect. 5.10.2.2), the number of EMR events (often
registered as the number of EMR peaks) per unit time, or the energy of EMR pulses
per unit time determined, for instance, by the summation of pulse amplitude squared
multiplied by pulse duration, per unit time, etc. All of these parameters characterize
only the EMR rate (number of EMR events per unit time), and can specify only the
cracking rate.
As noted above, the EMR activity is the number of crossings (per unit time) of the
EMR signal amplitude a specific voltage level of the counter. The number of crossings
formally depends on the EMR amplitude, i.e., the higher the EMR pulse amplitude,
the larger is the number of intersections with a defined voltage threshold. However,
 5.10 · EMR Induced by Underground Rocks under Stress 451

our experience shows that the EMR amplitude is relatively stable (5–50 mV m–1) and
does not deviate by more than one order of magnitude from its average level. This
phenomenon may be explained in the following manner: EMR amplitude is a func-
tion of crack area (Rabinovitch et al. 1998, 1999b) and this implies that this amplitude
is inversely proportional to the EMR frequency (Rabinovitch et al. 1999b). Hence, if
we select a specific frequency range to measure EMR activity (i.e., a definite crack
width), we limit the range of EMR amplitudes that will be registered (the EMR ampli-
tude span observed in laboratory conditions by Rabinovitch et al. (1999b) is also not
more than one order of magnitude for a given specific frequency).
The amplitude of EMR pulses quickly decreases after 2–3 oscillations (Fig. 5.58, Frid
2001). To differentiate the EMR data from mine industrial noise, we preset the voltage
level of our counter to register only the maximal half-wave of the signal (Fig. 5.58). The
maximal half-wave therefore intersects a set voltage level twice (Fig. 5.58), and the value
of the EMR activity (especially the limiting value) is twice as high as the cracking rate:

(5.29)

Hence, the theoretical criterion of RB hazard is the following: A RB hazard situa-

tion is predicted if the EMR activity, N, is larger than Nth = 157 pulse/10 s:

(5.30)

As seen, the calculated Nth=157 pulse/10 s and the empirical Nemp=115–120 pulse/10 s
criteria (see above following Eq. 5.22) are close enough.
Generalizing formulas (5.22, 5.28–5.30), we can formulate a theoretical EMR crite-
rion via the theoretical value of cracking rate for a given mine situation:

(5.31)

where N is the measured EMR parameter in a mine-working, Nth is the theoretical lim-
iting value of this EMR parameter, k is a coefficient that depends on the type of pa-
rameter (e.g., k for activity equals 2, but if we measure the number of EMR peaks, k = 1),
and nlim is the limiting cracking rate.
Our theoretical approach based on a minor number of parameters (limiting value
of gum volume, typical mine-working width, coal seam thickness, and coal uniaxial
compressive strength) enables us to estimate a value for EMR criterion that must be
verified by mine investigations. Its adoption would significantly reduce the time to
use the EMR criterion.

5.10.2.5
Precursors to Rock-Gas Outburst

Another dangerous phenomenon in mines is the “rock and gas outburst” in which a
burst is assisted by liberated gas (CH4 in coal mines, N2 in rock salt mines, etc.). The
gas is liberated by the rock fracturing and acts (like pore pressure) to decrease the
critical stress level needed for destruction of the mine-working walls (Petukhov and
Lin’kov 1983).
452 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

At present the standard geo-mechanical method of rock-gas outburst prognosis in

coal mines consists of measurement of volumes of gum and gas liberation rate
(Petukhov and Lin’kov 1983). Forecast boreholes are usually being drilled at intervals
of 1 m. After drilling of each interval, the volume of gum and the quantity of gas lib-
eration are measured. If both of these parameters exceed a definite limiting values,
drilling stops, and the given part of the mine-working is considered to be hazardous
to rock-gas outburst. An experience of mining works in North Kuzbass shows that a
formation of rock-gas outburst hazard zones is reliably fixed at the quantity of gum
of more than 5–6 l m–1 (liter per meter) and a gas liberation rate value of more than
5 l min–1 (liter per minute) at the fourth to sixth intervals (depths) of the forecast bore-
hole (from the fourth to the sixth meter from a mine-working face). This method is
widely used, although sometimes bore-holing can itself initiate rock-gas outburst.

EMR observations in coal mines. After registrations of natural EMR activity, the de-
gree of rock-gas outburst hazard at that point of mine-working was detected by the
standard method of measurement of the volumes of gum and gas liberation rate.
The first example of the EMR registration is shown in Fig. 5.68 (Frid 1997b). In this
case, drift was conducted in a zone of an intensively variable thickness of coal seam and a
sharp changeable inclination angle. Figure 5.68 shows that a decrease of seam thickness
from 2.6 m (18 April) to 1.1 m (from 17 May up to 30 May) corresponded to rock-gas out-
burst hazard increase (an increase of gum value of up to 6 l m–1 and gas liberation rate
quantity of up to 7–12 l min–1) with a simultaneous increase of the natural EMR activity
level. A restoration of coal seam thickness to its original value (15 June–29 June) caused a
decrease of gum and gas liberation rate parameters to their regular levels and a decrease
of EMR activity. An increase of EMR activity was also obtained during a drift conducted
through an inclination angle raise (of up to 60°) of the coal seam (5 April, Fig. 5.68).
An example of EMR registration in a zone of geological fault influence is shown in
Fig. 5.69 (Frid 1997b). At a distance of 8 m from a geological fault, an anomalous EMR
activity was registered (20 October). Controlled boreholing showed that the value of
gum was of the order of up to 6 l m–1 and the gas liberation rate was of the order of up
to 16 l m–1. A further conduction of this drift left the fault influence zone, and EMR
activity returned to normal (from 16 May up to 28 October).
EMR monitoring was also conducted in a mine-working where the influence of
pillars stretched along 150 m of a drift. Profiling of the mine-working showed an EMR
activity at the stressed region under the pillar where the values of gum and gas lib-
eration rate were also irregular: More than 7 l m–1 and 8 l min–1 respectively.

Fig. 5.68.
EMR monitoring in mine-work-
ing. Note that GLR on this figure
is gas liberation rate (
Gum
volume,
GLR, × seam thick-
ness,  seam dip,
EMR)
(Frid 1997b)
 5.10 · EMR Induced by Underground Rocks under Stress 453

Fig. 5.69. EMR monitoring in drift (
Gum volume,
GLR,
EMR) (Frid 1997b)

Fig. 5.70. EMR activity vs. Gum volume. Experimental EMR data are marked by
(Frid 1997b)

Electromagnetic radiation forecast criterion. We have conducted more than fifty in-
vestigations in rock-gas outburst dangerous and safe mine situations (Frid 1997b).
Figure 5.70 depicts a scatter of EMR activity data vs. gum values, while Fig. 5.71 shows
EMR activity vs. gas liberation rate values. Significant data dispersion is seen in both
cases, yielding regression coefficients in both cases of not more than 0.7. However,
both show a monotonic increase of EMR activity with a rise in both gum and gas
liberation rate. We hypothesized that the low correlation coefficients imply that the
value of EMR activity depends on both parameters together rather than on a single
one. To examine this hypothesis, we drew all experimental data (fifty-eight experi-
mental points) in the form of EMR activity vs. both gum volume and gas liberation
454 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

Fig. 5.71.
EMR activity vs. gas liberation
rate. Experimental EMR data
are marked by
(Frid 1997b)

Fig. 5.72. EMR activity surface vs. Gum (P) and gas liberation rate (Q) (Frid 1997b)

rate quantities (Fig. 5.72). A fit of this surface calculated by a least square method is
the following:

(5.32)

where N is the EMR activity, while P and Q are gum and gas liberation rate quantities,
respectively. A comparison of the experimental and the calculated data in Eq. 5.35
shows a very good correlation (Fig. 5.73). The correlation coefficient is 0.908. Thus,
Eq. 5.32 reflects the existence of a relationship between the EMR activity of a coal seam
and the gum and gas liberation rate quantities. Hence, an increase of EMR activity
can indicate rock-gas outburst hazard development.
We used this equation for an EMR criterion forecast. As was noted above, rock-gas
outburst hazard zones are correlated with a quantity of gum of more than 5–6 l m–1
and a gas liberation rate of more than 5 l min–1. The related EMR quantity by Eq. 5.32
is 67–73 pulse/10 s. Statistical parameters of gum, gas liberation rate and EMR activ-
ity data are shown in Table 5.5. Analysis of this table shows that the EMR criterion yields
mean and median values that are very close to the experimental data. We also carried
out a correlation analysis of rock-gas outburst hazard estimation by the gum-gas lib-
eration rate method and by the proposed EMR activity criterion. The correlation co-
 5.10 · EMR Induced by Underground Rocks under Stress 455

Fig. 5.73.
Regression between EMR ac-
tivity experimental (
) and
calculated data (Frid 1997b)

Table 5.5. Statistical analysis experimental and calculated EMR data

efficient is 0.966, indicating a very high forecast probability. Thus, if the EMR activity
is higher than 70 pulse/10 s in a zone of a mine-working, this zone should be declared
to be rock and gas outburst dangerous.
Equation 5.32 allows us to explain a relationship between the build-up of stress and
gas liberation in a mine-working face (and hence, an increase of rock and gas out-
burst hazard) and the increase of EMR activity. As is seen, EMR activity level is af-
fected by both stress level and gas liberation rate. An increase of stress level near a
mine-working causes an intensive fracturing of the extremely stressed coal. This frac-
turing process is the source of EMR activity increase. Fracturing of the coal seam in a
zone of higher stress excites an intensive process of gas liberation and in turn leads to
an additional internal (pore pressure like) pressure. This pressure is akin to a tensional
stress in the coal mass that enhances a fracturing catastrophe simultaneously leading
to a magnified EMR activity.

EMR control of rock-burst and gas-outburst hazard elimination. Both RB and gas
outburst hazards in a mine-working face are frequently eliminated by water infusion,
the performance of which is usually verified after its completion by the gum method.
During infusion, water “impregnates” coal near the mine-working face. If a control hole
intersects a water saturated zone, it excites water spouting out. Then gum turns into
slime, and it becomes impossible to measure its volume. This situation makes it neces-
sary to drill additional control holes. However, the existence of several control holes
(intended to estimate water infusion performance) diminishes the influence of the water
infusion itself because the increase of water pressure sometimes causes breaching to a
neighboring hole. Hence, we have a vicious cycle: To estimate performance we need a
hole, and the holes negatively influence performance. The only way out is to estimate
water infusion performance during its execution without drilling holes.
456 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

The analyses of safety conditions (see Sect. 5.10.2.1.), EMR features induced by
sample compression and associated with borehole drilling (see above), enables us to
state two principles of EMR used for water infusion control:

1. RB is impossible if the mine-working face is transferred back to the residual

strength.
2. If a material is under residual strength, its loading is not accompanied by EMR
excitation.

EMR-water infusion model. As is generally known, rock stress condition may be de-
scribed by the Coulomb-Mohr envelope (Twiss and Moores 1992). Let us assume that
the coal stress condition in the face of a mine-working is described by the Mohr circle
(Fig. 5.74, 1, Frid 2000) and by the Coulomb-Mohr envelope below which the material
is stable (Fig. 5.74, 1'). A water pressure increase forms a pore pressure, P, that is equiva-
lent to an additive tensional stress. This tensional stress shifts the Mohr circle to the
left by P. Hence, a part of the circle will now be above the Mohr-Coulomb envelope
(Fig. 5.74, 2), i.e., coal becomes unstable, and will fracture. This fracturing results in
an increase of EMR activity. As a result of coal fracturing, its strength decreases. How-
ever, if the stress level in a coal stratum is still high, a sharp decrease of water pressure
(for example, due to pump stoppage) will excite crack closure and EMR excitation.
Hence, a water pressure cycle (both increase and decrease) will affect EMR activity.
Thus, water pressure changes are accompanied by EMR activity up to the transition
of coal to its residual strength (Fig. 5.74, 3), and vice versa: An absence of EMR activity
variation during water pressure changes during the “loading-unloading” cycle could
be a criterion of the water infusion performance or in other words, a check of hazard.

EMR during water infusion in coal mines. EMR measurements during water infusion
were carried out on Koksoviy, Andreevsky, and Desyatiy coal stratums in North Kuzbass
(Russia) (Frid 2000). Water infusion was conducted in the following manner:
The length of the holes was 6 m and the sealed length was 5–5.5 m (a mining expe-
rience in North Kuzbass shows that the existence of a 5–6 m safeguard zone (residual
strength zone) in a mine-working face indicates a non-dangerous condition);
Water pressure was increased in steps of 0.2–0.3 γ H to provide coal “impregnation”
with water, where γ is the rock density (kg m–3) and H is the stratum depth (m).
A new step of water pressure increase took place only after EMR activity stabiliza-
tion. After reaching 0.8 γ H, water pressure was sharply decreased to 0.3 γ H; as our ex-
perience showed that if the water pressure exceeds a value of 0.8 γ H, the probability of
the water breaching to the mine-working face becomes very high.

Fig. 5.74.
Mohr-Coulomb failure envelope
around Mohr circles which
represents the different coal
conditions during water infu-
sion (Frid 2000)
 5.10 · EMR Induced by Underground Rocks under Stress 457

Fig. 5.75.
An example of EMR registra-
tion during water infusion. Full-
drawn curve is the water pres-
sure, while the hatched curve is
the EMR activity (Frid 2000)

Cycles of pressure increase–decreases were repeated as long as the water pressure

changes initiated an EMR activity. When EMR activity induced by water pressure
changes ceased, water infusion was terminated.
An example of EMR registration during water infusion is shown in Fig. 5.75. Water
infusion was carried out in the drift face of Andreevsky stratum (at 700 m depth, aver-
age rock density is about 2 × 103 kg m–3, the step size of water increase is 4 MPa (0.3 γ H),
and the maximal water pressure value is 11 MPa (maximal water pressure must be less
than 0.8 γ H, see above). Gum value measured at the drift face was 6 l m–1; hence, a RB
hazardous situation existed.
Figure 5.75 shows that an increase of water pressure of up to 4 MPa during the first
cycle is associated with EMR activity excitation (of 500 pulse/10 s); an increase of water
pressure to 8 MPa initiated an additional increase of EMR activity (to about 800 pulse/10 s).
The next decrease of water pressure to 4 MPa caused an insignificant decrease of EMR
activity, and a following water pressure increase to 11 MPa resulted in an additional EMR
activity jump (to 1 350 pulse/10 s). The next decrease of water pressure to 4 MPa led to a
significant decrease of EMR activity, but its level remained high. Therefore, the water pres-
sure rise during the second cycle also induced an increase of EMR activity. After the sec-
ond cycle of water infusion, EMR activity decreased to 50 pulse/10 s. Drilling of a forecast
borehole showed that gum value was about 4.5 l m–1. This value is a little lower than the
critical level, although it also shows that the zone of drift influence is still quite stressed.
The EMR activity was large up to the third cycle of water infusion, while the fifth and sixth
cycles of water infusion were not associated with any EMR appearance. Drilling of a forecast
borehole after the sixth cycle of water infusion showed that it was impossible to determine
gum value, due to very high coal humidity. However, measurement of natural EMR activity
in this zone showed that the average value of natural EMR activity was about 50 pulse/10 s;
that is significantly lower than the limiting value of general EMR activity (120 pulse/10 s,
Frid 1997a, 2001) and implied that the zone near the mine-working face was safe.
Note that the absence of EMR excitation due to water pressure changes is a qualitative
criterion. It does not determine, for example, the number of water pressure cycles still
needed for water infusion performance. Our research (more than 70 EMR investigations
during water infusion) has shown that the EMR criterion really characterizes water infu-
sion performance.
458 Chapter 5 · Electromagnetic Radiation Induced in Fractured Materials

5.11
Summary
A comprehensive model of EMR emission mechanism from cracks was developed. This
model can be described as follows: Following the breaking of bonds, the atoms on both
sides of the severed bonds are moved to “non equilibrium” positions relative to their
steady state ones and oscillate around them. Lines of oscillating atoms move together
and by being connected to atoms around them (in the forward direction and also at-
oms on their side of the two surfaces newly created by the fracture) the latter also
participate in the movement. The ensuing vibrations are similar to those calculated
in the Debye model for bulk oscillations. The larger is the number of cut bonds, the
larger is the area of excited atoms, and hence the greater becomes the EMR ampli-
tude. These oscillations behave like surface optical waves, where positive charges move
together in a diametrically opposite phase to the negative ones and decay exponen-
tially into the material like Rayleigh waves. The resulting oscillating electric dipole is
the source of the EMR. The pulse amplitude decays by an interaction with bulk
phonons.
The model leads to an invariant EMR ‘basic pulse shape’ that repeats itself in all
examined materials when failed by different testing techniques (including triaxial
compression, percussion drilling and large-scale rock explosion). The model enables
us to use EMR results for calculating the length, width and area of cracks and to com-
pare them to actual crack sizes measured by fractographic techniques. Thus, a “fin-
ger-print” of initial micro-cracking was established. The ‘basic shape’ appears also in
pilot experiments in open quarries during rock explosion.
An overall comparison of EMR signals that were excited by compression, drilling
and blasting shows that their basic shape is invariant under both dynamic and quasi-
static loadings. The similarity in shape of single EMR pulses, induced by the three
processes, therefore indicates that a unified EMR emitting mechanism is operating
behind them all and therefore enables us to analyze EMR signals induced by all three
processes by the same model.
Moreover, our investigations showed that global relationships (Gutenberg-Richter
and Benioff strain release) of EMR signals obtained during rock fracture in the labo-
ratory are similar to those measured in earthquakes.
EMR measurements were conducted in mines. Criteria for rock-bursts and rock-
gas outbursts were developed based on results from these measurements and on theo-
retical models. These criteria are easier and faster to implement than the existing ones,
which necessitate drillings and gum and GLR measurements.
Based on these achievements, possible experiments were proposed as guidelines
that can lead to earthquake forecasting via EMR detection and monitoring.
Chapter 6

Assorted Problems in Fracture Geology

This chapter consists of a variety of basic issues in fracture geology. Section 6.1 ex-
plains the importance of studying regional jointing. Section 6.2 outlines some neces-
sary procedures in studying joints, from data collection in field outcrops to joint set
discrimination and determination of paleostress directions. Section 6.3 relates to stud-
ies of paleostresses in the Israeli Negev and considers regional jointing in the broad
context of plate tectonics. Section 6.4 presents a new, perhaps controversial probe into
certain properties of joint length. Section 6.5 briefly mentions the method of local
paleostress estimation by fractographic techniques. Sections 6.6–6.9 concern four in-
vestigations from the Beer Sheva and Shephela synclines. Section 6.6 contains infor-
mation about a multi-fracture assemblage in a fault termination zone, where a joint
set exhibits an intricate pattern of the paleostress trajectories. Section 6.7 describes a
variety of fault-joint geometric and genetic relationships across the local stratigraphic
column. Section 6.8 explores the formation of “surface joints”. Section 6.9 relates to
studies on several aspects of fracture hydro-geology. Finally, Section 6.10 presents a
new approach to en echelon segmentation.

6.1
The Importance of Studying Regional Jointing
Regional vs. local joint distribution. Sections 6.1–6.3 relate to the discrimination of
systematic joint populations (Hodgson 1961b) into distinct sets, which is often termed
“the study of the regional joint pattern”. These joints fall into four groups (burial,
syntectonic, uplift and post-uplift) that generally have large, regional spreads and re-
flect slow, quasi-static fracture (Sect. 1.5), particularly in the early burial stage. They
are distinguished from dynamic joints that form rapidly in granites (Sect. 2.4.6.1) and
occassionally in connection with earthquakes, and have limited, local distribution.
This distinction is an approximation; it is made for convenience and is not always
accurate: Some joints may start slowly and reach high velocities (Sect. 4.9) and the
identification of dynamic joints is not always simple (Sect. 2.4.6.1).

Definitions and discrimination of joint sets and motivation for their study. There are
different definitions for sets. Whereas Twiss and Moores (1992, p. 38) consider that a
set consists of joints that have a “similar geometry”, for Ghosh (1993, p. 478) this term
applies to “an array of parallel joints”. It was shown that the task of distinguishing a
joint set from others may be simple in some areas (e.g., Gross 1993), but occasionally,
460 Chapter 6 · Assorted Problems in Fracture Geology

can be quite difficult (e.g., Nickelsen and Hough 1967; Bahat 1991a, p. 251–255; Younes
and Engelder 1999). However, a reliable division of joint sets in a fracture province
provides the key to identifying directions of paleo-principal directions and supplies a
major tool in the investigation of local as well as regional aspects of structural geol-
ogy. If it is fractographically clear that the joint surface forms normal to the direction
of the maximum principal tension or extension (Bahat 1979a), it becomes also clear
that a systematic joint set is coaxial with that of the direction of maximum paleo-
compression (Engelder and Geiser 1980).

The investigation of paleostress directions. It is quite remarkable that similar di-

lemmas regarding paleostress directions have intrigued investigators of regional frac-
ture in two remote fracture provinces: The Appalachian Plateau in the eastern USA
and the Beer Sheva syncline in the Israeli Negev (Chap. 3). This issue is discussed
below (Sect. 6.3) by elaborating on results from Israel, with reference to results from
the Appalachian Plateau. We note that in the context of identifying directions of paleo
principal directions, “cross fold joints” that strike approximately orthogonal to fold
axes (Engelder and Geiser 1980) are particularly important in these two provinces
and in other investigated areas that exhibit “regional joint patterns”.
At the heart of this investigation lies a formidable difficulty: Which methods should
one use in discriminating joint sets and what relative weight should be assigned to
the results obtained by the various methods? We deal below with this issue in two sec-
tions:

1. Qualitative, measurements, classification, plotting and division into sets.

2. Mapping joint distribution.

Experience has shown that the best results derive from a sensible utilization of all
qualitative and quantitative methods applicable to a specific research area.

6.2
The Procedures of Studying Systematic Joints
6.2.1
Qualitative, Measurements, Classification, Plotting and Division into Sets

Carrying out the various steps in the procedure of studying joints involves a some-
what complex procedure. The first efforts should be spent on acquiring a general fa-
miliarity with the scope of outcrops in the investigated area (e.g., different lithologies
and major structures). There seems to be no alternative to a good vehicle and a lot of
travelling in the search for appropriate outcrops. Air photographs of various magnifi-
cations can help but do not provide reliable substitutes in this pursuit. One starts with
a qualitative fracture study. Measurements would be the next step, as a prelude to clas-
sification. Classification, plotting and division into sets need to be done in alternating
steps when progress of each step is helping to examine the validity of estimations of
the preceding ones. A useful procedure of joint investigation is given by Goldstein and
Marshak (1988); we try to supplement their manual.
 6.2 · The Procedures of Studying Systematic Joints 461

6.2.1.1
Qualitative Description

The qualitative study consists of two parts. The first one involves a general descrip-
tion of the physical features of fracture. These include fracture (three) dimensions,
characteristic spacing, straightness vs. curvature, general deviation from verticality,
possible orthogonal relationships, anastomotic appearance as well as joint contacts
and fracture interaction styles. The second part relates to fractography. Numerous in-
vestigators report about the lack of fractographic features in their research areas. How-
ever, in many other areas rocks maintain a record of their fracture history by their frac-
ture surface morphology. While the mirror plane (Fig. 2.1) may disclose details of early
events, the fringe often tells about late developments of joints. Familiarity with the
fractographic descriptions in Chap. 2–4 may help in obtaining this information. First,
take photographs of the selected joints (close-ups at various extents), paying particular
attention to having the object lighted properly. Photographs taken under different light
conditions will show different fractographies of a given fracture surface. It is occasion-
ally advisable to wet the joint surface or increase the light on it. Do not forget to place a
scale that does not obscure the features you want to show. Second, make accurate draw-
ings of your selected photographs. Remember that drawing is far from being a trivial
undertaking; it is already an important part of the fractographic interpretation. The
radial features of the plume will lead to the fracture origin, provide information regard-
ing changes in propagation directions and tell about the sequence of fracture genera-
tions when cut by other fractures. The ripple marks will inform something about the
growth rhythm of the joint. Perhaps the most difficult task is marking the boundary
between the mirror plane and fringe; it can often be quite controversial (Sect. 2.2.6).

6.2.1.2
Measurements

Strike measurements. The number of strike measurements will often be a compro-

mise that takes into account field conditions, joint dips, the objectives of the study,
economy of time and method (Goldstein and Marshak 1988). Generally, no dip needs
be measured when joints deviate only several degrees from verticality. However, when
bedding dip exceeds five degrees, joint strikes may be evaluated only after rotation of
bedding to the horizontal position. We describe below several techniques that have
been used in studying joints cutting chalk horizontal layers in the Beer Sheva syn-
cline (Chap. 3) that are thought to be adequate.
At least a hundred strike measurements were taken in large outcrops in which thou-
sands of joints were exposed (e.g., stations 2, 3, 4 in Bahat and Grossmann 1988). In
such outcrops the “discrimination method” was used, whereby measurements were
carried out only on “reliable”, original joint surfaces that could be clearly identified
by fractographic ornamentation, yielding the accuracy of ±2–3°. This method provided
the most trustworthy results. The “comprehensive method” was used in many medium
size outcrops in which about a hundred or two hundred joints were exposed, whereby
all “reliable”, systematic joints were measured (e.g., stations 20 as above, and 83 from
present Fig. 3.37). This method provided quite dependable results. Sector intervals of 1°
462 Chapter 6 · Assorted Problems in Fracture Geology

were used in all the histogram plots, assuming that inaccuracies would balance each
other. The “selection method” (Goldstein and Marshak 1988) was used in twenty-nine
stations along Wadi Secher (Bahat and Shavit 1997). In this method, we visually scanned
each outcrop and selected representatives of the prominent joint sets. We then mea-
sured a few tens of joints in each station. This method provided a good “general pic-
ture” but its reliability for statistical analysis would be limited. An obvious question
that follows this description is: What about taking measurements in rocks where joints
are not fractographically decorated, fractures are not quite straight, strike and dip
spreadings are significant in suspected individual sets and joint surfaces are weath-
ered and/or splintered (like in many limestones and dolomites throughout Israel and
in some chalks in England, Bahat 1991a, Table 5.5)? The answer is yes, take measure-
ments, but be aware of the reduced reliability of the results and stay conservative in
data processing and interpretation of the results.

Length measurements. Scaneline (traverse) techniques (e.g., Goldstein and Marshak

1988) are often practiced in studying joint length distribution. We recommend how-
ever that a fractographic method should be used where possible. This can be accom-
plished on vertical exposures along road cuts and wadies. An example is given from
Wadi Naim, where the lengths of forty four joints were accurately measured from tip
to tip of unilateral and bilateral plumes that mark the joint surfaces along a horizontal
60 cm thick layer (Bahat 1988b). In this method, fracture length is actually measured
where the plume tips show the exact ends of the joint (Fig. 6.1). Scaneline techniques
are less accurate because they rely on measurements of many fractures even where a
large proportion of their traces terminates beyond the boundaries of the typical ex-
posure. This difficulty is overcome by making measurements only at the intersections
of the fractures with grids of parallel scanlines without attempting to identify the

Fig. 6.1. An exact joint tip is shown by the sharp distal boundary (arrow) of the plume, for measuring
joint lengths. Width of joint is 20 cm
 6.2 · The Procedures of Studying Systematic Joints 463

ends of the fracture traces. Fracture trace length is then estimated from the censored
data by statistical methods (e.g., Panek 1985).

Additional aspects of joint-set discrimination. There are situations where additional

methods are worth considering. “Strike superposition” results are also useful for set
discrimination, whereby results obtained in a given area by different investigators
(Bahat 1991a, p. 251) or by a given investigator (Sect. 3.4.3.1) are compared. Compar-
ing strikes of en echelon segments to strikes of the parent joints (Sect. 2.2.7.6) can also
improve the reliability of joint set differentiation. Furthermore, certain joint sets can
be identified by particular fracture markings (Fig. 3.36a). Thus, some initial results
on the regional joint pattern can be obtained by combined qualitative methods. Often,
new research areas require new approaches to the problem. In the Bristol Channel
(Sect. 3.2), the regional joint pattern has some unique complications that required the
application of sophisticated techniques specially designed for that area (Engelder and
Peacock 2001). The study of the gradual development of “surface joints” was carried
out by combined photographic and statistical techniques (Sect. 6.8).

A Genetic classification into joint groups. The genetic classification of joints that oc-
cur in sedimentary and plutonic rocks is important in fracture analysis, in order to
avoid mistakes in the resolution of the regional joint pattern and in the tectonophysics
interpretation of the investigated areas (e.g. Sects. 6.9.3 and 4.7.1.1). The genetic clas-
sification of joints must be carried out before plotting the results so that joints of dif-
ferent groups should not be plotted unintentionally together. The genetic characteriza-
tion of various fault-joint types (Sect. 6.7) can also help in understanding the fracture
history of investigated areas.

Plotting diagrams. Plotting the sum of measurements following separation into ge-
netic groups is the first step toward interpreting the results. The best way of plotting
results for vertical joints (either originally measured as such or after rotation) is by
histograms. Bahat (1997) showed that results displayed on a rose diagram may give a
different impression from the same results when plotted on a histogram. In extreme
cases, while a histogram can show “extra” peaks that possibly suggest a large number
of sets, a rose diagram for the same joint population may “hide” an essential peak or
two. Thus, there are various degrees of certainty regarding the division into joint sets
when all measurements are summarized in histograms or stereographic nets or in
both (e.g., Grossmann 1983; Bahat and Grossmann 1988), frequently due to unclear
maxima and minima on the plots. Additional statistical techniques are often used in
order to increase the credibility of joint set discrimination.

6.2.2
Statistical Analysis in Support of the Division of Joints into Sets

We refer below to two analytical tracks in support of the division of joints into sets;
first, methods of separation between two adjacent populations of joint azimuth, and
second, by a special reference to joint spacing. Often one selects the technique according
to the access to a computer distribution program, and these become more available
with time (e.g., Exell, jmp, Rock Works). A few choices are briefly mentioned below.
464 Chapter 6 · Assorted Problems in Fracture Geology

Fig. 6.2. a Fisher’s (1938) solution to separate joint populations of two sets when their measurements
partly overlap each other is by projecting the sample raw data to improve the definition of the boundary
between the two. After rotation of the coordinate system of the original π1 and π2 populations, their
distinction is significantly improved on the projection (from Johnson and Wichern 1988). b Equal-
area projection of synthetic directional data that were randomly selected from a Fisherian popula-
tion with true mean direction I = +90° and precision parameter k = 20; individual directions are shown
by black dots, mean direction is shown by a solid square with surrounding α95 confidence limit (circle),
where Dm, Im and k are, the calculated mean direction, the calculated mean inclination and the best
estimate of k, respectively (from Butler 1992)

Joint set discrimination. A recurrent problem is how to separate two joint popula-
tions when measurements of these populations partly overlap each other. Fisher’s
(1938) solution is by projecting the sample raw data, as illustrated schematically in
Fig. 6.2a. This projecting is based on the “Fisher sample linear discrimination func-
tion”, and the method is explained in textbooks of statistics (e.g., Johnson and Wichern
1988). Smoothing techniques (Wise and McCrory 1982; Goldstein and Marshak 1988)
can also help to make peaks stand out more clearly.
Additional contributions which are applicable in the earth sciences stemmed from
the need for statistical analysis of paleomagnetism (Fisher 1953; Fisher et al. 1987). Fisher
and coworkers developed a method by which each measured direction is given unit
weight and is represented by a point on a sphere of unit radius. This method is known
as the “Fisher distribution” and occasionally also termed “α95”. The confidence limit
α95 is a measure of the precision at which the true mean direction has been estimated.
One is 95% certain that the unknown true mean direction lies within α95 of the calcu-
lated mean. The inference is that there is a 5% chance that the true mean lies more
than α95 from the calculated mean (Fig. 6.2b). Thus, the “Fisher distribution” helps to
estimate the mean direction of given population. However, this method does not dis-
criminate between populations, i.e., the directions in Fig. 6.2b may represent more than
one population, which is disregarded by the small circle of α95 confidence limit.
Good computer distribution programs (e.g., jmp) help to calculate many statistical
parameters like the arithmetic mean, median, mode, standard deviation, skewness (sk)
and kurtosis (ku) (see Nomenclature). In addition, they can divide given collected data
of azimuths into sub-populations and offer alternative combinations of sub-popula-
tions for comparison. It is up to the researcher to correlate the various combinations
with additional qualitative (fracture physical properties and fractography) and quan-
titative (statistical parameters like kurtosis, spacing and FSR, see below) results in at-
tempting to achieve high resolution in joint set discrimination. Such careful proce-
dures can lead to satisfactory results and improve the credibility of the division into
sets, based on the interpretation of histograms.
 6.2 · The Procedures of Studying Systematic Joints 465

Fig. 6.3. Joint spacing distribution from the Mor Formation. a Set 343° (normalized joint spacing histo-
gram of joint set B ). b Set 350° (normalized joint spacing histogram of joint set C) (from Levi 2003)

Division into joint sets by a special reference to joint spacing. The importance of
joint spacing for various geological and other applications is well known (e.g., Price
and Cosgrove 1990, p. 54). Spacing is commonly measured between neighboring frac-
tures of the same set along a line orthogonal to the average orientation in the set. We
show below how a statistical analysis of joint spacing can improve both, the character-
ization of joint sets in relation to rock lithologies and the credibility of grouping into
sets, citing joint investigation in two quasi-synclines from the south Negev (Levi 2003).
Sets 343° (B) and 350° (C) from the Mor Formation (Fig. 6.3a,b) raise the suspicion
of a genetic linkage between the two (Levi 2003 and present Sect. 6.3.6). The high kur-
tosis value of set 350° suggests the possibility of a connection between the change in
joint direction (from 343° to 350°) and differences in mechanical properties of the rocks
cut by the two sets, a possibility raised by Goodwin (1995) (see also present Fig. 2.60).
Like in the Beer Sheva and the Shephela synclines (Bahat 1987a), the two sets do not
mix in the same chalk layer, rather they appear in separate layers (resembling Fig. 3.36a).
This enabled the following experiment to be run with a high degree of credibility. Two
sets of cylinders were drilled from the corresponding layers and were examined by an
ultrasonic method (ASTM: D2845-90) for their wave velocity vp, Young modulus E, shear
modulus Gm and dry density dd. The results (Fig. 6.4) clearly show a difference between
the mechanical properties of the two joint sets. It has therefore been suggested (Levi
2003) that both 343° and 350° sets were formed in one stress field. However, the joints
of set 343° formed in co-axial orientation with SHmax (the maximum horizontal com-
pressive stresses) in the rock of higher density, whereas the joints of set 350° that formed
in the rock of lower density deviated from parallelism to SHmax. This finding corresponds
quite well with additional sets of results (Fig. 2.61 and Tables 2.2 and 2.3).
Levi (2003) compares the FSR values (Gross 1993, see Nomenclature) of all the
investigated sets discriminated in his study area. The FSR value of set 326° is dis-
tinguished among all the other sets in having low values and a very small spread
(lower left corner of Fig. 6.5). The other sets have high FSR values that scatter con-
siderably. Levi (2003) considers that these results possibly reflect fracture under differ-
ent tectonic conditions. Set 326° formed during the burial stage under increasing over-
burden conditions, resulting in relatively low strains under approximately uniform con-
ditions. On the other hand, the characteristics of the other sets generally fit the criteria
of fracture during uplift (Tables 3.3 and 3.5). Single-layer uplift joints form consequent
to denudation under conditions of reduced overburden (Bahat 1999a) and are highly
466 Chapter 6 · Assorted Problems in Fracture Geology

Fig. 6.4.
Plots of elastic properties vs.
dry density for sets 343° (B)
and 350° (C) from the Mor
Formation, when wave velocity
is vp, Young modulus is E, shear
modulus is Gm and dry density
is dd (from Levi 2003)

strained, probably in tensile rather than extensile conditions. Such conditions may
result in high FSR values (Bai and Pollard 2000a). Fracture of uplift joints occurs in
shallow depths, in stress fields that often compromise between regional and local
stresses, due to influences from surface conditions (Bahat 1991a, p. 297). Therefore,
spread in FSR values (Fig. 6.5) may be expected to characterize uplift joints. Hence,
the determination of FSR values and their spread can help in the process of differen-
tiating joint genetic groups.
It was found that the ratio of joint length to joint spacing (L / S) (Bahat 1991a, p. 309)
characteristically varies according to joint grouping: Between 2 to 10 in burial joints
and in the 20 ≥100 range in syntectonic joints. In uplift joints, L / S values generally
range between those of the burial and syntectonic joints. Wu and Pollard (1995) dis-
tinguished two kinds of fracture patterns in sedimentary layers and in laboratory speci-
 6.2 · The Procedures of Studying Systematic Joints 467

Fig. 6.5.
Spread of FSR values in
different joint sets (see text for
explanation) (from Levi 2003)

mens. One, termed poorly-developed, is characteristic of the early stages in the devel-
opment of a fracture set when typical fracture lengths are roughly equal to or less than
typical spacings, i.e., L / S ≤ 1. The other, termed well-developed, is characteristic of the
later stages when fracture lengths are much greater than spacings, i.e., L / S >> 1. Wu
and Pollard (1995) suggest that characteristic fracture growths occur under various
degrees of strain. Fracture opening occurs during an advanced stage of jointing with-
out significant fracture propagation. As the applied strain increases, the spacing de-
creases, because existing fractures increase in length and new fractures begin to propa-
gate from flaws. However, when the applied strain reaches a certain limited value, the
spacing stops evolving and remains nearly constant as the strain continues to increase.
Levi (2003) found that the L / S values for sets 326° and 343° in the Mor Formation are 4.4
and 15.4, respectively, and that these values suggest that set 343° represents a more ad-
vanced strain than set 326°. This interpretation corresponds to the distinction made
above, based on the FSR criterion. Thus, FSR and L / S are complementary parameters
that correlate joint spacing characteristics with strain magnitudes.

6.2.3
Mapping Joint Distribution

Once the classification into joint sets is generally established, maps of joint set distri-
bution can be produced. “Trajectory maps” were successfully produced by Engelder
and Geiser (1980) and Geiser and Engelder (1983), showing the regional trajectories
of SHmax inferred by the extrapolation of joint directions of particular sets from sta-
tion to station (Fig. 3.1b,c). If in addition, one is interested in showing the relative
concentration of the various joint sets in particular areas on a structural map, “maps
of set distribution” may be used. Such maps may help in the interpretation of the ef-
fects of local and regional stress fields (e.g., due to neighboring faults) on the forma-
tion of the various sets (Levi 2003, p. 25).
468 Chapter 6 · Assorted Problems in Fracture Geology

6.3
Paleostresses in the Israeli Negev
We present below two different models regarding mesostructures and paleostresses in
the Israeli Negev. One model considers two major stress field regimes that fluctuated
through time in Israel (Eyal and Reches 1983; Eyal 1996; Eyal et al. 2001). The other model
maintains that the sum of systematic joint sets in the Israeli Negev reflects a system of
multi paleostresses (more than two major ones) that generally shows a clockwise rota-
tion with time (Bahat 1997, 1999b). The difference between the two concepts is a re-
minder of the difficulties involved in interpreting the regional joint distribution in the
Appalachian Plateau and the paleostress evolution in that region (Bahat 1991a, p. 251)
(Sect. 3.1). A similar dilemma is connected with the interpretation of joint mechanisms
in granites (Sect. 4.5.1). Hence, the two concepts represent a basic debate in fracture
geology. We present below brief accounts from various studies that relate to this debate
and then suggest our interpretation.

6.3.1
Jordan

6.3.1.1
Clockwise Rotation of Compression in Jordan

Burdon (1959) suggests three compression phases in Jordan. The first phase ‘a’ was
minor, started in the Maestrichtian and came from southeast, approximately in the
285° direction. These forces were light, and the stresses they induced were relieved by
the minor folds seen in the chert beds of the Maestrichtian of the Belqa Series. The sec-
ond phase ‘b’ was major, probably in the Lower Miocene, or even a little earlier, approxi-
mately in the 322° direction. In north Jordan the Ajlun dome was formed and in the
crystalline basement multi fracture systems occurred, including tension fractures in the
322° direction. The third phase ‘c’ was major. Compression came from the SSE, approxi-
mately in the 349° direction and took place in two sub-stages (C1 and C2), which were
separated by a comparative quiet stage in the Pleistocene. These two phases were asso-
ciated with a 107 km displacement along the Dead Sea rift, C1 of 62 km and C2 of 45 km.
Salameh and Zacher (1982) measured horizontal stylolites in limestones of the Upper
Cretaceous in Jordan. They found two dominant paleostress directions, 310–320°
and 350°, which seem to be close to the above phases ‘b’ and ‘c’.

6.3.1.2
Southwestern Jordan

Eldeen et al. (2000) present results obtained from kinematic slip data in the eastern
margins of the Dead Sea rift in southwest Jordan, and their results are cited below. Stress
inversion of fault slip data was performed by using an improved right-dieder method,
followed by rotational optimization (Delvaux, TENSOR program). Fault slip data (to-
taling 2 773 data) include fault planes, slickensides and sense of movements. These results
were obtained from rocks ranging in age from Precambrian crystalline basement to
Oligocene/Pleistocene sediments and were inverted to determine eighty-eight differ-
 6.3 · Paleostresses in the Israeli Negev 469

ent paleostress tensors. Fault slip data allowed several paleostress tensor groups (stages)
from the Precambrian to Post-Pleistocene period to be pointed out and were correlated
with the tectonic evolution of the Dead Sea rift. The first two tensor groups were ob-
tained from the Late Proterozoic (between 570 and 560 Ma) dike swarm, and the fault
slip data related to the dike emplacement. Both show a transtensional stress regime, the
first one (T1) with NE-SW SHmax corresponding to the major NE-trending dike system,
and the second one (T2) with N-S SHmax corresponding to a minor and later N-S trend-
ing dike system.
The other tensors are related to Cretaceous or younger deformation stages, roughly
coinciding with the three compression directions observed by Burdon (1959). The old-
est stage (T3) has a strike-slip stress regime with E-W SHmax that affected all Pre-Ter-
tiary series, but not the younger rocks present. The next stage (T4) is also character-
ized by a strike-slip stress regime, but with NW-SE SHmax, affecting the Palaeocene-
Eocene rocks but not Pleistocene rocks. Eldeen et al. (2000) relate NW-SE trending
dykes to the Miocene, including them in stage T4. A transpressional stage (T5), with
NNW-SSE to N-S SHmax affected rocks with an age up to the Pleistocene. In addition,
Late Pleistocene extensional stage (T6) with E-W SHmin (the minimum horizontal com-
pressive stresses) stress regime was evidenced from the fault slip data that affected
the Dana Conglomerate. The youngest tensor group obtained from the Dana Conglom-
erate points to an extensional regime with N-S SHmin (T7). Eldeen et al. (2000) suggest
that the Late Pleistocene-Holocene sinistral strike-slip deformation is still active.

6.3.2
The East Mediterranean

Letouzey and Tremolieres (1980) conducted an extensive investigation of paleostress

directions by measuring a large number of tectonic stylolites, joints, shear planes and
faults in sedimentary rocks around the Mediterranean Sea. They plotted their results
on a series of maps, displaying a complex pattern of paleostress directions that were
correlated with different stages of the structural evolution in the region. These tec-
tonic changes were linked to the Europe-African- Arabian indentation in the collision
zone since the Upper Cretaceous and closure of the Tethys Ocean (Fig. 6.6a–e). Particu-
larly relevant to the present discussion are the two maps from Turkey (Fig. 6.7a–d) and
from Israel (Fig. 6.8). Most intriguing is the great variety of paleostress directions found
in south Turkey that were categorized into three phases (we use below only upper semi-
circle azimuth numbers in describing directions of phases).
Letouzey and Tremolieres (1980) assign to the Late Cretaceous shortening (the “first
phase”) an early 075° compression direction associated with the Alpine zone that
changed to 280° west of the belt of Levantin faults in north Syria (along the continua-
tion of the Dead Sea rift up to south Turkey). The 315° shortening, taken to be the north
Arabian average direction, was prominent east of this belt (Fig. 6.7a). Letouzey and
Tremolieres (1980) consider that the first phase could correspond to a change in ori-
entation of the stress due to the indentation process or to sinistral rotation of the
Anatolian microplate during collision. The 280° corresponds reasonably well to the
290° compression direction found in Israel by them (Fig. 6.8).
The Late Eocene-pre-Middle Oligocene “second phase” is related to a tectonic event
well known from the Pontids, Taurus and north Cyprus Mountains. This phase involved
470 Chapter 6 · Assorted Problems in Fracture Geology

Fig. 6.6.
Sketch of evolution of the Medi-
terranean basins and paleo-
stress field orientations. On all
figures Europe was arbitrarily
fixed. Regional paleo-stress ori-
entations are in fossil positions.
a Late Cretaceous; b Eocene;
c Late Oligocene-Early Miocene;
d Late Miocene; e Plio-Quarte-
nary (from Letouzey and Tré-
molières 1980, see text for ex-
planation)
Fig. 6.7. Shortening phases in Eastern Turkey, where circle diagrams give the paleostress directions (PD) discussed in the text. a Late Cretaceous phase. At left PD from the
Alpine zone, at right PD from the Arabic platform. Note the belt of Levantin faults south of the star (*). b Late Eocene-Early Oligocene phase. At lower left PD from the north
Arabic platform (west), at lower right PD from the north Arabic platform (east). c Late Miocene phase (two small arrows show the 343° and 350° shortening directions).
d Plio-Quaternary phase. At left PD from the Alpine zone. At right PD from the Arabic platform (modified from Letouzey and Tremolieres 1980, see text for explanation)
6.3 · Paleostresses in the Israeli Negev
471
472 Chapter 6 · Assorted Problems in Fracture Geology

Fig. 6.8.
Measurements of shortening
phases in Upper Cretaceous
rocks and Eocene rocks from
Israel. The first phase (290°) is
interpreted to have preceded
the last phase (approximately
N-S) (from Letouzey and
Tremolieres 1980, see text for
explanation)

a 25° spatial change in the stress direction observed in the northern part of the Ara-
bian plate, along its collision contact with the Anatolian microplate. The difference is
between the 330° direction in the western side and the 355° direction in the eastern
side of the northern Arabian plate (Fig. 6.7b).
The tectonics of the Late Miocene (the “third phase”) affected all the geological
formations that Letouzey and Tremolieres (1980) studied. It was associated with fold-
ing, thrusting and breaking of continental blocks and exhibited large changes in ori-
entation of the stress field in different structural areas, showing changes in compres-
sion (or partition of) directions between E-W in the western side and 355° in the east-
ern side of the northern part of the Arabian plate (Fig. 6.7c).
The Plio-Quaternary phase is the continuation of the Late Miocene tectonics in the
north of the Arabian platform, including the large strike-slip Levantine and North
Anatolian fault systems (Fig. 6.7d). The recorded paleostress compression directions
from this tectonics are 025° in the Alpine zone, N-S in the western boundary of the
northern Arabian platform, and 025°, east of the boundary of the northern Arabian
 6.3 · Paleostresses in the Israeli Negev 473

platform. The sequence of the latter two directions is not clear, and it is not certain
that there was a continuation of the clockwise rotation during this phase. However,
Letouzey and Tremolieres (1980) implicitly integrate the influence of the Dead Sea
transform into the general pattern of the Europe-African-Arabian collision, not ear-
lier than the Plio-Quaternary (Fig. 6.6e), exhibiting the new plate boundary in the Plio-
Quaternary phase.

6.3.3
The Sinai-Israel Sub-Plate

Eyal (1996, Fig. 1) finds that almost all middle Miocene to Recent SHmax and SHmin data
compiled from the Sinai-Israel sub-plate cluster in two “distinct” orientations. This
observation is based on an investigation of both mesostructures (not including joints)
and macrostructures (including dikes and focal mechanism data). One SHmax trend
striking 344 ±3° (ninety-eight sites) is consistent with the Dead Sea stress field (DSS),
and the second, SHmax trend striking 282 ±5° (twenty-six sites), is consistent with the
Syrian Arc stress field (SAS). These trends correspond to the SAS and DSS by Eyal and
Reches (1983) and according to these authors, reflect an “outstanding homogeneity”
of the strain in the region. The SAS direction resembles the “first phase” (290°) by
Letouzey and Tremolieres (1980) (Fig. 6.8).

6.3.4
Two Synclines in the Northern Negev

Results from the Beer Sheva and Shephela synclines around Beer Sheva enabled Bahat
(1997) to show that the regional joint pattern in the Lower Eocene chalks differs sig-
nificantly from that of the Middle Eocene, indicating heterogeneous strain in Eocene
rocks in the area. In addition, the outstanding maxima 334° and 328° shown by Flexer
et al. (1984) and by Bahat and Grossmann (1988), respectively (which are based on thou-
sands of measurements in three synclines along the Syrian Arc), strongly point to a
NW SHmax, which is in variance with the 282° SAS direction discussed by Eyal (1996).
The maxima 328° and 334° populate the angular zone between the distinct SAS and DSS.
In addition, Bahat (1991a, Fig. 5.8) related to an intensely jointed outcrop of Santonian
chalk on the rift shoulder east of Arad (next to the Dead Sea) and found that by far the
dominant set was oriented about 320°. This direction corresponds far closer to the
NW SHmax than to the DSS. Thus, the combined observations made by Eyal (1996), Flexer
et al. (1984) and Bahat and Grossmann (1988) show more than two major paleostress
directions (Eyal 1996) since the Upper Cretaceous in Israel.
Combining the above observations with additional results by various authors, Bahat
(1999b) suggested an alternative model to the one pertaining to two distinct SAS and
DSS stress fields, as follows. A common feature of mesostructures that show WNW-
WSW directions of SHmax is that they generally occur in limestones and dolomites in
Cenomanian-Turonian monoclines, whereas those mesostructures that point to NW-
NNW SHmax directions are mainly joints that cut Eocene chalks in synclines. These
observations were interpreted according to a “stress rotation” model, which consid-
ered an early remote SHmax that was oriented WSW to WNW during the Cenomanian-
Turonian period and was later rotated clockwise to NW-NNW in the Senonian-Eocene.
474 Chapter 6 · Assorted Problems in Fracture Geology

6.3.5
The Neqarot Syncline in the Central Negev

Eyal et al. (2001) found four joint sets in carbonate beds from upper Turonian rocks
within the Neqarot syncline of south-central Israel:

1. A joint set with a mean strike of 160–340° and α95 < 3° found in forty beds at thirty-
two stations.
2. A joint set with a mean strike of 108–288° and α95 < 5° found in twenty-one beds at
nineteen stations.
3. A joint set with a mean strike of 036–216° and α95 < 9° found in nine beds at eight
stations.
4. A joint set with a mean strike of 136–316° and α95 < 6° found in six beds at five
stations, where α95 is the radius of confidence cone of Fisher (1953) vector distri-
bution.

Eyal et al. (2001) also found that the two most prominent joint sets observed in
carbonate beds of the Gerofit Formation in Nahal Neqarot (N in Fig. 6.12a) are aligned
NNW (340°) and WNW (288°) and are compatible with the DSS and SAS, respectively
(Eyal 1996 and Sect. 6.3.3). They found that cross-cutting relations of these joint sets
and the other less prominent joint sets imply different timing relationships. In some
beds, the WNW set consistently predated the NNW set, whereas in other beds (alto-
gether six stations out of some forty), the NNW set predates the WNW set. Based on
these findings Eyal et al. (2001) came up with the following paleostress model. First,
these results support the Eyal (1996) model of two major stress field regimes in Israel
that fluctuated through time since the middle Miocene, and second, the joint data sug-
gest abrupt changes rather than gradual rotations in stress field orientations. In seek-
ing to establish the validity of the stress model by Eyal et al. (2001), we revisited the
Nahal Neqarot outcrop with the objective of examining joint abutting contacts and their
sequence as well as their strike spreading and fracture properties. We concentrated on
stations that show the succession of DSS compatibles that formed before SAS com-
patibles as suggested by Eyal et al. (2001, Fig. 2 and Table 1, stations 12, 13, 15, M-4, 26
and 28). New measurements were carried out at specific locations where joints of dif-
ferent sets showed clear contacts (Fig. 6.9a–f), itemizing the results below.
Joints striking 032° arrest at joints that strike 292° in station 28; in these sets spac-
ing is limited to several cm (Fig. 6.9a). In a neighbor outcrop, some 40 m upstream
the wadi, a 090° striking joint arresting at a 330° joint and a 297° arresting at a 340°
striking joint, forming a 43° angle (Fig. 6.9b). In station 13 we found a 084° joint
arresting at a 356° striking joint, creating an 88° angle (Fig. 6.9c). In station M-2 we
found in bed 1 a joint striking 286° arresting in joints striking approximately 333°
at 47° angle, but the reverse abutting relationship seems to occur as well. While
spacing in the former set was some 80 cm, spacing in the latter set was very low, occa-
sionally about 1 cm (Fig. 6.9d). In station M-2, joint strikes in the NNW set range
between 325° and 338°. Contacts suggesting the two opposite sequences similar to
those shown in station M-2 are repeated in station 15 (Fig. 6.9e). A joint striking 270°
abuts two joints that strike 330° and 340° in station 14, forming 60° and 70° angles, re-
spectively (Fig. 6.9f).
 6.3 · Paleostresses in the Israeli Negev 475

Fig. 6.9. Joint traces in Nahal Neqarot (see text for descriptions). a The dominant joint set 292° ap-
pears from left to right; b A joint oriented 297° parallels to pen; c A 084° joint arresting at a 356° striking
joint; d A 286° joint (parallel to pen) arrests at joints striking approximately 333°; e A contact similar
to the one in d; f A joint striking 090° (parallel to pen) abuts two joints that strike 330° and 340°

There are indeed indications in Fig. 6.9a–f that sets grouped in the range 264(084)–297°
are younger than sets grouped in the range 330–356° as suggested by Eyal et al. (2001).
What remains unclear however is whether this sequence sufficiently justifies the con-
476 Chapter 6 · Assorted Problems in Fracture Geology

Fig. 6.10.
Summary equal area stereographic
projection of the WNW-ESE sets at
Nahal Neqarot. Each data point (i.e.,
pole) is the mean orientation for a
single set of systematic joints within an
individual bed. The mean orientation for
each combined plot is a great circle whose
pole is represented by a square. Rose
petals indicate the relative frequency
of the poles (after Eyal et al. 2001)

Fig. 6.11.
Rose diagram of joint orientations in
the vicinity of a scanline. Sector length
is proportional to number of joints
(from Becker and Gross 1996)

clusion that these two groups are SAS- and DSS-compatible. The two groups (264–297°
and 330–356°) show considerable spreading in Fig. 6.9a–f. Correspondingly, these joints
produce a wide range of angles at their contacts (from 43° to 88°). Angular spreading
is even greater in Fig. 2 and Table 1 from Eyal et al. (2001), e.g., in station 28, the aver-
age of set DSS compatible strikes 328°, which is far from the DSS average. Several joints
are oriented reasonably close the specified SAS and DSS directions (282 ±5° and 344 ±3°,
respectively, Eyal 1996). These include 286° (Fig. 6.9d), 279° (Fig. 6.9e), 340° (Fig. 6.9b)
and 340° (Fig. 6.9f). However, the orientations of most other joints deviate from the
specified directions by more than 10°. The strike spread of the WNW set (Fig. 6.9 and
6.10, also Eyal et al. 2001, Table 1) is significantly greater than the strike spread of the
SAS compatible set (293 ±3°) obtained by Becker and Gross (1996) (Fig. 6.11). Both data
were collected in carbonates of the Turonian Gerofit Formation in the same general area
(note the difference in number of measurements in Fig. 6.10 vs. 6.11). This raises the
possibility that the WNW oriented joints consist of more than one set. Furthermore,
there is a considerable heterogeneity in the physical properties of the joints that repre-
sent each of the above two groups (as reflected in Fig. 6.9a–f and other contacts). Par-
ticularly, joints of similar trends have considerable spacing differences in different lo-
cations in a given outcrop. For instance, in station M-2, joints trending 286° are spaced
approximately 80 cm in bed 1 (Fig. 6.9d), while joints trending about 285° have spac-
ing of some 15–20 cm in bed 5, albeit their occurrence in rock layers of a similar thick-
ness (about 40 ±10 cm) and lithology. In addition, the implication that each group is a
set, representing a unique paleostress field (Table 1 in Eyal et al. 2001), does not take
into account other genetic possibilities: Some of the joints could have been created by
influences exerted by the adjacent large Makhtesh Ramon fault (see the neighboring
Fig. 6.12a. Geological map of part of southern Israel. The black areas marked by N and S are northern and southern quasi-synclines, separated by the Paaran fault, P
(horizontal arrow). Vertical and inclined arrows relate to the approximate western boundary of the Dead Sea rift, and the Ramon fault respectively. E and N neighbor
the boundary with Egypt and the Neqarrot outcrop, respectively (from Levi 2003)
6.3 · Paleostresses in the Israeli Negev
477
478 Chapter 6 · Assorted Problems in Fracture Geology

Fig. 6.12b. Histograms of five joint sets (A–E) from the quasi-synclinal area south of the Paran fault
in Fig. 6.12a (modified from Levi 2003)

Neqarot outcrop N and the Ramon fault, marked by an inclined arrow in Fig. 6.12a),
or some of the DSS compatible joints could have been formed by additional stress fields
in the area, such as N-S SHmin (T7, Sect. 6.3.1.2). Finally, Eyal et al. (2001, Table 1) re-
lied on the α95 method in their set division, not taking into account the limitations of
this method (Sect. 6.2.2). In summary, the repeated observations that sets grouped in
the range 264–297° are younger than sets grouped in the range 330–356° support the
intriguing sequence suggested by Eyal et al. (2001). However, the strike spreading and
other fracture properties stress the possibility that these sets could be associated with
more than two unique stress fields.

6.3.6
Two Quasi-Synclines in the South Negev

Five joint sets. Levi (2003) investigated jointing in and around two quasi-synclinal
areas (folds that are structurally not well defined), north and south of the Paran fault
in south Israel, along the west side of the Dead Sea rift, some 30 km south of the
Ramon fault (Fig. 6.12a). His research concentrated on fracture in Santonian and
Eocene chalk formations and younger sediments. Based on 2646 strike measurements,
he distinguished five single-layer joint sets (A–E): A) 326 ±3°, B) 343 ±2°, C) 350 ±1°,
D) 360 ±2°, and E) 033 ±4° (Fig. 6.12b), and two multilayer sets linked to sets D and E.
He found that set 326° fractured Eocene Lower Eocene chalks in the burial stage, dur-
ing rock consolidation. Joint set 343° fractured after set 326°, and Levi (2003) linked it
to a clockwise stress rotation during the Miocene. The difference in strike between
sets B and C was assigned to lithological differences (Fig. 6.4). Both single-layer and
multi layer D and E sets were interpreted to be associated with Pliocene and younger
tectonics. However, set D was found to be younger than set E representing a local
counter clockwise stress rotation with time.
 6.3 · Paleostresses in the Israeli Negev 479

6.3.7
Summary of Results that Evidence SHmax Clockwise Rotation

The observations by Eldeen et al. (2000) show evidence for a general clockwise
rotation with time of the SHmax axis from E-W trend in the Cretaceous (their T3),
through NW-SE SHmax that affected the Palaeocene-Eocene rocks (T4) and NNW-SSE
to N-S SHmax affecting rocks with an age up to the Pleistocene (T5). This three-
stage (T3–T5) rotation from southwest Jordan fits previous observations quite
well. They approximately correspond to the clockwise rotation of the 285°, 322°
and 349° compression directions suggested by Burdon (1959) (Sect. 6.3.1.1). Stages T3
and T4 fit the paleo-stress pattern suggested by Bahat (1999b) in the northern and
central Negev parts of Israel (WNW-WSW SHmax in Cenomanian-Turonian mono-
clines and NW-NNW SHmax in Eocene chalks from synclines) (Sect. 6.3.4). Clock-
wise rotation of paleostress SHmax, was also demonstrated by Levi (2003) in southern
Israel (Sect. 6.3.6), with some similarities and a difference compared to the data given
by Eldeen et al. (2000). Levi’s set A, implying SHmax 326° in the Lower Eocene and
set D, implying SHmax 360° in Pliocene and younger tectonics correspond well to
T4 and T5, respectively. The difference relates to results that concern the Miocene.
While set B striking 343 ±2° and set C striking 350 ±1° are connected to the Mio-
cene by Levi (2003), Eldeen et al. (2000) assigned the more westerly direction NW-SE
to that period.
The NW-NNW SHmax (approximately the 326° direction) that correlates with the
tectonics that took place during the Eocene (Bahat 1999b; Levi 2003 and possibly Eldeen
et al. 2000 as well) matches reasonably well the 330° direction linked with Late Eocene-
pre-Middle Oligocene by (Letouzey and Tremolieres 1980) (Sect. 6.3.2, Fig. 6.7b). The
compression direction 355° in the eastern side of the northern part of the Arabian plate
(Letouzey and Tremolieres 1980) correlates fairly well with Levi’s (2003) set C (350 ±1°).
Sets B and C are oriented quite close to certain shortening directions measured in the
area along the Levantin faults (arrows in Fig. 6.7c). Quite significant are also the simi-
larities of the youngest stress directions. The recorded compression directions by
Letouzey and Tremolieres 1980) from the Plio-Quaternary tectonics are N-S in the
western side of the northern Arabian plate and 025° from the eastern side of the plate
(Fig. 6.7d). Both are almost precisely the same paleostress directions found by Levi
(2003) that were implied by sets 360° and 033°, respectively. The N-S compression
matches T5 by Eldeen et al. (2000).
In conclusion, the results obtained by Burdon (1959), Letouzey and Tremolieres
(1980), Bahat (1999b), Levi (2003) and Eldeen et al. (2000) that were conducted in
different areas, i.e., south Turkey as well as south Israel and Jordan, all point to a
clockwise rotation of the paleostress field from the Upper Cretaceous to Pleisto-
cene. This rotation possibly started and continued before and beyond these times.
Also, with a few exceptions, the linkages of compression directions to specific peri-
ods observed by the various authors are fairly close. These correlations support
the ideas formulated by Letouzey and Tremolieres (1980), pertaining to the con-
nection between the tectonics along a wide belt east and west of the Dead Sea rift
and the Europe-African-Arabian plate collision. These studies particularly demon-
strate a linkage between processes that took place in areas from South Turkey and
jointing in Israel.
480 Chapter 6 · Assorted Problems in Fracture Geology

6.3.8
Beyond the Mediterranean Area

6.3.8.1
The Appalachian Plateau

Three cross fold joint sets were distinguished in two outcrops near Watkins Glen in up-
state New York (Bahat 1991a, Tables 5.1 and 5.2). This suggested deviation from the previ-
ous concept of two major cross fold sets and demonstrated the considerable difficulties
that were involved in joint set differentiation. Clearly, any changes in the differentiation of
cross fold joint sets would alter the interpretation of both changes in paleostress mecha-
nisms and fracture history in the region. Younes and Engelder (1999) resolved this prob-
lem (Sect. 3.1). A review of this subject distinguishes three main concepts of regional cross-
fold joint patterns in the Appalachian Plateau (Sect. 3.1.6 and Fig. 3.10a–c). Two of these
concepts relate to rotational mechanisms of the horizontal remote compression, and the
third one concerns a fan pattern of radial joints that form contemporaneously by a given
stress field. The observations from the Appalachian Plateau are combined below with the
results from the Israeli Negev in formulating a model of regional jointing.

6.3.8.2
Systematic Jointing and Plate Collision

The lateral movements that took place along the Dead Sea transform ended at the col-
lision front along the Taurus belt in south Turkey. It appears from the above compari-
son of regional jointing that it is south of this collision front (Fig. 6.6) where strain
records from various times tell about the paleostress histories of different regions along
the transform, including Israel. Therefore, the sum of systematic joint sets reflecting
paleostresses in Israel and its vicinity is best explained via the history of the Europe-
African-Arabian collision and closure of the Tethys Ocean zone since the Upper Creta-
ceous (Letouzey and Tremolieres 1980). A similar explanation of jointing in the Appa-
lachian Plateau can be linked to the closure of the old Atlantic Ocean between the con-
tinents of Africa and North America (e.g., Twiss and Moores 1992, Fig. 6.11 and 6.12).
Quite possibly, such correlations may be made in other regions on the glob, where analo-
gous regional closures (e.g., the Canadian Cordillera) can be compared to recorded regional
joint patterns in their respective forelands, in accordance with the plate tectonic theory.
Ultimately, set directions derived from regional joint measurements in “large forelands”
(that formed by regional closures) and in “small forelands” (e.g., Arlegui and Simón 2001)
will be linked to a global map of paleostresses analogous to other global stress patterns
(Zoback 1989). Such linkages would be particularly relevant to burial and syntectonic joints
cutting sedimentary rocks and to lesser extents to uplift and post uplift joints. Repeatedly
reactivated burial joints are recognizable even in thrust-and-fold belts (possibly parts of
present Fig. 3.1d and Tokarski et al. 1999). Comparing the characteristics of these joints to
those associated with forelands would be quite an intriguing project.
A fascinating question is what are the mechanisms of paleostress changes and how
fast changes occur by shortening processes. Probably the modes of changes reflect the
variable boundary conditions in collision zones that may differ from time to time along
each zone. The interaction of two plates (or microplates) involves some rotation at the
 6.3 · Paleostresses in the Israeli Negev 481

contact, but this would be irregular, somewhere between rigid and plastic compliance
(Molnar and Tapponier 1975) and occasionally involve back and forth motions. Changes
in some strains recording these processes might be gradual, and others will be abrupt.
An abrupt change will occur when a previous process ends, enabling the new process
to take place. Deformations are observed far away from the collision zone and orienta-
tions of paleostresses deduced from structural analyses are surprisingly constant over
wide areas on continents (e.g., Sbar and Sykes 1973; Letouzey and Tremolieres 1980).
Accordingly, there is justification to look for possible connections between paleostress
records in Israel and in the northern part of the Arabian plate in south Turkey.
Is it appropriate to term the sequence described above, from the 326–330° range in
the Upper Cretaceous-Eocene through 343° in the Oligocene-Miocene to 360° and 033°
in the Plio-Quaternary, to be a clockwise stress rotation? Most probably yes. Was this
rotation a gradual progressive process? It probably was not, because it contained de-
viations from a simple process. Deviations may record local rotations in an opposite
direction from the general direction in a region. For instance, while a general clock-
wise rotation applies to Levi’s (2003) joint sets A–E, a local counter clockwise rotation
took place between sets D and E (set D was younger than set E) (Sect. 6.3.6). Quite in-
terestingly, gradual paleostress rotations are also recorded on much smaller scales in
the Beer Sheva syncline. These rotations were superimposed by deviations from the
general trends with back and forth motions (Sect. 3.4, Fig. 3.39b), suggesting diversions
from simple, gradual processes.

6.3.9
Open Questions

We consider below several open questions that relate to jointing and paleostresses,
including the rotation vs. conjugate concepts, the NNW (approximately 344°) compres-
sion, the Zoback (1992) model of global stress in the lithosphere, the timing of strike
parallel joints, and the rarity of fracture surface morphology on joints cutting limestone.

6.3.9.1
The Rotation versus Conjugate Concepts

The 309° and 344° joint sets were interpreted by Bahat (1987a) to be a conjugate couple
associated with the extensile set 328°, which indicated a regional compression, approxi-
mately along the bisector of the angle opened between 309° and 344° (sets 1c and 1d in
Table 3.3). On the other hand, the idea that the 309° set had been an expression of a “tran-
sient stage” of a stress clockwise rotation in the region was entertained as a possible
alternative interpretation (Bahat 1999b). This disagreement has been perplexing for
some time.
An interpretation that might resolve this disagreement is given below. This expla-
nation stems from results that pertain to the possible existence of extensile fracture
at deviating angles from α = 0 up to some 10° (Sect. 2.2.13.4). In the process of a clock-
wise regional rotation of the SHmax in Israel, from WSW-WNW, say 293° (Becker and
Gross 1996) through 309° to 328° and 344° (Bahat 1987a, 1999b), joints in the follow-
ing orientations were formed sequentially (Fig. 6.13). Set 293° was formed parallel
to SHmax. Set 309° formed during the period of rotation from 293° to 328° (and was
482 Chapter 6 · Assorted Problems in Fracture Geology

Fig. 6.13.
Formation of a series of joint sets
by a hypothetic clockwise stress
rotation model. Two arrows mark
paleostress directions that were
coaxial with the joint sets 293°
and 328°, while the other sets
formed at small angles to rotating
horizontal compression

mostly decorated by horizontal plumes). Set 328° was created parallel to SHmax (and
was mostly decorated by concentric rib marks). Set 344° formed during the rotation
from 328° to 344° (and was mostly decorated by horizontal plumes). Perhaps fracture
of the latter set took place when SHmax approximated the 330° direction (e.g., the sec-
ond phase by Letouzey and Tremolieres 1980) or 334° (Flexer et al. 1984). The rate of
this rotation is not known, but it is estimated that sets 309° and 328° were created in
the Lower Eocene and sets 334° and 344° formed somewhat later. Thus, the similar
angular relations between sets 309°and 344° with respect to SHmax at 328° do not nec-
essarily imply that these three sets formed simultaneously (Fig. 3.10). The speculation
depicted in Fig. 6.13 needs to be further investigated in additional fracture provinces
before becoming a sound model. The investigation is particularly necessary because
this scheme considers creation of joints at small angles with respect to principal stress
directions, which is a deviation from the conventional concept of co-axiality of joint
formation with respect to the direction of SHmax.

6.3.9.2
The Significance of FSR Values

It is not really clear what is the genetic significance of the high FSR values and the
high scatters of FSR values obtained for set 344° by Levi (2003) (Sect. 6.2.2). One ex-
planation relates them to accumulated strains after repeated stress events (earth-
quakes) along an extended period of time (Eyal et al. 2001). Levi (2003) shows high
FSR values for both B (set 344°) and D (set N–S) (Fig. 6.5). He considers that the lat-
ter two sets originated in the early Miocene and Plicetocene, respectively. This raises
the question of how the high FSR value could be formed in set D during such a short
time, since the Plicetocene, if the mechanism is repeated stresses along an extended
period of time (Eyal et al. 2001). In addition, both Eyal et al. (2001) and Levi (2003)
found high scatters of FSR values for set 344° (Fig. 6.5), implying that non uniform
processes exerted on the rocks and perhaps suggesting more than a single mechanism.
Perhaps an alternative explanation is more plausible (Levi 2003). Set A (326°)
formed in the burial stage under increasing overburden conditions, while the rock was
not fully solid, and therefore, displayed a fairly uniform FSR distribution. On the other
hand, the other sets (B–E) developed during uplift processes (Levi 2003). The uplifts
took place under reduced overburden loads and increased tension (Bahat 1999a), so
that the rock encountered lower resistance to jointing, with consequent increase of
both the FSR values and their scattering.
 6.3 · Paleostresses in the Israeli Negev 483

6.3.9.3
The Global Stress Model

Zoback (1992) presented a model of global stress in the lithosphere consisting of first-
and second-order patterns. The first-order midplate stress fields are believed to be
largely the result of compressional loads applied at plate boundaries, primarily ridge
push and continental collision. These regional uniform stress orientations often extend
20–200 times the approximately 20–25 km thickness of the upper brittle lithosphere.
The second-order stresses are local perturbations that are associated with specific geo-
logic or tectonic features, including lithospheric flexure, lateral strength contrasts, as
well as lateral density contrasts, which give rise to buoyancy forces. These second-or-
der stress patterns typically have wavelengths ranging from 5 to 10 times the thickness
of the brittle upper lithosphere. Often the second-order stress field results in a rotation
of the horizontal stresses. The distinction between first- and second-order scales and
mechanisms is appealing. Such ideas have been entertained by various investigators in
resolving “misalignments” in the field (Currie and Reik 1977).
It appears that on account of great scale differences, there is no clear correlation
between the rotation model linked with the continental collisions in the Mediterra-
nean region (Sect. 6.3.2) and the global stress model. The scale of the first-order, which
may fit that of the Mediterranean region, does not involve a rotation of the horizontal
stresses, and the Mediterranean region is far greater than the scale of the second or-
der. However, a “compromise model”, pertaining to a “local perturbation” the size of
the Mediterranean region may, with some terminological adaptations, fit the global
concept by Zoback (1992)?

6.3.9.4
The Timing of the Strike Parallel Joints

Much emphasis has been placed in previous sections on various aspects of cross fold joint
formation. Also important is the relative timing of cross fold and strike parallel joints in a
given fold (F1 and F2 joints, respectively in Fig. 3.20b) or a slightly folded area, like the
Appalachian Plateau or the Beer Sheva syncline. It is generally accepted that F1 joints form
in response to σ1 compression parallel to layering and normal to the fold axis, during early
stages of folding in both (approximately) symmetric (Burger and Hamill 1976) and asym-
metric anticlines (Reches 1976). What about the timing of F2 joints? Burger and Hamill
(1976) argue that calcite data that reveal a state of stress associated with the time of
F1 joints formation on the Teton anticline (Friedman and Stearns 1971) contain no evi-
dence for stresses related to F2 fractures that were also developed. Therefore, it is pos-
sible that either the stresses associated with the development of F2 joints were main-
tained for too short a time for calcite to undergo appreciable twin gliding or that
F2 fractures formed at very low mean stress. There are, on the other hand, indications
of time alternation of F1 and F2 joints in the two sets, in slightly folded synclines (Bahat
1991a, Fig. 4.8), and late reoccurrence of F2 joints when local stresses were resumed
(Gross et al. 1997). Many F2 joints probably formed during relaxation of the remote stress
(Price 1974; Bahat 1989, see also Dunne and North 1990). It seems that generally, local
principal stress rotation during advanced folding of asymmetric anticlines (Fig. 3.20a)
may be sufficient to open joints against the remote compression (i.e., strike parallel
484 Chapter 6 · Assorted Problems in Fracture Geology

joints). Hence, the timing and sequence of F2 joints can be quite variable in a fold, and
decisions regarding their timing should be taken with great caution until more deter-
ministic criteria become available.

6.3.9.5
Jointing in Limestone

It is our impression that the frequent inferior qualities of joints in limestone, in terms
of joint straightness and flatness, azimuth spreading in a given set and rarity of frac-
ture surface morphology, seems to be significant compared with those in chalk. If this
impression is sound, it would remain a challenge for combined efforts of experts in
structural geology and limestone petrology to resolve this issue.

6.4
Analysis of the “Fat Tail” of Joint Length Distributions
6.4.1
Joint Length Distributions

A “fat tail” is a characteristic of statistical distribution, where the decay of the distri-
bution for “large values” shows a power law behavior and does not conform to either
negative exponential or Gaussian forms. Although the statistical “fat tail” data treat-
ment characteristically consists of small populations, these populations are exception-
ally important because they are represented by the largest individuals in the system
that for certain applications are the dominant ones. For instance, Bahat and Adar (1993)
showed that from a series of joint sets around Wadi Naim in the Beer Sheva syncline,
Israel, one particular set (that was oriented at azimuth 062°) provided by far the best
water flow conditions. This set consisted of the longest fractures in the area, through
which most of the water drainage took place (Sect. 6.9). What is the connection be-
tween the Naim set 028° (Table 3.5 and Fig. 3.36a,b) and the fat tail (FT) distribution?
We study the fracture length histograms of two joint sets in detail: First, a set cut-
ting Middle Eocene chalks along Wadi Naim in the Beer Sheva syncline, Israel
(Fig. 6.14a), and second, a set from Ward lake, California (Fig. 6.14b). We focus on cer-
tain characteristics of the longest joints at the “end” of the distribution curve. The joints
of the first set that were formed by extension during Tertiary uplifts (Bahat 1999a)
are fully revealed along Wadi Naim, showing a bimodal distribution. Bahat (1988b)
demonstrated bimodal distributions in additional joint sets, compared to previous
suggestions that joint length in rocks had Weibull, normal, or power-law distribution
(e.g., Segall and Pollard 1983; Kelner et al. 1999; Zhao et al. 2000 and references therein),
and this issue remained an open question until now.
The second distribution (Fig. 6.14b) was considered by Segall and Pollard (1983)
to conform to a power-law distribution with powers of ~1.8. The fits of their curve to
the actual distribution is not clear. Trying to fit the whole distribution by a power law
has two shortcomings:

1. The decrease of frequency for small values of joint lengths is ignored.

2. The bimodal character of the distributions is overlooked.
 6.4 · Analysis of the “Fat Tail” of Joint Length Distributions 485

Fig. 6.14. Three histograms of joint lengths (note lithology and scale differences). a Set 028° from
Middle Eocene chalks (Bahat 1988b); b From granodiorite Ward Lake (Segall and Pollard 1983, Fig. 8b).
c Lineaments (modified from Nur (1982) after Kowalik and Gold (1976))
486 Chapter 6 · Assorted Problems in Fracture Geology

It is the second problem that is considered here. Apparently, a similar bimodality

of fracture length population characterizes findings in much larger scales (Fig. 6.14c).
This distribution was considered to be exponentially dependent (Nur 1982). As is pres-
ently shown, the second peak of the distribution marks the fat tail, which is indeed
given by a power law distribution. The latter, however, has different power values than
those used by Segall and Pollard and pertains only to the fat tail and not to the com-
plete joint length distribution.
Drawing for the Wadi Naim set, the cumulative distribution of length, P(x), i.e., the
number of all crack lengths larger than or equal to x as a function of the fracture
length x, on a log-log plot, one obtains (Fig. 6.15a) a reasonable FT (albeit only for one
decade) with a slope of around 2.8. A similar graph (Fig. 6.15b) of the Segall and Pol-
lard (1983) results shows a similar FT behavior with slopes of ~3.3 (for their Fig. 8a
results) and ~2.8 (for Fig. 6.14b). Although FT results only continue along a decade or
so, they definitely do not conform to exponential or to faster decreases.

Fig. 6.15.
The cumulative distribution
of length, P(x) = number of
all crack lengths larger than
or equal to x, as a function of
the crack length x, on a log-
log plot. Straight lines: Linear
regression. a Results from
Wadi Naim (length in cm);
b results from Ward Lake
(length in m)
 6.4 · Analysis of the “Fat Tail” of Joint Length Distributions 487

6.4.2
The “Fat Tail” Phenomenon

There exist many situations that show distributions having FT. These include phenom-
ena in economics (Mandelbrot 1960; Mirowski 1995), physics (Krapivski and Majumdar
2000), earth sciences (Sornette et al. 1996; Caers et al. 1999a,b; Kelly et al. 1994) and
are distributed in books and review articles (Arnold 1983; Adler et al. 1998). The con-
nection of FT with fractals has also been extensively discussed (Takayasu 1989). How-
ever, the exact origin of the FT behavior is hitherto unknown. Sand dune avalanche
distribution was explained by the theory of self organized criticality (Jensen 1998),
while some distributions with FT encountered a purely economic model for their ori-
gin. None of these models however can be used for the distribution of fracture lengths,
which we consider here.
The term FT relates to the following phenomenon. Consider the frequency distri-
bution of lengths of fractures found in a specific rock. The expected behavior of this
(or any) distribution is that it should decrease (for large lengths) at least exponen-
tially (say for a Poissonian distribution) or even normally (large numbers rule, Feller
1971, p. 219). There are, however, quite a few distributions for which the frequency,
especially for large values (tails) of the measured quantity (large fracture lengths for
the example), decreases as some power of this quantity – a FT. It is the purpose of this
section to address such a behavior of the distribution of fracture lengths measured in
rock layers and to discuss the possible origin of this FT.

6.4.3
The Model

According to Wu and Pollard (1992) (see also Sect. 1.2, second scenario), fractures in
a layer develop in two stages. In the first stage, fractures appear at random sites until
a certain distribution of spacing saturation (Sect. 1.5) is obtained. In the second stage,
almost no new fractures are created but the already existing ones grow in length. We
adapt this two-stage model (with some modifications) and treat the second stage here.
Consider a fracture of length x. Suppose its closest neighbor is another fracture of
length x'. According to Segall and Pollard (1983, Fig. 11) and (Nur 1982), the fracture
of length x can (and will) grow if x > x', while the x' crack will not grow. This interac-
tion between x and x' is valid, provided both cracks belong to a set and the spacing
between them is less than ~5 times the average inter-crack spacing in the material.
Since we treat parallel cracks after they have reached saturation, we adhere to the above
assumption: Namely, x grows if larger than x', its nearest neighbor (nn).
The probability that x will grow is therefore

(6.1)

where p(y)dy is the probability _that the nn be at a distance (spacing) of between y and
y + dy from the x fracture, and g is the conditional probability that if the fracture is of
length x and if its nn is at a distance y from it, than this nn is of length x'. We now as-
488 Chapter 6 · Assorted Problems in Fracture Geology

_
sume that g(x, y, x') is independent of either _ x or y. This is a plausible assumption but is
not always valid. Under this assumption, g(x, y, x') = g(x'), the probability that a frac-
ture is of length x'. Note that the integrals on y and on x' of Eq. 6.1 become disconnected
and the first integral becomes equal to 1. The probability g(x) evidently depends on time
during the process, a dependence suppressed here but to be discussed later on.
Denote by

(6.2)

which under the previous assumption gives the probability that a fracture of length x
grows. Note that

Next, we calculate the amount of growth of an x-long fracture. The exact details of
this calculation are not of major importance. It is only the scheme that is relevant.
Assume that there is a minimal value of extension stress σ00 that only when σ > σ00
the fracture can grow. Since the change in crack length ∆x is small with respect to the
fracture length itself, the criterion σ > σ00 is equivalent to KI > KI0 where KI0 is a mini-
mal stress intensity factor below which fractures do not grow at all (K0 in Fig. 4.34,
see also Sect. 1.5). Note that we speak about extreme subcritical conditions in the sense
that KI (and, a fortuori, KI0) are very much smaller than Griffith’s KIc, the value of KI
above which the crack starts to run at great speeds. During a time interval ∆t (which
can be very long, in a geological process), the stress field has changed. In this interval,
there are assumed to have occurred N subintervals, ∆ti, during which σ was larger than
σ00 and cracks grew.

6.4.4
The Fractal Nature of the Tail

Suppose that the crack growth velocity (v) (Charles 1958, Sect. 1.5) is given by

(6.3)

where A and n are constants, and x is the crack length at the time t.
This is the accepted form for the velocity changes during creep for a large series of
materials. Note that the most important parameter in Eq. 6.3 is n, which characterizes
crack growth for a specific material and can be obtained experimentally from the slope
of a log v vs. log KI plot of actual data. Assuming that the crack growth, ∆xi, during
the subinterval ∆ti, is small, namely ∆xi << x, (Eq. 6.3) can easily be integrated. Since
KI is proportional to σ √⎯x , we have

(6.4)
 6.4 · Analysis of the “Fat Tail” of Joint Length Distributions 489

and the total increase in length of a crack of length x during the complete time
interval ∆t is

(6.5)

where

The balance equation for the probability (per unit time and per unit length) g(x, t)
to find at time t (measured along the process) a crack of length x, is as follows:

(6.6)

Here, the first term on the right denotes the probability that the crack had already
the length x at time t and stayed at this length, while the second term describes the prob-
ability that the crack had a length x – ∆x at time t, where ∆x is given by Eq. 6.5 and was
able to grow. This probability is proportional to S at that time. From Eq. 6.6 one gets,

(6.7)

Since by Eq. 6.2,

and using Eq. 6.5 the following nonlinear partial differential equation is obtained for S:

(6.8)

A separation of variables S = X(x)T(t) then yields

(6.9)

where γ is a constant and prime denotes a derivative with respect to the correspond-
ing variable. The time-dependent part is therefore

(6.10)

where t0 is the initial time and a = 1 / T (t = 0). Equation 6.10 is evidently valid as long as
t < t1, where t1 – t0 = a / u. For “short” times (t – t0) << a / u, T increases linearly with time:
T ≅ 1 / a + (t – t0)u / a. This linear increase is in agreement with Wu and Pollard (1992,
Fig. 6c). The fractal nature of the tail of the distribution is obtained from the x–dependent
part X(x). Assuming X(x) = η xm, where η is a constant, Eq. 6.9 can be fulfilled only if
490 Chapter 6 · Assorted Problems in Fracture Geology

n = 2 (–m + 1) (6.11)

A solution of Eq. 6.8 is therefore

(6.12)

or, for “short” times, choosing t0 = 0 as the starting time,

S(x,t) = (c1 + c2t)xm (6.13)

hence

g(x, t) = (c1' + c2't)xm–1 (6.14)

Equations 6.11–6.14 establish for this model the “power” character of the distribu-
tion g(x). Note that this character stems from the nonlinearity of Eq. 6.7, whose ori-
gin is the interdependence of growth of one fracture on its neighbors. Another con-
tributing factor is the Charles “subcritical” fracture velocity-KI dependence, Eq. 6.3.

6.4.5
Discussion and Conclusion

So far the final bimodal distribution (Fig. 6.14a–c) was considered. Here we explain
the process of its creation, starting from a “one peak” distribution. We consider a simple
example as follows. Assume that the normalized joint length distribution function at
the time of saturation is of the form shown in Fig. 6.16a and expressed as

(6.15)

Fig. 6.16.
a The crack length distribution
example of Eq. 6.15 with l0 = 1.
b The probability per unit
length that a crack of length x
and with the distribution of
Eq. 6.15 would grow (Eq. 6.16)
with ∆y = l0 = 1
 6.4 · Analysis of the “Fat Tail” of Joint Length Distributions 491

Here l0 is a scaling length. Note that the maximum of this distribution lies at

The probability p that a fracture of length between y and y + ∆y grows is the prod-
uct of the probability that the fracture be of such a length, i.e., g(y)∆y and the prob-
ability that its neigbours are shorter than y, namely,

For the distribution of Eq. 6.15,

and therefore,

(6.16)

This relation is depicted in Fig. 6.16b. It is easy to calculate (numerically) that the
maximum of p occurs at y ~ 1.03l0. Fractures of this length have the highest proba-
bility to grow. One therefore expects that with time, a decrease of population (dip)
would occur in the vicinity of this length and an increase of population would appear
beyond this point leading to a bimodal distribution (Fig. 6.17), which eventually
assumes the slope of Eq. 6.14. Note that for the distribution of Eq. 6.15 the ratio be-

Fig. 6.17.
Full line: The crack length
distribution example of
Eq. 6.15 with l0 = 1. Dashed
line: Schematic bimodal
distribution obtained after
length increase beyond
saturation, according to the
present model (see Fig. 6.16
and text)
492 Chapter 6 · Assorted Problems in Fracture Geology

tween the position of population dip (1.03 l0) and the position of the first maximum
(0.707l0) is 1.46. This ratio is characteristic of the specific joint length distribution and
is not general.
Returning now to the field results, firstly it is clear that a power of –2.8 for P(x)
implies a power of m = –3.8 for g(x) (Eq. 6.14). Hence, for the results of Segall and
Pollard (1983), as well as for the results of Bahat (1988b), n of Eq. 6.3 is ~9.6, by Eq. 6.11.
It is definitely within the lower range of n values obtained for rocks (Atkinson 1987,
Table 11.6). Let us stress that this is an indirect method to obtain the characteristic
creep value of n for these rocks. Since geological creep conditions cannot be repro-
duced in the laboratory, this is apparently the only method hitherto known to obtain
these values.
Note also that the values of n obtained in this method for chalk and for granite are
very similar in spite of rock differences. It might be argued that for such prolonged
creep processes, n values are smaller than those obtained in the laboratory (Atkinson
1987), and probably these values are within the same range for different rocks. This
general result should be tested for other bimodal crack length distributions.
In summary, the present section provides a new interpretation of the profile of joint
length distributions. Here, two sets of fracture length measurements are discussed. It
is shown that both show a FT behavior with a similar power. Next, a model based on
creep growth of cracks is developed to account for the FT behavior and simultaneously
explains the appearance of the bimodal behavior of the distribution. We hypothesize
that this model should be applicable to additional joint length distributions.

6.5
Estimation of Paleo Fracture-Stress
The methods of paleo-fracture stress estimation on joint surfaces were summarized
(Bahat and Rabinovitch 1988; Bahat 1991a, p. 229 and reference therein), and gener-
alized for outcrops that do not display the fracture origin (Bahat et al. 1999). Weinberger
(2001a) applied the method for estimating the paleostresses on joints from certain
dolomite layers of the Judea Group in central Israel. When the fracture origin is un-
known, the mirror radius (Fig. 2.1a,b) cannot be measured directly and needs be cal-
culated. The location of the fracture origin of the joint that is represented by the relict
fringe of en echelon cracks is uncertain (Fig. 6.18a) so that the mirror radius, r is de-
termined trigonometrically (Fig. 6.18c). The measured distances are given by 2y be-
tween the two tips of the visible fringe and by x (Fig. 6.18c). Other relationships are
z / sinα = 2r, where α is the angle produced by 2y and z; z2 = x2 + y2 and r = x2 + y2 / 2x.
For the derivation of r, two distances were determined at random locations along the
fringe boundary (Fig. 6.18a), resulting in two r values differing 8% from each other,
suggesting the error range in measurement. The error in r determination related to
distance measurement and to a slight ellipticity deviation of the mirror from circu-
larity is estimated as 15% (see detailed calculations in the above mentioned references).
In these calculations, the local paleo-fracture stress is obtained, i.e., the stress applied
on the entire joint surface. It is assumed that the stress is unchanged throughout the
fracture process. This stress differs from both the (regional) remote stress and the
magnified stress acting at the tip of the joint.
 6.5 · Estimation of Paleo Fracture-Stress 493

Fig. 6.18. The calculation of mirror radius when it is unmeasurable on the outcrop, applied on ex-
foliation joints striking N 27° E and dipping 42° NW on the southwestern side of Half Dome at Yose-
mite National Park. a Photograph of a curved fringe of en echelon cracks at left representing
a remnant of an exfoliation joint (whose origin has been at the right side of the outcrop) on a slice
that has been removed. b Scaled diagram and grid of a. The removal of the external slice exposed the
surface of a second exfoliation joint whose origin is identified at P by the meeting area of radial barbs
of a large plume. c Diagram showing the curved inner boundary of the fringe of en echelon cracks
and the angle α between 2y and z (from Bahat et al. 1999)
494 Chapter 6 · Assorted Problems in Fracture Geology

6.6
Fracture Variability in a Fault Termination Zone
6.6.1
Introduction

A fault termination zone is a segment boundary, often related to earthquake dynamics

(Schwartz and Sibson 1989), and is a potential major component in seismic activities (Aki
1979; King 1986). The maximum displacement in the middle of a strike-slip fault trans-
forms into other manifestations of strain around the tip of the fault where the slip ap-
proaches zero (e.g. Muraoka and Kamata 1983). Chinnery (1966) shows that following this
transformation, new regions develop around the tip of the fault in which the maximum
shear stress, instead of being reduced, it may increase considerably. Changes lead to a com-
plex geometry in the fault termination zone, including jogs, secondary faults, folds and
rotation of fracture components (e.g. Trudgill and Cartwright 1994). A left-stepping bend
with right-lateral shear produce contractional (anti-dilational) jogs, whereas stepping and
lateral shear of the same sense produce extensional jogs. Contractional jogs may control
rupture initiation and may impede fault growth (Sibson 1985; Scholz 1990, p. 155). Hence,
the special conditions that develop in the fault termination zone may create both stabili-
ties and instabilities, and understanding fault termination zones provides a key insight into
rupture reoccurrence, fault growth, and growth impedance.
Detailed field characterizations of fault termination zones are rare, mostly because
full exposures of such terminations are uncommon (e.g. Cruikshank et al. 1991;
Bayasgalan et al. 1999 and references therein). This section concerns a detailed study of
a termination zone of a contractional jog, characterized in chalk, which is a weak rock
and its failure has resulted in a unique combination of a primary fault and eight sec-
ondary structures. The relative ages and interactions of joints and faults are often enig-
matic. Previous analyses of curved joints near faults were presented by Rawnsley et al.
(1992) in a scale of tens of meters, and by Castaing et al. (1996, Fig. 6) in scales of hun-
dreds of meters. Castaing et al. (1996) observed that joints are both earlier and possibly
contemporaneous with the faults. The present section contains a detailed account of
relative ages of various components of a strike-slip fault system and a joint set on a
small scale (length of 10–20 m in plan). This exposure may provide a template for ter-
minations of larger strike-slip faults.

6.6.2
Description of the Fault Termination Zone

The fault termination zone is confined to a single layer (layer 1 in Fig. 3.39b and 6.19). The
map of this zone (Fig. 6.20) is a product of a mosaic of forty overlapping photographs.
The pictures were taken at 1 m height at an angle of 90° to the outcrop face. The resolution
of the map is ≤10 cm. Details of fracture relationships are supplemented by photographs
(Fig. 6.21a–f). The fault termination zone consists of eight fracture elements:

1. A straight fault and a gouge zone (Pr in Fig. 6.19, 6.20 and 6.21a).
2. A curved secondary fault S1.
3. Two left-stepping, secondary faults, S2 and S3.
 6.6 · Fracture Variability in a Fault Termination Zone 495

4. A bulge caused by a local uplift in the northeastern area that surrounds the tip of Pr.
5. A set of curvilinear joints that vary in orientation sympathetically with the faults.
6. Cataclastic fault rock along certain parts of the faults.
7. Transverse cracks perpendicular to Pr in the bulged area.
8. En echelon cracks cutting the curved fault.

Fig. 6.19. A general view of the fault termination zone. The primary fault Pr (is 3.75 m long), the
three secondary faults S1, S2 and S3, set 344°, layers 1 and 2, joint set 028° in layers 1 and 2, and set set
062°: θ is the branching angle of S1 from Pr

Fig. 6.20. A map of various components of the fault termination zone in layer 1: Curvilinear joint set 344°,
the primary fault Pr, traverse cracks t, three segments (S1–S3) of the secondary faults and the termina-
tion at T. A single joint that crosses the termination zone is highlighted by two lateral arrows. The fault
traces are slightly emphasized by rock splintering. The dominant joint set 344° crossed by younger joint
set 062°. Note frames (4a–f) of Fig. 6.21a–f on map. Frames for Fig. 6.21a and b almost coincide
496 Chapter 6 · Assorted Problems in Fracture Geology

Fig. 6.21. Close-up photographs of parts of the fault termination zone. a, b The end of primary fault Pr,
the curved fault S1 and a gap (black shadow) at the contact area between the two. The shear zone along the
primary fault is populated by shear fractures (on the right side of 30 cm scale). Joints of set 344° are paral-
lel to the primary fault. At the right side of the primary fault is a straight joint j that sub parallels to the
joint J. At right-center of picture is a medium-sized joint trending approximately 344° and curves towards
joint j. Transverse cracks t sub orthogonal to set 344° around the contact area of Pr with S1. Photographs a
and b were taken from different angles. S1 starts at the end of Pr where the maximum uplift of 18 cm (in
shadow) occurs. S1 is cut by two en echelon crack sets e (some of them sub parallel to the meter scale).
Joint j next to Pr differs from Joint J next to S1. Transverse cracks t occur at left side of end of primary fault.
c Two manifestations of cataclastic fault rock along the second part of S1 (diagonal in picture). Rock frag-
mentation occurs between walls of a fault zone (between inclined short arrows) and many short joints (Sf)
occur at right of fault but do not cross it. Right-lateral 2.5 cm displacement of an early straight, long joint
along the fault (vertical arrows). An earlier fault would have prevented the long joint from crossing the fault
zone. 30 cm scale parallels the 344° joints. d The curved fault S1 terminates at an early joint where slight
rock fragmentation ends (below 30 cm scale). e The initial part of the first of two left-stepping faults S2.
Note intense rock mosaics of short joints and rock fragmentation along the fault. Scale is 30 cm. f A pen
(13.5 cm long) points to the fault termination location T (see Fig. 6.20). S2 and S3 approach this location
from left to right along joints. They bound a stretch of gradually reduced width with loss of cataclastic fault
rock and increased joint spacing. At the right of pen jointing of set 344° initiates a “normal” distribution
 6.6 · Fracture Variability in a Fault Termination Zone 497

The primary fault (Pr) and the gouge zone. This is a straight, vertical fault, which
strikes 318° along 3.75 m. The extent of the fault continuation into the cliff is unknown
(Fig. 6.19); it curves along the strike before entering into the cliff. There is a gouge
zone along Pr that reaches the width of 7 cm at its tip (Fig. 6.21a). The gouge zone is
sheared, resulting in slightly curved left-stepping R-shears resembling the right-lat-
eral shear zones obtained experimentally by Bartlett et al. (1981, Fig. 6).

The fault S1: Its gap with Pr and the bulge. The curved fault S1 is vertical. It forms a
sickle-like fracture in combination with the primary fault, such that S1 appears to be
right stepping with respect to Pr, in a left-stepping system (see S2, S3 and Fig. 6.23
below). The northeastern fault wall (Fig. 6.20) is folded upward, bulging along the end
of Pr and the beginning of S1 such that there is a gradual increase of vertical offset in
layer 1 from 0 to 18 cm along the two faults, producing a crest at the contact area be-
tween them (at the Pr-S1 contact in Fig. 6.22). This crest is restricted to the “uplifted
side” (note that layer 1 is higher on the left side of Pr than on its right side (Fig. 6.19)),
as distinguished from the “lowered side” of the fault (in the southwestern side of the
sickle-like fracture). Thus, bulging on one side of the fault introduced a distinction
between uplifted and lowered sides along Pr and S1 and also along S2 and S3 (Fig. 6.22).
The angle at branching initiation of S1 from Pr, θ, is 50 ±5° (Fig. 6.19).
There is a gap between the fault walls that continues laterally along S1 until it gradu-
ally dies out (Fig. 6.21a,b). There are some remnants of secondary material in this area
(gouge?), suggesting that much more of it had filled the gap before being washed away.
The gap is irregular and difficult to reconstruct to its original condition. This recon-
struction may be achieved in two alternative ways. A good fit is obtained when the
down thrown side is brought in contact with Pr by laterally overlapping the shear zone.
This would close S1 and would leave about 0.5 cm right lateral displacement along Pr.
On the other hand, a left lateral slip of 2.5 cm of the down thrown side along the gouge
would also close S1 quite well. Both reconstructions would indicate an original right
lateral displacement along Pr and S1.
S1 is divided into three parts that expose different styles of interaction with the
joints of set 344° (see below). The first part, the one closer to the tip of Pr is 2.9 m long
along its curved contour and has a smooth surface. Joints of set 344° seem to arrest at
the first part of S1. Joint j (at the lower part of Fig. 6.21a) is straight while joint J at the
upper part of the figure wiggles somewhat, and they do not appear to have been a single
joint before separation by the fault. The second part of S1 is 2.4 m long and partially
appears as a fault zone containing cataclastic fault rock (Fig. 6.21c) that becomes less
abundant towards its end (Fig. 6.21d). A joint of set 344° appears to be right-laterally
displaced 2.5 cm along the second part of S1. This joint is straight on both sides of the
fault and does not show any interaction with the fault that would indicate any “aware-

Fig. 6.22. A simplified perspective of fault termination zone. The vertical displacements along Pr, S1,
S2 and S3 are shown by black areas. Note bending along Pr and S1 and rhythmic maxima for the
various faults
498 Chapter 6 · Assorted Problems in Fracture Geology

ness” of the fault presence (Fig. 6.21c). Thus, the fracture contacts reveal that the
first part of S1 preceded the jointing next to it, while jointing occurred before the
second part of S1, implying intermittent growth of either faulting or jointing there.
There are in addition short joints (see below) that arrest at the second part of S1
(Fig. 6.21c). Along its third part, which is about 35 cm long, S1 makes a slight right
turn towards its end, cutting a few medium-sized joints (see below) before coinciding
with the direction of one of the joints (Fig. 6.21d). Cataclastic fault rock is minimal
along the third part of the curved fault.

Two left-stepping faults. Two left-stepping faults (S2 and S3 in Fig. 6.19 and 6.20)
occur near the curved fault. S2 starts about 4 m and 0.5 m from the closest points
of Pr and S1, respectively. S2 is slightly concave toward the southwest. It overlaps about
0.7 m with the curved fault. At 2.15 m distance from its northwestern end, the fault
splays with a branch about 1.30 long at a 20° angle. S2 crosses joints of set 344° near
its early part (Fig. 6.21e), and then it coincides with a joint. The splay of this fault also
coincides with a joint. S3 is almost straight, overlaps S2 and runs entirely along a joint.
The distance between S2 and S3 varies from about 40 cm at their northwestern ends
to about 30 cm at their southeastern termination.
The two left-stepping faults terminate in close proximity to each other (Fig. 6.21f,
at T). The intensity of cataclastic fault rock declines to almost zero beyond T. The num-
ber of joints diminishes from some ten before T to three beyond this termination. Also,
the opening along the fractures reduced at T from 1–2 cm to 1 mm or less.
The folding to bulges along S2 and S3 have maxima (11 cm and 8.5 cm, respectively)
at the northeastern up-thrown part of each fault, and they vanish along the verti-
cal fault planes towards the southwestern ends (Fig. 6.22). S2 and S3 differ signifi-
cantly in orientation from S1. Whereas S1 strongly deviates from the direction of the
primary fault Pr, S2 and S3 start to propagate along strikes, approximately in continu-
ation with Pr (Fig. 6.20 and 6.21). While fault S1 terminates with only little cataclastic
deformation along a joint of set 344°, fault S2 initiates with abundant cataclastic fault
rock (Fig. 6.21e).

The joint set 344°. This set displays a strong curvilinearity along which growth oc-
curred in several stages (generations). The joint set is spatially restricted to layer 1 in
a 7 m wide area around the faults. The area is bounded by talus along the northeastern
side and by a joint set oriented 028° along its southwestern side (at the right of Fig. 6.19
and 6.20). The joints of set 344° change strike in “four zones” of the termination zone
as follows. The joints are approximately straight and parallel to Pr along the fault.
Beyond the tip of Pr the joints curve sympathetically with respect to the secondary
faults, S1, S2 and S3, forming the “curvilinear zone”. On the southeastern side, beyond
the final fault termination (T in Fig. 6.21f) and throughout the sourthwestern “low-
ered side” of the fault all joints are straight with uniform 344° orientation.
A joint oriented 344° in the “lowered side” curves and “hooks” into a straight joint
that parallels to the primary fault (Fig. 6.21a), suggesting a fracture sequence between
the joints, because the former joint “knew” about the existence of the straight joint. At
least one joint approximately follows the shape of S1: First, the concavity of S1 towards
northeast and then, its concavity towards southwest, towards the termination of S1.
 6.6 · Fracture Variability in a Fault Termination Zone 499

(Fig. 6.19 and 6.20). Some joints however are crossed by the secondary faults at acute
angles ~40° or at smaller angles ~5–15°. Individual joints of set 344° may be up to 10 m
long with only very little interaction with each other. Hence, although these joints curve
and vary in orientation, they are termed set 344°.
We divide the joints of set 344° into three groups according to their lengths. The
“long”, “medium-sized” and “short” fractures are up to 10 m, 4 m and 1 m long, re-
spectively. The long and medium-sized joints are parallel to each other. Near S1 and
along S2 and S3, medium-sized joints fill gaps between long joints (Fig. 6.20), reduc-
ing spacing to about 2 cm. However, further away from the faults S2 and S3, towards
the southeast end of the outcrop where long joints are oriented 344°, their spacing
becomes larger and ranges from 10 to 25 cm, not having the medium-sized joints
among them. Thus, the medium-sized joints appear to postdate the long joints.
The long and medium-sized joints cut by S1 are distinguished from the short joints
not cut by S1 (Fig. 6.21c). The medium-sized joints appear to be quasi-systematic
(straight and parallel to each other), whereas the short joints are non-systematic and
somewhat distorted. Joints of set 344° are generally quite closed (an aperture of
about 0.1 mm or less). However, some of them widen their apertures up to about 1 cm
along the second part of S1, and particularly, along the stretch between S2 and S3.

Cataclastic fault rocks. In certain locations, the country rock is intensely ruptured
into fragments that range in size from several millimeters to several centimeters. The
“mosaic style” occurs along the second part of S1 (Fig. 6.21c), where rock fragmenta-
tion occurs between two walls of a fault zone. A mosaic is also spatially associated
with short joints that either start or arrest at the southwestern side of the fault zone,
on which perpendicular younger cracks may arrest, creating a rock mosaic of square
and angular splinters. This style is also in association with S2, but the mosaic next
to S2 (Fig. 6.21e) is more intensely populated. Mosaics extend up to 30 cm from a fault,
mostly occurring along the concave side of the faults (Fig. 6.21e), where the second-
ary faults cross the 344° joints, particularly where joint spacing is small. Fragmenta-
tion also occurs as rock splintering not associated with rock mosaics. They are par-
ticularly found between S2 and S3. This splintering also extends to distances of up to
about 12 cm across the two sides of the faults.

Transverse cracks perpendicular to Pr in the bulged area. The bulged area close to
the tip of Pr is intensely fractured by transverse cracks. The cracks are short (about
12 cm), trending approximately orthogonal to the primary fault (t in Fig. 6.20 and
6.21a), and generally arrest at joints of set 344°. The transverse cracks cut part of the
gouge along the primary fault (they are not visible in Fig. 6.21a).

En echelon cracks. Two sets of en echelon cracks cut the bulged area of the first part
of S1 at its continuation from the tip of Pr (e in Fig. 6.21b). Their azimuths are 330°
and 350°, and their crack length averages are about 22 cm and 14 cm, respectively.
Cracks of set 350° generally arrest at cracks of set 330°, implying that set 330° gener-
ally was older. Rock splintering associated with these cracks is not as intense and
irregular as the cataclastic fault rock along the second part of S1 and the left-stepping
faults.
500 Chapter 6 · Assorted Problems in Fracture Geology

Table 6.1. Ratios of fracture parameters in the fault termination zone

6.6.3
Fracture Parametrization in the Fault Termination Zone

Ratios of fracture parameters. Assuming similarity of strike-slip faults in various scales

and therefore similarity of their termination zones, ratios of fracture parameters are
possibly more important than absolute values. Some ratios of fracture parameters are
summarized in Table 6.1. It turns out that the “straight” lengths of S1 and S2 (ignoring
curvature along the fault) and the distance from the tip of Pr to the initiation of S2
(Fig. 6.20) are about the same.

Fracture sequence in the fault termination zone. The great versatility of fracture
styles and their clear exposure enables one to decipher the sequence of most fracture
processes in the fault termination zone (Table 6.2). The right-lateral offset along Pr
resulted in creating a wing crack (the first part of S1) and a gap at the contact area
between Pr and S1. More than a single offset along Pr (repeated S2 and S3 faults and
e cracks) resulted in the deformation in the termination zone. Bulging in the area
close to the tip of Pr has resulted in transverse cracks oriented 025° (characteristic to
crests of outer arcs, Fig. 3.19a). These cracks arrest at the 344° joints that parallel to Pr
in the “uplifted side” of the fault, suggesting that maximum bulging occurred after the
formation of 344° joints. On the other hand, joints of set 344° seem to arrest at the first
part of S1, suggesting that the first part of S1 preceded the joints. Also, in the “lowered
side” a joint oriented 344° curves and “hooks” into a straight joint that parallels Pr,
suggesting a time interval between these joints. Assuming that the straight parallel
joints on both sides of Pr were formed approximately simultaneously, it should follow
that the lateral component of deformation creating S1 preceded the vertical offset
associated with the bulging. That is, when the lateral offset(s) was exhausted, further
stress release resulted in vertical stresses producing the bulging.
The onset of cataclastic deformation along the second part of S1 suggests a change
in mode of fracture at the transition from the first part of S1. This change is not sur-
prising, considering the change in orientation along S1 with respect to the 344° azi-
muth, which is taken to be the σ1 trajectory of the stress field that created Pr (Fig. 6.20).
In analogy to previous observations (e.g., Brace and Bombolakis 1963; Jayatilaka and
Trustrum 1978), the second part of S1 was formed by a stable process under condi-
tions of increasing stresses. This part of S1 postdated a 344º joint (Fig. 6.21c). The im-
plication is that the second part of S1 developed discontinuously (incrementally) with
 6.6 · Fracture Variability in a Fault Termination Zone 501

Table 6.2. Sequence of fracture episodes of joints and faults in the fault termination zone

respect to the first part of S1. The curved fault S1 terminates with only little cataclastic
deformation along a joint of set 344° (Fig. 6.21d), whereas the initiation of S2 is asso-
ciated with intense cataclastic deformation (Fig. 6.21e). This suggests that an early epi-
sode of secondary faulting ended with the termination of S1, but the creation of S2
manifests a new fracture episode that occurred when high stresses were attained.
Many short joints arrest at S2. Since S2 postdated S1, the joints that interact with S2
quite likely postdated those that are associated with S1. Spacing of set 344° in various
areas adjacent to the secondary faults is reduced down to about 2 cm. The reduced spac-
ing reflects multi-stage jointing of medium-sized and short joints: Late joints fractured
between and parallel to previously long joints. It has been observed in many outcrops
of syntectonic joints (joints associated with intense deformation) cutting chalks that
these fractures generally form by more than a single failure event (Bahat 1991a, p. 282).
Generally, cataclastic fault rock is thought to form concurrently with faulting. How-
ever, it should not be ruled out that the mosaics that occur close to one boundary of a
second fault (center of Fig. 6.21c) could also reflect post fault energy release. The ques-
tion marks in Table 6.2 indicate that certain parts of the fracture sequence are not cer-
tain. Episode 7 is marked by ‘?’, because only one medium-sized joint is cut by the sec-
ond part of S1. The reason for marking ‘?’ in relation to episodes 9, 12 and 15 is that
short joints may be contemporaneous with the adjacent faults but could also be post
faulting.
502 Chapter 6 · Assorted Problems in Fracture Geology

Two gap types. The gap continuation along the first part of S1 (Fig. 6.21a,b) is a result of
a mode I fracture. Closing the gap into its original condition before faulting by a left lateral
slip of the down thrown side along the gouge would create an appearance of a conti-
nuation of j- and J-joints across S1 and may suggest that the joint predated S1. On the
other hand, closing the gap by contacting sidewise the down thrown side with Pr (over-
lapping across the shear zone) may result in an offset of joints j and J across S1. This would
fit the model of joint j arresting at the earlier S1, becoming the favored explanation, imply-
ing that the shear zone developed only after the separation of the two walls of Pr.
It may be speculated that the growth of some primary faults and their termination zones
to large (continental) strike-slip faults can be explained by a stepwise process. This takes
place by reoccurring bridging of the gaps between the tip of the primary fault and the ini-
tiation of the first or second en echelon faults (Pr, S2 and S3, respectively, in Fig. 6.20). Such
a process might shift the location of maximum stress intensity from the tip of Pr to the
termination of S2 and/or S3 (point T) before the next earthquake would take place.

Models of termination zones and experimental results. There is an important dif-

ference between two existing models of termination zones. In one model of a right
lateral offset, all secondary faults maintain left-stepping configuration, splaying to-
wards one side of the continuation of the primary fault (Fig. 6.23a) (Hobbs et al. 1976,
Fig. 7.18d and Twiss and Moores 1992, Fig. 7.7). Chinnery (1966) suggests two pos-
sible theoretical combinations of straight primary and curved secondary faults. One
of them consists of a sickle-like structure that resembles the present combination

Fig. 6.23. Comparison of previous models of fault termination zones to the present observations for
right lateral strike-slip faults. a Splay faults branching from primary fault near its terminus (modified
after Hobbs et al. 1976). b Trajectories of maximum principal stress direction shown by dashed lines at the
termination zone of a strike-slip fault (from Means 1979, after Chinnery 1966). c Miniature of fault
pattern from present map in Fig. 6.20. d Compression applied at a small angle to primary fault P, resulting
in secondary fractures w and e (see text for explanation) (modified after Petit and Barquins (1987, 1988)
 6.6 · Fracture Variability in a Fault Termination Zone 503

of Pr and S1: The trajectories of the maximum principal stress at the termination zone
of a strike-slip fault (Means 1979) resemble the contours of the present joint set 344°
(Fig. 6.23b and c, respectively). In the latter, the curved fault S1 appears to be the re-
sult of right stepping, apparently in conflict with the requirement of an exclusive left-
stepping arrangement of the first model (Fig. 6.23a).
Petit and Barquins (1987, 1988) demonstrate experimental results obtained from
PMMA and sandstone under uniaxial compression, which strongly resemble the frac-
ture geometry in the termination zone (Fig. 6.23d, see also Fig. 2.58a). In their experi-
ments, the “sickle” formed by Pr and S1 and the secondary fault S2 are represented
by P, w and e, respectively. The location and the angular relationship with respect to
the primary fault Pr suggest that S2 and S3 are analogous to the “shear zone” e that
Petit and Barquins (1987, 1988) produced under maximal stress conditions. They also
found that the wing crack resulted from mode I operation, which shows opening, and
the shear zone resulted from mode II loading (see Sect. 2.2.13). Also relevant is the
experiment by Petit and Barquins (1987, 1988) on sandstone that revealed cataclastic
deformation along the en echelon cracks at e, resembling the cataclastics found along S2
and S3. There is an additional resemblance between these experimental results and
the fault termination zone, in that w deviates from the direction of P, and e strikes ap-
proximately in continuation with P (Fig. 6.23d).

Bulging and vertical offset. Spatial constraints cause shear stresses around the tip of the
primary fault that are partly released by lateral displacement and partly by vertical offset.
The area in the termination zone may be divided into compressional and tensional re-
gimes. In a contractional jog, a vertical upthrow along one side of the fault would occur in
the compressional regime and a downthrow would occur in the tensional one (Chinnery
and Petrak 1968). Accordingly, bulging along the primary and secondary faults is ex-
hibited in the upthrown block (Fig. 6.19) as expected. The rhythmic maxima in the verti-
cal displacements along S1, S2 and S3 imply that these bulges occurred concurrently with
each individual fault. A single vertical displacement at a later time would be expected to
have been monotonous with a gradual change from maximum to minimum through-
out the termination zone rather than a rhythmic one. This rhythmic mechanism sup-
ports the suggestion of an incremental growth of the secondary faults (Table 6.2).

6.6.4
Open Questions

The student may like to ask several questions that need to be further explored:

1. How do existing left-stepping configurations (Fig. 6.23a) reconcile with the field
and experimental data shown in Fig. 6.23c,d?
2. How similar or different to this fault termination zone are termination zones of
other contractional jogs?
3. To what extent fault termination zones change with size?
4. Set 344° was induced by a perturbation at the tip of the fault (rather than along it,
Fig. 3.12a). There is a need to characterize the secondary stress field that formed
around the tip of the primary fault and induced the three secondary fractures (far
away from the Pr tip).
504 Chapter 6 · Assorted Problems in Fracture Geology

6.7
Fault-Joint Relationships in the Beer Sheva Syncline, Israel
6.7.1
Introduction

This section relates to eight distinct fault-joint systems in the Beer Sheva syncline from
south Israel that differ from each other in their genetic affiliation and their geometric
and age relationships, citing Bahat (2000, 2004).

6.7.1.1
Joints and Fault Classifications

Joints are the most ubiquitous structures in sedimentary rocks. This is because joints
may be formed under small differential stresses, and therefore most environments are
conducive to their development, both systematically and non systematically (Hodgson
1961b). Nelson (1979), Engelder (1985) and Bahat (1991a) proposed genetic classifi-
cations of joints. According to the latter classification, joints may be divided into four
groups: Burial, syntectonic, uplift and post uplift. These can be distinguished in the
field by both geometric and fractographic criteria. The four groups reflect four char-
acteristic stages in basin evolution.
The last fifty years of fault investigation enjoyed the benefit of the pioneering clas-
sification of faults into three classes by Anderson (1951). The great power of Anderson’s
grouping stemmed from its simplicity. It helped to demonstrate the dependence of the
three main fault types on systematic changes in the stress field that also reflected on
the most likely tectonic conditions. However, fault classification cannot be applied to
timing of basin histories. On the other hand, extended research on the temporal rela-
tionship between fault-joint relationships (FJR) (e.g., Sterns 1968; Pohn 1981; Cruiks-
hank et al. 1991; Gross 1995; Martel 1997; Kattenhorn et al. 2000; Wilkins et al. 2001; Eyal
et al. 2001) has accumulated a vast amount of knowledge that can be used in refining
the fracture histories of basins. The work by Peacock (2001a) is a recent summary of
some key observations related to this topic.

6.7.1.2
The Importance of Temporal Fault-Joint Relationships

The temporal fault-joint relationships are important for a variety of reasons that are
specified below.

1. Sequences between faults and joints may help to decipher the tectonic history of
the region (including earthquakes), and temporal relationships.
2. They also provide information on the stress history of the region.
3. Faults may affect later joint development and joints may transform to faults (Segall
and Pollard 1983), and these processes are mechanically intriguing.
4. Distinct joint/fault genetics may be linked with different fracture parameters (aper-
ture, length, spacing, interaction, etc.).
 6.7 · Fault-Joint Relationships in the Beer Sheva Syncline, Israel 505

5. Fluid flow (water, hydrocarbon liquids and gas) depends on fault-joint relation-
ships and their fracture parameters. For instance (Bahat 1991a, p. 239).
a Burial pre-fault joints may be closed due to their development in a process of
increasing lithostatic pressure;
b Some syntectonic, syn-fault joints may be sealed into veins, while others strongly
enhance water flow, like the strike parallel joints (062°) in the Beer Sheva syn-
cline;
c Uplift joints that form after burial-normal faults are opened fractures.

Unlike many other folds that exhibit relatively uniform structures, the Beer Sheva
Eocene syncline is divided into four distinct stratigraphic units that display different
fracture characteristics and store a record of many (more than twenty) different frac-
ture episodes since the Lower Eocene (Sect. 3.4). One manifestation of this multi-frac-
ture complex is the occurrence of at least eight distinct FJRs. Various aspects of FJRs in
this syncline were introduced in previous publications, including pre and post-fault
burial joints, syn-fault syntectonic joints and post-fault uplift joints that partly trans-
form into faults (e.g., Bahat 1991a, pp. 241, 275 and 282; 1998a, 1999a; Gross et al. 1997).
A synthesis of the previous observations constitutes the major part of this section, with
the objective of demonstrating the extent to which the estimation of timing of fracture
development, relative to the structural history of a syncline can be achieved. Price’s
(1974) model of fracture in basins is adapted here in analyzing fracture in the Beer Sheva
syncline, as in previous publications (e.g., Bahat 1989). A distinction is made in the fol-
lowing description of the various FJR types between the “primary faults” and the vari-
ous joints (or faults in one case) that are associated with them, which are collectively
termed “secondary fractures”. This distinction is not relevant to type 8 (see below).

6.7.2
Eight Types of Fault-Joint Relationships

Eight FJR types of fault-joint relationships from the Beer Sheva syncline (Fig. 3.33b) are
characterized below. Table 6.3 summarizes FJR that take into account different parameters:

1. Temporal relationships.
2. Geometric contacts.
3. Tectonic setting.
4. Stresses.
5. Fracture system.

Six types relate to normal faults, one type concerns the terminal zone of a strike-
slip fault and the eighth type is about vertical faults.
Type 1 joints occur in the Mor Formation of the Lower Eocene and is associated
with a normal fault from the burial stage of the syncline (Bahat 1985). The fault
strikes ~292° and dips ~45° N, cutting alternating chalk layers (about 90 cm thick) with
chert beds (about 10 cm thick) (Fig. 6.24a,b), while the chalk is dissected by two or-
thogonal, vertical, single-layer joint sets oriented 328° and 059°. Joints of the cross-fold
set 328° arrest at the boundary of the chalk layers with chert beds. The fault that dis-
506 Chapter 6 · Assorted Problems in Fracture Geology

Table 6.3. Eight types of fault-joint relationships (FJR) in the Beer Sheva syncline, Israel

places these joints contains in its fracture zone fragments of chert that had been bro-
ken during early offsets of the fault and unfractured nodules that precipitated after
the cessation of the fault activities. The precipitation of chert is associated with the
diagenetic (burial) stage of the chalk (Knauth and Lowe 1978). Chert does not occur
in the overlying Middle Eocene (Horsha Formation). Therefore, both the cross-fold
joints and the normal fault occurred before the sedimentation of the Middle Eocene,
i.e., they were formed during the burial stage by a process controlled by the tectonic
regime that prevailed in the Lower Eocene (Gross et al. 1997). The joint orientations
are unrelated to the fault attitude, requiring that the remote stresses had to change
from the conditions under which the joints were formed to new conditions that in-
duced the fault. Hence, during the burial stage of the syncline, differential stresses
gradually increased, such that the formation of the cross-fold joints and some strike-
parallel joints preceded the creation of the fault (Bahat 1985, 1991b; Gross et al. 1997).
Generally, joint orientation and spacing is almost the same around the fault.
Type 2 relates to the same fault that is shown in Fig. 6.24a,b but it concerns a local
spacing reduction in the hanging wall compared to the uniform spacing elsewhere
around the fault. Gross et al. (1997) found that due to dragging of one of the chert beds,
an unrelieved strain remained stored in the hanging-wall from the early stages of the
fault slip. This unrelieved strain was subsequently released in the form of ~055° ori-
ented joints. Hence, the fracture sequence during the burial stage was early cross-fold
and strike-parallel orthogonal joint sets, normal faulting and subsequently some ad-
ditional fold strike-parallel joints in the hanging wall.
Fig. 6.24.
Structural relationships of
faults and associated joints
from the Beer Sheva syn-
cline. a Two conjugate faults.
b An enlargement of the
fault dipping towards left, to
which types 1 and 2 relate.
The faults cross vertical
single-layer joints in Lower
Eocene chalks near Beer
Sheva. The two orthogonal
joint sets and the faults strike
in three distinct directions.
Thus, joints and faults were
formed in different remote
stress fields (after Bahat
1985). c Types 3 and 4. Type 3
is represented by inclined
joints that parallel the dip of
a normal fault in Wadi Secher
at the upper hanging wall
(marked j). Type 4 is shown
as inclined secondary faults
(f1–f3, marked 1–3, respecti-
vely) at various acute angles
to the primary fault in the
lower footwall. d Type 7.
Spacing of the vertical joint
set decreases considerably as
the joints come closer to the
normal fault, which is a char-
acteristic feature of syntec-
tonic joints. Therefore, the
primary fault and joints de-
veloped in the same remote
stress field. The black rect-
angle is 10 cm long
6.7 · Fault-Joint Relationships in the Beer Sheva Syncline, Israel
507
508 Chapter 6 · Assorted Problems in Fracture Geology

Two more joint spacing peculiarities were observed in the same outcrop (Bahat
1991a, p. 255). First, joint spacing is wider in thinner chalk beds, i.e., the layer thick-
ness-joint spacing plot shows a negative slope for both cross-fold joints and strike
joints, which is in contrast to most observations by other investigators that show posi-
tive plots (e.g., Price and Cosgrove 1990, p. 54). An exceptional finding of a negative
slope was also made by Eyal et al. (2001) in the chalky limestone beds Nahal Neqarot
in southern Israel. Second, for a given layer thickness, joint spacing is a little wider in
the strike-parallel joints (oriented 055°) than in the cross-fold joints (oriented 326°).
An explanation for these two observations requires further studies. It is possible that
spacing is wider in the strike-parallel joints because their formation was associated
with stress relaxation, compared to the cross-fold joints that reflected regional stresses
(Bahat 1989).
Types 3 and 4 joints outcrop along Wadi Secher (Fig. 3.33c) and occur in the Middle
Eocene chalks of the syncline. The joints of Type 3 are inclined and are approximately
parallel to the Secher fault dip (Fig. 6.24c). Fracture distribution around the fault is
explained by Fig. 6.25a–e. The remote stresses (σ1∞, σ2∞, σ3∞) have induced the normal
fault that dips towards left (Fig. 6.25a). A new, local stress field (σ1, σ2, σ3 in the shape
of a stress ellipsoid) was created by the slip (or slips) along the fault (Chinnery 1966).
A damage volume was formed along the fault by its initial formation and by its subse-
quent slips. The damage volume is the rock volume that contains secondary structures
(mostly fractures) that are confined to the fault, and it approximates the shape of an
ellipsoid (Fig. 6.25b). Two orthogonal planes can schematically describe the distribu-

Fig. 6.25. Diagram of primary normal fault and secondary fractures in three dimensions. a The nor-
mal plane (ellipse) that describes the arrangement of local principal stresses is superposed on one of
two conjugate faults in a rectangular block that describes the remote principal stresses. Orientations
of remote and local principal stresses differ significantly. These figures contain geometric elements
from Means (1979) and Park (1983) (see explanation to types 3 and 4). b Stress ellipsoid in perspec-
tive, showing its two principal planes, the fault plane σ1–σ2, which is termed the “parallel plane” with
its elliptical tip line boundary, and the “normal plane” σ1–σ3. c The “parallel plane” that contains the
fault plane with its elliptical tip line boundary. d The “normal plane” is divided into four sectors I–IV.
e The approximate shape of sector I in the normal plane of the Secher fault A
 6.7 · Fault-Joint Relationships in the Beer Sheva Syncline, Israel 509

tion of secondary fractures in a damage volume. The tip line forms an elliptical line
(Walsh and Watterson 1988; Knott et al. 1996) that marks the loci on the fault plane
(approximately the “parallel plane”) along which displacement has decreased to zero
(Muraoka and Kamata 1983; Walsh and Watterson 1988) (Fig. 6.25c). The “normal
plane” is orthogonal to the parallel plane and is divided into four sectors (I–IV) by
the trace of the fault and the tip line (Fig. 6.25d). For normal faults the extensional
sectors I and III are widest (where rock is weak) and narrowest in the contractional
sectors II and IV (where rock is strong) (e.g., Knott et al. 1996).
The direction of maximum shear stress in the initial field of remote stress
(σ1∞, σ2∞, σ3∞) is transformed to the direction of σ1 (Fig. 6.23b) in a simple 45° rotation
exercise (Means 1979, problem 12.4) during the formation of the local stress field. This
explains how the local minimum principal stress (σ3) was normal to the fault plane,
creating type 3 joints (Fig. 6.24c). Some twelve joints of type 3 are spatially confined
to sector I in the damage volume of the Secher fault (one or two joints occur in sector IV
as well). Unexpectedly, the width of sector I of the Secher fault was found to be about
60% of the length of the fault, implying a strong deviation of the normal plane from
an elliptical shape (Fig. 6.25e). Accordingly, the schematic shape of the damage vol-
ume (shown two dimensionally in Fig. 6.25d) is modified to a more realistic “quasi-
ellipsoid damage volume” (Fig. 6.25e). Type 3 joints resemble the (unspecified) frac-
tures that are parallel to the dip of the normal fault from the known results described
by Stearns (1968) (the fractures dipping towards left in Fig. 6.26). They also resemble
the FJR of large strike-slip faults in which joints cut parallel to the dip and strike of
the primary fault close to the fault termination zone (Chinnery 1966, Fig. 5).
Type 4 joints consist of a series of three fractures (f1–f3) that formed in sector III
of the normal Secher fault (Fig. 6.24c and 6.25d). Fractures f1 and f3 exhibit changes
in orientation relative to that of the fault. The fault strikes 040° and dips 51° to NW,
f1 strikes 035° and dips 68° to NW, f2 strikes 038° and dips 69° to NW and f3 strikes 088°
and dips 85° to NW. The changes reflect combined strike twist and dip tilting from an
attitude relatively close to that of the fault to an almost vertical fracture that totally
divorced from the trend of the fault. This reflects a response to local perturbed stresses

Fig. 6.26.
Rose diagram showing the dis-
tribution of dips of two sets of
fractures (unspecified) associ-
ated with a normal fault (after
Stearns 1968, and Twiss and
Moores 1992)
510 Chapter 6 · Assorted Problems in Fracture Geology

along the fault (Rawnsley et al. 1992) that were effective in initiating fractures f1–f3,
combined with a gradual increase of the influence of the remote stresses with distance
of the fractures from the fault (up to about 3.5 m). There is some shear displacement
along these fractures, which distinguishes them from the joints of the other types.
Type 4 resembles the fracture set that occurs at a large angle (>45° but <90°) relative
to the fault in the Stearns model (dipping to the right in Fig. 6.26). However, in the
latter the angles created between the fault and joints are larger than in type 4 (see more
about the model depicted in Fig. 6.25 in the discussion section). Type 4, like type 3 is
thought to be genetically related to displacements along the fault, in association with
an early uplift stage in the Beer Sheva syncline (see below).
Type 5 joints occur at a fault termination zone (Sect. 6.6), cutting a Middle Eocene
chalk layer during the uplift stage. The termination zone consists of a primary fault,
secondary faults, and a joint set in Wadi Naim (Fig. 6.19 and 6.20). The primary fault is
a vertical right-lateral strike-slip fault, striking 318°. A set of three partly curved sec-
ondary faults initiate at the tip of the primary fault, and a joint set striking 344° curves
sympathetically with the primary and secondary faults in a close proximity with the
faults. Therefore, the 344° striking joints are considered to be genetically associated with
the faults, and they are termed syntectonic joints. The 028° striking joint set is ubiqui-
tous in Wadi Naim. However, where these joints approach the termination zone they
arrest or interact (hooking style) with the 344° striking joint set, but they never pen-
etrate to the termination zone. Therefore, the 028° striking joint set postdated the above
three structures in the same chalk layer, i.e., they are younger than type 5 FJR joints.
Thus, sets 344° and 028° belong to two separate fracture events: An early and an ad-
vanced one during the uplift process (Bahat 1999a).
Type 6 joints are cut by the Naim normal fault A (Fig. 3.36g). There are no visible
differences in the behavior (spacing) of the joints from both sides of the fault. The strike
of this set forms an angle larger than 45° with the strike of the fault. Unlike the system
described by Kattenhorn et al. (2000) there are no genetic relations between the fault and
joints. These are single-layer uplift joints that were associated with the advanced uplift
in the syncline, but they had been preceded by the earlier 344° uplift set (type 5). There-
fore, the fault of type 6 was associated with an advanced uplift stage of the syncline.
Type 7 joints are adjacent to Naim normal fault B that occurs some 125 m west of
Naim fault A. The joints show significant decrease in spacing close to the lower part of
fault B (Fig. 6.24d), and as such they comply with a basic criterion of syntectonic joints
(Bahat 1991a, p. 279). Compared to types 1 and 6, which reflect joints and faults that
formed by different stress fields, type 7 represents an assemblage of joints that were
formed in the same local stress field as the fault. Thus, even if the faults from types 6
and 7 cut the same chalk layers and appear to belong to the same fault system, these
two types seem to show different FJR manifestations. Whereas the fault of type 6 is as-
sociated with joints that formed before faulting, the fault of type 7 created a local stress
field that formed syntectonic joints adjacent to it.
Type 8 joints occur in the Lower Eocene chalks around Beer Sheva in the form of
vertical multilayer joints that often traverse the entire exposed outcrops; they may run
alongside or across burial normal faults from type 1 (Fig. 3.35b). Often multilayer joints
are associated with vertical multilayer faults in irregular 5–20 m spacing. These verti-
cal fractures are considered to have been formed after the burial stage during a single
or multi uplift episodes of the area (Fig. 5.15 in Bahat 1991a).
 6.7 · Fault-Joint Relationships in the Beer Sheva Syncline, Israel 511

6.7.3
Discussion

6.7.3.1
Genetic Relationships and Relative Ages of Fractures

Types 1, 2 and 8 occur in Lower Eocene chalks. While types 1 and 2 are products of the burial
stage, type 8 formed much later by uplift (Bahat 1985, 1991a). The primary faults of types 3,
4, 5 and 7 were created by remote stresses, but their secondary fractures resulted from local
stresses that were associated with slip (or slips) along the respective faults, i.e., the primary
fault and secondary fractures of these types are genetically related. On the other hand, type 6
concerns a fault and neighbor joints that were genetically unrelated, because they were
formed by two distinct remote stresses. The genetically unrelated style of type 6 resembles
that of type 1 from the Lower Eocene, whereas the genetically related style of types 3, 4, 5
and 7 resembles that of type 2. Hence, the joints of types 1 and 6 are pre-fault secondary
fractures, whereas the joints of types 2, 3, 4, 5 and 7 are both post and syn-fault fractures.
These genetic relationships can help in working out the relative ages of types 3–7. The
secondary fractures associated with types 3 and 4 formed in pristine chalks and predated
the regional joints along Wadi Secher. On the other hand, types 6 and 7 occurred in chalk
layers that had already been jointed in previous fracture episodes. The five FJR types 3
to 7 occur in Middle Eocene chalks and originated during uplift processes that took
place in the syncline (Bahat 1999a). Also, types 3 and 4 are associated with the Secher
fault as they cut older chalk layers at the flank of the fold (Wadi Secher, Fig. 3.33c), while
types 5–7 are exposed in younger layers closer to the synclinal center (along Wadi Naim).
Therefore, types 3 and 4 preceded types 6 and 7. Types 3 and 4 belong to an early uplift
stage, whereas types 6 and 7 are correlated with an advanced uplift. Since type 7 joints
are younger than type 6 joints, they may be categorized as late uplift products.
There are limitations in estimating the relative ages of certain FJRs. Type 5 preceded
types 6 and 7, but no correlation can be made regarding the relative ages of types 3–4
and type 5 because they occur in remote locations and lack cutting relationships. Also,
no comparison is possible regarding the timing of types 5–7 and type 8. It may how-
ever be suggested that the multilayer fracture of type 8 were products of stronger ten-
sile stresses than types 5–7 (Bahat 1991a).

6.7.3.2
Limitations of the Model Depicted in Figure 6.25

The set of joints of type 3 formed in a principal plane sub-parallel to the plane of the pri-
mary fault (Fig. 6.24c) and the model in Fig. 6.25 explains how this was possible. This model
is simple for normal faults that dip 45° or close to it, because for such faults the direction
of maximum shear stress in the initial field of remote stress is easily transformed to the
direction of σ1 in the local stress field (Fig. 6.25a). The validity of this model would be veri-
fied if type 3 joints were exactly parallel primary faults in other fracture provinces. For
normal faults whose dips deviate from 45°, additional studies will show the required
changes and expected spreads in orientation of the secondary fractures of type 3. Perhaps
the relatively large dip spread of the set that resembles type 3 joints in Fig. 6.26 is a good
realistic example.
512 Chapter 6 · Assorted Problems in Fracture Geology

6.7.3.3
Speculation on Future Implications

Fractures in the Beer Sheva syncline display two FJR systems that are linked to burial,
whereas six others relate to various syntectonic/uplift categories. Joint sets within these
systems are categorized into pre-, syn- and post-fault associations. Early and late events
that relate to pre-faulting and post-faulting, respectively can be distinguished in the
burial system. In the uplift category, syn-fault early uplift and post-fault early uplift
events can be distinguished from pre-fault and syn-fault late uplift events. Future stud-
ies may possibly show fuller ranges of FJR than the one described in the Beer Sheva
syncline. These would include pre, syn and post fault joints in the three categories, burial,
syntectonic and uplift and to lesser extents in the post-uplift category. Complex FJR
structures in association with combined structures like burial-syntectonic or
syntectonic-uplift can also be expected.

6.8
The Path from Geological Joints to Soil
through a Transitional Layer of Surface Joints
This section concentrates on a “transitional layer” that lies above the uneroded chalk lay-
ers of the Mor Formation and below the soil cover that lies at the top of many outcrops
around Beer Sheva, in south Israel (Sect. 3.4). The transitional layer reveals various extents
of erosion in different locations and is characterized by an abundance of “surface joints”
that were added to the earlier single-layer joints, which formed in the chalk of the Mor
Formation (Fig. 2.63a) during the Lower Eocene (Bahat 1985). The surface joints may
interest soil specialists, hydrologists and environmentalists, because they represent an
important stage in the transition of rock to soil. In addition, these joints add signifi-
cantly to the fracture population in the transitional layer and therefore increase the
penetration of running water to the ground. An outcrop that shows the succession of
several chalk layers, the transitional layer and the soil cover (Fig. 6.27a,b) was selected
in the Shephela syncline (Fig. 3.33b) for detailed investigation, as described below.

6.8.1
Methods

For fracture analysis in Fig. 6.27b, photographs using Kodak Gold 200ASA film were taken,
digitalized using an Epson Perfection 1600 scanner and panoramas built up using the photo-
merge utility of Adobe Photoshop Elements. These were entered into ImageJ (Rasband 2003).
The scale was calibrated using a stick of known length (130 cm) that had been included in
the photographs for this purpose. A grid 200 × 200 cm was then overlain on the pictures
and then returned to Photoshop. Individual layers were divided into quadrants of 200 cm
long. The photographs were again saved and re-entered into ImageJ. Using a Wacom graph-
ics pad and pen, the vertical and horizontal fractures of each quadrant were outlined and
measured separately using the software measuring utility. The area of each quadrant and
the distance of the middle of the quadrant from the surface were similarly obtained. The
results were recorded in Microsoft Excel and the total length of fractures was divided by
the area to give fracture intensity (fracture length, given in cm, per unit area, given in m2).
Fig. 6.27a. An outcrop that consists of six rock layers which are overlaid by a soil bed. a The examined outcrop, showing the frame which is detailed in b, where all
fractures are traced
6.8 · The Path from Geological Joints to Soil through a Transitional Layer of Surface Joints
513
514 Chapter 6 · Assorted Problems in Fracture Geology

Fig. 6.27b. An outcrop that consists of six rock layers which are overlaid by a soil bed. Frame with
traced fractures. Numbers are marked on each quadrant. The first digit marks the layer number and
the other two relate to sequential quadrants to which the outcrop is divided

6.8.2
Results

Figure 6.27a shows a panorama of the examined outcrop, which is marked by a frame. All
the fractures that are confined to individual chalk layers and the transitional layer are traced
within the frame (Fig. 6.27b), revealing two distinct major fracture groups: Vertical and
sub-vertical (VSV) joints and horizontal and sub-horizontal (HSH) joints. Figure 6.28a–c
show plots of fracture intensity (cm m–2) per quadrant vs. distance from the surface, i.e.,
every point on the graph represents a quadrant average. The curve that connects the aver-
ages of fracture intensities of the vertical fractures (VFI) undulates throughout the six
layers (Fig. 6.28a), suggesting that vertical fractures were not significantly affected by the
process leading to the development of the “transitional layer”. On the other hand, a
gradual, strong increase in fracture intensity of the horizontal fractures (HFI) is seen
in layers 3–6 (Fig. 6.28b): From the average 122 cm m–2 (60.5 standard deviation) to
 6.8 · The Path from Geological Joints to Soil through a Transitional Layer of Surface Joints 515

Fig. 6.28. Fracture parameters of single-layer fractures vs. vertical distance from the surface above
the soil bed. a Fracture intensity of vertical joints. b Fracture intensity of horizontal joints

541 cm m–2 (85.5 standard deviation). The latter curve suggests that only the two upper
chalk layers (4–5) were somewhat affected by the process leading to the development
of the “transitional layer” in layer 6. Hence, the curve in Fig. 6.28b provides a “fracture
profile” that characterizes the transition from rock (chalk) to soil in this outcrop
(Fig. 6.27a).
516 Chapter 6 · Assorted Problems in Fracture Geology

6.8.3
Discussion

The increase of the population of the horizontal fractures in layers 4–6 and particu-
larly in layer 6 suggests that fracture was induced by a stress field in which the mini-
mum principal stress was vertical. This stress field was created by the reduction of the
compression exerted by the weight of the overburden. The distinct, very strong con-
centration of HSH in layer 6, i.e., closest to the soil surface, implies that the change in
stress field is much younger (Pleistocene or younger) than the latest uplift processes in
the region (Sect. 3.4). Three mechanisms seem to be relevant here. First, the gradual
transition of the upper rock layer into soil reduced the density of the latter, which re-
sulted in some decrease in the vertical compression on layer 6, leading to the develop-
ment of a vertical tension, much as it follows uplift and denudation processes in larger
dimensions. The decrease in the vertical compression was strong in the transitional layer,
moderate in layers 4 and 5, and practically absent in deeper layers. Second, what is par-
ticularly striking is that the transitional layer is almost exclusively limited to layer 6.
This is such a common feature in all the outcrops of the region that it is inconceivable
that the “right” amount of overburden was removed to produce cracking only in this
layer. Thus, there must be some additional factor acting locally in the transitional layer.
Haiaty (1975) shows that the tensile strength of the Eocene (Yoqneam) chalk drops
dramatically from 4.3–27.5 kg cm–2 in the dry zone to 2.3–8.5 kg cm–2 in the water satu-
rated zone. More recently Bahat et al. (2001a) showed that even a small amount of wa-
ter markedly reduces the (compression) strength of chalk. Thus the exclusive contact of
layer 6 with the water saturated loess soil was sufficient to critically reduce the rock strength,
causing fracture. Third, rhythmic changes in temperature were particularly effective in the
transitional layer: Diurnal/nocturnal variation of temperature does not occur deeper
than 30 cm in these rocks (commun. from Howard C. B. on unpublished work). Perhaps
a combination of these three mechanisms explains the formation of the surface joints.

6.9
Hydro-Geological Research in the Beer Sheva Syncline
6.9.1
Introduction

The well-exposed outcrops in the Negev desert enabled detailed studies to be carried out
on fracture-geology in the Beer Sheva syncline (Sect. 3.4). This syncline also provides good
grounds for hydro-geological investigation. Most of this investigation is concentrated in a
relatively small area of some 25 km2 in the center of the syncline around the large Ramat
Hovav complex of chemical industry and the National Site for Treatment and Isolation
of Hazardous Waste (Fig. 6.29). According to Nativ et al. (1995), the aridity of the area
(180 mm yr–1 rainfall) and the low-permeability chalk matrix (~2 millidarcy; Dagan
1977a,b) were considered a natural barrier to potential groundwater contamination
resulting from the industrial activities. However, the high concentrations of heavy metals
and organic pollutants found in the local groundwater (Nativ and Nissim 1992) invali-
dated the concept of a natural barrier to contaminant migration across the vadose zone.
These observations as well as others, such as seasonal fluctuations in groundwater lev-
Fig. 6.29. Location map for the chemical industrial complex, facilities, ephemeral streams and observation holes. Old wells are represented by squares and new wells
are represented by triangles. Contours describe topography and their interval is in meters. Straight lines are inferred fractures interpolated between measurement
points along the ephemeral streams and man-made cuts. The trace of Wadi Secher is marked by double dashed line. A, B and C are locations of water level mounds
(from Nativ et al. 1999)
6.9 · Hydro-Geological Research in the Beer Sheva Syncline
517
518 Chapter 6 · Assorted Problems in Fracture Geology

Fig. 6.30. Schematic geological cross section for the northern Negev Desert (from Nativ et al. 1999)

els, water table responses to flood events, the occurrence of tritium in the groundwater,
as well as the behavior of additional chemical and isotopic tracers (Nativ and Nissim
1992), indicate that the groundwater is rapidly affected by on-surface activities, and
infiltrating water and contaminants bypass the low-permeability chalk matrix via pref-
erential flow paths. Nativ and Nissim (1992) suggested that the numerous fractures cross-
ing the chalk matrix (Bahat 1988a) serve as preferential flow paths for contaminant
migration from land surface to groundwater (Dahan et al. 1999) (Fig. 6.29). Although
the Eocene (also termed Avdat) chalk is not a major water source (groundwater is brack-
ish), the hydrological relationships between the chalk and the adjacent aquifers could
result in the contamination of fresh groundwater (Fig. 6.30). The underlying Judea aqui-
fer in the Cenomanian rocks provides over 30 × 106 m3 of water annually to the city of
Beer Sheva and its vicinity. Kronfeld et al. (1993) have documented leakage between
adjacent aquifers(as a result of excessive pumping). This section presents a brief ac-
count of recent hydro-geological studies in Ramat Hovav.

6.9.2
Summary of the Results

Selected summaries are grouped into fluid flow and transport sections, ending with a
section on open questions that stresses the awaiting challenges.

6.9.2.1
Fluid Flow through Fractures Intersecting Chalk

Preferred flow directions. A regional hydrogeological investigation of the flow pat-

tern in the Eocene chalk formations in the area indicates a general subsurface flow
in the 270° direction in the upper eastern basin following the topographic inclina-
tion (Nativ and Nissim 1992). However, toward the lower part (further west) of the
 6.9 · Hydro-Geological Research in the Beer Sheva Syncline 519

Beer Sheva syncline the groundwater flow is directed to the northwest (310°).
Rophe et al. (1992) measured water levels in three observed wells in Ramat Hovav.
The best fit with the obtained data indicates that the maximum hydraulic conduc-
tivity lies at an azimuth of 238° in which direction the IG (imaginary gradients) is
minimal. Rophe et al. (1992) hypothesized the existence of a major set of fractures
at this azimuth. Apparently, the 238° azimuth corresponds quite well to the direction
of set 062° (242°), which consists of long and deep penetrating fractures in the syn-
cline (Bahat and Adar 1993, present Fig. 6.29 and group 13 in Sect. 3.4.2.5). This find-
ing has recently been confirmed by many hydraulic conductivity measurements (Nativ
et al. 2003).

Water recharge and percolation through fractures. Four dry-drilling holes were bored
by Nativ et al. (1995) through the vadose zone. Core and auger samples, collected at
30- to 50-cm intervals, were used for chemical and isotopic analyses, enabling the con-
struction of the following:

1. A tritium profile to estimate the rate of water flow through the unsaturated zone.
2. Oxygen-18 and deuterium profiles to assess the evaporation of water at land sur-
face before percolation and in the upper part of the vadose zone after infiltration.
3. Chloride and bromide profiles as tracers for inert solutes and pollutants.

The tritium and bromide profiles showed the rate of infiltration through the un-
saturated matrix to be very slow (1.6–11 cm yr–1). A conceptual model was proposed
in which a small portion of the rainwater percolates downward through the matrix,
while a larger percentage of the percolating water moves through preferential path-
ways in fractures. The water flowing through the fractures penetrates the matrix across
the fracture walls where it increases the tritium concentrations, depletes the stable
isotopic composition, and dilutes the salt concentrations. The observed rapid down-
ward migration of tritium and heavy metals through the profuse fractures makes the
chalk inefficient as a hydrologic barrier.

Hydraulic properties of fractures in chalk. One of the challenges of monitoring net-

work design in a fractured rock setting is the heterogeneity of the rocks. Nativ et al.
(1999) summarize the activities and problems associated with the monitoring of con-
taminated groundwater in porous, low-permeable fractured chalk in Ramat Hovav
(Fig. 6.29). The fracture systems and their hydraulic properties were characterized by
Nativ et al. (2003) to estimate contaminant release from Ramat Hovav. These included
identification of the prevailing directions of fracture systems in outcrops, in cores
retrieved from inclined coreholes, in coreholes using video logs, and in trenches. The
orientation and inclination of these fracture systems were determined, and evidence
of groundwater flow on the fracture surfaces was described and ranked. Their hy-
draulic conductivity was determined through slug and pumping tests performed at
discrete intervals. Temperature, electrical conductivity, caliper, gamma and heat-pulse
logs were conducted in the same coreholes. The results from the logs, tests, and core
descriptions were compared to identify reliable and cost-effective tools for investigat-
ing the hydraulic characteristics of fracture systems. Nativ et al. (2003) concluded that
in the study area:
520 Chapter 6 · Assorted Problems in Fracture Geology

Fig. 6.31a.
Schematic cross section of the
field setup used to determine
solute and particle flux through
a single natural fracture (after
Dahan et al. 1998 and Weisbrod
et al. 2000a)

1. Fracture mapping in outcrops and coreholes (including downhole video and cali-
per logs) must be supplemented by hydraulic testing of the mapped fracture sets in
the coreholes.
2. Inclined coreholes provide information regarding the orientation of the hydrauli-
cally active fracture systems that cannot be obtained from vertical boreholes.
3. Hydraulic testing of unpacked holes provides a reasonable estimate of the maxtmum
hydraulic conductivity.
4. The hydraulic conductivity distribution with depth is log normal and all signifi-
cant ground water flow takes place within the upper 25 m.

Percolation experiments in natural discrete fractures. The mechanisms controlling

fluid flow through fractures intersecting chalk in the vadose zone were studied by
Dahan et al. (2000) through water percolation experiments (Fig. 6.31a) in natural dis-
crete fractures and by close examination of the inner structure of fracture voids. The
percolation experiments showed that the flow is focused in dissolution channels along
the fracture plane and that fluxes and flow trajectories within that net vary in both
time and space. The locations of the dissolution channels, the main potential flow
paths within the fracture plane, were generally associated with fracture intersections.
The flow through these channels was governed primarily by the mineralogical com-
position of the filling material and the inner structure of the fracture voids. Salt dis-
solution, solid-particle migration and clay swelling were found to be the predominant
processes controlling flow through the dissolution channels. These physical changes
in the structure of the filling material in the dissolution channels accounted for the
observed unstable flow behavior. The results suggest that models aimed at simulating
water percolation through fractures in unsaturated chalk should consider the map-
ping of fracture intersections in addition to the commonly used mapping of fracture
lineaments. Moreover, the detailed characterization of fracture apertures may not be
the key parameter determining fracture flow, because in such formations the flow is
controlled primarily by filling material. These materials undergo significant physical
variations during wetting and drying cycles (Weisbrod et al. 1999).
 6.9 · Hydro-Geological Research in the Beer Sheva Syncline 521

Fig. 6.31b. Schematic description of experimental setup. The solutions were drained from the drip-
ping point into the gradual cylinders and then to the four flow cells (from Weisbrod et al. 2000b)

6.9.2.2
Transport of Solutes and Particles

Groundwater salinity. The processes affecting salinization of precipitation, surface

water, vadose water and groundwater in the chalk aquitard were studied by Nativ et al.
(1997a,b). Observations spanning eighteen years included the collection of rainfall at
three rain sampling stations, flood water at six flood stations, vadose water from four
coreholes penetrating chalk formations, and groundwater from sixteen monitoring wells
tapping the chalk aquitard. Dissolved carbonate dust and evaporation of the falling
raindrops result in Ca(HCO3)2 facies and increased ion concentration of the rainwater
with respect to inland, more humid regions. Based on isotopic compositions, the expo-
sure of flood water to evaporation during flood events is minimal. The observed
Ca(HCO3)2 facies and salt enrichment by a factor of three to five in the flood water with
respect to precipitation, results primarily from interactions of the flood water with the
chalk and limestone bedrock, including ion exchange on Na- and K-bearing minerals
and the dissolution of calcite, gypsum and halite. The presence of these salts at and
near land surface results from the complete evaporation of rainwater in land surface
depression storage areas following most rain events. Water percolation and salt disso-
lution account for the NaCl facies of vadose water and the variable rates of isotopic
depletion and salt dilution observed in the underlying heterogeneous groundwater in
the saturated zone. Although the variable mixing with low-saline, isotopically depleted
water percolating from the fractures accounts for the depleted isotopic composition of
the groundwater, its relatively low solute content cannot modify the groundwater NaCl
522 Chapter 6 · Assorted Problems in Fracture Geology

facies. Consequently, only groundwater salinity in the chalk is reduced by the preferen-
tially flowing water, but the Ca(HCO3)2 facies prevailing in the rainwater and flood
water disappears, and the NaCl imprint from the vadose zone prevails.

Fracture surface erosion. Two coated and two uncoated slices from the fracture sur-
face of an unsaturated chalk core were exposed by Weisbrod et al. (1999) to short flow
events (twenty-four, eight, and nine hours) of industrial wastewater and/or synthetic
rainwater, followed by long drying periods (weeks) (Fig. 6.31b). The topography of
the fracture surface was shown to be unstable due to the detachment of colloidal and
large-sized particles during the first three to seven hours of flow. Following rainwater
flow, erosion was more pronounced on the coated than on the uncoated surface (mean
erosion of 0.313 and 0.134 mm, respectively) (Fig. 6.32). Interaction with industrial
wastewater generated a skin of organic matter and gypsum that collapsed following
contact with rainwater, leading to a deeper erosion of the uncoated surface than of the
coated one (1.238 and 0.549 mm on the average, respectively). Erosion of the fracture
surface was measured using a laser-scanning system and was calculated from high-
resolution topographical maps (elevation z ≤ ±0.01 mm) generated by Geographic
Information System (GIS, ARCInfo) prior to and following the flow experiments. The

Fig. 6.32. a The naturally fractured core used for the experiments (5 m long, 25 cm diameter) show-
ing the sampling location where two slices were sawed out from the coated areas (by surface weather-
ing and precipitates of secondary mineralization) and two from the uncoated areas. The area of each
slice was 42–50 cm2, with a thickness of 4.5 cm. b The drilling site and drilling machine of the hori-
zontal corehole from which the core was retrieved (from Weisbrod et al. 1999)
 6.9 · Hydro-Geological Research in the Beer Sheva Syncline 523

mean thickness of the eroded surface was found to be strongly correlated with the
calculated total accumulated mass of particles and soluble salts released from the frac-
ture surface. This relationship can be used to evaluate fracture surface erosion in large
field and laboratory experiments.

Flow experiments on fracture surface properties. The surface topography of a natu-

ral fracture intersecting a chalk formation was mapped by Weisbrod et al. (2000b)
(see also Weisbrod et al. 1998), using the aforementioned high-resolution laser-scan-
ning device, and its roughness was evaluated using five different criteria. Two coated
and two uncoated slices (Fig. 6.32) of the natural fracture surface were encased in
flow cells and exposed to short (twenty-four and eight hours) interactions with per-
colating synthetic rainwater and industrial wastewater, followed by long drying peri-
ods (weeks) (Fig. 6.31b). These flow experiments simulated the intermittent infiltra-
tion of surface runoff and industrial effluents typical of the study area and resulted in
erosion of the fracture surfaces and modification of their roughness. The temporal
variations in surface roughness were evaluated using commonly used criteria. The
surfaces coated by secondary mineralization (Avigur and Bahat 1990; Weisbrod et al.
1999) were significantly rougher, had greater relief, more heterogeneous topography,
and a larger surface area than the uncoated surfaces (matrix mineralogy). The obser-
vations suggest that coating and fast weathering (caused by intermittent flow events)
significantly change roughness, surface area, relief, and the heterogeneity of a frac-
ture surface. These in turn may affect the spatial distribution of flow paths across that
fracture surface.

Particle detachment and transport from fractures. A series of field and laboratory
experiments were conducted by Dahan et al. 1999 and Weisbrod et al. 1999), respec-
tively, to study the mechanisms of particle detachment and transport from fractures
in vadose chalk. Experiments of intermittent flow events along fracture surfaces were
carried out in the laboratory. In the field (Fig. 6.31a), water was percolated from land
surface via a discrete fracture into a compartmental sampler installed inside a
horizontal corehole located below the surface. The mass, size distribution and com-
position of the particles drained with the water from the fracture voids were exam-
ined along with flow rates and salt dissolution. Two boreholes penetrating the under-
lying saturated zone were sampled and analyzed for colloidal concentration and
composition. The results of these experiments summarized by Weisbrod et al. (2002)
suggest that:

1. Particle detachment causes flow-rate variability in the unsaturated fracture.

2. The mechanisms of particle detachment and salt dissolution within the fracture
are linked.
3. Although most of the detached particles are large and likely to accumulate inside
fractures, some colloidal particles also eroded from the fracture void and are likely
to be transported to the groundwater.

Most of the detached particles had a mean diameter of >2 µm, while the mobile
colloidal phase in the groundwater had a mean diameter of ~1 µm.
524 Chapter 6 · Assorted Problems in Fracture Geology

6.9.3
Open Questions Related to Water Flow in Heterogeneous Fracture Systems

The aforementioned summaries of recent studies from Ramat Hovav reveal a commend-
able progress in understanding water conduct (i.e., water flow and solute and particle
transport) in the chalk aquitard. We return now to the work by Nativ et al. (2003) and
take the risk of extrapolating it into six major challenges that are still awaiting the in-
vestigators in this area. The first challenge is differentiating the contributions of local
geological structures to water conduct. The second is “getting the general picture” on
water conduct by synthesizing these contributions. The third is minimizing the gaps
along the interface between this “general picture” and adapted or newly devised con-
ceptual and network fracture models on liquid flow (e.g., Nelson 1987; Berkowitz et al.
1988; Odling et al. 1999; Bonnet et al. 2001; Connoly and Cosgrove 1999). Contribution
in this direction has been made by Berkowitz et al. (2001). The fourth is characterizing
the differences between quasi-static and dynamic systems (such as intermittent changes
in the apertures of joints) and applying them to new conceptual models. The fifth is
devising engineering methods for eliminating water pollution, locally and regionally,
and monitoring their execution. Finally, the sixth is comparing the Ramat Hovav sys-
tem to other systems and models (e.g. Long et al. 1996).
We limit our comments to addressing the first challenge. This would involve dif-
ferentiating the water conduct when flowing through different structures, as demon-
strated by the two examples below. First, it is generally accepted that the major con-
duits in the area are the NE-ENE uplift multilayer joints (Fig. 6.29). It should be no-
ticed, however, that the NE-ENE uplift fracture category consists of a series of frac-
ture sets that formed in different geological associations when cutting Lower and
Middle Eocene chalks (Table 3.3), and correspondingly, would have different drain-
age properties (including different affinities to coating and uncoating properties that
were demonstrated above). Second, while many joints and faults provide poor drain-
age systems to water flow due to secondary minerals that often form along them (e.g.,
Avigur and Bahat 1990; Dahan et al. 1999), many other faults enhance rock permeability
and water flow (e.g., Witherspoon and Newman 1967; Birkeland and Larson 1978).
Quite important is the observation that liquid drainage may be considerably improved
along fracture intersections (e.g., Hsieh et al. 1985; Dahan et al. 2000). The Beer Sheva
syncline is profused with faults (Nativ et al. 2003) and joints (Sect. 3.4). Furthermore,
there are numerous intersections and many styles of contact between faults and joints
in the syncline (Sect. 6.7), and these should contribute significantly to the “general pic-
ture” of water conduct (see also Bahat 2004).
Finally, the investigation in Ramat Hovav involved recently an international par-
ticipation of teams from Ben Gurion University of the Negev, The Hebrew University
of Jerusalem (Israel), Bonn University, The University Of Karlsruhe (Germany), Penn
State University, The University of California (Berkeley, USA), The Geological Survey
of Denmark and Greenland (GEUS), The British Geological Survey (BGS) and The
Volcani Research Center (Bet Dagan, Israel). These teams issued a series of annual
reports (2000–2003), which were partly covered under the name FRACFLOW, a project
promoted by the European Commission (see also Bloomfield 1999). These reports
concerned various aspects of water flow through fractured rocks and contamination
(note particularly Adar and Nativ 2003).
 6.10 · A New Fringe Characterization and Analysis 525

6.10
A New Fringe Characterization and Analysis
The mechanism of en echelon segmentation by local stress rotation (Fig. 2.40) via the
fracture slanting model (Sect. 2.2.5.2) is workable (see also Preston’s (1931) model in
Bahat 1991a, Fig. 2.5). However, some people still debate the exact gear that drives the
local stress rotation (e.g., Lazarus 2001a,b). Ramsay and Lisle (2000) suggest a new
mechanism, which differs from the model by Pollard et al. (1982) by exposing the rock
to superposing remote shear stress on normal stress rather than local shear stress su-
perposing on normal stress. The model by Ramsay and Lisle (2000, p. 948) is cited below
with minor modifications, presenting one of their solutions (Fig. 6.33a–d).
When the en echelon segments have the same sense on either side of the parent joint
(Fig. 6.33b), the situation is such that the layer boundaries are subjected to a shear
strain (shear plane parallel to the bed boundaries, shear direction oblique to the main
central fracture, and similar shear sense on both boundaries). The fracture tip propa-
gates from its point of origin in such a way that the incremental fracture surface is
oriented perpendicular to the maximum deviatoric tensile stress direction (and maxi-

Fig. 6.33.
Three-dimensional block model
of en echelon segmentation,
combining layer parallel exten-
sion, layer normal shortening
with layer boundary shear. a The
various k-parameters, see text
for their significance. b, c and d
show the predicted joint surface
patterns arising with differing
types of boundary shear (from
Ramsay and Lisle 2000)
526 Chapter 6 · Assorted Problems in Fracture Geology

mum extensive elastic strain) that are normal to the parent joint. The layer is oriented
parallel to the x-y-plane, with boundaries at z = ±1.0 (Fig. 6.33a). It has been subjected
throughout to a uniform stretch parallel to the y-direction of k1 and a uniform shorten-
ing in z of k2, with no change in the x-direction (plane strain for this deformation com-
ponent). The layer has then been displaced by a shear displacement with shear plane xy,
with vector components at each level of k3 parallel to x and k4 parallel to y and which
vary in shear intensity (γ ) with distance from the central plane (i.e., with z) according
to γ = az3.
The odd power of z accords with the observation of shear sense symmetry based
on similar en echelon symmetry on either side of the parent fracture. In this model,
k1, k2, k3, k4 and a are all constant terms chosen to suit individual problems: k1 and k2
are paired parameters which describe the background bed-parallel, bed-perpendicu-
lar strains (and which include volume change ∆v = 1 – k1k2):k3, k4 and a define the value
of the shear at z = ±1 (γ = a(k12 + k22)1/2), and orientation (α) of the shear direction from
the x-axis (tan α = k4 / k3). The equations which define the displacement of any point
with initial coordinates (x, y, z) and final coordinates (x', y', z') are:

(6.17)

The displacement gradient matrix is (from Eq. 36.62 by Ramsay and Lisle 2000):

(6.18)

We cite Ramsay and Lisle’s (2000) solution for shear at 45° to the x-(and y-)axis, for
the parameters k1 = 1.10, k2 = 0.90, k3 = k4 = –0.10, a = 1.00, as applied to the fracture
in Fig. 6.33b. These parameters give small strains with an increase in volume
(∆v = +0.01) and Ramsay and Lisle (2000) consider that they are compatible with an
uplift type of overall displacement and with small shear strains along the boundary
of the layer (γ = ±0.14 at z = ±1.0). The values and orientations of the principal strains
are set out in Table 6.4.
The special interest in these data is the systematic change in the plane normal to
the greatest extension as one moves from the central line of the layer, because this plane
would be the one that would guide the orientation of the frontal fracture propagation
plane. In the center of the layer the extension fracture would propagate parallel to the
xz-surface of the coordinate system (strike 0, dip vertical), but as it propagates towards
edge of the layer the fracture systematically changes both its strike and dip through
the values, 0/1 to E, 179/5 to E, 175/10 to E, 166/18 to E, and 156/25 to E. This orienta-
tion change is exactly that which one observes in the en echelon segments in joint
fringes (Fig. 6.33b). The stresses at the tips of the en echelon segments will be changed
from that of the general situation, and there will be a strong tendency for interlinking
of the tips of one segment with the surface of the next segment by a step (bridge)
(Fig. 2.39).
 6.10 · A New Fringe Characterization and Analysis 527

Table 6.4.
The values and orientations of
the principal strains

Although the values of the strains used in this example are perhaps a little large for
elastic deformations, the orientation changes depend almost entirely upon the pro-
portional values of the pairs of constants governing the relative values of the bulk dis-
tortion and the shear components, so the conclusions as to the main features of the
geometry are valid for a large range of parameters. Finally, the program of Appendix
36.B TWISTHACKLES.BAS (given in Ramsay and Lisle 2000) enables the constant
parameters to be varied as one wishes to explore the properties of different models.
Nomenclature

Physical Terms
Creep. A slow incremental crack increase under a constant or slowly varying load that is
below that of the the Griffith criterion.
Critical dynamic stress intensity factor KID. The value that if the stress intensity factor is
larger than it, dynamic (fast) fracturing would continue. KID is usually larger than KIC (see
Eq. 1.137).
Critical or post critical conditions. Are associated with crack growth at KI = KIC.
Critical stress intensity factor. A material-dependent property KIC. When KI = KIC, the
Griffith criterion is fulfilled and fracturing can commence.
Effective saturation. An assumption that beyond a certain value, crack density increases
very slowly (logarithmically) with load increase.
Fatigue. Material failure by cracking under periodic or repeated stresses.
Flaws. General name for small heterogeneities in rocks (fossils, concretions, pores) that
may initiate crack growth.
Griffith criterion. A necessary condition for fracturing of materials: The stress intensity
factor KI, becomes equal to the critical stress intensity factor KIC.
Griffith flaws. Flaws in brittle material (glass) that have an effective length of ≈2 µm (Lawn
1993, p. 13), while in rocks they may range up to lengths of grain boundaries.
Linear elastic fracture mechanics (LEFM). The framework theory for linearly calculating
stress fields in different configurations and deducing KI values. Very valuable for static
fracture problems, but fails for dynamic or extremely slow fracturing.
Saturation of crack spacing. An assumption that the crack spacing (No. of cracks per unit
length) is limited. Once the spacing reaches this value, no increase is possible under any load.
Stable crack growth. Occurs under loads in which KI decreases with the increase in crack
length, and the crack arrests when the load is removed (also termed ‘controlled’).
Stress concentration. A situation where near to an abrupt geometrical change in the mate-
rial, such as a flaw, crack, pore, vacancy, etc., a local increase of stress occurs in the stress field.
Sress intensity factor KI. A measure of the ratio of the maximal stress concentration near
a flaw to the nominal stress. In an infinite plate under remote tension, σ0, having a sharp
flaw of length c, expressed as KI = σ0 (πc)½.
530 Nomenclature

Sub-critical conditions. Are associated with crack growth below the critical stress inten-
sity factor KIC.
Time to failure. The time interval from the minute of application of a load, smaller than
that needed for instantaneous fracture, until the sample breaks.
Unstable crack growth. Occurs ‘catastrophically’ at KI = KIC even when unloaded (also
termed ‘uncontrolled’).
v vs. KI diagrams. Correlate the changes in the instantaneous crack velocity v with values
of the instantaneous tensile stress intensity factor KI.

Nomenclature of Main Fractographic Features

Arrest marks. Ripple marks that form on the fracture surface and signify slow crack propa-
gation down to arrest, or crack restarting growing.
Barbs. See plumes.
Dynamic mirror plane. Forms when fractured primarily by high tensile stresses; it propa-
gates rapidly and exhibits characteristic morphological features.
En echelon-fringe. A fringe populated by en echelon cracks.
En echelon segmentation. Cracks that systematically relate to each other in a ‘shingle-like’
arrangement and are confined to the fringe.
Fractography. The branch of science that analyses fracture surface morphology and re-
lated features and their causes, as well as mechanisms, in technological materials.
Fringe. A zone that surrounds the mirror plane and contains cracks that display rough
morphologies.
Hackle cracks. A series of cracks that relate to each other three dimensionally in various
directions and are confined to the fringe.
Hackle-fringe. A fringe populated by hackle cracks.
Mirror boundary. A narrow morphological rim separating the mirror from the fringe.
Mirror plane. The crack surface that contains the initial flaw, the critical flaw, the striae
(plume) and the ripple marks.
Oscillating cracks. Bends of the mirror plane around the axis normal to the direction of
crack propagation.
Plumes. Form by slow fracture on the mirror plane due to stress fields which rotate the
fracture front about the axis of the direction of crack propagation, and often are identified
by their dendrite-like morphologies consisting of many splayed barbs.
Quasi-static mirror plane. Forms when fractured under low stresses; it propagates sub-
critically and exhibits characteristic morphological features.
Ripple marks. Form on the mirror plane due to stress fields that intermittently bend the
crack front around the axis normal to the direction of propagation, when both axes are on
the fracture surface. Two types of ripple marks occur, undulations and arrest marks.
 Nomenclature 531

Striae. Form by rapid fracture on the mirror plane due to stress fields which rotate the
fracture front about the axis of the direction of crack propagation, and often are identified
by sharp, continuous cuts in the material (glass or rock).
Undulations. Ripple marks that form on the fracture surface and signify crack propaga-
tion in various velocities.

Nomenclature of Joints
Burial joints, or subsidence joints. Form during the sedimentation and diagenetic pro-
cesses, and often are single-layer joints. They may form in great depths.
Cooling joints. Form during various cooling stages of the pluton (or volcanic rock).
Crack. In the present text is almost synonymous with fracture. It is generally applied to
small fractures. These are general terms, regardless of loading mode.
Cross-fold joints. Form sub-orthogonal to fold axes.
Cross-joints. A set of younger, nonsystematic joints that arrest at earlier, systematic, single-
layer joints, seen in plan view (Gross 1993).
Dihedral angle. Produced between two genetically linked fractures or fracture sets
(joints or faults) that maintain a conjugate relationship.
Fault. Of a primary nature, that originated by slip(s), which resulted in lost cohesion
in the rock. Of a secondary nature, that formed by slip(s) on an existing joint.
Fractography. Fracture markings that describe the surface morphology that forms
essentially by an opening mode. Every such surface (joint) has a unique fracto-
graphy.
Group. This term is occasionally applied (rather than set or system) when fracture
affinities are debatable.
Joint. A surface on which the rock has lost cohesion and does not show a discernible
lateral slip.
Joint set. Consists of a series of joints of similar strikes that exhibit similar fracture sur-
face morphologies (if present). This similarity often reflects on the same fracture condi-
tions, even if in different fracture episodes.
Joint system. Consists of various joint sets that probably are genetically linked.
Multi-layer joints. Cut many layers, often dissecting the entire outcrop. They are widely
and irregularly spaced.
Orthogonal joint sets. Form perpendicular to each other (in a map), but may deviate up
to 10° from this relationship.
Post uplift joints. Post uplift fractures that form in association with erosion.
Single-layer joints. Cut vertically individual layers (or less commonly, two layers) but ar-
rest at the layer boundaries. They maintain small, regular spacing.
Strike joints, or strike-parallel joints. Form sub-parallel to fold axes.
532 Nomenclature

Surface joints. Recent fractures that are confined to the upper layer, below the soil cover.
Syntectonic joints. Form during intense deformation, occasionally associated with fault-
ing. They may form in great depths.
Tectonofractography. The application of joint fractography to the interpretation of
tectonophysical processes.
Uplift joints. Form during uplift. They can be both single-layer and multi-layer fractures
and form in shallow depths.

Selected Supplementary Parameters

Skewness. The degree of asymmetry, or departure from symmetry, of a distribution. If
the frequency curve of a distribution has a longer “tail” to the right of the central maxi-
mum than to the left, the distribution is said to be skewed to the right or to have positive
skewness. If the reverse is true, it is said to be skewed to the left or to have negative skew-
ness (Spiegel 1961).
Kurtosis. The degree of peakedness (between high and flat-topped peaks) of a distribu-
tion, usually taken relative to a normal distribution (Spiegel 1961).
Statistics of extremes. The statistics of the largest (or smallest) members of a population,
e.g., the severest earthquakes, the highest waves, etc.
FSI. The fracture spacing index, defined as the slope of the best-fit line when joint spacing
is plotted versus layer thickness (Narr and Suppe 1991).
FSR. The fracture spacing ratio defined as the layer thickness divided by the median joint
spacing for an individual bed (Gross 1993).
L / S. The ratio of joint length to joint spacing (Bahat 1991a, p. 309; Wu and Pollard 1995).
References

Abraham FF, Brodbeck D, Ridge WE, Xu X (1997) A molecular dynamics investigation of rapid frac-
ture mechanics. J Mech Phys Solids 45:1595–1619
Abraham FF, Walkup R, Gao HJ, Duchaineau M, De la Rubia TD, Seager M (2002) Simulating materi-
als failure by using up to one billion atoms and the world’s fastest computer: Brittle fracture.
Proceedings of the National Academy of Sciences of the United States of America 99:5777–5782
Acocella V, Gudmundsson A, Funiciello R (2000) Interaction and linkage of extension fractures and
normal faults: Examples from the rift zone of Iceland. J Struct Geol 22:1233–1246
Adar E, Nativ R (2003) Contaminant transport, monitoring and remediation strategies in fractured
chalk in the northern Negev. Annual Report for the year 2002, submitted to the Ramat Hovav
Industrial Council
Adler RJ, Feldman RE, Taqqu MS (1998) A practical guide to heavy tails, statistical techniques and
applications. Birkhauser, Boston
Ahlfors LV (1979) Complex analysis, 3rd edn. McGraw Hill Book Co, Aucland
Aitchison J, Brown JAC (1976) The lognormal distribution. Cambridge University Press, Cambridge
Aki K (1979) Geometric features of a fault zone related to the nucleation and termination of a earth-
quake rupture. U.S. Geological Survey Open-File Report 89–315, pp 1–9
Allen CR, Castle RO, Clark MM, Grantz AA, Sharp RV, Theodore TG, Wolfe EW, Young TL (1972) Map
showing surface ruptures created at the time of and after the Borrego Mountain Earthquake of
April 9, 1968 (GMT). U.S. Geological Survey, Professional Paper 78, Plate 1
Anderson EM (1951) The dynamics of faulting, 2nd edn. Oliver & Boyd, Edinburgh
Anderson TL (1995) Fracture mechanics: Fundamentals and applications. CRC Press, London
Anton L (1999) Simple equation for earthquake distribution. Phys Rev E 59(6):7213–7215
Arakawa K, Takahashi K (1987) Dependence of crack acceleration of the dynamic stress intensity
factor in polymers. Exp Mech 27:195–200
Arakawa K, Takahashi K (1991) Relationships between fracture parameters and fracture surface rough-
ness of brittle polymers. Int J Fracture 48:103–114
Arakawa K, Mada T, Takahashi K (2000) Correlations among dynamic stress intensity factor, crack
velocity and acceleration in brittle fracture. Int J Fracture 105:311–320
Arkin Y, Hamaoui M (1967) The Judea Group (Upper Cretaceous) in central and southrn Israel. Isr
Geol Surv Bull 42
Arlegui L, Simón JL (2001) Geometry and distribution of regional joint sets in a non-homogeneous
stress field: Case study in the Ebro basin (Spain). J Struct Geol 23:297–313
Arnold BC (1983) Pareto distributions. International Co-operative publishing house, Fairland, Mary-
land, USA
Arthur MJ (1989) The Cenozoic evolution of the Lundy pull-apart basin into the Lundy rhomb horst.
Geol Mag 126:187–198
Ashcroft NW, Mermin ND (1981) Solid state physics. CBS Publishing, Tokio, pp 398
Atkinson BK (1984) Subcritical crack growth in geological materials. J Geophys Res 89:4077–4114
Atkinson BK (1987) Introduction to fracture mechanics and its geophysical applications. In: Atkinson
BK (ed) Fracture mechanics of rocks. Academic Press, pp 1–26
Atkinson C, Eshelby JD (1968) The flow of energy into the tip of a moving crack. Int J Fracture Mech
4:3–8
Atkinson BK, Meredith PG (1987a) The theory of subcritical crack growth with applications to the
minerals and rocks. In: Atkinson BK (ed) Fracture mechanics of rock. Academic Press, London,
pp 111–166
Atkinson BK, Meredith PG (1987b) Experimental fracture mechanics data for rocks and minerals. In:
Atkinson BK (ed) Fracture mechanics of rock. Academic Press, London, pp 477–525
534 References

Avigour A, Bahat D (1990) Chemical weathering of fractured Eocene chalks in the Negev, Israel. Chem
Geol 89:149–156
Babcock CL (1977) Silicate glass technology methods. John Wiley & Sons Ltd., New York
Bahat D (1968) Structural changes at room temperature of heated anorthoclases. Can Mineral 9:531–538
Bahat D (1970) Kinetic study on the hexacelsian-celsian phase transformation. J Mater Sci 5:805–810
Bahat D (1977) Thermally-induced wavy Hertzian fracture. J Am Ceram Soc 60:118–120
Bahat D (1979a) Theoretical considerations on mechanical parameters of joint surfaces based on
studies on ceramics. Geol Mag 11:81–92
Bahat D (1979b) On the African rift system, theoretical and experimental study. Earth Planet Sc Lett
45:445–452
Bahat D (1979c) Interpretation on the basis of Hertzian theory of a spiral carbonatite structure at
Homa Mountain, Kenya. Tectonophysics 60:235–246
Bahat D (1982) Extensional aspects of earthquake induced ruptures determined by an analysis of
fracture bifurcation. Tectonophysics 83:163–183
Bahat D (1983) New aspects of rhomb structures. J Struct Geol 5:591–601
Bahat D (1985) Low angle normal faults in Lower Eocene chalks near Beer-Sheva, Israel. J Struct Geol
7:613–620
Bahat D (1986a) Joints and en echelon cracks in Middle Eocene chalks near Beer-Sheva, Israel.
J Struct Geol 8:181–190
Bahat D (1986b) Criteria for the differentiation of en echelons and hackles in fractured rocks.
Tectonophysics 121:197–206
Bahat D (1987a) Correlation between styles of fracture markings and orientation of cross-fold joints.
Tectonophysics 136:323–333
Bahat D (1987b) Jointing and fracture interactions in Middle Eocene chalks near Beer-Sheva, Israel.
Tectonophysics 136:299–321
Bahat D (1988a) Early single-layer and late multi-layer joints in the Lower Eocene chalks near Beer-
Sheva, Israel. Annal-Tecton 2:3–11
Bahat D (1988b) Fractographic determination of joint length distribution in chalk. Rock Mech Rock
Eng 21:79–94
Bahat D (1988c) Fractographic determination of joint length distribution in chalk. Rock Mech Rock
Engin 21:79–94
Bahat D (1989) Fracture stresses at shallow depths during burial. Tectonophysics 169:59–65
Bahat D (1991a) Tectonofractography. Springer-Verlag, Heidelberg
Bahat D (1991b) Plane stress and plane strain fracture in Eocene chalks around Beer-Sheva. Tectono-
physics 196:61–67
Bahat D (1997) Mechanisms of dilatant en echelon crack formation in jointed layered chalks. J Struct
Geol 19:1375–1392
Bahat D (1998a) Four joint genetic groups and their distinct characteristics. In: Rossmanith HP (ed)
Mechanics of jointed and faulted rock, MJFR-3, Vienna, Austria. Balkema, Rotterdam, pp 211–216
Bahat D (1998b) Quantitative tectonofractography – an appraisal. In: Rossmanith HP (ed) Mechanics
of jointed and faulted rock, MJFR-3, Vienna, Austria. Balkema, Rotterdam, pp 59–67
Bahat D (1998c) On joints and paleostress associated with folds along the Syrian Arc in the Negev.
Israel J Earth Sci 48:29–36
Bahat D (1999a) Single-layer burial joints vs. single-layer uplift joints in Eocene chalk from the Beer-
Sheva syncline in Israel. J Struct Geol 21:293–303
Bahat D (1999b) On paleostresses associated with the Syrian Arc in Israel. Isr J Earth Sci 48:29–36
Bahat D (2000) The quasi-ellipsoid damage zone of faults and five types of fault-joint relationship in
the Beer Sheva syncline, Israel. IX Congresso Geologico Chileno, pp 749–753
Bahat D (2004) Eight distinct fault-joint geometric/genetic relationships in the Beer Sheva syncline,
Israel. In: Cosgrov JW, Engelder T (eds) The initiation, propagation and arrest of joints and other
fractures. Spec Publ Geol Soc London (in press)
Bahat D, Adar EM (1993) Comment on “Analysis of subsurface flow and formation anisotropy in a
fractured aquitard using transient water level data” by Rophe B, Berkowitz B, Magaritz M, Ronen
D. Water Resour Res 29:4171–4173
Bahat D, Engelder T (1984) Surface morphology on cross fold joints of the Appalachian Plateau, New
York and Pennsylvania. Tectonophysics 104:299–313
Bahat D, Grossmann HNF (1988) Regional jointing and paleostresses in Eocene chalks around Beer-
Sheva. Israel J Earth Sci 37:1–11
Bahat D, Rabinovitch A (1988) Paleostress determination in a rock by a fractographic method. J Struct
Geol 10:193–199
Bahat D, Rabinovitch A (2000) New fractographic aspets of natural and artificial fractures in chalks,
from the Upper Galilee, Israel, and experimental fracture in Perspex. J Struct Geol 22:1427–1435
 References 535

Bahat D, Shavit R (1997) Fracture in the Beer-Sheva Syncline (Ramat Hovav). Unpublished report
submitted to the Ramat Hovav Industrial Municipality
Bahat D, Leonard G, Rabinovitch A (1982) Analysis of symmetric fracture mirrors in glass bottles. Int
J Fracture 18:29–38
Bahat D, Grossenbacher K, Karasaki K (1995) Investigation of exfoliation joints in Navajo Sandstone
at the Zion National Park and in granite at the Yosemite National Park by tectonofractographic
techniques. LBL-36971; UC-400, 67 pp
Bahat D, Grossenbacher K, Karasaki K (1999) Mechanism of exfoliation joint formation in granitic
rocks at Yosemite National Park. J Struct Geol 21:85–96
Bahat D, Rabinovitch A, Frid V (2001a) Fracture characterization of chalk in uniaxial and triaxial
tests by rock mechanics, fractographic and electromagnetic radiation methods. J Struct Geol
23:1531–1547
Bahat D, Bankwitz P, Bankwitz E (2001b) Joint formation in granite plutons: En echelon-hackle series
on mirror fringes (Example: South Bohemian Pluton, Czech Republic). Zeitschrift der deutschen
geologischen Gesellschaft 152:593–609
Bahat D, Bankwitz P, Bankwitz E (2001c) Changes of crack velocities at the transition from the parent
joint through the en echelon fringe to a secondary mirror plane. J Struct Geol 23:1215–1221
Bahat D, Frid V, Rabinovitch A, Palchik V (2002a) Exploration via electromagnetic radiation and
fractographic methods of fracture properties induced by compression in glass-ceramic. Int J Frac-
ture 116:179–194
Bahat D, Bankwitz P, Bankwitz E, Rabinovitch E (2002b) Comparison of the new fracture areas created
by the formation of en echelon and hackle fringes on joint surfaces. Z Geol Wissenschaft 30:1–12
Bahat D, Bankwitz P, Bankwitz E (2003) Preuplift joints in granites: Evidence for subcritical and
postcritical fracture growth. Geol Soc Am Bull 115:148–165
Bahat D, Bankwitz P, Bankwitz E (2004) Title?. In: Cosgrov JW, Engelder T (eds) The initiation, propa-
gation and arrest of joints and other fractures. Spec Publ Geol Soc London (in press)
Bai TX, Pollard DD (2000a) Fracture spacing in layered rocks: A new explanation based on the stress
transition. J Struct Geol 22:43–57
Bai TX, Pollard DD (2000b) Closely spaced fractures in layered rocks: Initiation mechanism and
propagation kinematics. J Struct Geol 22:1409–1425
Bak P, Tang C (1989) Earthquakes as a self-organized criticality. J Geophys Res 94:15635–15637
Bak P, Tang C, Wiesenfield K (1987) Self-organized criticality – An explanation of 1/f noise. Phys Rev
Lett 59:381–384
Bak P, Tang C, Wiesenfield K (1988) Self-organized criticality. Phys Rev A 38:364–374
Balk R (1937) Structural behavior of igneous rocks. Geological Society of America Memoir
Bank-Sills L, Schur D (1989) On the influence of crack plane orientation in fatigue crack propagation
and catastrophic failure. American Society for Testing and Materials, Standard Technical Publica-
tion 1020:497–513
Bankwitz P (1965) Über Klüfte I. Beobachtungen im Thüringischen Schiefergebirge. Geologie 14:241–253
Bankwitz P (1966) Über Klüfte II. Die Bildung der Kluftoberfläche und eine Systematik ihrer Struk-
turen. Geologie 15:896–941
Bankwitz P, Bankwitz E (1984) Die Symmetrie von Kluftoberflächen und ihre Nutzung für eine Paläo-
spannungsanalyse. Z Geol Wissenschaft 12:305–334
Bankwitz P, Bankwitz E (1994) Event related jointing in rocks on Bornholm Island (Denmark). Z Geol
Wissenschaft 22:97–114
Bankwitz P, Bankwitz E (1995) Aspekte der Entwicklung von Klüften in postkinematischen Graniten
des Erzgebirges (speziell Eibenstocker Massiv). Z Geol Wissenschaft 23:777–793
Bankwitz P, Bankwitz E (1997) Fractographic features on joints in KTB drill cores as indicators of the
contemporary stress orientation. Geol Rundsch 86:S34–S44
Bankwitz P, Bankwitz E (2004) The relationship of tilt and twist of fringe-cracks in granite plutons.
In: Cosgrov JW, Engelder T (eds) The initiation, propagation and arrest of joints and other frac-
tures. Spec Publ Geol Soc London (in press)
Bankwitz P, Bankwitz E (1994) Event related jointing in rocks on Bornholm usland (Denmark). Z Geol
Wiss 22:97–114
Bankwitz P, Bankwitz E (1997) Fractographic features on joints in KTB drill cores as indicators of the
contemporary stress orientation. Geol Rundsch 86 Suppl:S34-S44
Bankwitz P, Bankwitz E, Thomas R, Wemmer K, Kämpf H (2004) Age and depth evidence for pre-
exhumation joints in granite plutons: Fracturing during the early cooling stage of felsic rock. In:
Cosgrov JW, Engelder T (eds) The initiation, propagation and arrest of joints and other fractures.
Spec Publ Geol Soc London (in press)
Bansal GK (1977) On fracture mirror formation in glass and polycrystalline ceramics. Phil Mag 35:
935–944
536 References

Bartlett WL, Friedman M, Logan MJ (1981) Experimental folding and faulting of rocks under confin-
ing pressure, Part IX. Wrench faults in limestone layers. Tectonophysics 79:255–277
Barton NR, Choubey V (1977) The shear strength of rock joints in theory and practice. Rock Mech
10:1–54
Bateman PC, Sawka WN (1981) Raymond Quadrangle, Madera and Mariposa counties, California;
analytic data. U.S. Geological Survey Professional Paper 16:107–172
Bateman PC, Wahrhaftig C (1966) Geology of northern California. Bulletin – California, Division of
Mines and Geology, pp 107–172
Bayasgalan A, Jackson J, Ritz JF, Carretier S (1999) Field examples of strike-slip fault terminations in
Mongolia and their tectonic significance. Tectonics 18:394–411
Bazant ZP, Planas J (1998) Fracture and size effect in concrete and other quasi-brittle materials. CRC
Press, Boca Raton
Beall GH (1989) Design of glass-ceramics. Rev Solid State Sci 3:333–354
Beall GH, Karstetter BR, Rittler HL (1967) Crystallization and chemical strengthening of stuffed quartz
glass-ceramics. J Am Ceram Soc 50:99–108
Beauchamp EK (1995) Crack front stability and hackle formation in high velocity glass fracture.
J Am Ceram Soc 78:689–697
Beauchamp EK (1996) Mechanics for hackle formation and crack branching. In: Varner JR, Frechette
VD, Quinn GD (eds) Fractography of glasses and ceramics III. The American Ceramic Society,
Ceramic Transactions 64:409–446
Beck ME, Rojas C, Cembrano J (1993) On the nature of buttressing in margin-parallel strike-slip fault
system. Geology 126:755–758
Becker GF (1891) The structure of a portion of the Sierra Nevada of California. Geol Soc Am Bull
2:49–74
Becker A, Gross MR (1996) Mechanism for joint saturation in mechanically layered rocks: An ex-
ample from southern Israel. Tectonophysics 257:223–237
Behr H-J, Dürbaum H-J, Bankwitz P, DEKORP Research Group (B) (1994) Crustal structure of the
Saxothuringian Zone: Results of the deep seismic profile MVE-90 (East). Z Geol Wissenschaft
22:647–769
Beinkafner KJ (1983) Terminal expression of decollement in Chautauqua County, Now York. North-
eastern Geology 5:160–171
Bella F, Bjadji PF, Della Monica J, Zilpimiani DO, Manguladze PV (1992) Observations of natural
electromagnertic radiation during earthquakes of central Italy. Izv-Phys Solid Earth 28:88–94
Benes K (1971) Structure of plutonic bodies in the Bohemian Massif. Upper Mantle Project (UMP),
Final Report Geology, Praha, pp 111–119
Bentor YK, Vroman A (1960) The geological map of Israel, scale 1 : 250 000, sheet 16, Mt. Sodom, 2nd
edn, with explanatory notes. Government Printer, Jerusalem
Bentor YK, Vroman A, Zak A (1970) Geological map of Israel, 1:250 000. Survey of Israel
Bergerat F, Bouroz-Weil C, Angelier J (1992) Palaeostresses inferred from macrofractures, Colorado
Plateau, Western USA. Tectonophysics 206:219–243
Berkowitz B, Bear J, Braester C (1988) Continuum models for contaminant transport in fractured
porous formations. Water Resour Res 24:1225–1236
Berkowitz B, Nativ R, Adar E (2001) Evaluation of conceptual and quantitative models of fluid flow
and chemical transport in fractured media. In: Panel on Conceptual Models of Flow and Trans-
port in the Fractured Vadose Zone, U.S. National Committee for Rock Mechanics, Board on Earth
Sciences and Resources, National Research Council (eds) Conceptual models of flow and trans-
port in the fractured vadose zone. National Academy Press, Washington, pp 115–147
Bernstein N, Ness DW (2003) Lattice trapping barriers to brittle fracture. Phys Rev Lett 91:025501–1–4
Berry JP (1960) Some kinetic considerations of the Griffith criterion for fracture I. Equations of motion
at constant force. J Mech Phys Solids 8:194–206
Bevan TG, Hancock PL (1986) A late Cenozoic regional mesofracture system in southern England
and northern France. Journal of the Geological Society London 143:355–362
Bilichenko SV, Iljin FS, Kim EF (1990) ULF response of the ionosphere for earthquake preparation
processes. Dokl Akad Nauk SSSR+ 311:1077–1080
Birkeland PW, Larson EE (1978) Putnam’s Geology, 3rd ed. Oxford Univ Press New York
Bleaney BI, Bleaney B (1965) Electricity and magnetism. Oxford University Press, London
Bloomfield JP (1999) Fracflow-geological state-of-the art review. British Geol. Surv., Natural Envir.
Res. Coun., Technical Report WD/99/13, 45 pp
Bobet A, Einstein HH (1998) Fracture coalescence in rock-type materials under uniaxial and biaxial
compression. Int J Rock Mech Min 35:863–888
Bogosh R, Bourne J, Shirav M, Harnois L (1997) Petrochemistry of a Late Precambrian granitiferous
granite, pegmatite and aplite, southern Israel. Mineral Mag 61:111–122
 References 537

Bonn D, Kellay H, Prochnow M, Ben-Djemiaa K, Mounier J (1998) Delayed fracture of an inhomoge-

neous soft solid. Science 280:265–267
Bonnet E, Bour O, Odling NE, Davy P, Main I, Cowie P, Berkowitz B (2001) Scaling of fracture system
in geological media. Rev Geophys 39:347–383
Brace WF (1964) Brittle fracture of rocks. In: Judd WR (ed) State of stress in the earth’s crust. Elsevier,
New York, pp 111–180
Brace WF, Bombolakis EG (1963) A note on brittle crack growth in compression. J Geophys Res 68:
3709–3713
Brace WF, Paulding JBW, Scholz C (1966) Dilatancy of the fracture of crystalline rocks. J Geophys Res
71:939–953
Bradley WC (1963) Large-scale exfoliation in massive sandstones of the Colorado Plateau. Geol Soc
Am Bull 74:519–528
Bradley WH, Pepper JF (1938) Structure and gas possibilities of the Oriskany Sandstone in Steuben,
Yates, and parts of the adjacent counties. New York Geological Survey Bulletin 899-A
Braun M, Flexer A, Lasman N (1977) Geological report – Ramat-Hovav underground oil storage project.
Geoprospect LTD
Breiter K (2001) South Bohemian pluton. Overview. Exkursionsführer und Veröffentlichungen der
GGW, Berlin, 212:130–140
Breiter K, Koller F (1999) Two-mica granites in the central part of the South Bohemian Pluton.
Abhandlungen der Geologischen Bundesanstalt Wien 56:201–212
Breiter K, Seltmann R (1995) Ore mineralizations of the Krušné Hory Mts. (Erzgebirge). Third Bien-
nial SGA Meeting, Excursion Guide, pp 1–200, Czech Geological Survey, Prague
Breiter K, Sokol A (1997) Chemistry of the Bohemian granitoids: Geotectonic and metallogenetic
implications. Sbornik geologickych ved, 31, Praha, pp 75–96
Brereton NR, Evans CJ (1987) Rock stress orientations in the United Kingdom from borehole breakouts.
Report of the Regional Geophysics research Group, British Geological Survey, no. R6 87/14
Broek D (1984) Elementary engineering fracture mechanics, 3rd edition. Martinus Nijhoff Publish-
ers, Boston, p 469
Broek D (1991) Elementary engineering fracture mechanics. Kluwer Academic Publisher, Dordrecht
Brooks M, Trayner PM, Trimble TJ (1988) Mesozoic reactivation of Variscan thrusting in the Bristol
Channel area, UK. J Geol Soc London 145:439–444
Buchbinder B, Zilberman E (1997) Sequence stratigraphy of Miocene-Pliocene carbonate-siliciclastic
shelf deposits in the eastern Mediterranean margin (Israel): Effects of eustasy and tectonics. Sedi-
mentary Geology 112:7–32
Buchbinder B, Benjamini C, Mimran Y, Gvirtzman G (1988) Mass transport in Eocene pelagic chalk
on the northwestern edge of the Arabian platform, Shefela area, Israel. Sedimentology 35:257–274
Burdon D (1959) Handbook of the geology of Jordan. Benham & Co, Colchester
Burger HR, Hamill MN (1976) Petrofabric stress analysis of the Dry Creek Ridge anticline, Montana.
Geol Soc Am Bull 87:555–566
Butler RF (1992) Paleomagnetism: Magnetic domains to geologic terranes. Blackwell Scientific Pub-
lications, Oxford, 319 p
Büttner S (1997) Die spätvariszische Krustenentwicklung in der südlichen Böhmischen Masse: Meta-
morphose, Krustenkinematik und Plutonismus. Frankfurter Geowissenschaftliche Arbeiten, A 16
Büttner S, Kruhl JH (1997) The evolution of the late-Variscan high-T/low-P region: The south-east-
ern margin of the Bohemian Massif. Geol Rundsch 86:21–38
Cadman J (1969) The origin of exfoliation joints in granitic rocks. Unpublished Ph.D. thesis. Univer-
sity of California, Berkeley, USA
Caers J, Beirlant J, Maes MA (1999a) Statistics for modeling heavy tailed distributions in geology.
Part I. Methology. Geology 31:391–410
Caers J, Beirlant J, Maes MA (1999b) Statistics for modeling heavy tailed distributions in geology.
Part II. Application. Geology 31:411–434
Carlson J, Langer J (1989) Mechanical model of an earthquake fault. Phys Rev A 40:6470–6484
Carlson J, Langer J, Shaw B, Tang C (1991) Intrinsic-properties of a Burridge-Knopoff model of an
earthquake fault. Phys Rev A 44(2):884–897
Carmichael ISE, Turner FJ, Verhoogen J (1974) Igneous petrology. Mc-Graw Hill Book Company, New
York, p 739
Carter NL, Friedman M (1965) Dynamic analysis of deformed quartz and calcite from dry Creek
Ridge Anticline Montana. Am J Sci 263:747–785
Castaing C, Halawani MA, Gervais F, Chilès JP, Genter A, Bourgine B, Ouillon G, Brosse JM, Martin P,
Genna A, Janjou D (1996) Scaling relationships in intraplate fracture systems related to Red Sea
rifting. Tectonophysics 261:291–314
Castillo E (1988) Extreme value theory in engineering. Academic Press, Boston
538 References

Charles RJ (1958) Dynamic fatigue of glass. J Appl Phys 29:1657–1662

Chinnery MA (1966) Secondary faulting I. Theoretical aspects. Can J Earth Sci 3:163–174
Chinnery MA, Petrak JA (1968) Dislocation fault model with a variable discontinuity. Tectonophysics
5:513
Chmyrev VM, Isaev VN, Bilichenko SV (1986) Electric fields and hydromagnetic waves in the iono-
sphere above an earthquake center. Geomagn Aeronomy+ 26:1020–1022
Ciliberto S, Guarino A, Scorretti R (2001) The effect of disorder on the fracture nucleation process.
Physica D 158:83–104
Clark MM (1972) Surface rupture along the Coyote Creek fault. US Geol Surv Prof Pap 787:190–207
Cloos H (1921) Der Mechanismus tiefvulkanischer Vorgänge. Friedrich Vieweg, Braunschweig
Cloos H (1922) Über Ausbau und Anwendung der granit-tektonischen Methode. Abh Preuss Geol
Landesanstalt 89:1–18
Cohen AJB (1993) Hydrogeologic characterization of a fractured granitic rock aquifer, Raymond,
California. Lawrence Berkeley Laboratory – 34838, UC-400
Coleman BD (1957) Time dependence of mechanical breakdown in bundles of fibers. I. constant total
load. J Appl Phys 28:1058–1067
Connoly P, Cosgrove J (1999) Prediction of fracture-induced permeability and fluid flow in the crust
using experimental stress data. AAPG Bull 83:757–777
Conrad W, Behr HJ, Trzebski R (1996) Die LINSSER-Filterung des Schwerefeldes der Süddeutschen
Grossscholle und ihre Interpretation. Z Geol Wissenschaft 21:21–40
Cooke ML, Pollard DD (1996) Fracture propagation paths under mixed mode loading within rectan-
gular blocks of polymethyl metha-crylate. J Geophys Res 101:3387–3400
Cooke ML, Underwood CA (2001) Fracture termination and step-over at bedding interfaces due to
frictional slip and interface opening. J Struct Geol 23:223–238
Cosgrove JW (1995) The expression of hydraulic fracturing in rocks and sediments. In: Ameen MS
(ed) Fractography: Fracture topography as a tool in fracture mechanics and stress analysis. Geo-
logical Society Special Publication 92:187–196
Cotterell B, Rice JR (1980) Slightly curved or kinked cracks. Int J Fracture 16:155–169
Cowie PA, Scholtz CH (1992a) Displacement-length scaling relationship for faults: Data synthesis
and discussion. J Struct Geol 14:1149–1156
Cowie PA, Scholtz CH (1992b) Growth of faults by the accumulation of seismic slip. J Geophys Res
97:11085–11095
Cox HL (1952) The elasticity and strength of paper and other fibrous materials. Brit J Appl Phys 3:72–79
Coyle DA, Wagner GA (1995) Spaltspuren – Ein Beitrag zur postvariszischen Tektonik und Abtragung
des KTB-Umfelds. Geowissenschaften 13:142–146
Cramer T, Wanner A, Gumbsch P (1999) Dynamic failure of glass and single-crystalline silicon.
Z Metallkd 90:675–686
Cramer T, Wanner A, Gumbsch P (2000) Energy dissipation and path instabilities in dynamic frac-
ture of silicon single crystals. Phys Rev Lett 85:788–791
Cress GO, Brady BT, Rowell GA (1987) Sources of electromagnetic radiation from fracture of rock
samples in the laboratory. Geophys Res Lett 14(4):331–334
Cruikshank KM, Aydin A (1994) Role of fracture localization in arch formation, Arches National
Park, Utah. Geol Soc Am Bull 106:876–891
Cruikshank KM, Zhao G, Johnson AM (1991) Analysis of minor fractures associated with joints and
faulted joints. J Struct Geol 13:865–886
Currie JB, Reik GA (1977) A method of distinguishing regional directions of jointing and of identify-
ing joint sets associated with individual geologic structures. Can J Earth Sci 14:1211–1228
Dagan G (1977a) Hydrological analysis of BS3, BS4, TH4 and TH5 exploratory wells data. Petrol Ser
Ltd, Tel Aviv
Dagan G (1977b) Hydrological analysis of BS5 and BS10 exploratory wells data and summary of
hydrological results for Ramat Hovav Area. Petrol Ser Ltd, Tel Aviv
Dahan O, Nativ R, Adar EM, Berkowitz B (1998) A measurement system to determine water flux and
solute transport trough fractures in the unsaturated zone. Ground Water 36:444–448
Dahan O, Nativ R, Adar EM, Berkowitz B, Ronen Z (1999) Field observation of flow in a fracture
intersecting unsaturated chalk. Water Resour Res 35:3315–3326
Dahan O, Nativ R, Adar EM, Berkowitz B, Weisbrod N (2000) On fracture structure and preferential
flow in unsaturated chalk. Ground Water 38:444–451
Dahmen K, Ertas D, Ben-Zion Y (1998) Gutenberg-Richter and characteristic eathquake behaviour is
simple mean-field modes of heterrogeneous faults. Phys Rev E 58(2):1494
Dale TN (1923) The commercial granites of New England. U.S. Geological Survey Bulletin 738
Dally JW, Fourney WL, Irwin GR (1985) On the uniqueness of the stress intensity factor-crack veloc-
ity relationship. Int J Fracture 27:159–168
 References 539

Dart CJ, McClay K, Hollings (1995) 3D analysis of inverted extensional fault systems, southern Bristol
Channel Basin, UK. In: Buchanan JG, Buchanan PG (eds) Basin inversion. Geological Society Spe-
cial Publication 88:393–413
Davis JC (1973) Statistical and data analysis in geology. John Wiley & Sons Ltd., New York
Davis GH (1999) Structural geology of the Colorado Plateau region of southern Utah, with special
emphasis on deformation bands. Geol Soc Am, special paper 342
Davis K, Baker DM (1965) Ionospheric effects observed around the time of the Alaskan earthquake
of 28 March 1964. J Geophys Res 70:2251–2253
Davis GH, Reynolds SJ (1996) Structural geology of rocks and regions. John Wiley & Sons Ltd., New
York
De Freminville MCh (1914) Recherches sur la fragilité –l’éclatement. Rev Metall-Paris 11:971–1056
De Sitter LU (1964) Structural geology. McGraw Hill, New York
Dean S, Baranoski M, Bertoli L, Kribbs G, Stephan T, Kulander B, Sochman D, Mumpower D (1984)
Regional fracture analysis in western Valley and Ridge and adjoining plateau’ West Virginia and
Maryland. AAPG Bull 67:448
Dean SL, Kulander BR, Skinner JM (1988) Structural evolution of the Alleghanian orogeny in south-
eastern West Virginia. Geol Soc Am Bull 100:299–310
Dennis JG (1972) Structural geology. The Ronald Press Company, New York, p 532
Dennis JG (1987) Structural geology: An introduction. Wm C Brown Publishers, Dubuque
Derr JS (1986) Luminous phenomena and their relationship to rock fracture. Nature 321:470–471
Dieterich JH, Carter N (1969) Stress history of folding. Am J Sci 267:129–154
Döll W (1976) Application of an energy-balance and an energy method to dynamic crack-propaga-
tion. Int J Fracture 12:595–605
Doremus RH, Johnson WA (1978) Depths of fracture-initiating flows and initial stages of crack propa-
gation in glass. J Mater Sci 13:855–858
Drakopoulos J, Stavrakakis GN, Latoussakis J (1993) Evaluation and interpretation of thirteen official-
telegrams for the period September 10, 1986 to April 28, 1988. Tectonophysics 224(1–3):223–236
Drobdev VI, Krasnov VM, Salihov RM (1978) About ionospheric perturbation associated with earth-
quake and explosions. Izvestia Vysshih Uchebnih Zavedeniy 21:1862–1978 (in Russian)
Dulaney EN, Brace WF (1960) Velocity behavior of a growing crack. J Appl Phys 31:2233–2236
Dunne WM, North CP (1990) Orthogonal fracture systems at the limits of thrusting: an example
from southwestern Wales. J Struct Geol 12:207–215
Dyer R (1988) Using joint interactions to estimate paleostress ratios. J Struct Geol 10:685–699
Ehlers EG, Blatt H (1982) Petrology: igneous, sedimentary and metamorphic. WH Freeman and Com-
pany, San Francisco, 732 pp
Eidam J, Hammer J, Korich D, Krauss M (1992) Zur Abgrenzung altersverschiedener Granitoide
innerhalb des Lausitzer Granodiorit-Massivs. Z Geol Wissenschaft 20:289–293
Eldeen ZU, Delvaux D, Jacobs P (2000) Tectonic and paleostress evolution in the Wadi Araba segment
of the Dead Sea rift, SW Jordan. The first Stephan Müller Conference of European Geological
Society (EGS). From Continental Breakup to Collision. Israel, Dead Sea. Abstract
Elgazzar AS (1998) An inhomogeneous self-organized critical model for earthquakes. Physica A 251:303
Ellsworth WL, Beroza GC (1995) Seismic evidence for a seismic nucleation phase. Science 268:581–858
Engelder T (1979a) Mechanisms for strain within the Upper Devonian clastic sequence of the Appa-
lachian Plateau, western New York. Am J Sci 279:527–542
Engelder T (1979b) Nature of deformation within the outer limits of the Central Appalachian foreland
fold and thrust belt in New York State. Tectonophysics 55:289–310
Engelder T (1985) Loading paths to joint propagation during cycle: An example of the Appalachian
Plateau, USA. J Struct Geol 7:459–476
Engelder T (1989) The analysis of pinnate joints in the Mount Desert Island granite: Implications for
post-intrusion kinematics in the coastal volcanic belt, Maine. Geology 17:564–567
Engelder T (1993) Stress regimes in the lithosphere. Princeton University Press, Princeton, N.J., 457 pp
Engelder T (1999) Transitional-tensile fracture propagation: A status report. J Struct Geol 21:1049–1055
Engelder T, Engelder R (1977) Fossil distortion and decollement tectonics of Appalachian Plateau.
Geology 5:547–460
Engelder T, Geiser P (1979) The relationship between pencil cleavage and lateral shortening within
the Devonian section of the Appalachian Plateau, New York. Geology 7:460–464
Engelder T, Geiser P (1980) On the use of regional joint sets as trajectories of paleostress fields dur-
ing the development of the Appalachian Plateau, New York. J Geophys Res 85:6319–6341
Engelder T, Gross MR (1993) Curving cross joints and the lithospheric stress-field in eastern North-
America. Geology 21:817–820
Engelder T, Peacock DCP (2001) Joint development normal to regional compression during flexural-
flow folding: The Lilstock buttress anticline, Somerset, England. J Struct Geol 23:259–277
540 References

Engelder T, Geiser P, Bahat D (1987) Alleghanian deformation within shales and siltstones of the
Upper Devonian Appalachian Basin Finger Lakes district, New York. Geol Soc Am Continental
Filed Guide, Northeastern Section, pp 113–118
Engelder T, Fischer MP, Gross MR (1993) Geological aspects of fracture mechanics: A short course
manual notes. Geological Society of America Annual Meeting, Boston
Erdogan F (1983) Stress intensity factors. J Appl Mech-T ASME 50:992–1002
Erdogan F, Sih GC (1963) On the crack extension plates under plane loading and transverse shear.
J Basic Eng-T ASME 85:519–527
Etheridge MA (1983) Differential stress magnitudes during regional deformation and metamorphism:
Upper bound imposed by tensile fracturing. Geology 11:231–234
Ettensohn FR (1985) The Catskill Delta complex and the Acadian orogeny: A model. In: Woodrow DL,
Sevon WD (eds) The Catskill Delta. Geological Society of America Special Paper 201:39–49
Evans AG (1974) Role of inclusions in fracture of ceramic materials. J Mater Sci 9:1145–1152
Evans AG (1978) Microfracture from thermal expansion anisotropy – I. Single phase system. Acta
Metall 26:1845–1853
Evans MA (1994) Joints and decollement zones in Middle Devonian shales: Evidence for multiple
deformation events in the central Appalachian Plateau. Geol Soc Am Bull 106:447–460
Evans AG, Johnson H (1975) The fracture stress and its dependence on slow crack growth. J Mater Sci
10:214–222
Evans AG, Wiederhorn SM (1974) Proof testing of ceramic materials – An analytical basis for failure
prediction. Int J Fracture 10:379–392
Eyal Y (1996) Stress field fluctuations along the Dead Sea rift since the Middle Miocene. Tectonophysics
15:157–170
Eyal Y, Reches Z (1983) Tectonic analysis of the Dead Sea rift region since the Late Cretaceous based
on mesostructures. Tectonophysics 2:167–185
Eyal Y, Gross MR, Engelder T, Becker A (2001) Joint development during fluctuation of the regional
stress field in southern Israel. J Struct Geol 23:279–298
Fabrocini A, Fantoni S, Eckhard K (eds) (2002) Introduction to modern methods of quantum many
body theory and their applications. World Scientific
Faill RT (1973) Kink-band folding, Valley and Ridge-Province, Pennsylvania. Geol Soc Am Bull 84:1289–1213
Fairhurst C, Cook NGW (1966) The phenomenon of rock splitting parallel to the direction of maxi-
mum compression in the neighborhood of a surface. In: Proceedings of the 1st Congress Interna-
tional Society for Rock Mechanics, Lisbon, pp 687–692
Feller W (1971) An introduction to probability theory and its applications. John Wiley, New York
Ferney BD, DeVary MR, Hsia KJ (1999) Oscillatory crack growth in glass. Scripta Materialia 41:275–281
Fifolt DA, Petrenko VF, Schulson EM (1993) Preliminary study of electromagnetic radiation from
cracks in ice. Phil Mag B 67(3):289–299
Fineberg J, Marder M (1999) Instability in dynamic fracture. Phys Rep 313:1–108
Fineberg J, Gross SP, Marder M, Swinney HL (1991) Instability in dynamic fracture. Phys Rev Lett
67:457–460
Fineberg J, Gross SP, Marder M, Swinney HL (1992) Instability in the propagation of fast cracks. Phys
Rev B 45:5146–5154
Finkel VM, Golovin YI, Sereda VE (1975) Electrical effects accompaying fracture of LiF crystals and
problem of rock control. Sov Phys Sol-State 17(3):492–495
Finkelstein D, Powell J (1970) Earthquake lighting. Nature 228:759–760
Fisher RA (1938) The statistical utilization of multiple measurements. Ann Eugenic 8:376–386
Fisher RA (1953) Dispersion on a sphere. P Roy Soc Lond A Mat 217:295–305
Fisher NI, Lewis T, Embleton BJJ (1987) Statistical analysis of spherical data. Cambridge, London
Flexer A, Diamant E, Polishook B, Livnat A (1984) Relation of joints in the 'Avedat group (Eocene) to
the tectonic pattern of Israel. Israel J Earth Sci 33:12–25
Freund LB (1972a) Crack propagation in an elastic solid subjected to general loading – I. Constant
rate of extension. J Mech Phys Solids 20:129–140
Freund LB (1972b) Crack propagation in an elastic solid subjected to general loading – II. Non uni-
form rate of extension. J Mech Phys Solids 20:141–152
Freund LB (1973) Crack propagation in an elastic solid subjected to general loading – III. Stress wave
loading J Mech Phys Solids 21:47–61
Freund LB (1974) Crack propagation in an elastic solid subjected to general loading – IV. Obliquely
incident stress pulse J Mech Phys Solids 22:137–146
Freund LB (1990) Dynamic fracture mechanics. Cambridge Univ Press
Freund F (2000) Time resolve study of charge generation and propagation in igneous rocks. Geophys
Res Lett 105(5):11001–11019
Freund F (2002) Charge generation and propagation in igneous rocks. J Geodyn 33(4–5):543–570
 References 541

Frey MG (1973) Influence of Salina salt on structure in the New York – Pennsylvania part of the
Appalachian Plateau. AAPG Bull 57:1027–1037
Frid V (1990) Rockburst hazard forecast of coal seams by theirs electromagnetic radiation. Ph.D.
thesis, The state research institute of mining geomechanics and mine surveying (VNIMI), St. Pe-
tersburg (in Russian)
Frid V (1997a) Rock-burst hazard forecast by electromagnetic radiation excited by rock fracture.
Rock Mech Rock Eng 30(4):229–236
Frid V (1997b) Electromagnetic radiation method for rock and gas outburst forecast. J Appl Geophys
38:97–104
Frid V (2000) Electromagnetic radiation method water-infusion control in rockburst-prone strata.
J Appl Geophys 43:5–13
Frid V (2001) Calculation of electromagnetic radiation criterion for rockburst hazard forecast in
coal mines. Pure Appl Geophys 158:931–944
Frid V, Shabarov A, Proskurjakov V (1992) Formation of electromagnetic radiation in coal stratum.
J Min Sci+ 28(2):139–145
Frid V, Rabinovitch A, Bahat D (1999) Electromagnetic radiation associated with induced triaxial
fracture in granite. Phil Mag Lett 79:79–84
Frid V, Bahat D, Goldbaum J, Rabinovitch A (2000) Experimental and theoretical investigation of
electromagnetic radiation induced by rock fracture. Israel J Earth Sci 49:9–19
Frid V, Rabinovitch A, Bahat D (2003) Fracture induced electromagnetic radiation. J Phys D Appl
Phys 36:1620–1628
Friedman M, Logan JM (1973) Lüders bands in experimentally deformed sandstone and limestone.
Geol Soc Am Bull 84:1465–1476
Friedman M, Stearns DW (1971) Relations between stresses inferred from calcite twin lamellae and
macrofractures, Teton Anticline, Montana. Geol Soc Am Bull 82:3151–3162
Frost HJ (2001) Mechanisms of crack nucleation in ice. Eng Fract Mech 68:1823–1837
Frost HJ, Gupta V (1993) Crack nucleation mechanisms and fracture toughness measurements in
freshwater ice. In: Proceedings of the AMD Symposium on Ice Mechanics held at the Joint SES-
ASC-ASME Conference in Virginia (1993), June 6–9, AMD-Vol. 163, ASME, New York, p 235
Fujinawa Y, Takahashi K (1998) Electromagnetic radiations associated with major earthquakes. Phys
Earth Planet In 105(3–4):249–259
Fuller KNG, Fox PG, Field JE (1975) Temperature rise at tip of fast-moving cracks in glassy polymers.
P Roy Soc Lond A Mat 341:537–550
Gabrielov A, Keilis-Borok V, Zaliapin I, Newman WI (2000a) Critical transitions in colliding cascades.
Phys Rev E 62(1):237–249
Gabrielov A, Zaliapin I, Newman WI, Keilis-Borok VI (2000b) Colliding cascades model for earth-
quake prediction. Geophys J Int 143:427–437
Galperin YI, Jorjro NV, Kavrazhkin RA, et al. (1985) VLF and ELF effects in the upper ionospheric
caused by large-scaled acoustic waves in the lower ionosphere observed from Aurol-3 satellite. In:
Results of the ARCARD-3 Project, Cepadues, Toulouse, pp 661–664
Gardiner CW (1990) Handbook of stochastic methods for physics, chemistry and natural sciences,
2nd edn. Springer-Verlag, Berlin
Geiser PA (1988) The role of kinematics in the construction and analysis of geological cross section
in deformed terranes. In: Mitra G, Wojtal G (eds) Geometries and mechanisms of thrusting. Geo-
logical Society of America Special Paper 222:47–76
Geiser P, Engelder T (1983) The distribution of layer parallel shortening fabrics in the Appalachian
foreland of New York and Pennsylvania – Evidence for 2 non-coaxial phases of the Alleghanian
Orogeny. Geol Soc Am Mem 158:161–175
Gerdes A, Wörner G, Finger F (1998) Late-orogenic magmatism in the southern Bohemian Massif –
Geochemical and isotopic constraints on possible sources and magma evolution. Acta Universita-
tis Carolinae Geologica Praha 42:41–45
Germanovich LN, Dyskin AV (2000) Fracture mechanisms and instability of openings in compres-
sion. Int J Rock Mech Min 37:263–284
Germanovich LN, Ring LM, Carter BJ, Ingraffea AR, Dyskin AV, Ustinov KB (1995) Simulation of
crack growth and interaction in compression. Proc Inter Cong Rock Mech Balkema, Rotterdam,
Brookfield, pp 1:219–226
Gershenzon N, Zilpimiani D, Maguladze P (1985) Electromagnetic radiation from crack tip during
ionic crystals fracture. Dokl Akad Nauk SSSR+ 248:1077–1081
Gershenzon N, Gokhberg M, Morgunov V (1987) Sources of electromagnetic emissions preceding
seismic events. Izv-Phys Solid Earth 23(2):96–101
Gershenzon N, Gokhberg M, Karakin A (1989) Modeling the connection between earthquake prepa-
ration process and crystal electromagnetic emission. Phys Earth Planet In 57(1–2):129–138
542 References

Ghahremani F, Hutchinson JW, Tvergaard V (1990) Three-dimensional effects in microcrack nucle-

ation in brittle polycrystals. J Am Ceram Soc 73:1548–1554
Ghosh SK (1993) Structural geology: Fundamentals and modern developments. Pergamon Press,
Oxford
Gilbert GK (1904) Domes and dome structures of the High Sierra. Geol Soc Am Bull 15:29–36
Gillespie P, Walsh JJ, Watterson J (1992) Limitations of displacement and dimension data for single
faults and the consequences for data analysis and interpretation. J Struct Geol 14:1157–1172
Gilman JJ (1958) Creation cleavage steps by dislocation. Trans Metall Soc AIME 212:310–315
Gokhberg M, Morgunov V, Aronov E (1979) High frequency electromagnetic radiation during seis-
mic activity. Dokl Akad Nauk SSSR+ 248:1077–1081
Gokhberg M, Yoshino T, Morgunov V (1982) Results of recording operative electromagnetic Earth-
quake precursor in Japan. Izv-Phys Solid Eart+ 18(2):144–146
Gokhberg M, Gufel’d I, Gershenzon N (1985) Electromagnetic effects during rupture of the Earth’s
crust. Izv-Phys Solid Eart+ 21:52–62
Gokhberg M, Morgunov V, Matveyev V (1986) On the observation of anomalous electromagnetic
emission in the epicentral zone of an Earthquake. Izv-Phys Solid Eart+ 22(8):676–678
Gol’d RM, Markov G, Mogila PG (1975) Pulsed electromagnetic radiation of minerals and rocks sub-
jected to mechanical loading. Izv-Phys Solid Eart+ 7:109–111
Gold DP, Parizek RR (1976) Field guide to lineamer and fractures in central Pennsylvania. Interna-
tional Conference on New Basement Tectonics, 2nd edn. Newark University of Delaware
Goldbaum J, Frid V, Rabinovitch A, Bahat D (2001) Electromagnetic radiation induced by percussion
drilling. Int J Fracture 111:L15-L20
Goldstein A, Marshak S (1988) Analysis of fracture array geometry. In: Marshak S, Mitra G (eds)
Basic methods of structural geology. Prentice Hall, Englewood Cliffs, NJ, pp 249–267
Golovin Y, Shibkov A (1986a) Dynamics of dislocation pile-ups and pulsed polarization of LiF single
crystals in the case of single slip. Sov Phys Sol-State 28(9):1625–1626
Golovin Y, Shibkov A (1986b) Fast electrical processes and dynamics of dislocations in plastically
deformed alkali halide crystals. Sov Phys Sol-State 28(11):1965–1968
Golubovic L, Feng S (1991) Rate of microcrack nucleation. Phys Rev A 43:5223–5227
Goncharov A, Korjakov VP, Kuznetzov VM (1980) Acoustic emission and electromagnetic radiation
during uniaxial compression. Dokl Akad Nauk SSSR+ 255(4):821–824
Goodwin AM (1995) Spatial change in joint geometry in the Chalk of eastern England. In: Ameen MS
(ed) Fractography: Fracture topography as a tool in fracture mechanics and stress analysis. Geo-
logical Society Special Publication 92:197–213
Gradshteyn IS, Ryzhik IM (1980) Table of integrals, series and products, fourth edn. Academic Press,
New York, pp 80ff
Grater RK (1945) Landslide in Zion Canyon, Zion National Park, Utah. J Geol 53:116–124
Gray M, Mitra G (1993) Migration of deformation fronts during progressive deformation – Evidence
from detailed structural studies in the Pennsylvania anthracite region, USA. J Struct Geol 15(3–5):
435–459
Gregory HE (1950) Geology and geography of the Zion National Park region Utah and Arizona. Geol
Surv Prof Paper 220
Griffith AA (1920) The phenomena of rupture and flow in solids. Philos T Roy Soc A 221:163–198
Griffith AA (1924) The theory of rupture. Proceedings of the First International Conference of Ap-
plied Mechanics, Delft, pp 55–63
Griggs DT, Handin J (1960) Observations on fracture and hypothesis of earthquakes. In: Griggs DT,
Handin J (eds) Rock deformation. Geol Soc Am Mem 79:347–364
Gross MR (1993) The origin and spacing of cross joints: Examples from the Monterey Formation,
Santa Barbara coastline, California. J Struct Geol 15:737–751
Gross MR (1995) Fracture partitioning: Failure mode as a function of lithology in the Monterey For-
mation of coastal California. GSA Bull 107: 779–792
Gross MR, Engelder T (1991) A case for neotectonic joints along the Niagara Escarpment. Tectono-
physics 10:631–641
Gross MR, Fischer MP, Engelder T, Greenfield RJ (1995) Factors controlling joints spacing in interbedded
sedimentary rocks: Integrating numerical models with field observations from the Monterey For-
mation, USA. In: Ameen MS (ed) Fractography: Fracture topography as a tool in fracture mechanics
and stress analysis. Geological Society (London) Special Publication 92, pp 215–233
Gross MR, Bahat D, Becker A (1997) Relations between jointing and faulting based on fracture-spacing
ratios and fault-slip profiles: A new method to estimate strain in layered rocks. Geology 25:887–890
Grossenbacher K, Bahat D, Karasaki K (1996) Triangulator: Excel spreadsheets for converting rela-
tive bearings to XYZ coordinates, with applications to scaling photographs and orienting sur-
faces. Comput Geosci 22:1053–1059
 References 543

Grossmann NF (1983) A numerical method for the definition of discontinuity sets. Proc. 5th Congress
International Society for Rock Mechanics, Melbourne, Balkema, 1:B17–B21
Guarino A, Garcimartin A, Ciliberto S (1998) An experimental test of the critical behavior of fracture
precursor. Eur Phys J B 6:13–24
Guarino A, Ciliberto S, Garcimartin A (1999) Failure time and microcrack nucleation. Europhys Lett
47:456–461
Guarino A, Ciliberto S, Garcimartin A, Zei M, Scorretti R (2002) Failure time and critical behaviour of
fracture precursors in heterogeneous materials. Eur Phys J B26:141–151
Gucer DE, Gurland J (1962) Comparison of statistics of two fracture models. J Mech Phys Solids
10:365–373
Gudmundsson A, Homberg C (1999) Evolution of stress fields and faulting in seismic zones. Pure
Appl Geophys 154:257–280
Gueguen Y, Palciauskas V (1994) Introduction to the physics of rocks. Princeton University Press,
Princeton
Guin JP, Wiederhorn SM (2003) Crack growth threshold in soda lime silicate glass: Role of hold-time.
J Non-Cryst Solids 316:12–20
Gutenberg B, Richter CF (1954) Seismicity of the Earth and associated phenomena. Princeton Uni-
versity Press, Princeton
Gvirtzman G (1969) The Saqiye Group (late Eocene to early Pleistocene) in the coastal plain and
Hashephela regions. Geol Surv Isr Bull 51(2)
Hahn GT, Rosenfield AR (1965) Local yielding and extension of a crack under plane stress. Acta
Metall 13:293–306
Hamblin WK (1965) Origin of Òreverse dragÓ on the downthrow side of normal faults. Geol Soc Am
Bull 76:1145–1164
Hancock PL (1972) The analysis of en-echelon veins. Geol Mag 3:269–276
Hancock PL (1985) Brittle microtectonics: Principles and practice. J Struct Geol 7:437–457
Hancock PL (1986) Joint spectra. In: Nichol I, Nesbitt RW (eds) Geology in the real world – the Kingsley
Dunham volume. Institution of Mining and Metallurgy, London, pp 155–164
Hancock PL, Al-Kadhi A (1978) Analysis of mesoscopic fractures in the Dhurma-Nisah segment of
the central Arabian Graben system. J Geol Soc 135:339–347
Hancock PL, Engelder T (1989) Neotectonic joints. Geol Soc Am Bull 101:1197–1208
Hancock PL, Dunne WM, Tringham ME (1982) Variscan structures in south-west Dyfed. In: Bassett
MG (ed) Excursion-guide on geological excursions in Dyfed, South-west Wales 12:215–248
Handin J, Hager RV Jr, Friedman M, Feather JN (1963) Experimental deformation of sedimentary
rocks under confining pressure: Pore pressure tests. AAPG Bull 47:717–755
Hansen EC, Borg IY (1962) The dynamic significance of deformation lamellae in quartz of a calcite-
cemented sandstone. Am J Sci 260:321–336
Harlow DG (1985) The pure flaw model for chopped fiber composites. P Roy Soc Lond A Mat397:211–232
Hatheway AW, Kiersch GA (1982) Engineering properties of rock. In: Carmichael RC (ed) Handbook
of physical properties or rock. CRC Press Inc, Boca Raton, Florida, pp 289–311
Hauch JA, Marder MP (1998) Energy balance in dynamic fracture, investigated by a potential drop
technique. Int J Fracture 90:133–151
Hayakawa M, Tomizawa I, Ohta K (1993) Direction finding of precursory radio emissions associated
with earthquakes: a proposal. Phys Earth Planet In 77:127–135
Hayati G (1975) The engineering geology of some chalks in Israel. Ph.D. Thesis, Technion, Israel Inst
of Technol, (in Hebrew)
Helgeson DE, Aydin A (1991) Characteristics of joint propagation across layer interfaces in sedimen-
tary rocks. J Struct Geol 13:897–911
Hermann HJ, Roux S (eds) (1990) Statistical models for the fracture of disordered media. North Hol-
land, Amsterdam
Hertzberg RW (1976) Deformation and fracture mechanics of engineering materials. John Wiley &
Sons Ltd., New York, p 605
Hidalgo RC, Moreno Y, Kun F, Herrmann HJ (2002) Fracture model with variable range of interaction.
Phys Rev E 65:046148–1–8
Hirata T, Satoh T, Ito K (1987) Fractal structure of spatial distribution of microfracturing in rock.
Geophys J Roy Astr S 90:369–374
Hirschmann G (1966) Assyntische und variszische Baueinheiten im Grundgebirge der Oberlausitz.
Freiberger Forschungsh C 212:1–146
Hoagland RG, Hahn GT, Rosenfield AR (1973) Influence of microstructure on fracture propagation
in rock. Rock Mech 5:77–106
Hobbs DW (1967) The formation of tension joints in sedimentary rocks: An explanation. Geol Mag
104:550–556
544 References

Hobbs BE, Means WD, Williams PF (1976) An outline of structured geology. John Wiley & Sons Ltd.,
New York, p 571
Hodgson RA (1961a) Classification of structures on joint surfaces. Am J Sci 259:493–502
Hodgson RA (1961b) Regional study of jointing in comb ridge-Navajo Mountain area, Arizona and
Utah. AAPG Bull 45:1–38
Hoek E, Bieniawski ZT (1966) Fracture propagation mechanism in hard rock. C R Congr Soc Int Mec
Roches 3:243–249
Holloway DG (1986) The fracture behaviour of glass. Glass Technol 27:120–133
Holloway S, Chadwick RA (1986) The Sticklepath-Lustleigh fault zone: Tertiary sinistral reactivation
of a Variscan dextral strike slip fault. J Geol Soc London 143:452–477
Holman WR (1976) The origin of sheeting joints. A hypothesis. Unpublished Ph.D. thesis. University
of California, Los Angeles, USA
Holzhausen GR (1989) Origin of sheet fracture, 1. Morphology and boundary conditions. Eng Geol
27:225–278
Holzhausen GR, Johnson AM (1979) Analyses of longitudinal splitting of uniaxially compressed rock
cilinders. Int J Rock Mech Min 16:163–177
Horii H, Nemat-Nasser S (1985) Compression-induced microcrack growth in brittle solids: Axial split-
ting and shear failure. J Geophys Res 90:3105–3125
Hovestadt M (1902) Jena glass and its scientific and industrial applications. Macmillan & Co, London
Hsieh PA, Newman SP, Stiles GK, Simpson ES (1985) Field determination of the three-dimensional
hydraulic conductivity tensor of anisotropic media. 2. Methodology and application to fractured
rocks. Water Resour Res 21:1667–1676
Huber NK (1987) The geologic story of Yosemite National Park. U.S. Geological Survey Bulletin 1595
Hudak P (1992) Terminal decollement in the Appalachian plateau of northwestern PA. Northeastern
Geology 1:108–112
Hull D (1999) Fractography: Observing, measuring, and interpreting fracture surface topology. Cam-
bridge University Press, Cambridge
Hunt CB (1956) Cenozoic geology of the Colorado Plateau. U.S. Geological Survey Professional Pa-
pers 279, 99 pp
Huntington HE (1966) The Yosemite story. Doubleday & Company, New York
Ipatov Y (1989) Fundamentals of electromagnetic radiation method for rockburst forecast on Khibin
apatite mines, Ph.D. thesis, The State Research Institute of Mining Geomechanics and Mine Sur-
veying (VNIMI), St. Petersburg (in Russian)
Irwin GR (1957) Analysis of stresses and strains near the end of a crack traversing a plate. J Appl
Mech-T ASME 24:361–364
Irwin GR (1962) Crack-extension force for a part-through crack in a plate. J Appl Mech-T ASME 29:
651–654
Issar A, Bahat D, Wakshal E (1988) Occurrence of secondary gypsum veins in joints in chalks in the
Negev, Israel. Catena 15:241–247
Ito K, Matsuzaki M (1990) Earthqaukes as self-organized critical phenomena. J Geophys Res 95:
6853–6860
Ivanov VV, Pimonov AG, Yegorov PV, Kolpakova LA (1990) Determination of the constants of termo-
fluctuational equations for strength and crack parameters from electromagnetic pulse emissions
of rocks. Izv-Phys Solid Earth 26(7):597–601
Jaeger JC, Cook NGW (1979) Fundamentals of rock mechanics. Chapman and Hall, London, pp 593
Jagasivamani V, Iyer KJL (1988) Electromagnetic emission during the fracture of heat-treated spring
steel. Mater Lett 6(11–12):418–422
Jayatilaka A de S, Trustrum K (1978) Fracture of brittle materials in uniaxial compression. J Mater Sci
13:455–457
Jensen HJ (1998) Self-organized criticality. Cambridge Univ Press, Cambridge
Johnson RM (1970) Physical processes in geology. Freeman and Company, San Francisco
Johnson E (1992) Process region changes from rapidly propagating cracks. Int J Fracture 55:47–63
Johnson JW, Holloway DG (1966) On the shape and size of the fracture zones on glass fracture sur-
faces. Phil Mag 14:731–743
Johnson RA, Wichern DW (1988) Applied multivariate statistical analysis. Prentice Hall, Engelwood
Cliffs, New York
Johnston AC (1991) Light from seismic waves. Nature 354:361
Kalthoff JF (1973) On the propagation direction of bifurcated cracks. In: Sih GC (ed) Dynamic crack
propagation. Noordhoff, Leyden, pp 449–458
Kanninen MF, Popelar CH (1985) Advanced fracture mechanics. Oxford University Press, New York
Karasaki K, Freifeld B, Cohen A, Grossenbacher K, Cook P, Vasco D (2000) A multidisciplinary frac-
tured rock characterization study at Raymond field site, Raymond, CA. J Hydrol 236:17–34
 References 545

Kattenhorn SA, Aydin A, Pollard DD (2000) Joints at high angles to normal fault strike: An explana-
tion using 3-D numerical models of fault-perturbed stress fields. J Struct Geol 22:1–23
Kay SM, Snedden WT, Foster BP, Kay RW (1983) Upper mantleand crustal fragments in the Ithaca
Kimberlites. J Geol 91:277–290
Kelly PM, Leach CA, Pawlings RD (1994) Acoustic, emission studies of fatigue loaded SiC platelet-
reinforced Y-TZP. J Mater Sci Lett 13:211–212
Kelly PG, Sanderson DJ, Peacock DCP (1998) Linkage and evolution of conjugate strike-slip fault
zones in limestones of Somerset and Northumbria. J Struct Geol 20:1477–1493
Kelner S, Bouchon M, Coutant O (1999) Characterization of fractures in shallow granite from the
modelling of the anisotropy and attenuation of seismic waves. B Seismol Soc Am 89:706–717
Kerkhof F (1973) Wave fractographic investigation of brittle fracture dynamics. In: Sih GC (ed) Dy-
namic crack propagation. Noordhoff, Int Pub Leyden, pp 3–35
Kerkhof F (1975) Bruchmechanische Analyse von Schadensfällen an Gläsern. Glastech Ber-Glass
48:112–124
Kerkhof F, Richter H (1969) Investigation of the influence of water vapour on crack velocities in glass
by ultrasonic fractography. Proc. 2nd Intern. Conference on Fracture, Brighton 40:463–473
Khatiashvili N (1984) The electromagnetic effect accompanying the fracturing of alkaline halide crys-
tals and rocks. Izv-Phys Solid Eart+ 20:656–661
Khesin B, Alexeyev V, Eppelbaum L (1996) Interpretation of geophysical fields in complicated envi-
ronments. Kluwer Academic Publishers, London
Khesin B, Alexeyev V, Eppelbaum L (1997) Rapid methods for interpretation of induced polarization
anomalies. J Appl Geophys 37:117–130
King CY (1983) Electromagnetic emissions before earthquakes. Nature 301:377
King GCP (1986) Speculation on the geometry of the initiation and termination process of earth-
quake rupture and its relation to morphology and geological structure. Pure Appl Geophys 124:
567–585
King PJ, Sheard FW (1970) Unharmonic attenuation of surface waves on crystals at low temperatures.
P Roy Soc Lond A Mat 320:175–179
Kittel C (1987) Quantum theory of solids. John Wiley & Sons Ltd., New York
Kloster M, Hansen A, Hemmer PC (1997) Burst avalanches in solvable models of fibrous materials.
Phys Rev E 56:2615–2625
Knapp RB, Norton D (1981) Preliminary numerical analysis of crystallization and stress evolution in
cooling pluton environments. Am J Sci 281:35–68
Knauss WG (2000) Prospectives in experimental solid mechanics. Int J Solids Struct 37:251–266
Knauth LP, Lowe DR (1978) Oxygen isotope geochemistry of cherts from Onverwacht group (3.4 bil-
lion years), Transvaal, South-Africa, with implications for secular variations in isotopic composi-
tion of cherts. Earth Planet Sc Lett 41:209–222
Knott SD, Beach A, Brockbank PJ, Brown JL, McCallum JE, Welbon AI (1996) Spatial and mechanical
controls on normal fault populations. J Struct Geol 18:356–372
Kobayashi AS, Mall S (1978) Dynamic fracture toughness of Homalite-100. Exp Mech 18:11–18
Kobayashi A, Ohtani N, Sato T (1974) Phenomenological aspects of viscoelastic crack propagation.
J Appl Polym Sci 18:1625–1638
Kornfel’d MI (1971) Electrical charges on the surface of alkaline-halide crystals. Sov Phys Sol-State
13(2):381–385
Kornfel’d MI (1977) Mechanism of electrification of cleaved crystals. Sov Phys Sol-State 19(4):648–649
Kronfeld J, Rosenthal R, Weinberger G, Flexer A, Berkowitz B 1993. The interaction of two major old
water bodies and its implication for the exploitation of groundwater in the multiple aquifer sys-
tem in the central and northern Negev. Journal of Hydrology 143: 169-190
Kossmat F (1927) Gliederung des varistischen Gebirgsbaues. Abh Sächs Geol Landesamt 1:1–39
Kossobokov VG, Keilis-Borok VI, Cheng B (2000) Similarities of multiple fracturing on a neutron star
and on the Earth. Phys Rev E 61(4):3529–3533
Kotz S, Nadarajah S (2000) Extreme value distributions theory and applications. Imperial College
Press, London
Kowalik WS, Gold DP (1976) The use of Landsat-1 imagery in mapping lineaments in Pennsylvania.
Proc. 1st Int. Conf. on the New Basement Tectonics. Utah Geol Soc, pp 236–249
Krapivsky PL, Majumdar SN (2000) Traveling waves, front selection and exact nontrivial exponents
in a random fragmentation problem. Phys Rev Lett 85:5492–5495
Krauss M, Eidam J, Hammer J, Korich D (1992) Die cadomisch-variszische Entwicklung des Lausitzer
Granodiorit-Komplexes. Zbl Geol Paläontol Teil I 1/2:71–85
Krenkel E (1924) Der Syrische Bogen. Centrail Mineral 9:274–281, 10301–10313
Kuksenko VS, Mansurov VA (1986) Fracture localization in rocks at different scale levels. Physical-
technical Problems of Economical Resources Mining 3:49–55
546 References

Kuksenko VS, Lyashkov AI, Mirzoev KM (1982) Connection between the sizes of cracks generated
under loading and duration of elastic energy emission. Dokl Akad Nauk SSSR 246(4):846–848
Kuksenko VS, Manzikov V, Mansurov VA (1985) Regularities in the development of microfocal rup-
ture. Izv-Phys Solid Earth 21(7):553–556
Kuksenko VS, Ingevatkin IE, Mahgikov BC (1987) Physical and methodical foundations of rockburst
forecast. Physical-technical Problems of Economical Material Mining 1:9–21
Kuksenko VS, Kilkeev RS, Lyashkov AI (1990) Electrification of quartz glass by mechanical loading.
Sov Phys Sol-State 32(8):1320–1322
Kulander BR, Barton CC, Dean SC (1979) The application of fractography to core and outcrop frac-
ture investigations. Report to U.S. D.O.E. Morgantown Energy Technology Center, METC SP-79/3
Kurie AE (1966) Recurrent structural disturbance of Colorado Plateau margin near Zion National
Park Utah. Geol Soc Am Bull 77:867
Kustov VA, Liperovsky T (1988) About perturbation in ionospheric F-region begore strong earth-
quake. Izv-Phys Solid Eart+ 4:12–20
Lacazette A, Engelder T (1992) Fluid-driven cyclic propagation of a joint in the Ithace siltstone, Ap-
palachian Basin, New York. In: Evans B, Wong TF (eds) Fault mechanics and transport properties
of rocks. Academic Press Ltd, London, pp 297–324
Ladeira FL, Price NJ (1981) Relationship between fracture spacing and bed thickness. J Struct Geol
3:179–183
Lake SD, Karner GD (1987) The structure and evolution of the Wessex Basin, southern England – an
example of inversion tectonics. Tectonophysics 137:347
Landau LD, Lifshitz EM (1963) Theory of elasticity. Pergamon Press, London
LaPointe PR, Hudson JA (1985) Characterization and interpretation of rock mass joint patterns. Geo-
logical Society of America Special Paper 199
Laube GC (1876) Geologie des böhmischen Erzgebirges I. Archiv f Naturwiss Landesdurchforsch,
pp 1–208
Lawn BR (1993) Fracture of brittle solids, 2nd edn. Cambridge Solid State Science Series, Cambridge
University Press, Cambridge
Lawn BR, Wilshaw TR (1975a) Fracture of brittle solids. Cambridge University Press, London, pp 1–204
Lawn BR,Wilshaw TR (1975b) Review indentation fracture: Principles and applications. J Mater Sci
10:1045–1081
Lazarus V, Leblond JB, Mouchrif SE (2001a) Crack front rotation and segmentation in mixed mode
I + III or I + II + III. Part I: Calculation of stress intensity factors. J Mech Phys Solids 49:1399–1420
Lazarus V, Leblond JB, Mouchrif SE (2001b) Crack front rotation and segmentation in mixed mode
I + III or I + II + III. Part II: Comparison with experiments. J Mech Phys Solids 49:1421–1443
Lee FT, Diehl SF (1989) Geologic and geomechanical properties of the Mount Valdo granite, Maine.
Maine Geological Survey, vol 2, pp 187–200
Lee CC, Liu JY, Pan CJ (2000) The height of sporadic-E layer simultaneously observed by the VHF
radar and ionosondes in Chung-Li. Geophys Res Lett 27:641–644
Letouzey J, Tremolieres P (1980) Paleo-stress fields around the Mediterranean derived from
microtectonics: Comparison with plate tectonic data. Rock Mech 9:173–192
Levi T (2003) Joint sets as a tool for analyses the tectonic deformation in the central Arava western
margins. Ms Sc Thesis, Ben Gurion University, Israel (in Hebrew)
Li Z, Yang J, Lee H (1988) Temperature-fields near a running crack tip. Engng Fracture Mech 30:791–
799
Li L, Aubertin M, Simon R (1999) Multiaxial failure criterion with time and size effects for intact
rocks. In: Amadei B, Kranz RL, Scott GA, Smeallie PH (eds) Rock mechanics for industry. Balkema,
Rotterdam, pp 653–659
Lindner H (1972) Ergebnisse der Gravimetermessungen im Bereich des Lausitzer Massivs und seiner
Randgebiete. Geologie 21:927
Link B, Epstein RI (1996) Thermally driven neutron star glitches. Astrophys J 457:844–854
Linsser H (1967) Investigation of tectonics by gravity detailing. Geophys Prosp XV/3:480–515
Liu C, Knauss WG, Rosakis AJ (1998) Loading rates and the dynamic initiation toughness in brittle
solids. Int J Fracture 90:103–118
Liu JY, Chu YH, Chen MQ (2000) Modeling and ground observations of the ionosphere related to the
COSMIC project. Terr Atmos Ocean Sci 11:349–364
Lobst R (2001) Der Intrusionsverband cadomischer Granodiorite am Klosterberg bei Demitz-Thumitz.
Exkursionsführer und Veröffentlichungen der GGW 212:67–69
Lockner DA (1993) The role of acoustic emission in the study of rock fracture. Int J Rock Mech Min
7:883–889
Lockner DA (1996) Brittle fracture as an analog to earthquakes: Can acoustic emission be used to
develop a viable prediction strategy. J Acoustic Emission 14(3–4):S88–S101
 References 547

Lockner DA, Byerlee JD, Kuksenko V, Ponomarev A, Sidorin A (1991) Quasi-static fault growth and
shear fracture energy in granite. Nature 350:39–42
Lockwood JP, Moore JG (1979) Regional deformation of the Sierra Nevada, California, on conjugate
microfault sets. J Geophys Res 84:6041–6049
Loewy S (1995) The post-Alleghanian tectonic history of the Appalachian Basin based on joint pat-
terns in Devonian black shales. University Park, Pennsylvania, Pennsylvania State University,
(Master’s thesis)
Long JCS, et al. (1996) Rock fractures and fluid flow: Contemporary understanding and applications. Com-
mission on Geosciences, Environment and Resources (CGER), The National Academies Press, 551 pp
Loosveld RJH, Franssen RCMW (1992) Extensional vs. shear fractures: Implications for reservoir
characterization. Society of Petroleum Engineers 25017:23–30
Maccagno TM, Knott JF (1989) The fracture behaviour of PMMA in mixed modes I and II. Eng Fract
Mech 34:65–86
Mandelbrot B (1960) The Pareto-Levy law and the distribution of income. Int Econ Rev 1:79–106
Mansurov VA (1994) Acoustic emission from failing rock behavior. Rock Mech Rock Eng 27(3):173–182
Marder M (1993) Simple models of rapid fracture. Physica D 66:125–134
Marder M (1996a) Energetic developments in fracture. Nature 381:275–276
Marder M (1996b) Statistical mechanics of cracks. Phys Rev E 54:3442–3454
Marder M, Gross SP (1995) Origin of crack tip instabilities. J Mech Phys Solids 43:1–48
Marder M, Liu XM (1993) Instability in lattice fracture. Phys Rev Lett 71:2417–2420
Markov GA, Ipatov Y (1986) Method of electromagnetic radiation for rockburst forecast on apatite
mines. Eng Geol 3:54–57 (in Russian)
Mart Y, Horowitz A (1981) The tectonics of the Timna region in Southern Israel and the evolution of
the Dead-Sea Rift. Tectonophysics 79:165–199
Martel SJ (1997) Effects of cohesive zones on small faults and implications for secondary fracturing
and fault trace geometry. J Struct Geol 19:835–847
Martens JH (1924) Igneous rocks of Ithaca, New York and vicinity. Geol Soc Am Bull 3:305–320
Matejka D, Janousek K (1998) Whole-rock geochemistry and petrogenesis of granites from the north-
ern part of the Moldanubian batholith (Czeck Republic). Acta Universitatis Carolinae Geologica
42:73–79
Matsuzaki M, Takayasu H (1991) Fractal features of the earthquake phenomenon and a simple me-
chanical model. J Geophys Res 96:19925–19931
Matte PH, Maluski H, Rajlich P, Franke W (1990) Terrane boundaries in the Bohemian Massif: Result
of large-scale Variscan shearing. Tectonophysics 177:151–170
Mayo EB (1941) Deformation in the interval Mt. Lyell–Mt. Whitney, California. Geol Soc Am Bull
94:563–575
McConaughy DT, Engelder T (1999) Joint interaction with embedded concretions: Joint loading con-
figurations inferred from propagation paths. J Struct Geol 21:1637–1652
Means WD (1987) A newly recognized type of slickensise striation. J Struct Geol 9:585–590
Means WD (1979) Stress and strain. Springer-Verlag, Heidelberg
Means WD (1987) A newly recognized type of slickenside striation. J Struct Geol 9:585–590
Mecholsky JJ (1991) Quantitative fractography: An assessment. In: Frechette VD, Varner JR (eds)
Fractography of glasses and ceramics II. Ceramic Transactions 17:413–451
Mecholsky JJ, Freiman SW (1979) Determination of fracture mechanics parameters through
fractographic analysis of ceramics. In: Freiman SW (ed) Fracture mechanics applied to brittle
materials. ASTM Spec Tech Pub 678:136–150
Menzel D, Schröder B (1994) Geologische Kriterien zur Unterbau-Exhumierung im Naab-Gebirge.
KTB Report 94–3:179–184
Michalske TA (1977) The stress corrosion limit: Its measurement and implications. In: Bradt RC,
Evans AG, Hasselman DPH, Lange FF (eds) Fracture mechanics of ceramics: Surface flaws, statis-
tics, and microcracking, vol 5. Plenum Press, New York, p 277
Michalske TA (1984) Fractography of slow fracture in glass. In: Mecholsky JJ Jr, Powell SR Jr (eds)
Fractography of ceramic and metal failures. Philadelphia, American Society for Testing and Ma-
terials Special Technical Publication 827:121–136
Miroshnichencko M, Kuksenko V (1980) Emission of electromagnetic pulses during nucleation of
cracks in solid insulators. Sov Phys-Solid State 22(5):895–896
Mirowski P (1995) Mandelbrot’s economics after a quarter century. Fractals 3:581–600
Misra A (1975) Electromagnetic effects at metallic fracture. Nature 254:133–134
Misra A (1977) Physical model for the stress-induced electromagnetic effect in metals. Appl Phys
16:195–199
Misra A, Ghosh S (1980) Electromagnetic radiation characteristics during fatigue propagation and
failure. Appl Phys 23:387–390
548 References

Mogi K (1962) Study of elastic shocks caused by the fracture of heterogeneous materials and its
relation to earthquake phenomena. B Earthq Res I Tokyo 40:125–173
Mogi K (1985) Earthquake prediction. Academic Press, Tokyo
Molchan G, Kronrod T, Panza G (1997) Multiscale seismicity medel for seismic risk. Bull Seismol Soc
Am 87:1220–1229
Molchanov OA, Hayakawa M (1998a) Subionospheric VLF perturbations possibly related to earth-
quakes. J Geophys Res 103(A8):17489–17504
Molchanov OA, Hayakawa M (1998b) On the generation mechanism of ULF seismogenic electromag-
netic emissions. Phys Earth Planet In 105:201–210
Molnar P, Tapponier P (1975) Cenozoic tectonics of Asia: Effects of a continental collision. Science
189:419–426
Molotskii MI (1980) Dislocation mechanism for Misra effect. Sov Tekh Phys Lett 6(1):22–23
Moore DE, Lockner DA (1995) The role of microcracking in shear-fracture propagation in granite.
J Struct Geol 17:95–114
Morgunov V (1985) Electromagnetic emission during seismic activity. Izv-Phys Solid Earth 21(3):
220–226
Morgunov V, Matveyev V (1990) Electromagnetic emission in aftershocks of Spitak earthquake. Izv-
Phys Solid Earth 26:457–461
Morgunov V, Matveyev V (1991) Electric and electromagnetic effects in the epicentral zone of Spitak
aftershocks. Izv-Phys Solid Earth 27:1002–1005
Morgunov V, Gerasimovich E, Matveyev V (1988) Deformation of the Earth surface and anomalous
electromagnetic emission. Izv-Phys Solid Earth 24:924–929
Mortimore RN (1990) Chalk or chalks. Proc Intern Chalk Symp, Brighton Politechnic, Thomas Telford,
pp 15–45
Mott NF (1948) Fracture of metals: Theoretical considerations. Engineering-London 165:16–18
Muehlberger WR (1961) Conjugate joint sets of small dihedral angle. J Geol 69:211–219
Müller A (2000) Cathodolumuniscence and characterization of defect structures in quartz with
applications to the study of granitic rocks. Thesis, University Göttingen, pp 1–229, http://
webdoc.sub.gwdg.de/diss/2001/mueller/index.html
Müller G, Dahm T (2000) Fracture morphology of tensile cracks and rupture velocity. J Geophys Res
105:723–738
Müller A, MŸller B, Behr HJ (2001) Structural constants in granitic rocks of the Lusatian Granodior-
ite Complex and the Erzgebirge, Germany – in commemoration of Hans Cloos. Z Geol Wissenschaft
29:521–544
Murakami Y (ed) (1987) Stress intensity factors handbook, vol 1, 2. The Society of Materials Sci, Ja-
pan, Pergamon Press, Oxford
Murakami Y (ed) (1992) Stress intensity factors handbook, vol 3. The Society of Materials Sci, Japan,
Pergamon Press, Oxford
Murakami Y (ed) (2001) Stress intensity factors handbook, vol 4, 5. The Society of Materials Sci, Ja-
pan, Pergamon Press, Oxford
Muraoka H, Kamata H (1983) Displacement distribution along minor fault traces. J Struct Geol 5:483–495
Murgatroyd JB (1942) The significance of surface marks on fractured glass. J Soc Glass Technol 26:
155–171
Murphy PJ (1981) Detachment structures in south central NY. Northeastern Geology 3:105–116
Nabaro B (1967) Theory of crystal dislocations. Oxford University Press, Oxford
Nanjo K, Nagahawa H, Satomura M (1998) Rates of aftershocks decay and the fractal structure of
active fault systems. Tectonophysics 287:173–186
Narr W, Suppe J (1991) Joint spacing in sedimentary rocks. J Struct Geol 13:1037–1047
Nativ R, Nissim I (1992) Characterization of a desert aquitard – hydrologic and hydrochemical con-
siderations. Ground Water 30:598–606
Nativ R, Adar E, Dahan O, Geyh M (1995) Water recharge and solute transport through the vadose
zone of fractured chalk under desert conditions. Water Resour Res 31:253–261
Nativ R, Adar E, Dahan O, Nissim I (1997a) Water salinization in arid regions – Observations from the
Negev desert, Israel. J Hydrol 196:271–296
Nativ R, Halleran A, Hunley A (1997b) Evidence for ground-water circulation in the brine-filled
aquitard, Oak Ridge, Tennessee. Ground Water 35:647–659
Nativ R, Adar EM, Becker A (1999) Designing a monitoring network for contaminated ground water
in fractured chalk. Ground Water 37:38–47
Nativ R, Adar E, Assaf L, Nygaard E (2003) Characterization of the hydraulic properties of fractures
in chalk. Ground Water 41:532–543
Nelson RA (1979) Natural fracture systems: Description and classification. Bull Amer Assoc Petrol
Geolog 63:2214–2221
 References 549

Nelson RA (1987) Fracture reservoirs: Turning knowledge into practice. J Petrol Technol 16470:407–414
Nemcok M, Bayer R, Miliorizos M (1995) Structural analysis of the inverted Bristol Channel Basin:
Implications for the geometry and timing of fracture porosity. In: Buchanan JG, Buchanan PG
(eds) Basin inversion. Geological Society Special Publication 88:355–392
Nesbitt AC, Austin BA (1988) The emission and propagation of electromagnetic energy from stressed
quartzite rock underground. The Trans of the SA Inst of Electr Eng 79:53–57
Newman WI, Phoenix SL (2001) Time dependent fiber bundles with local load-sharing. Phys Rev E
63:0215071–19
Newman WI, Turtcotte DL, Gabrielov AM (1995) Log-periodic behavior of a hierarchical failure model
with applications to precursory seismic activation. Phys Rev E 52(5):4827–4835
Nickelsen RP (1979) Sequence of structural stages of the Allegheny orogeny at the Bear Valley strip
mine, Shamokin, Pennsylvania. Am J Sci 279:225–271
Nickelsen RP (1988) Structural evolution of folded thrusts and duplexes on a first-order anticlinorium
in the Valley and Ridge Province of Pennsylvania. In: Mitra G, Wojtal S (eds) Geometries and
mechanics of thrusting with special reference to the Appalachians. Geological Society of America
Special Paper 222:89–106
Nickelsen RP (1996) Alleghanian sequential deformationon on the SW limb of the Pennsylvania Sa-
lient in Fulton and Franklin Counties, south-central Pennsylvania: Pennsylvania Geology Survey.
Annual field conference of Pennsylvania geologists, 61st, guidebook, pp 1–9
Nickelsen RP, Engelder T (1989) Fold-thrust geometries of the Juniata culmination, Central Appala-
chians of Pennsylvania. In: Engelder T (ed) Structures of the Appalachian foreland fold-thrust
belt. International Geological Congress, 28th, Fild Trip Guidebook T166
Nickelsen RP, Hough ND (1967) Jointing in the Appalachian Plateau of Pennsylvania. Geol Soc Am
Bull 78:609–630
Nikiforova NN, Yadakhin TN (1989) Studies of electromagnetic emission of tectonic origin in the
Kirghiz SSR. Phys Earth Planet In 57:68–74
Niles WH (1871) Peculiar phenomena observed in quarrying. Proc Boston Soc Nat Hist vol 14
Nitsan V (1977) Electromagnetic emission accompanying fracture of quartz-bearing rocks. Geophys
Res Lett 4(8):33–336
Nur A (1982) The origin of tensile fracture lineaments. J Struct Geol 4:31–40
O’Keefe SG, Thiel DV (1995) A mechanism for the production of electromagnetic radiation during
fracture of brittle materials. Phys Earth Planet In 89:127–135
Odling NE, Gillespie P, Bourgine B, Castaing C, Chils J-P, Christensen NP, Fillion E, Genter A, Olsen C,
Thrane L, Trice R, Aarseth E, Walsh JJ, Watterson J (1999) Variations in fracture system geometry
and their implications for fluid flow in fractured hydrocarbon reservoirs. Petrol Geosci 5:373–384
Ogawa T, Oike K, Miura T (1985) Electromagnetic radiation from rocks. J Geophys Res 90(d4):
6245–6249
Olami Z, Feder HJS, Christensen A (1992) Self-organized criticality in a continuous, non-conserva-
tive cellurar automaton modeling earthquakes. Phys Rev Lett 68:1244–1247
Olson JE (1993) Joint pattern development: Efects of subcritical crack growth and mechanical crack
interaction. J Geophys Res 98:12251–12265
Olson JE, Pollard DD (1989) Inferring paleostress from natural fracture patterns: A new method.
Geology 17:345–348
Olson JE, Pollard DD (1991) The initiation and growth of en-echelon veins. J Struct Geol 13:595–608
Orkan N, Voight B (1985) Regional joint evolution in the Valley and Ridge Province of Pennsylvania
in relation to the Alleghany Orogeny: Guidebook for the 50th annual field conference of Pennsyl-
vania geologists. Bureau of Topography and Geological Survey, Harrisburg, Pa, pp 144-164
Ouellet M (1990) Earthquake lights and seismicity. Nature 348:492
Panek LA (1985) Estimating fracture trace length from censored measurements on multiple scanlines.
Proceedings of the International Symposium on Fundamentals of Rock Joints, Björkliden,
pp 13–23
Paris P, Erdogan F (1963) A critical analysis of crack propagation laws. J Basic Eng-T ASME 85:528–534
Park RG (1983) Foundations of structural geology. Chapman and Hall, New York, p 135
Parker JM (1942) Regional systematic jointing in slightly deformed sedimentary rock. Geol Soc Am
Bull 53:381–408
Parrot M, Achache J, Berthelier J, Blanc E (1993) High frequency electromagnetic effects. Phys Earth
Planet In 77:65–81
Pauchard L, Meunier J (1993) Instantaneous and time-lag breaking of a two-dimensional solid rod
under a bending stress. Phys Rev Lett 70:3565–3568
Pauchard L, Meunier J (1998) Experimental study of the breakage of a two-dimensional crystal. Phil
Mag B78:221–224
Peacock DCP (2001a) The temporal relationship between joints and faults. J Struct Geol 23:329–341
550 References

Peacock DCP (2001b) Are the lower planes of double seismic zones caused by serpentine dehydra-
tion in subducting oceanic mantle? Geology 29:299–302
Peacock DCP, Sanderson DJ (1991) Displacements, segment linkage and relay ramps in normal-fault
zones. J Struct Geol 13:721
Peacock DCP, Sanderson DJ (1992) Effects of layering and anisotropy on fault geometry. J Geol Soc
London 149:793–802
Peacock DCP, Sanderson DJ (1994) Geometry and development of relay ramps in normal-fault sys-
tems. AAPG Bull 78:147–165
Peacock DCP, Sanderson DJ (1995) Strike-slip relay ramps. J Struct Geol 17:1351–1360
Peng S (1970) Fracture and failure of Chelmford Granite. Ph.D. thesis, Stanford University, Stanford, USA
Peng S, Johnson AM (1972) Crack growth and faulting in cylindrical specimens of Chelmsford Gran-
ite. Int J Rock Mech Min 9:37–86
Peterson RE (1974) Stress concentration factors. John Wiley & Sons Ltd., New York
Petit J-P (1988) Normal stress dependent rupture morphology in direct shear tests on sandstone
with applications to some natural fault surface features. Int J Rock Mech Min 25:411–419
Petit J-P, Barquins M (1987) Formation de fissures en milier confine dans le PMMA (polymetachrylate
de methyle): modele analogique de formation de structures tectoniques cassantes. C R Acad Sci
Paris 305:55–60
Petit J-P, Barquins M (1988) Can natural faults propagate under mode II conditions? Tectonophysics
7:1243–1256
Petrenko VF (1993) On the nature of electrical polarization of materilas caused by cracks. Applica-
tion to ice electromagnetic radiation. Philos Mag B 67(3):301–315
Petrov VA, Gorobetz LZ (1987) Size effect of the concentration threshold of destruction. Izv-Phys
Solid Eart+ 23(1):75–77
Petukhov IM (1972) Rockbursts on coal mines. Nedra, Moscow
Petukhov IM, Lin’kov LM (1983) Mechanics of rock bursts and outbursts. Nedra, Moscow
Petukhov IM, Lin’kov LM, Sidorov VS (1976) Theory of protective stratums. Nedra, Moscow
Picu RC (1996) Singularities of an interface crack impinging on a triple grain junction. Int J Solids
Struct 33:1563–1573
Picu RC (1997) Three-dimensional stress concentration at grain triple junctions in columnar ice. Phil
Mag Lett 76:159–166
Picu RC, Gupta V (1996) Singularities at grain triple junctions in two-dimensional polycrystals with
cubic and orthorhombic grains. J Appl Mech-T ASME 63:295–300
Pohn HA (1981) Joint spacing as a method of locating faults. GLGYB 9:258–261
Poirier JP (1985) Creep of crystals. Cambridge University Press, Cambridge, UK
Politi A, Ciliberto S, Scorretti R (2002) Failure time in the fiber-bundle model with thermal noise and
disorder. Phys Rev E 66:026171–026176
Pollard DD (2000) Strain and stress: Discussion. J Struct Geol 22:1359–1367
Pollard DD, Segall P (1987) Theoretical displacements and stresses near fractures in rock: With ap-
plications to faults, joints, veins, dikes, and solution surfaces. In: Atkinson BK (ed) Fracture me-
chanics of rock. Academic Press, London, pp 277–349
Pollard DD, Segall P, Delaney PT (1982) Formation and interpretation of dilatant echelon cracks. Geol
Soc Am Bull 93:1291–1303
Pomeau Y (1992) Brisure spontanée de cristaux bidimensionnels courbés. C R Acad Sci Paris, t314,
Série II, pp 553–556
Pook LP (2000) Linear elastic fracture mechanics for engineering: Theory and applications. WIT
Press, Southampton, Boston
Preston FW (1931) The propagation of fissures in glass and other bodies with special reference to the
split-wave front. J Am Ceram Soc 14:419–427
Price NJ (1966) Fault and joint development in brittle and semi-brittle rock. Pergamon Press, New
York, pp 110–161
Price NJ (1974) The development of stress systems and fracture patterns in undeformed sediments.
Int Soc Rock Mech Cngr Proc pp 487–496
Price NJ (1977) Aspects of gravity tectonics and the development of listric faults. J Geol Soc London
133:311–327
Price NJ, Cosgrove JW (1990) Analysis of geological structures. Cambridge University Press, Cam-
bridge, England
Rabinovitch A (1994) On the Mott derivation of crack velocity. Phil Mag Lett 70:231–233
Rabinovitch A, Bahat D (1979) Catastrophe theory: A technique for crack propagation analysis.
J Appl Phys 50:321–334
Rabinovitch A, Bahat D (1999) Model of joint spacing distribution based on shadow compliance.
J Geophys Res 104:4877–4886
 References 551

Rabinovitch A, Belizovsky G, Bahat D (1993) On the direction of secondary cracks in brittle fracture.
Int J Fracture 63:R25–R30
Rabinovitch A, Belizovsky G, Bahat D (1994) Length distribution of secondary cracks in brittle mate-
rials. Phil Mag Lett 70:277–288
Rabinovitch A, Bahat D, Frid V (1995) Comparison of electromagnetic radiation and acoustic emis-
sion in granite fracturing. Int J Fracture 71(2):R33–R41
Rabinovitch A, Bahat D, Frid V (1996) Emission of electromagnetic radiation by rock fracturing.
Z Geol Wissenschaft 24(3–4):361–368
Rabinovitch A, Frid V, Bahat D (1998) Parametrization of electromagnetic radiation pulses obtained
by triaxial fracture in granite samples. Phil Mag Lett 77(5):289–293
Rabinovitch A, Bahat D, Melamed Z (1999a) A note on joint spacing. Rock Mech Rock Eng 32:71–75
Rabinovitch A, Frid V, Bahat D (1999b) A note on the amplitude – frequency relation of electromag-
netic radiation pulses induced by material failure. Phil Mag Lett 79:195–200
Rabinovitch A, Belizovsky G, Bahat D (2000a) Origin of mist and hackle patterns in brittle fracture,
Phys Rev B 61:14968–14974
Rabinovitch A, Zlotnikov R, Bahat D (2000b) Flaw length distribution measurement in brittle materi-
als. J Appl Phys 87:7720–7725
Rabinovitch A, Frid V, Bahat D, Goldbaum J (2000c) Fracture area calculation from electromagnetic
radiation and its use in chalk failure analysis. Int J Rock Mech Min 37:1149–1154
Rabinovitch A, Bahat D, Frid V (2002a) Similarity and dissimilarity of electromagnetic radiation from
carbonate rocks under compression, drilling and blasting. Int J Rock Mech Min 39(1):125–129
Rabinovitch A, Frid V, Bahat D (2002b) Gutenberg-Richter type relation for laboratory fracture in-
duced electromagnetic radiation. Physical Review E 65:011401–011404
Rabinovitch A, Frid V, Bahat D, Goldbaum J (2003a) Decay mechanism of the fracture induced elec-
tromagnetic pulses. J Appl Phys 93(9):5085–5090
Rabinovitch A, Frid V, Goldbaum J, Bahat D (2003b) Polarization-depolarization process in glass dur-
ing percussion drilling. Phil Mag 83(25):2929–2940
Ramsay JG (1980) Crack-seal mechanism of rock deformation. Nature 284(5752):135–139
Ramsay JG, Huber MI (1987) The techniques of modern structural geology, vol 2: Folds and frac-
tures. Academic Press, London
Ramsay JG, Lisle R (2000) The techniques of modern structural geology, vol 3: Applications of con-
tinuum mechanics in structural geology. Academic Press, New York
Rasband W (2003) Image J 1.30V. National Institutes of Health, USA
Ravi-Chandar K (1998) Dynamic fracture of nominally brittle materials. Int J Fracture 90:83–102
Ravi-Chandar K, Knauss WG (1984a) An experimental investigation into dynamic fracture – I. Crack
initiation and crack arrest. Int J Fracture 25:247–262
Ravi-Chandar K, Knauss WG (1984b) An experimental investigation into dynamic fracture –
II. Microstructural aspects. Int J Fracture 26:65–80
Ravi-Chandar K, Knauss WG (1984c) An experimental investigation into dynamic fracture – III. Steady
state crack propagation and crack branching. Int J Fracture 26:141–154
Ravi-Chandar K, Knauss WG (1984d) An experimental investigation into dynamic fracture – IV. On
the interaction of stress waves with propagating cracks. Int J Fracture 26:189–200
Ravi-Chandar K, Knauss WG (1987) On the acceleration of the transient stress field near the tip of a
crack. J Appl Mech-T ASME 54:72–78
Ravi-Chandar K, Knauss W (1999) Processes controlling fast fracture of brittle solids. Comput Sci
Eng 1999 Sept/Oct:24–31
Ravishankar SR, Murthy CRL (2000) Characteristics of AE signals obtained during drilling compos-
ite laminates. NDT&E International 33:341–348
Rawnsley KD, Rives T, Petit JP, Hencher SR, Lumsden AC (1992) Joint development in perturbed stress
fields near faults. J Struct Geol 14:939–951
Rawnsley KD, Peacock DCP, Rives T, Petit JP (1998) Joints in the Mesozoic sediments around the
Bristol Channel Basin. J Struct Geol 20:1641–1661
Reches Z (1976) Analysis of joints in 2 monoclines in Israel. Geol Soc Am Bull 87:1654–1662
Reches Z, Lockner DA (1994) Nucleation and growth of faults in brittle rocks. J Geophys Res 99:
18159–18173
Red’kin V, Kuprijanov AS, Bufalov VV (1985) Geophysical devices for distance rock burst control. In:
Smirnov V (ed) Geophysical methods of stress and deformation control. Novosibirsk, pp 81–82
(in Russian)
Regel’ VR, Slutsker AI, Tomashevskii EE (1972) The kinetic nature of the strength solids. Sov Phys
Uspekhi 1:45–65
Rehrig WA, Heidrick TL (1972) Regional fracturing in Laramide stocks of Arizona and its relation-
ship to porphyry copper mineralization. Econ Geol 67:198–213
552 References

Rehrig WA, Heidrick TL (1976) Regional tectonic stress during the Laramide and late Tertiary intru-
sive periods, Basin and Range Province, Arizona. Arizona Geological Digest 10:205–228
Renshaw CE, Schulson EM (2001) Universal behavior in compressive failure of brittle materials. Na-
ture 412:897–900
Rice RW (1974) Fracture topography of ceramics. In: Frechette VD, La Course WC, Burdick VL
(eds) Surfaces and interfaces of glass and ceramics. Plenum Publishing Corpration, New York,
pp 439–472
Rice JR (1979) Theory of precursory processes in the inception of earthquake rupture. Gerl Beitr
Geophys 88:91–127
Rice RW (1984) Ceramic fracture features, observations, mechanisms and uses. In: Mecholsky JJ, Powell
SR (eds) Fractography of ceramic and metal failures. ASTM, Philadelphia, pp 1–51
Richter HG, Kerkhof F (1994) Stress wave fractography. In: Bradt RC, Tressler RE (eds) Fractography
of glass. Olenum Press, New York pp 75–109
Rikitake TJ (1997) Nature of electromagnetic radiation precursory to an earthquake. J Geomagn
Geoelectr 49:1153–1163
Rives T, Razack M, Petit JP, Rawnsley KD (1992) Joint spacing: Analogue and numerical simulations.
J Struct Geol 14:925–937
Roberts JC (1974) Jointing and minor tectonics of the vale of Glamorgan between Ogmore-by-Sea
and Lavernock Point, South Wales. Geol J 9:97–114
Roberts JC (1979) Jointing and minor tectonics of the South Gower Peninsula between Mumbles
Head and Rhossilli Bay, South Wales. Geol J 14:1–14
Roberts DG (1989) Basin inversion in and around the British Isles. In: Cooper MA, Williams GD (eds)
Inversion tectonics. Spec Publ Geol Soc London 44:131–150
Roberts JC (1995) Fracture surface markings in Liassic limestone at Lavernock Point, South Wales In:
Ameen MS (ed) Fractography: Fracture topography as a tool in fracture mechanics and stress
analysis. Geological Society (London) Special Publication 92:175–186
Robinson ES (1970) Mechanical disintegration of the Navajo Sandstone in Zion Canyon, Utah. Geol
Soc Am Bull 81:2799–2806
Rodgers J (1970) The tectonics of the Appalachians. Wiley Interscience, New York
Rophe B, Berkowitz B, Magaritz M, Ronen D (1992) Analysis of subsurface flow and anisotropy in a
fractured aquitard using transient water level data. Water Resour Res 28:199–207
Rossmanith HP, Knasmillner RE, Daehnke A, Mishnaevsky L (1996) Wave propagation, damage evo-
lution and dynamic fracture extension. 1. Percussion drilling. Mater Sci+ 32:350–358
Rossmanith HP, Daehnke A, Knasmillner RE, Kouzniak N, Ohtsu M, Uenishi K (1997) Fracture me-
chanics applications to drilling and blasting. Fatigue Fract Eng M 20:1617–1636
Roy A, Saha TK (1995) Interaction of a penny-shaped crack with an elliptic crack. Int J Fracture 73:51–65
Rudnicki JW, Olsson WA (1998) Reexamination of fault angles predicted by shear localization theory.
Int J Rock Mech Min 35: Paper No 88
Ryan MP, Sammis CG (1978) Cyclic fracture mechanisms in cooling basalt. Geol Soc Am Bull 89:
1295–1308
Rzhevskii VV, Novik GY (1978) Foundations of rock physics. Nedra, Moscow
Sadovskii MA, Sobolev GA, Migunov NI (1979) Changes of natural radiowaves radiation during
Karpatian earthquake. Dokl Akad Nauk SSSR+ 244(2):316–319
Sakai H, Oda H, Nakayama T, Doi H (1992) Electromagnetic changes at explosion seismic experi-
ment. J Phys Earth 40:447–458
Salameh E, Zacher W (1982) Horizontal stylolites and paleostress in Jordan. Neues Jahrb Geol P M
8:509–512
Sbar ML, Sykes LR (1973) Contemporary compressive stress and seismicity in eastern North America:
An example of intra-plate tectonics. Geol Soc Am Bull 84:1861–1882
Scharbert S (1998) Some geochronological data from the South Bohemian Pluton in Austria: A criti-
cal review. Acta Universitatis Carolinae Geologica Praha 42:114–118
Schiff LI (1986) Quantum mechanics. McGraw Hills, London
Schoeck G, Pichl W (1990) Bond trapping of cracks. Phys Stat Solidi 118:109–115
Scholz CH (1990) The mechanics of earthquakes and faulting. Cambridge, Cambridge University Press
Scholz CH (1998) Earthquakes and friction laws. Nature 391:37–42
Schwartz DP, Sibson RH (1989) Fault segmentation and controls of rupture initiation and termina-
tion. U.S. Geological Survey Open File Report 89–13
Scitovich VP, Lazarevich LM (1985) Estimation of stress condition of rock massive by EMR. In: Smirnov
V (ed) Geophysical methods of stress and deformation control. Novosibirsk, pp 65–66 (in Russian)
Scorretti R, Ciliberto S, Guarino A (2001) Disorder enhances the effects of thermal noise in the fiber
bundle model. Europhys Lett 55:626–632
Scott DR (1996) Seismicity and stress rotation in a granular model of the brittle crust. Nature 381:592–595
 References 553

Scott WH, Hansen EC, Twiss RJ (1965) Stress analysis of quartz deformation lamellae in a minor fold.
Am J Sci 263:729–746
Secor DT Jr (1965) Role of fluid pressure in jointing. Am J Sci 263:633–646
Secor DT Jr (1969) Mechanics of natural extension fracturing at depth in the Earth’s crust. In: Re-
search in tectonics. Geol Surv Canada Paper 3–47
Segall P, Pollard DD (1983) Joint formation in granitic rock of the Sierra Nevada. Geol Soc Am Bull
94:563–575
Segall P, McKee EH, Martel SJ, Turrin BD (1990) Late Cretaceous age of fractures in the Sierra Nevada
Batholith, California. Geology 18:1248–1251
Shainin VE (1950) Conjugate sets of en echelon tension fractures in the Athens limestone at Riverton,
Virginia. Geol Soc Am Bull 61:509–517
Shand Shand EB (1954) Experimental study of fracture of glass: 2 Experimental data. J Am Ceram
Soc 37:559–572
Shand EB (1959) Breaking stress of glass determined from dimensions of fracture mirrors. J Am
Ceram Soc 42:474–477
Sharon E, Finberg J (1996) Microbranching instability and the dynamic fracture of brittle materials.
Phys Rev B 54:7128–7139
Sharon E, Fineberg J (1998) Universal features of the micro-branching instability in dynamic frac-
ture. Phil Mag B78:243–251
Sharon E, Fineberg J (1999) Confirming the continuum theory of dynamic brittle fracture for fast
cracks. Nature 397:333–335
Sharon E, Gross SP, Fineberg J (1995) Local crack branching as a mechanism for instability in dy-
namic fracture. Phys Rev Lett 74:5096–5099
Sheldon P (1912) Some observations and experiments on joint planes. J Geol 20:164–183
Shen B, Stephensson O, Einstein HH, Ghahreman B (1995) Coalescence of fractures under shear stresses
in experiments. J Geophys Res 100:5975–5990
Sibson RH (1985) Stopping of earthquake ruptures at dilatational jogs. Nature 316:248–251
Simmons G, Todd T, Baldridge WS (1975) Toward a quantitative relationship between elastic proper-
ties and cracks in low porosity rocks. Am J Sci 275:318
Smekal A (1953) Zum Bruchvorgang bei sprödem Stoffverhalten unter ein- und mehrachsigen Bean-
spruchungen. Oesterr Ing-Arch 7:49–70
Smith RL (1982) The asymptotic-distribution of the strength of a series-parallel system with equal
load-sharing. J Appl Probab 10:137–171
Smith RL, Phoenix SL (1981) Asymptotic distributions for the failure of fibrous materials under se-
ries-parallel structure and equal load-sharing. J Appl Mech-T ASME 48:75–82
Sneddon IN (1951) Fourier transforms. McGraw Hill Company Inc., New York, p 542
Sobolev GA, Semerchan AA, Salov BG (1982) Precursors of the destruction of large rock sample. Izv-
Phys Solid Eart+ 18(8):572–580
Sommer E (1967) Das Bruchverhalten von Rundstäben aus Glas im Manteldruckversuch mit
Überlagerter Zugspannung. Glastech Ber-Glass 40:304–307
Sommer E (1969) Formation of fracture “lances” in glass. Eng Fract Mech 1:539–546
Sornette D, Knopoff L, Kagan YY, Vanneste C (1996) Rank-ordering statistics of extreme events: Ap-
plication to the distribution of large earthquakes. J Geophys Res 101:13883–13893
Spiegel MR (1961) Statistics. McGraw-Hill Book Company, N.Y., 359 pp
Spiker EC, Gray M (1997) A study of progressive deformation at the Allegheny Front, Lycoming County,
Pennsylvania. Geological Society of America Abstracts with Programs, vol 29, no 1
Sprunt ES, Brace WF (1974) Direct observations of microcavities in crystalline rocks. Int J Rock Mech
Min 11:139–150
Spyropoulos C, Scholz CH, Shaw BE (2002) Transition regimes for growing crack populations. Phys
Rev E 65:0561051–05610510
Srivastava GP (1990) The physics of phonons. Adam Hilger, Bristol
Stearns DW (1968) Certain aspects of fracture in naturally deformed rocks. In: Riecker RE (ed) Rock
mechanics seminar, vol 1. Special rept., Terrestrial Sciences Lab Air Force, Cambridge Research
Labs, Bedford, Mass, pp 27–116
Stearns DW, Friedman M (1972) Reservoirs in fractured rock. In: King RE (ed) Stratigraphic oil and
gas fields. AAPG M Em 16:82–106
Stearns DW (1968) Certain aspects of fracture in naturally deformed rocks. In: Riecker RE (ed) Rock
mechanics seminar, vol. 1, special rept. Terrestrial Sciences Lab Air Force, Cambridge Research
Labs, Bedford, Mass, pp 27–116
Stern TW, Bateman PC, Morgan BA, Newell MF, Peck DL (1981) Isotopic U-Pb ages of zircon from
the granitoids of the central Sierra Nevada, California. U.S. Geological Survey Professional Paper
17 pp
554 References

Stewart ME, Taylor WJ (1996) Structural analysis and fault segment boundary identification along
the Hurricane fault in southwestern Utah. J Struct Geol 18:1017–1029
Suess FE (1926) Intrusionstektonik und Wandertektonik im variszischen Grundgebirge. Gebrüder
Borntraeger, Leipzig, 268 pp
Suk M (1984) Geological history of the territory of Czech Socialist Republic. Geological Survey, Prague
Suppe J (1985) Principles of structural geology. Prentice-Hall, Englewood Cliffs, NJ
Suresh S (1998) Fatigue of materials, 2nd edn. Cambridge University Press, Cambridge
Swaby PA, Rawnsley KD (1997) An interative #D fracture modeling environment. Soc Petroleum Eng
9:88–92
Swallowe GM, Field JE, Horn LA (1986) Measurements of transient high-temperatures during the
deformation of polymers. J Mater Sci 21:4089–4096
Takayasu H (1989) Fractals in the physical sciences. John Wiley & Sons Ltd., New York
Tetelman AS, McEvly AJ (1967) Fracture of structural materials. John Willey & Sons Ltd., New York
Thomas R, Klemm W (1997) Microthermometric study of silicate melt inclusions in Variscan gran-
ites from SE Germany. J Petrology 38:1753–1765
Thomas AL, Pollard DD (1993) The geometry of echelon fractures in rock: Implications from labora-
tory and numerical experiments. J Struct Geol 15:323–334
Timoshenko SP, Goodier JN (1951) Theory of elasticity. McGraw-Hill, New York
Timoshenko SP, Goodier JN (1984) Theory of elasticity, 3rd edn. McGraw-Hill Int Book Co, Auckland
Tobisch OT, Paterson SR, Saleeby JB, Geary EE (1989) Nature and timing of deformation in the Foothills
terrane, central Sierra Nevada, California: Its bearing on orogenesis. Geol Soc Am Bull 101:401–413
Tokarski AK, Zuchiewicz W, Swierczewska A (1999) The influence of early joints on structural devel-
opment of thrust-and-fold belts: A case study from the Outer Carpathians (Poland). Geol Carpath
50:178–180
Tomizawa I, Yamada I (1995) Generation mechanism of electric impulses observed in explosion seis-
mic experiments. J Geomagn Geoelectr 47:313–324
Tomizawa I, Hakayawa M, Yoshino T (1994) Observation of ELF/VLF electromagnetic variations as-
sociated with seismic experimental explosion. In: Hayakawa M, Fujinawa Y (eds) Electromag-
netic phenomena related to earthquake prediction. Terra Scientific Publishing Company, Tokyo,
pp 337–347
Trudgill B, Cartwright J (1994) Relay-ramp forms and normal-fault linkages, Canyonlands National
Park. Utah Bull Geol Soc Am 106:1143–1157
Trzebski R (1997) Morphogenesis, tectonic setting and intrusion dynamics of the late-Variscan gran-
ites at the northwest-margin of the Bohemian Massif. Göttinger Arb Geol Paläont 69:1–66
Tsang YW (1984) The effect of tortuosity on fluid-flow through a single fracture. Water Resour Res
20:1209–1215
Tschegg EK (1983) Mode 3 and mode 1 fatigue crack propagation behaviour under tortional loading.
J Mater Sci 18:1604–1614
Turcotte DL (1992) Fractals and chaos in geology and geophysics. Cambridge University Press, Cambridge
Turcotte DL, Newman WI, Shcherbakov R (2003) Micro and microscopic models of rock fracture.
Geophys J Int 152:718–728
Tvergaard V, Hutchinson JW (1988) Microcracking in ceramics induced by thermal expansion or
elastic anisotropy. J Am Ceram Soc 71:157–166
Twidale CR, Vidal Romani JR, Campbell EM, Centeno JD (1996) Sheet fractures: Response to ero-
sional offloading or to tectonic stress? Z Geomorph 106:1–24
Twiss RJ, Moores EM (1992) Structural geology. Freeman & Company, New York
Umeda Y (1990) High-amplitude seismic-waves radiated from the bright spot of an earthquake.
Tectonophysics 175:81–92
Urusovskaja AA (1969) Electric effects associated with plastic deformation of ionic crystals. Sov Phys
Uspekhi 11:631–643
Van Hise CR (1896) Principles of North-American Pre-Cambrian geology: U.S. geology. U.S. Geologi-
cal Survey 16th Annual Report, pp 581–874
Van Tyne A (1983) Natural gas potential of the Devonian black shales of New York. NoE Geol 5:209–216
Varotsos P, Sarlis, N, Lazaridou M (1999) Interconnection of defect parameters and stress-induced
electric signals in ionic crystals. Phys Rev B 59:24–27
Ver Steeg K (1942) Jointing in the coal beds of Ohio. Econ Geol 37:530–509
Vozoff K, Frid V (2001) Electromagnetic emissions monitoring to warn of wind blasts and gas outs.
Results of ACARP Project at Moonee Colliery: C9005. In: Symposium on Geological Hazards,
Rafferety’s Resort Newcastle, 15–16 November 2001
Wagner GA, Van Den Haute P (1992) Fission track dating. Enke-Verlag, Stuttgart
Walmann T, Malthe-Sorenssen A, Feder J, Jossang T, Meakin P, Hardy HH (1996) Scaling relations for
the lengths and widths of fractures. Phys Rev Lett 77:5393–5396
 References 555

Walsh JJ, Watterson J (1988) Analysis of the relationship between displacements and dimensions of
faults. J Struct Geol 10:239–247
Warwick JW, Stoker C, Meyer TR (1982) Radio emission associated with rock failure: Possible appli-
cation to the great Chilean earthquake of May 22, 1960. J Geophys Res 87(b4):2851–2859
Watznauer A (1954) Die erzgebirgischen Granitintrusionen. Geologie 3:688–706
Wawrzynek PA, Ingraffea AR (1987) Interactive finite element analysis of fracture processes: An inte-
grated approach. Theor Appl Fract Mec 8:137–150
Wedel AA (1932) Geological structures of the Devonian strata of southcentral New York, New York
State. Mus Bull 294:74
Weinberger R (1999) Initiation and growth of cracks during desiccation of stratified muddy sedi-
ments. J Struct Geol 21:379–386
Weinberger R (2001a) Joint nucleations in layered rocks with non-uniform distribution of cavities.
J Struct Geol 23:1241–1254
Weinberger R (2001b) Evolution of polygonal patterns in stratified mud during desiccation: The role
of flaw distribution and layer boundaries. Geol Soc Am Bull 113:20–31
Weisbrod N, Nativ R, Ronen D, Adar E (1998) On the variability of fracture surfaces in unsaturated
chalk. Water Resour Res 34:1881–1887
Weisbrod N, Nativ R, Adar E, Ronen D (1999) Impact of intermittent rainwater and wastewater flow
on coated and uncoated fractures in chalk. Water Resour Res 35:3211–3222
Weisbrod N, Nativ R, Adar E, Ronen D (2000a) Salt accumulation and flushing in unsaturated frac-
tures in an arid invironment. Ground Water 38:452–461
Weisbrod N, Nativ R, Adar E, Ronen D, Ben-Nun A (2000b) Impact of coating and weathering on the
properties of chalk fracture. Geophys Res Lett 105:27853–27864
Weisbrod N, Dahan O, Adar EM (2002) Particle transport in unsaturated fractured chalk under arid
conditions. J Contam Hydrol 56:117–136
Westergaard HM (1939) Bearing pressure and cracks. Trans Am Soc Mech Eng 61A:49–53
White WS (1946) Rock-bursts in the granite quarries at Bare, Vermont. U.S. Geological Survey, Circular 13
Whittaker A, Green GW (1983) Geology of the country around Weston-super-Mare. Memoir of the
Geological Survey of Great Britain. Sheet 279 and parts of 263 and 295
Wibberley CAJ, Petit JP, Rives T (2000) Micromechanics of shear rupture and the control of normal
stress. J Struct Geol 22:411–427
Widom B (1966) Random sequential addition of hard spheres to a volume. J Chem Phys 44:3888
Wiederhorn SM (1967) Influence of water vapor on crack propagation in soda-lime glass. J Am Ceram
Soc 50:407–414
Wiederhorn SM (1978) Mechanisms of subcritical crack growth in glass. In: Bradt RC, Hasselman
DPH, Lange FF (eds) Fracture mechanics of ceramics, vol 4. Plenum, New York, pp 549–580
Wiederhorn SM, Bolz LH (1970) Stress corrosion and static fatigue of glass. J Am Ceram Soc 53:543–548
Wiederhorn SM, Johnson H (1973) Effect of electrolyte PH on crack propagation in glass. J Am Ceram
Soc 56:192–197
Wiederhorn SM, Johnson H, Diness AM, Heuer AH (1974) Fracture of glass in vacuum. J Am Ceram
Soc 57:336–341
Wiederhorn SM, Dretzke A, Rodel J (2002) Crack growth in soda-lime-silicate glass near the static
fatigue limit. J Am Ceram Soc 85:2287–2292
Wilkins SJ, Michael RG, Wacker M, Eyal Y, Engelder T (2001) Faulted joints: Kinematics, displace-
ment-length scaling relations and criteria for their identification. J Struct Geol 23:315–327
Willemse EJM, Pollard DD (1998) On the orientation and patterns of wing cracks and solution sur-
faces at the tips of a sliding flaw or fault. J Geophys Res 103B:2427–2438
Williams ML (1957) On the stress distribution at the base of a stationary crack. J Appl Mech-T ASME
24:109–114
Wise DU, McCrory TA (1982) A new method of fracture analysis: Azimuth versus traverse distance
plots. Geol Soc Am Bull 93:889–897
Witherspoon PA, Newman SP (1967) Evaluating a slightly permeable caprock in aquifer gas storage.
I. Caprock of infinite thickness. J Petrol Technol 19:949
Wong RHC, Chau KT (1998) Peak strength of replicated and real rocks containing cracks. Key Eng
Mat 145–149:953–958
Wong RHC, Chau KT, Wang P (1996) Microcracking and grain size effect in Yuen Kong marbles. Int
J Rock Mech Min 33:479–485
Woodworth JB (1895) Some features of joints. Science 2:903–904
Woodworth JB (1896) On the fracture system of joints, with remarks on certain great fractures. Bos-
ton Society Natural History Proceedings 27:63–184
Wu H, Pollard DD (1992) Propagation of a set of opening-mode fractures in layered brittle materials
under uniaxial strain cycles. J Geophys Res 97:3381–3396
556 References

Wu H, Pollard DD (1995) An experimental study of relationship between joint spacing and layer
thickness. J Struct Geol 17:887–905
Wyss M (2001) Why is earthquake prediction research not progressing faster? Tectonophysics 338:
217–223
Xu XP, Needleman A (1994) Numerical simulations of fast crack growth in brittle solids. J Mech Phys
Solids 42:1397–1434
Yamada I, Masuda K, Mizutani H (1989) Electromagnetic and acoustic emission associated with rock
fracture. Phys Earth Planet In 57:157–168
Yang B, Ravi-Chandar K (2001) Crack path instabilities in a quenched glass plate. J Mech Phys Solids
49:91–130
Yatomi C (1981) The effect of surface-energy on temperature rise around a fast running crack. Eng
Fract Mech 14:759–762
Yepez E, Angulobrown F, Peralta J (1995) Electric field pastterns as seismic precursors. Geophys Res
Lett 22(22):3087–3090
Yoffe EH (1951) The moving Griffith crack. Phil Mag 42:739–750
Yokobori T, Uozomi M, Ichikawa M (1971) Interaction between non-coplanar parallel straggered elastic
cracks. Rep Res Inst Strength and Fract of Mater 7:25–47
Yoon KJ, Wiederhorn SM, Luecke WE (2000) Comparison of tensile and compressive creep behavior
in silicon nitride. J Am Ceram Soc 85:2017–2022
Yoshino T, Tomizawa I (1989) Observation of low-frequency electromagnetic emissions as precur-
sors to the volcanic eruption at Mt. Mihara during November, 1986. Phys Earth Planet In 57:32–39
Yoshino T, Tomizawa I, Sugimoto T (1993) Results of statistical analysis of low-frequency seismogenic
EM emissions as precursors to earthquakes and volcanic eruptions. Phys Earth Planet In 77:21–31
Younes AL (1996) Tectonic implications from joints and fold history in the Appalachian Plateau, cen-
tral New York. Geological Society of America Abstracts with Programs, vol 28, no 3, A-111
Younes AI, Engelder T (1995) Paleostress rotation as indicated by joints within the Appalachian Pla-
teau: Clockwise, counterclockwise, or both? American Geological Society, Abstracts of the Annual
Meeting, Session 135
Younes AI, Engelder T (1999) Fringe cracks: Key structures for the interpretation of the progressive
Alleghanian deformation of the Appalachian Plateau. Geol Soc Am Bull 111:219–239
Yukawa S, Timo DP, Rubio A (1969) Fracture design practices for rotating equipment. In: Liebowitz H
(ed) Fracture. Academic Press, New York, pp 65–157
Yuse A, Sano M (1993) Transition between crack patterns in quenched glass plates. Nature 362:329–331
Yuse A, Sano M (1997) Instabilities of quasi-static crack patterns in quenched glass plates. Physica D
108:365–378
Zawislanski P (1994) Surface fracture distribution at Raymond field site, Raymond, California,
Lawrence Berkeley National Laboratory Report, LBNL-42675
Zhao M, Jacobi RD (1997) Formation of regional cross-fold joints in the northern Appalachian Pla-
teau. J Struct Geol 19:817–834
Zhao YX, Gao Q, Wang JN (2000) The evolution of short fatigue crack lengths and crack density: Two
approaches. Fatigue Fract Eng M 23:929–941
Zhurkov SN (1965) Kinetic concept of the strength of solids. Int J Fracture Mech 1:311–323
Zoback ML (1989) Global patterns of tectonic stress. Nature 34:291–298
Zoback ML (1992) First- and second-order patterns of stress in the lithosphere: The world stress
map project. J Geophys Res 97:11703–11728
Zoback MD, Healy JH (1984) Friction, faulting, and in situ stress. Ann Geophys 2:689–698
Zurkov SN, Kuksenko VS, Slutsker AI (1969) Formation of submicroscopic cracks in polimers under
load. Sov Phys Sol-State 11(2):238–245
Subject Index

A –, side-walls 239, 244, 245, 346

–, symmetric 347
acoustic emission 26, 28, 29, 380, 407, 410, 434, arrest marks 91, 97, 98, 126–130, 144, 147,
435, 440 162–165, 239, 240, 245, 249, 261, 270, 296, 298,
AE 28, 380, 407, 408, 411, 436 299, 305, 316, 359, 360–362
age 178, 180, 195, 205, 206, 214, 224, 226, 236, 258, Arrhenius 14, 15, 17, 25
266, 267, 276, 278, 279, 287, 288, 294, 295, 319, ASTM 110, 145
320, 322, 324, 325, 352, 494, 429, 468, 469, 479, Austria 319
504, 511 avalanche 30–32, 487
Airy function 36 axis 26, 47, 48, 74, 91, 92, 102, 103, 108, 109, 111,
Alleghanian 195, 197, 198, 201–207, 279, 280 115–117, 125, 137, 143, 165, 166, 178, 179, 192,
Alpine 213–215, 217, 220, 222, 226, 227, 229–231, 208, 233, 245, 251, 265, 266, 272–274, 285, 288,
278–280, 469, 471, 472 292, 296, 298, 316, 317, 336, 349, 415, 418, 479,
amorphous 58, 114, 408 483, 526
amplitude 113, 195, 232, 233, 365, 382, 383, 385,
388–391, 398, 403, 405, 407–409, 412, 414, 423, B
430–432, 435–438, 441, 445, 450, 451, 458
anastomotic 180, 265, 461 barrier 14–18, 21–23, 516, 519
Anatolian microplate 469, 472 basement 287, 468
angle basin
–, critical 175 –, Ebro 177, 179
–, dihedral 107, 108, 166, 170, 171, 173–178 –, of attraction 23
–, high 217, 220, 222, 227, 277 –, range 235, 236, 322
–, kink 150–152, 156 batholith 285, 286, 295, 322–327, 329, 331–354
–, low 173, 233, 260, 277 bedding
–, of tilt 141, 156, 157, 160, 202, 313 –, contacts 124
–, of twist 110, 111, 116, 135–137, 156, 157, –, interface 90, 91, 95
160, 313, 368 belt 105, 194, 202, 208, 294, 469, 471, 479, 480
anhydrite 90 bend 100, 125, 188, 213, 227, 250, 336, 337, 343,
anisotropy 11, 13 347, 362, 494, 497
anticline 191, 214, 217–233, 271, 278, 280, 281, 483 Benioff 437, 458
antiform 276, 294 bifurcation 58–60, 113, 114, 167, 183, 189, 190,
antiplane (mode III) 9, 34 358, 371
Appalachian Plateau 84, 150–158, 160, 177, 178, bi-lateral 84, 88, 109, 132, 144, 146, 148, 298, 319,
191–209, 279–281, 359, 460, 468, 480, 483 462
aquifer 518 biotite 189, 295, 316, 327–331, 387
Arabian blasting 379, 380, 405, 420–432, 445
–, plate 472, 479, 481 Blue Lias 213, 220, 222, 229, 231
–, platform 177, 179, 472 Boltzmann 15, 17
arch bond
–, asymmetric 244, 347 –, atomic 382, 388
–, back-wall 242, 244, 245, 247 –, breaking 407, 458
–, ceiling 244–247, 249, 338–340, 346, 347 –, severed 382, 388, 407, 458
–, full 244, 246, 334, 338–340 borehole 314, 432, 439, 440, 445, 447, 448, 452,
–, National Park 456, 457, 520, 523
–, partial 241, 243, 244, 246, 334, 336, boudinage 296
338–340, 347 Bouguer gravity field 287, 292
558 Subject Index

boundary –, Europe-African-Arabian plate 469, 472,

–, diffraction 129 473, 479, 480
–, zigzag 299, 305 colloidal 522, 523
box, semi- 128–130, 305 Colorado Plateau 234–236, 279
brackish 518 compatibility equation 5, 35, 36
branch, micro- 365, 366 composite 26, 84, 105, 123, 135, 144, 147, 175, 177,
branching 55, 56, 59, 82, 98, 99, 118–122, 125, 167, 230, 239, 243, 245, 246, 250, 261, 270, 277, 307,
183, 189, 190, 327–332, 345, 360–366, 374, 495, 311, 312, 317, 341, 343, 344
497, 502 compression
Brazil, Chapada dos Guimaraes 249 –, Alpine 213–215, 231, 280
breakdown –, biaxial 104, 185, 186, 247
–, continuous 132, 134, 135, 142, 148, 150 –, stiff 380
–, discontinuous 132–135, 142, 147, 148, 160, –, triaxial 104, 105, 107, 387, 407, 418, 458
270 –, uniaxial 104, 108, 114, 115, 168, 170, 185, 246,
–, incipient zone 107 387, 391, 396, 398, 404, 410–412, 414, 417, 503
–, parabolic 111 concave 104, 133, 243–245, 250, 338, 340, 498, 499
Bristol Channel 192, 210, 211, 213–215, 217, 219, cone, quasi- 102
221–233, 278–281, 463 confidence 93, 145, 170, 217, 369, 370, 374
bromide 519 contemporaneous 170, 217, 480, 494, 501
buckling 117, 184, 228, 248, 266, 274, 343–345, 363 cooling 162, 234, 283, 286, 296, 316, 320–322, 324,
bulging 336, 337, 487, 500, 503 325, 349–352, 354, 355, 363
coplanar 121, 131, 183, 239, 344
C core 163, 208, 246, 288, 294, 296, 361, 378, 519–523
country rock 214, 284, 294, 295, 341, 354, 499
Cadomian 189, 288, 292 Cox-Hobbs 63, 67, 70–72, 74, 76, 78
calcite 106, 167, 194, 213–215, 222, 233, 483, 521 CPP (car parking process) 64, 65, 71
carbonate 87, 189, 220, 229, 474, 476, 521 crack
Carboniferous 292 –, array 85
cataclastic 168, 495–501, 503 –, closure 149, 366
caustics 118 –, dynamic 1, 8, 45, 46, 50, 53, 112
cavity –, early circular 341, 345
–, critical 91, 93, 94 –, elliptical 41, 162, 239, 245
–, eccentric 92 –, flaky 394
–, rim 92 –, heat 403
–, sizes 93, 94 –, initiation 8, 55, 58, 83, 86, 87
–, smallest 93 –, length 13, 15, 19, 412, 43–45, 49, 53, 54, 60,
ceramic 10, 58, 77, 83, 84, 86, 108, 109, 125–127, 62, 105, 118, 119, 136, 151, 331, 371, 374,
129, 130, 140, 173, 174, 176, 357, 387–389, 392, 382, 392–395, 403, 412, 414, 450, 486, 488,
394, 396, 398, 400, 401, 403, 404 490–492, 499
chalk 84, 85, 100–102, 105–108, 117, 121, 134, –, mud 87, 89, 157
139–142, 144, 145, 150, 158, 159, 171–176, 180, –, oscillating 99, 100
181, 191, 231–233, 251, 253–255, 257, 259, 261, –, penny shaped 42, 43, 84
263–280, 345, 366, 377, 387–391, 395–404, 407, –, primary 48, 59, 60, 100, 120, 365
419, 431, 435–437, 461, 462, 465, 473, 478, 479, –, quasi-circular 239, 241
484, 485, 492, 494, 501, 505–508, 510–512, –, resistance 54
514–524 –, restarting 98
charge 404–408, 411, 422, 425, 427, 458 –, secondary 48, 53, 59, 120, 160, 183, 184,
chert 141, 158, 159, 181, 251, 254, 468, 505, 506 337, 365
chilled margin 325 –, splay 183
chloride 519 –, tip 2, 8, 32, 33, 46–50, 52, 53, 56, 57, 118,
circular 7, 60, 85, 94–96, 120, 143, 160, 162, 163, 137, 138, 184, 336, 357, 385, 406
165, 171, 173, 234, 238, 239, 241–247, 250, 296, –, velocity 43, 48, 53–59, 62, 63, 99, 113, 118,
298, 299, 305, 318, 340, 341, 345, 359, 361, 366, 356–358, 362, 366, 382, 402, 427, 450
492 –, width 118, 136, 383, 385, 388, 389, 395,
classification 81, 83, 158, 175, 254, 279, 280, 285, 396, 400, 402, 408, 451
325, 350, 355, 357, 380, 413, 460, 463, 467, 504 –, wing 103, 104, 109, 167, 168, 183–186, 500,
clay 87, 520 503
cluster 32, 94, 199, 201, 210, 221, 231, 381, 405, Cranz-Shardin 118
473 creep
coalescence 17, 26, 85, 105, 343, 344, 399, 414, 431 –, primary 77
collision –, secondary 77
–, continent 273, 295, 483 –, tertiary 77
 Subject Index 559

Cretaceous 90, 140, 232, 273, 321, 350, 468–473, displacement 3–5, 9, 33–38, 46, 52, 104, 106, 107,
479–481 151, 152, 167, 202, 213, 214, 223, 229, 286, 322,
criterion 325, 328, 330, 331, 384, 385, 427, 468, 494, 496,
–, EMR 446, 451, 454, 457 497, 503, 509, 510, 526
–, Griffith 1–10, 15–17, 23, 333, 38, 44, 62, 70, dissipation 55, 58, 365, 366, 382
371 distribution
–, rockburst hazard 438 –, bimodal 484, 490, 491
critical –, Fisher 464, 474
–, crack 10, 16–18, 56, 62, 87, 357 –, power-law 29, 31, 389, 401, 434, 435, 437,
–, velocity 56, 365 440, 484, 486
crust 87, 88, 105, 295, 319, 350, 351, 381, 432–436 –, stress 89, 94, 440, 447, 449
crystal –, Weibull 23, 30–32, 484
–, silicon single 113 domain 101–105, 222, 233, 283, 286, 287, 324, 325
–, single 149, 380, 409 drilling 295, 327, 377, 380, 405, 418–432, 439, 440,
crystallization 286, 327 445, 447, 448, 452, 455–458, 519, 522
curve 32, 65, 74, 86, 107, 109, 110, 117, 118, 130, dry density 171, 173, 400, 465, 466
141, 144, 162, 166, 170, 173, 176, 178, 208, 209, ductile 83, 149, 229, 391, 405, 415, 418
213, 215, 222, 234, 247, 254, 258, 260, 265, 283, dynamic 1, 2, 5, 8, 34, 35, 43–63, 112, 113, 118,
305, 330, 331, 336, 356–362, 370, 385, 385–389, 121, 125, 130, 160, 162, 167, 189, 190, 232, 283,
403, 414, 415, 417, 438, 439, 447, 457, 484, 496– 286, 307, 317, 318, 332, 348, 362, 365, 366, 375,
500, 514, 515 377, 379, 382, 402, 402, 430–434, 458, 459, 494,
cusp 84, 100, 128, 163 524
cylindrical axis 102, 103, 117
Czech Republic 162, 286, 296, 363, 377 E

D earthquake (EQ) 85, 105, 170, 179, 271, 296, 380,

381, 432-438, 440, 449, 458, 459, 482, 494, 502,
damage 504
–, volume 508, 509 edge effect 112
–, zone 85, 418 Egypt 251, 355, 477
Dana conglomerate 469 El Capitan 84, 245, 248, 317, 332, 334, 337–348
Danube 294, 295 Elbe 287
Dead Sea Rift 269, 468, 469, 478, 479 electron
décollement 195, 205 –, charged 404
decoupling 148 –, conduction 404
defect 10, 13, 14, 87, 106, 109, 149, 182, 186, 408, elasticity 2, 3, 7, 8, 12, 71, 414, 415, 417
418, 421, 430 electromagnetic radiation 105, 108, 109, 117,
deformation 85, 94, 106, 123, 137, 168, 203, 204, 379–458
208, 213, 214, 233, 236, 251, 280, 322, 351, 371, electronic, total station 341
372, 404, 406, 410, 413–418, 433, 439, 440, 447, ellipticity 162, 165, 312, 318, 492
469, 481, 498, 510, 503, 526, 527 ELS (uniform (equal) loading sharing) 27
delay 21, 27, 45, 362, 426 EMR
dendrite 88, 129, 130, 134 –, profiling 442, 443, 445
depolarization 423–428 –, pulse 105, 106, 380–400, 404–421, 427,
depth 82, 84–88, 92, 112, 115, 125, 149, 176, 179, 430–432, 435, 436, 441, 450, 451, 458
180, 182, 189, 230, 247, 250, 280, 281, 285, 287, –, spectrum 405, 433
288, 295, 296, 313, 315, 318–321, 332, 359, 363, en echelon
364, 381, 422, 433, 438, 440, 441, 447, 448, 452, –, bridged 118
456, 457, 466, 520 –, fringes 95, 96, 114, 117, 130–132, 136, 139,
desiccate 87, 88 141, 144, 145, 152, 153, 155–158, 160, 162,
detachment 155, 195, 198, 199, 201–207, 522, 523 164, 239, 251, 261, 262, 265, 266, 266, 268,
Devonian 84, 158, 178, 191–209, 279, 290 269, 298, 299, 307, 348, 362, 366, 372, 374,
diagram, v vs KI 356–359 377
dike –, segmentation 96, 113, 114, 117, 121, 130–
–, pegmatite 325, 332 133, 139, 141–149, 153, 154, 158, 176, 164,
–, system 469 262, 265, 269, 272, 277, 298, 311, 313, 337,
dilatancy 105 339, 348, 357, 360–362, 366, 369, 372, 459,
dipole 383, 404, 406, 407, 412, 422, 423, 425 525
discharge 405 end member 124, 125, 130, 131, 162, 361, 373
discrimination 195, 250, 264, 268, 390, 459, 462– energy
464 –, balance 7, 32, 33, 43, 45, 50
dislocation 2, 10, 11, 149, 404–408 –, barrier 14–17, 21–23
560 Subject Index

–, elastic 3, 6, 7, 17, 28, 43, 370 –, -fold basin 251

–, excess 58, 113, 371 –, foot-wall 108
–, flux 48–50 –, hanging-wall 188, 506
–, high 139 –, Hurricane 235, 236
–, kinetic 3, 43, 44, 49–51, 58, 371, 434, 435 –, -joint relationships 504–506, 509, 511
–, liberated 43 –, Levantin 469, 479, 472, 479
–, plastic 371 –, Makhtesh Ramon 476
–, potential 43, 49, 50, 438 –, micro 322, 324
–, release rate 7, 28, 38, 40, 41, 53, 58, 113, –, normal 188, 213–215, 217–219, 222, 229,
114, 370–372 236, 251, 255, 258, 260, 265, 268, 272, 273,
–, relieved 6 277, 278, 280, 323, 505–511
–, stored strain 367 –, North Anatolian 472
–, surface 7, 8, 18, 33, 43, 84, 89 –, orthogonal set 68, 272, 273, 277, 293, 322,
–, total 15, 18, 19, 43, 44 324
England 139, 140, 167, 171–173, 175, 177, 191, –, Paunsagunt 236
210, 214, 222, 232, 462 –, Pfahl 295
environmental 10, 182, 251, 361, 512 –, pre- 505, 511, 512
Eocene 84, 106, 134, 141, 144, 171, 173, 176, 180, –, primary 188, 190, 265, 494–499, 502, 503,
181, 191, 214, 231, 232, 245, 251–255, 257, 259, 505, 509–511
261, 263, 265–269, 271–273, 275–277, 279, 280, –, reactivated 182, 213, 215, 480
322, 350, 387, 419, 469–473, 478, 479, 481, 482, –, reverse 112, 214, 215, 217, 226, 227, 231
484, 485, 505–508, 510–512, 516, 518, 524 –, Sevier 235, 236
epoxy 118, 119, 327, 378 –, strike-slip 139, 152, 214, 217, 219, 227, 229,
erosion 10, 189, 214, 234, 245, 247, 248, 265, 274, 258, 267, 272, 322, 469, 494, 500, 502, 503,
295, 347, 512, 522, 523 509, 510
Erzgebirge 284–287, 292, 313, 319, 320 –, syn- 505, 511, 512
Europe 162, 214, 226, 283, 286, 287, 292, 294, 295, –, termination zone 168, 182, 183, 188, 190,
313–315, 354, 355, 469, 470, 473, 479, 480, 524 265, 277, 459, 494–503, 509, 510
exfoliation 84, 127, 139, 177, 180, 191, 234–250, 286, –, vertical 255, 258, 260, 267, 277, 497, 498, 505
288, 322, 324, 332, 334, 337, 338, 340–348, 493 FBM 26
experiment 8, 11, 14, 17, 19, 20, 23–29, 34, 45, 48, feldspar 319, 327, 329–330, 387, 427
53–56, 58, 60–73, 76–79, 82–86, 89, 94, 96–101, fiber 10, 11, 17, 20, 24, 26–30, 32, 83, 227
104, 107–108, 110–119, 123, 126, 128, 129, 131, flaw
134, 139, 142, 149, 150, 168–171, 174, 176, 177, –, critical 16, 82, 83, 86, 87, 91, 96, 125, 126,
182, 183, 185, 246, 278, 296, 344, 345, 356–359, 140, 141, 371
362, 364–366, 380–382, 385, 386, 391, 392, 396– –, Griffith 10, 83, 120
405, 408, 410–413, 417–422, 429, 431, 449, 453– –, initial 15, 17, 82, 83, 86, 87, 125, 126, 182,
455, 458, 465, 488, 497, 502, 503, 520–523 185
extensile 108, 139, 149, 167, 170, 171, 173, 174, –, length 82
177, 271, 273, 280, 296, 318, 466, 481 –, nucleation 9
extension 1, 8, 33, 35, 40, 41, 49, 81, 82, 84, 85, 103, –, surface 21
119, 136, 139, 150, 152, 158, 160, 166, 171, 173, flood 287, 518, 521, 522
175, 176, 178, 183, 188, 203, 213, 219, 222, 227, flow
230, 231, 233, 236, 250, 286, 295, 322, 324, 343, –, experiment 523
354, 362, 460, 484, 488, 494, 509, 525, 526 –, flexural 214, 227–232
extinction 327, 330 –, -line 286
–, rate 118, 523
F fold 195, 202, 204, 205, 208, 210, 213, 214, 217,
222, 226, 227, 230–233, 254, 255, 258, 271, 273,
failure 16, 17, 19–32, 83, 84, 94, 104, 106–109, 275–278, 280, 352, 460, 468, 472, 478, 480, 483,
111, 115–117, 134, 138, 149, 166, 167, 175, 179, 484, 494, 497, 498, 505
185, 188, 380, 381, 387, 390, 392, 398, 404, 407, –, advance 233, 483
413–418, 432–437, 440, 441, 450, 456, 494, 501 –, amplitude 195, 232
fan 84, 178, 201, 203, 208, 210, 245, 248, 299, –, anticlinal 191, 210, 271, 276
341–345, 360, 363, 480 –, asymmetric 270
fat tail 484–487, 489, 491 –, axis 192, 194, 195, 202, 204, 205, 208, 209,
fatigue 62, 64, 83, 97, 139, 349, 366, 367 213, 233, 274, 285, 352, 483
fault –, buttress 214, 217
–, blocks 213, 223 –, cross, set 195, 258, 266, 480, 483, 505, 506
–, cataclastic rock 495–503 –, drag 217
–, classification 504 –, flank 207, 217, 231, 258, 261, 267, 272, 277,
–, conjugate 170, 171, 175, 508 511
 Subject Index 561

–, flexural flow 228–231 –, plume 9, 84, 87–91, 96, 102, 103, 105, 107,
–, flexural split 229 108, 110, 112–119, 130–148, 152–166, 171,
–, fracture 271, 272 173, 176, 186, 187, 241, 251, 261, 296, 298,
–, hinge 231 299, 305, 307, 308, 317, 326, 334, 336–347,
–, -joint 251, 272 359–365, 461, 462, 482, 493
–, kink 208 –, rib marking 171
–, limb 230 –, ripple mark 86, 95, 96, 100, 125, 126,
–, mechanism 228 128–130, 163, 166, 239, 240, 243, 299, 305,
–, parallel 227 337, 341–344
–, rotation 207, 233 –, sequence 121
–, sigmoid 251 –, striae 9, 82, 84, 98, 103, 108, 109, 125–130,
–, strain 231 176, 238, 241, 305, 337, 342, 359, 362
–, symmetric 232, 233, 270 –, undulation 9, 98–100, 121, 125, 127–130,
–, syncline 483 239, 240, 246, 250, 299, 305, 337, 338, 362
–, thrust 194 –, Wallner line 98, 125, 127
–, unfolding 275, 276, 278 fracture
foliation 285, 316 –, area 7, 96, 104, 182, 283, 356, 367–372, 397,
force, buoyancy 483 399, 422
forecast 380, 381, 433, 440, 446–448, 453–458 –, asymptotic 338
foreland –, azimuth 145
–, large 480 –, blasted 139
–, small 480 –, breakdown 136, 137
Formation –, brittle 1, 101, 245
–, Blue Lias 213, 220, 222, 229, 231 –, classification 325
–, Genesee 195 –, curviplanar 250, 347
–, Gerofit 474, 476 –, delayed 21, 231, 251
–, Horsha 106, 251, 253, 255, 258–262, 265, –, diagonal 286, 325, 332, 354
269, 272, 277 –, dip 102, 105
–, Ithaca 153–155, 198, 201–203 –, downward propagation 144, 341
–, Machias 190 –, dynamic 43–61, 113, 167, 189, 348, 362, 366
–, Middlesex 195 –, dynamics 2, 58, 375
–, Mor 159, 251, 253, 255, 258, 264, 272, 273, –, early 162, 285
277, 465–467, 505, 512 –, flat-lying 286
–, Rhinestreet Shales 195, 201 –, group 175, 182, 258, 266, 272, 276, 354, 514
–, Soreq 90, 91, 94 –, history 362, 461, 463, 480
fossil, deformed 195, 205 –, horsetail 183, 186, 188
fractal 17, 20, 21, 395, 434, 435, 437, 487–489 –, intergranular 127, 327, 329, 330
fractionation 349 –, internal 343–345
fractographic –, intersection 123, 124, 520, 524
–, arrest marks 91, 97, 98, 126–130, 144, 147, –, late 160, 270, 284, 285
162–165, 239, 240, 245, 249, 261, 270, 296, –, linear elastic 1, 8, 32–41, 150
298, 299, 305, 316, 359–362 –, long 488
–, barbs 88, 103, 105, 129, 132, 144, 147, 176, –, mechanics 1, 2, 8, 32–41, 83, 110, 131, 150,
270, 336, 339, 343, 347, 360, 493 271, 296, 357
–, box 126, 128–130, 226, 304 –, medium 330, 449
–, circular markings 173 –, meso 101, 104, 105, 107
–, examination 107, 174, 399, 422 –, mode 9, 34, 135, 138, 284, 305, 332, 390,
–, fringe 81–83, 95, 96, 110, 112–117, 120–134, 431, 524
136, 139–166, 186, 195, 199–206, 238, 239, –, nucleation 15, 16, 81, 84, 85, 317
242, 245–247, 251, 261, 262, 266, 268–270, –, oblique extensile 167, 170
283, 296, 298, 299, 305–307, 310, 311, 313, –, origin 83, 84, 87, 125, 126, 165, 189, 296,
317–319, 334, 336, 339, 348, 349, 356, 343, 461, 492
360–377, 461, 492, 493, 525–527 –, polarization 183, 188
–, hackle 2, 95–99, 112–114, 118–131, 139, 140, –, post consolidation 324
152, 158, 160, 162, 165, 283, 299, 307, 317, 318, –, primary 121, 152, 182, 183, 188, 190, 286,
345, 348, 349, 356, 358–362, 367–377, 527 327, 328, 330–332
–, mark overprinting 317 –, profile 515
–, mirror 58–60, 81–83, 86, 87, 96, 111–113, –, rapid 125, 127–130, 285, 305, 363, 372
117–130, 133, 140, 141, 157, 158, 160–166, –, secondary 81, 121, 152, 182, 183, 185, 186,
242, 245, 247, 296, 298, 299, 305, 307, 188–190, 327, 328, 330
311–313, 316–318, 359–371, 374, 375, 461, –, sequence 104, 117, 143, 274, 331, 498, 500,
492, 493 501, 506
562 Subject Index

–, shear 100–107, 113, 114, 166, 182, 186, 188, glass 82, 84, 86, 100, 120, 125, 129–131, 140, 142,
331, 390, 391, 403, 418, 496 149, 173, 319, 320, 339, 356–359, 362, 366, 377,
–, short 158, 499 408, 410, 426, 427
–, sickle-like 497 –, borosilicate 99, 360, 366
–, side 406 –, bottle 60, 95, 112, 167
–, slanting 95, 136, 139, 186, 525 –, ceramics 58, 84, 108, 129, 130, 173, 174,
–, step-over 124, 125 176, 357, 387–389, 392–404
–, stress estimation 492 –, fiber 29
–, surface 113–115, 118, 120, 124, 125, –, polymeric 73, 149
128–130, 158, 160, 176, 210, 270, 295, 296, –, soda lime 8, 10, 15, 54–58, 82, 97, 98, 114–117,
305, 316, 317, 330, 337, 341, 348, 349, 357, 121, 174, 388, 396, 400–404, 419–422, 425, 426
359, 366–368, 372–374, 461, 481, 484, 519, gneiss 295, 292, 294
522, 523, 525 gouge 494, 497, 499, 502
–, tensile 81, 102, 104, 105, 107, 166–168, 170, grain boundary 11, 26, 330
173, 182, 183, 186, 188, 190, 233, 246, 295, grainy 10, 11, 13, 16, 17, 26, 127, 130
305, 390, 391, 403, 481 granite
–, transgranular 327–330 –, Bohus 314
–, transitional 264 –, Borsov 84, 162–166, 286, 294–296, 298,
–, vertical 160, 186, 244, 246, 247, 294, 308, 299, 305, 313, 314, 316–321, 356, 357,
311, 325, 332, 334, 341, 342, 509, 510, 514 359–361
–, zone 248, 251, 264, 307, 308, 317, 318, 325, –, Capanne 314
363, 391, 409, 506 –, Cimer 295, 320, 349
fracturing, fluid-driven 316 –, Donegal 314
fragmentation 430, 431, 496, 499 –, Eibenstock 292, 293, 313, 315, 319
frame 86, 150, 276, 292, 299, 336, 337, 341, 342, –, Eilat 387, 419
358, 360–362, 385, 386, 495, 513, 514 –, Eisgarn 295, 349, 352
France 177 –, El Capitan 84, 245, 248, 317, 332, 334,
frequency 93, 94, 98, 222, 225, 231, 379, 382, 337–348, 363
384–389, 392, 395–398, 401, 402, 404–411, 414, –, Erzgebirge 284–287, 292, 313, 314, 319, 320
418, 420, 421, 428, 429, 431–435, 440–442, 446 –, Flossenbürg 314
friction 29, 64, 85, 184, 230, 366, 433, 434 –, Freistadt 295
fringe –, Hudcice 314
–, angularity 375 –, Lasenice 349
–, characterization 525, 527 –, Mauthausen 295
–, complex 162, 163, 361, 368 –, Melechov 349, 352, 353
–, crack 113, 128, 141, 148, 152–157, 195, 199, –, monzo 287
200–206, 373, 375, 377 –, Mrákotin 84, 162, 284, 286, 294–296,
–, en echelon 82, 95, 96, 114, 117, 130–132, 307–321, 349, 367, 368, 372, 373, 375, 377
136, 139, 141, 144, 145, 152–158, 160, –, Niederbobritzsch 292
162–164, 186, 239, 245, 246, 251, 261, 262, –, porosity 363, 403
265, 266, 268, 269, 298, 299, 307, 348, 361, –, Solopysky 314
362, 368, 369, 371, 372, 374 –, Strzegon 314
–, hackle 95, 114, 120, 122, 129–131, 139, 140, –, Sunn 314
158, 160, 162, 283, 307, 317, 348, 349, –, Weinsberg 295, 313, 315, 319, 321
366–375, 377 –, Westerly 331, 357–359
–, spectrum 130 –, Zvule 320, 349, 352, 353
–, thickness 374 granodiorite
–, triangle 166, 298, 318 –, Knowles 286, 324, 325, 331
–, uniform width 130, 131, 313, 348, 366, 367, –, Lusatian Complex 284–292
374 –, Mt. Givens 363, 364
–, width 130, 313, 374 gravity 4, 207, 247, 287–292, 341, 354
–, zone 112, 113, 365 Griffith 1–3, 5–8, 10, 11, 15–17, 23, 33, 34, 38, 41,
FSI 74 44, 62, 70, 83–85, 89, 120, 371, 488
FSR 74, 464–467, 482 group 2, 10, 11, 16, 19, 20, 26, 32, 37, 55, 81, 90,
131, 161, 162, 175–179, 182, 195, 197, 204, 222,
G 226, 243, 255, 258, 260–268, 272–278, 280, 295,
322, 325, 342, 345, 350, 361, 381, 405, 413, 414,
gabbroic 420, 421, 426, 427, 459, 463, 465, 467, 469, 475,
–, intrusion 288, 292 476, 478, 492, 499, 504, 514, 518, 519
–, layer 288 growth
geochemical 349, 352 –, quasi-static 349, 367
Germany 127, 286, 287, 313, 419, 524 –, supercritical 356
 Subject Index 563

gum 438–457, 460 –, circular 96, 296

Gumbel 23 –, composite 123, 147, 270, 341, 344
Gutenberg-Richter 434, 436, 437, 440, 458 –, concentric 143, 147, 269, 270, 332
gypsum 107, 180, 267, 521, 522 –, conjugate 178, 179, 210
–, cross 90, 177–179, 213, 265
H –, cross-fold 171, 178, 179, 192, 208, 209, 251,
258, 265, 268, 506, 508
hackle –, cross-strike 192
–, cuspate 95, 126, 131, 139, 140, 349, 367 –, curved 132, 135, 269, 494
–, flake 375 –, dynamic 63, 121, 162, 189, 307, 459
–, fringe 95, 114, 120, 122, 129–131, 139, 140, –, exfoliation 238, 248, 332, 334, 338, 341,
158, 160, 162, 283, 307, 317, 318, 348, 349, 344, 347
356, 366–372, 375, 377 –, hybrid 81, 166, 171, 177, 272, 273
–, index of raggedness 283, 356, 372, 373, –, L- 284, 286, 288
375–377 –, length 307, 459, 462, 466, 484–492
–, twist 152 –, multi-layer (m.l.) 251, 255, 258, 260, 261,
Half Dome 246, 341, 345, 348, 363, 364, 493 264–268, 272, 277, 278, 510, 524
halite 521 –, oblique 324, 325
hazard rate 27 –, of termination-zone 183, 188
hazardous waste 516 –, orthogonal 182, 215, 249, 288, 506
Hertzian 209, 210, 404 –, parent 83, 110, 112, 132–139, 141–161, 166,
heterogeneous 176, 182, 186, 195, 199, 201–206, 265, 266, 269,
–, material 10, 11, 16 270, 307, 311, 313, 336, 365, 463, 525, 526
–, nucleation 10, 21 –, pinnate 157, 183, 186, 188
homogeneous –, polygonal 214, 215
–, material 4, 10, 83 –, post-uplift 81, 177, 179, 245, 261, 272, 274,
–, nucleation 11, 17, 18 280, 345, 355, 363, 480
Hook’s law 2–4, 32, 37 –, primary 81, 188, 288, 292, 332, 366
hydraulic conductivity 519, 520 –, Q- 286, 288, 292
hydro –, regional 170, 171, 179, 191, 192, 194, 195,
–, fracturing 296 235, 237, 245, 248, 354, 363, 460, 461, 463,
–, geology 459 468, 473, 480, 511
–, thermal 286, 332, 354 –, S- 90, 177, 178, 213, 265, 286, 288, 292, 354
hydrological 234, 518 –, secondary 332
–, semi-elliptical 96
I –, sheet 177, 180, 201, 355
–, single-layer (s.l.) 77, 95, 96, 126, 130, 140,
ice 13, 380, 384, 406 171, 176, 231, 251, 258, 260–269, 272–275,
impact 10, 83, 100, 423, 425 277, 278, 465, 478, 505, 510, 512, 515
indentation 209, 210, 406, 469 –, spacing 63–66, 70–72, 74, 78, 79, 215, 261,
infiltration 519, 523 288, 463–467, 496, 499, 508
initial –, stable 345
–, defect 106 –, steep 322, 324, 354
–, flaw 15, 17, 82, 83, 86, 87, 125, 126, 182, 185 –, stepping 168, 183
–, growth 17, 87, 90–92, 94 –, surface 81, 121, 130, 132, 140, 141, 144, 146,
instability 57, 85, 113, 114, 188 158, 160, 162, 171, 173, 177, 179–182, 195, 251,
intrusion 286–288, 292, 295, 296, 319–321, 332, 349 270, 313, 314, 334, 357, 359, 361, 367, 369,
isolated 94, 95, 168, 176, 223, 245, 295 371, 372, 459–463, 492, 497, 512–516, 525
isotopic 325, 518, 519, 521 –, syntectonic 260, 261, 266, 277, 280, 345,
Israel 74, 87, 90, 171, 180, 181, 192, 232, 233, 248, 351, 355, 466, 480, 501, 505, 510
251–279, 387, 419, 429, 459, 460, 462, 468–484, –, systematic 81, 90, 125, 133, 157, 166, 170,
492, 504–514 178, 179, 182, 191, 194, 216, 220, 221, 223,
232, 276, 355, 459–468, 476, 480
J –, Trefoil 163, 165, 305, 306, 319, 361, 365, 366
–, type 245, 307, 463
jamming 65 –, uplift 81, 141, 177, 179, 245, 258, 261, 264,
jog 123, 494, 503 268, 269, 272, 274, 276, 277, 280, 284, 296,
joint 313, 316, 317, 345, 351, 355, 363, 465, 466,
–, aperture 327, 331 480, 505, 510
–, bifurcation 189, 190 Jordan 133, 248, 468, 479
–, burial 182, 264, 266, 268, 269, 272, 277, Juniata Culmination 203–205, 207, 208
280, 355, 466, 480, 505 Jurassic 167, 191, 201, 206, 213, 231, 236, 249, 279, 354
564 Subject Index

K magma 210, 234, 284, 287, 288, 295, 296, 320, 349,
350
kinematic 151, 208, 468 map trajectory 468
kink material
–, curved 110, 150, 152, 183 –, amorphous 58
–, sharp 110 –, anisotropic 11, 271
–, straight 150, 152, 157 –, brittle 2, 10, 16, 53, 54, 56, 58, 63, 77, 81–83,
kyanite 294 185, 186, 271, 404, 430, 435
–, continuous 158–160
L –, isotropic 11, 162, 271
–, polymeric 119, 125, 357
ladder pattern 214–216 –, strong 83, 174, 175, 177
lagging bay 360 –, synthetic 313
Lamé constants 4, 5, 33 –, technological 81, 83, 95
Laplace equation 35, 37 –, weak 83, 107, 177
laser, range-finder 341 measurement 17, 19–21, 54, 56, 58, 64, 67, 72,
layer 74–77, 92, 94, 111, 115, 141, 145, 147, 179, 210,
–, boundaries 110, 123, 124, 134, 139, 141, 221, 239, 277, 313, 334, 357, 363, 368, 370, 373,
149, 158, 160, 161, 176, 271, 277, 278, 289, 375, 377, 380, 391, 392, 399, 400, 403, 405, 407,
295, 296 412, 425–427, 433, 435, 439, 441, 442, 445, 446,
–, group 262 448, 452, 457, 458, 460–464, 472–474, 476, 478,
–, normalized thickness 76 480, 492, 517, 519
–, -parallel 125, 141, 195, 203–205, 229 mechanical paradox 227, 232
–, transitional 180–182, 512–516 mechanism
length distribution 462, 466, 484–492 –, breakdown 132, 136
LGC 287, 288 –, fold 228
limestone 76, 107, 133, 159, 160, 167, 178, 204, –, twist 307
205, 210, 213–216, 221, 227–233, 276, 279, 280, Mediterranean 273, 276, 469, 470, 480, 483
404, 419, 429, 462, 468, 473, 481, 484, 508, 521 melt inclusion 319, 320
lineament 207, 322, 323, 485, 520 merge zone 334, 336, 338, 339, 347
linear elastic racture mechanics (LEFM) 1, 8, 32, Mesozoic 201, 206, 213, 217, 235, 279, 322, 350
33, 35, 37, 39, 41, 49, 53, 55, 150 metal 11, 62, 125, 138, 149, 357, 366, 372, 380, 404,
linked 105, 107, 141, 160, 170, 178, 214, 244, 274, 405, 408, 409, 438, 516, 519
275, 277, 280, 316, 366, 395, 396, 411, 469, 478, metamorphism 294, 295
479, 483, 504, 512, 523 method
Linsser-filtering 366 –, comprehensive 262, 461
lithology 150, 152, 156, 157, 209, 255, 273, 276, –, discrimination 461
279, 476, 485 –, selection 261, 462
lithostatic 149, 167, 168, 171, 173, 175, 177, 179, –, shadow optical 118
180, 184, 190, 319, 332 –, statistical 463
loading mica 293, 328, 330, 331
–, dynamic 431, 458 microcrack
–, quasi-static 431, 458 –, effective 22, 44, 45, 167
local 16, 29, 31, 32, 57, 83, 85, 89, 93, 108, 112, –, nucleation 10, 11, 16
114, 117, 123–125, 132, 136, 137, 140, 143, 144, millidarcy 516
146–149, 153, 166–168, 177–179, 182, 186, 188, mine 182, 197, 198, 379, 381, 438–457
203–209, 215, 227–234, 243, 245, 247, 248, 251, mineral 286, 327, 329–331, 354, 357, 363, 414, 427,
258, 261, 267, 271, 273, 274, 277, 280, 289–291, 522, 524
313, 316, 317, 320, 330, 332, 335, 339, 340, 343, Miocene 213, 214, 231, 236, 276, 279, 322, 468–474,
344, 348, 354, 360, 362–367, 373, 385, 433, 459, 479–482
460, 466, 467, 478, 481, 483, 492, 495, 506, mirror
508–511, 516, 524, 525 –, boundary 59, 112, 122, 125–127, 140, 141,
longitudinal splitting 101–105, 108, 170, 177, 248, 160, 163, 242, 299, 305, 307, 361, 366,
280, 343, 344, 392 368–371, 375
Luders’ band 101 –, plane 60, 81–83, 86, 87, 96, 121, 122, 125–133,
Lusatian Granodiorite Complex 284, 286, 287, 140, 141, 157, 158, 160–165, 242, 245, 296,
292 298, 299, 305, 307, 313, 317, 362, 365, 375, 461
–, primary 121, 122, 125, 305, 365
M –, radius 82, 125, 141, 492, 493
–, secondary 121, 125, 305, 365, 377
macroscopic 90, 95, 115 mist 56, 58–61, 82, 98, 112, 119–122, 125, 127, 358,
Maestrichtian 481 371
 Subject Index 565

mode orogeny 195, 197, 279, 280

–, I 9, 39, 47, 81, 98, 100, 108, 110, 111, 114, 117, orthogonal 139, 166, 182, 205, 209, 215, 230, 233,
118, 131, 134, 136–138, 149–152, 155, 161, 249, 258, 265, 267, 272, 273, 277, 286, 288, 292,
166–168, 175, 176, 190, 245, 271, 296, 328, 293, 305, 322, 324, 325, 338, 341, 363, 369, 370,
330, 331, 348, 366, 367, 369, 372, 502, 503 386, 460, 461, 465, 496, 499, 505–509
–, II 9, 39, 98, 111, 112, 152, 167, 168, 176, oscillation, optical 408
177, 184, 186, 190, 296, 337, 503 overburden 70, 149, 176, 180, 250, 276, 279–281,
–, III 9, 34, 35, 39, 111, 112, 114, 137, 138, 149, 318, 344, 364, 465, 482, 516
176, 337, 366, 369, 372 overcoring 194
–, mixed 85, 98, 100, 110, 111, 114, 117, 125, overlapping 136, 140, 143, 145, 147, 201, 208, 221,
127, 128, 131, 134–136, 148–150, 160, 161, 270, 327, 371, 420, 494, 497, 502
166, 167, 173, 175–177, 271, 273, 348, 357,
366, 367, 369, 408 P
–, shear 9, 175, 176, 391
–, tensile 108, 190, 344 Palaeozoic 292, 295
model paleomagnetism 464
–, 2-D 185 paleostress 59, 83, 121, 131, 140, 145, 207, 208,
–, 3-D 337 269, 273, 274, 277, 278, 280, 319, 324, 357, 363,
–, capacitor 406–408 364, 459, 460, 468, 469, 472, 473–477, 479,
–, Cloos 285, 286, 288, 354 480–483, 492
–, EMR 380–399 paradox 176, 277, 232
–, fiber bundle 11, 17, 20, 26, 28 parametrization 392
–, Hull 337, 339 Paris law 62
–, polarization 425 pegmatite 324, 325, 332
–, Stearns 510 percolate 84, 519, 523
Mohr-Coulomb 415, 456 perimeter 22, 32, 132, 305, 347, 360
Molasse 295 permeability 422, 516, 518, 524
monitoring 117, 380, 438, 443–445, 452, 453, 458, Permian 213, 352
519, 521, 524 perturbation 1, 95, 214, 216, 432, 433, 483, 503
Mrákotin 84, 162, 284, 286, 294–296, 307, 308, petrographic 295, 316, 317, 324, 327, 377
310–314, 317–321, 349, 367, 368, 372, 373, 375, petrological 313, 324
377 phonon 382, 384, 385, 407, 458
muscovite 295, 316, 319–321, 327, 328, 330–332 pit density 118
plagioclase 329, 331, 414
N plate tectonics 459
Pleistocene 180, 236, 267, 276, 279, 468, 469, 479, 516
neotectonic 177–181, 265 Plexiglas 110, 168
New York State 123, 178, 192, 195, 197, 201, 204, Pliocene 279, 324, 478, 479
207 plume
noise 381, 386, 405, 408, 411, 429, 432, 433, 436, –, bilateral 132, 148, 462
438, 441, 442, 451 –, concentric 146, 334, 336–340, 347
Nomenclature 11, 64, 81, 464, 465 –, curved-concentric 334, 336–339, 347
nonlinear 1, 55, 415, 489, 490 –, dendrite-like 88, 129
Norton equation 77 –, intensity 166, 361
nucleation –, multi-texture 163
–, multiple 16, 26 –, radial 91, 162, 163, 241, 296, 299, 341, 343,
–, zone 84, 85, 109 359, 361, 363
–, rhythmic 359, 360
O –, sub-vertical 338
–, trajectories 340
offset 90, 152, 167, 217, 219, 251, 265, 266, 277, –, unilateral 148
278, 322, 330–332, 377, 378, 497, 500, 503, 506 –, wavy 298, 361
Oligocene 213, 214, 231, 276, 279, 468–471, 479, plumose structure 91, 133
481 pluton
optical surface wave 383 –, emplacement 313
order 1, 2, 4–6, 8, 10, 11, 13, 16, 28, 36, 41, 47, 52, –, North Bohemian 286
62, 71–75, 94, 102, 105, 110, 119–121, 130, 140, –, South Bohemian 125, 161, 190, 209,
146, 150, 162, 179, 180, 184, 187, 208, 213, 238, 284–286, 294–321, 349, 351, 354, 358, 377
250, 258, 287, 325, 327, 330, 331, 352, 371, 385, PMMA 54–59, 110, 113, 114, 119, 139, 149, 168,
391, 395, 398–400, 404–406, 413, 414, 616, 418, 185, 357, 359, 365, 372, 419, 427, 503
425, 428 Poisson ratio 3, 5, 37, 84, 113, 209, 229, 230, 391,
organic 516, 522 395, 400, 416–418
566 Subject Index

polarization 183, 188, 380, 383, 406, 421–428 quasi

Pollard-Segall 63, 67, 78 –, ellipsoid 509
polymethyl methacrylate 110, 357 –, -elliptical joint 96
pore radii 102, 106, 403 –, -static 63, 98, 99, 110, 112, 113, 126, 130, 162,
pre-cut 104, 114–117, 168, 174, 182, 183, 185, 186 348, 349, 363, 365–367, 379, 431, 459, 524
press stiff 385, 446, 447 –, -triangle 313, 318
pressure
–, confining 104, 105, 107, 108, 184, 331, 385, R
387, 388, 415–418
–, fluid 84, 149, 189, 230, 266, 277, 278, 316–318, ratio
354 –, KIII/KI 110, 138, 139, 144, 313, 339, 348, 366
–, internal 121, 137, 138 –, Poisson 35, 37, 84, 113, 209, 229, 230
–, lithostatic 149, 319, 505 reactant 357
–, pore 63, 71, 84, 167, 261, 316, 348, 359, 363, rectangularity 373
364, 451, 455, 456 relaxation 113, 205, 213, 214, 232, 273, 278, 384,
primary 48, 59, 60, 77, 81, 100, 120–122, 125, 133, 398, 399, 411, 425–427, 483, 508
152, 182, 183, 185–190, 217, 260, 265, 277, 284, remote 7, 17, 18, 23, 38, 42, 49, 59, 68, 72, 74, 84,
286, 288, 292, 305, 327, 328, 330–332, 365, 366, 114, 125, 132, 136–139, 143, 146–148, 158, 167,
494, 495–499, 502–511 168, 176–179, 182, 186, 203, 224, 227–232, 266,
procedure 1, 6, 16, 23, 44, 70, 76, 93, 98, 102, 108, 271, 316, 340, 362–368, 460, 473, 480, 483, 492,
122, 222, 327, 363, 371, 373, 378, 389, 392, 397, 506–511, 525
459–467 rhythmic 261, 298, 316, 359, 360, 497, 503, 516
profile 89, 100, 101, 104, 121–123, 126, 141, 307, 311, ridges and grooves 101, 102, 106
319, 328, 329, 344, 373–375, 421, 492, 515, 519 ripple mark 86, 95, 96, 100, 125–130, 163, 166, 239,
propagation 1, 8–10, 40, 44–46, 48, 50, 53, 55–60, 240, 243, 299, 305, 377, 341, 343, 344, 359, 363, 461
83–89, 91, 96–100, 104, 105, 107, 110, 113, 117, rock
118, 121, 124–131, 135, 137–139, 143–160, 165, –, and gas outburst 380, 451, 455
167, 176, 180, 185, 186, 202, 205–207, 230, 239, –, mosaic 496, 499
240, 243, 250, 258, 261, 274, 284, 296, 305, 316, –, sample 381, 410, 411, 415
318, 334, 337, 339, 341, 343–346, 357, 359–363, –, strength 94, 171, 276, 516
365, 366, 383, 404, 406, 411, 425, 461, 467, 526 –, strong 105, 108
Proterozoic 287, 288, 292–294, 469 –, weak 104, 105, 108, 141, 173, 276, 494
province 81, 84, 150, 160, 191–283, 285, 287, 322, rockburst hazard 438–452
325, 332, 349, 351, 354, 460, 482, 511 root 118, 134, 148, 180, 295, 319, 320, 373, 437
pulse rose diagram 463, 476, 509
–, classification 413 rotation 89, 132, 135, 137, 138, 142–148, 153–156,
–, envelope maximum 382 168, 177, 186, 195, 197–210, 214, 222, 232, 233,
–, extended chain 420 261–264, 267, 280, 307, 324, 336, 339, 461,
–, individual 393, 401, 405, 414, 429, 431 463–473, 478–483, 494, 509, 525
–, lengthy 393, 414 roughness root mean square 118
–, parameter 394 Royal Arches 332, 334, 341, 345–348
–, short 393, 414 rupture 16, 17, 27, 29, 32, 62, 85, 87, 88, 105, 112,
–, uneven shape 413, 414 113, 357, 366, 494, 499

Q S
quadrant 150, 183, 188, 323, 512, 514 saddle 121, 122
quarry sandstone 107, 123, 139, 147, 157, 168, 175, 184,
–, blasting 379, 380, 405, 429–432, 445, 458 191, 195, 234–250, 261, 274, 279, 334, 340, 345,
–, Borsov 84, 162–166, 286, 294–299, 305, 347, 348, 409, 410, 414, 503
313–321, 356–361 Santonian 232, 233, 266, 276, 277, 473, 478
–, Friepeâ 319 saturation, effective 65, 70–74, 77–79, 487, 490, 491
–, Klosterberg 189, 288 SBP 294, 295, 313, 314, 319–321
–, Mrákotin 84, 162, 284, 286, 294–296, 307, scale 1, 2, 20, 56–61, 72, 81–91, 100–103, 120, 126,
308, 310–314, 317–321, 349, 367, 368, 372– 127, 131–133, 140, 143, 154, 167, 183, 189, 191,
377 208, 209, 215, 239, 241, 242, 254, 255, 259, 264,
–, Shaw 323–326, 332 270, 271–277, 288, 296, 298, 308, 326, 328, 336
Quartenary 470 scanline 65, 462, 476
quartz scenario 16, 17, 26, 32, 62, 146, 147, 275, 366, 407,
–, diorite 348 487
–, monzonite 348 schist 294
–, xenomorphic 30 schlieren 286
 Subject Index 567

secondary 48, 53, 55, 56, 59, 60, 77, 81, 100, 113, spectrum 56, 130, 175, 178, 254, 296, 317, 377,
120, 121, 125, 152, 160, 176, 180, 182–190, 217, 380, 405, 408, 420, 433
260, 265, 277, 305, 307, 325, 327, 328, 330–332, spherical 92
337, 365, 368, 372, 377, 494, 495, 497–508 splay 183, 190, 332, 498, 502
section 2, 6, 8, 10, 17, 18, 21, 26, 34, 44, 62–64, split
74–76, 83, 87, 93, 96, 98, 100, 101, 105, 112, 128, –, discontinuity 98
131, 136, 140, 141, 149, 150 –, longitudinal 81, 101–109, 114–117, 166,
sector 158, 261, 299, 305, 461, 481, 508, 509 170–177, 248, 280, 343–345, 392, 396
sedimentary 63, 81, 85, 87, 95, 107, 113, 123–125, –, wiggly 170, 177
130, 140, 141, 150, 152, 158, 160, 161, 166, 177, splitting 59, 81, 101–108, 166, 170, 177, 248, 280,
189 343, 344, 363, 392, 405, 414, 415
segment 64, 77, 78, 89, 95, 96, 110, 113–117, stable 10, 16, 33, 85, 86, 168, 185, 343, 345, 365,
127–149, 152, 153, 155, 159–190, 200, 204, 238, 404, 442, 451, 456, 500, 519, 520, 522
242, 265, 269, 277, 298, 299, 305–307, 311, 313, stair 103, 133, 236
318, 336, 337, 362–377, 463, 494, 525 starch 112, 113, 357, 365, 366
seismic 85, 112, 260, 319, 433–440, 494 static fatigue limit 97
self-correction 271 statistics of extreme 23
Senonian 261, 276, 473 step 142, 143, 149, 360, 408, 409
set 2, 5, 6, 30, 34–36, 41, 43, 59, 82, 83, 89, 90, 100, stepping 2, 52, 60, 92, 102, 104, 107, 114, 123–125,
103–109, 145, 148, 156, 162, 168, 170, 171–180, 132–141, 168, 183, 215, 238, 245, 299, 305–307,
188, 193–282, 288, 294–325, 386, 389, 432, 451, 336, 337, 368, 369, 378, 380, 450, 456, 457, 460,
459–498, 500–510, 519–521 463, 494, 496–499, 502, 503, 526
shadow 64, 66–72, 78, 118, 154, 215, 496 stereographic projection 221, 271, 293, 476
shale 76, 84, 107, 123, 154–160, 178, 191, 195, 197, stock 287, 349
201–207, 213, 215, 216, 227–231, 279 strength 20, 30, 31, 83, 84, 94, 101, 107, 108, 116,
shear 117, 124, 171, 173–175, 177, 186, 246, 275–279,
–, area 104, 391 348, 387–390, 415, 418, 438–440, 446–451, 456,
–, conjugate 186 483, 516
–, couple 187, 205, 227, 229 stress
–, flexural 228, 229 –, Alpine 241, 278
–, incipient 104, 106, 107 –, compressive 70, 184, 227, 229–231, 234,
–, left-lateral 142, 147, 150 245, 347, 465, 469
–, modulus 35, 68, 465, 466 –, concentration
–, R- 497 –, concentration 10, 11, 13, 22, 89, 94, 158,
–, Riedel 186, 188 184, 188, 213, 217, 247, 250
–, right-lateral 142, 147, 150, 152, 494, 497 –, Dead Sea 473
–, zone 105, 168, 186, 496, 497, 502, 503 –, deviatoric 233
sheeting 180, 234, 250, 295, 322, 324, 348 –, effective 95, 130, 280
shingle-like arrangement 348, 349, 366, 367 –, factor (SCF) 94
shortening 195, 203–205, 233, 469, 471, 472, 479, –, field 507
480, 525, 526 –, field 86, 149, 210
Sierra Nevada 285, 286, 322–327, 331–333, 335, –, global model 89, 95, 274
337, 341–352, 363 –, gradient 89, 95, 274
sigmoidal 288 –, heterogeneous
silicon 77, 113, 114, 388 –, inversion 468
sillimanite 294 –, lithospheric field 214
siltstone 84, 107, 153, 155, 158, 175, 191, 195, 203, 279 –, local 132, 143, 144, 146, 147, 179, 186, 205,
Silurian 192, 195, 203 215, 227, 232, 267, 271, 273, 274, 280, 316,
simultaneous 55, 113, 121, 142–145, 166, 170, 171, 337, 373, 466, 483, 508, 525, 509–511
208, 210, 277, 278, 332, 452 –, maximum 13, 74, 502
Sinai 284, 355, 363, 364, 473 –, midplate field 483
slickenside 106, 229, 322, 468 –, minimum 137, 138, 152, 316, 525
smoothing 464 –, normal 137, 138, 152, 316, 525
soil 180–182, 512–516 –, paleo- 470, 479
Somerset 167, 191, 213, 217 –, peak 106, 107, 168, 397–399, 415, 436, 439,
South Bohemian Pluton 125, 161, 190, 209, 446
284–286, 294–321, 349, 354, 358, 377 –, perturbed 213, 509
South Moldanubian Massif 294 –, plane 35
spacing 63–79, 124, 153, 155, 179–182, 203, 214, –, principal 48, 84, 89, 107, 108, 110, 137, 138,
215, 237, 250, 261, 265, 269, 288, 298, 307, 317, 146, 147, 170, 173, 178, 180, 186, 209, 228,
322, 325, 334, 345, 355, 395, 414, 442, 461–467 229, 232–234, 250, 258, 274, 284, 316, 344,
Spain 177 386, 482, 483, 502, 503, 508, 509, 516
568 Subject Index

–, quasi-static 98, 431 synclinorium 352, 354

–, rate 24, 361 Synform 275, 276, 278
–, ratio 138, 229, 230 Syria 251, 273, 469
–, redistribution 144 Syrian arc 251, 252, 278, 473
–, relaxation 113, 273, 278, 425, 426, 508 systematic 81, 97, 117, 125, 130, 157, 166, 168, 170,
–, remote 7, 17, 18, 23, 38, 49, 59, 68, 72, 74, 137, 177–180, 182, 189, 191, 194, 202, 216, 220, 221,
146–148, 168, 177, 179, 203, 224, 227, 271, 223, 232, 274, 276, 322, 324, 331, 332, 351, 355,
316, 340, 363, 365, 483, 492, 506, 508–511 438, 459–468, 476, 480, 499, 504, 526
–, residual 234
–, sequence 203, 280 T
–, singularity 12, 13, 18
–, -strain 2, 107, 398, 399, 414, 415, 417, 425, Taughannock Falls State Park 153, 154
438, 439 technique 77, 83, 87, 98, 100, 114, 121, 194, 221,
–, syn-intrusive 292 287, 296, 319, 341, 372, 380, 432, 458, 459, 461–
–, tensile 18, 41, 42, 63, 87, 113, 118, 125, 137, 464
150, 153, 158, 167, 176, 180, 182, 213, tectonic 63, 83, 177, 195, 197, 198, 201, 203, 205,
227–231, 233, 245–247, 296, 318, 332, 339, 214, 222, 232, 234, 235, 251, 261, 267, 271, 276,
356, 363, 365, 511, 525 279, 280, 322, 324, 351, 355, 465, 466, 469, 472,
–, thermal 99, 100 480, 483, 504–506
–, trajectory 192, 207–209, 213, 234, 246, 247, –, neo 177–181, 265
459 –, plate 236, 459
–, uniform 313, 318, 483 –, relaxation 205
–, wave 98, 430, 431 –, syn 81, 177, 205, 232, 245, 251, 260, 261,
stress intensity factor 266, 272, 277, 280, 324, 345, 351, 355, 363,
–, critical 33 459, 466, 480, 501, 504, 505, 507, 510, 512
–, dynamical 47, 54, 58, 118 tectonics 83, 195, 205, 207, 230, 231, 250, 459, 472,
–, Kc 33 478, 479
–, KI 33, 40, 59, 60, 97, 138, 184, 356, 357, 359, tectonofractography 81, 83, 85–175, 337
361, 488 temperature 15, 17, 20, 23–26, 54, 63, 76, 77, 99, 234,
–, KIc 33, 59, 60, 356, 357, 359 319–321, 361, 385, 400, 403, 405, 427, 516, 519
–, KII 33, 40 tension
–, KIII 33, 40, 138 –, area 391
–, static 58, 59 –, gash 168, 186
striae 9, 82, 84, 98, 103, 108, 109, 125–130, 176, –, pure 143, 362
238, 241, 305, 337, 342, 359, 362, 390 termination 120, 123, 124, 150, 167, 168, 182, 183,
strike-slip 139, 152, 214, 217, 219, 227, 229, 258, 188, 199, 258, 265, 277, 322, 336, 459, 494–504,
267, 272, 322, 469, 472, 494, 500, 502, 503, 509, 510
510 Tertiary 77, 121, 213, 232, 265, 290, 316, 321, 322,
structure 1, 10, 32, 91, 117, 128, 129, 133, 139, 147, 469, 484
149, 166, 168, 177, 186, 188, 190, 191, 195, 197, test 76, 85, 93, 100, 101, 104, 107, 116, 117, 167,
198, 204, 207, 208, 217, 222, 233, 250, 252, 258, 183, 229, 230, 246, 337, 387, 391, 392, 397, 403,
271–273, 276, 278, 286, 287, 296, 305, 307, 331, 410, 415–421, 435, 436, 442, 458, 492, 519, 520
336, 341, 345, 346, 349–354, 361, 388, 435, 460, theory, Secor 84, 85, 316
468, 473, 494, 502, 504, 505, 508, 510, 512, 520, thermal
524 –, expansion 11, 13
subcritical 62–79, 356, 488, 490 –, gradient 316, 359
surface –, nucleation 11, 14, 17, 21, 26, 28
–, hidden 368, 370 thrusts 213, 287
–, morphology 81–83, 87, 88, 90, 127, 152, time
155, 210, 236, 261, 305, 461, 481, 481 –, decay 400, 41,, 412, 427
SVOW 384, 408 –, fall 382, 384, 385, 401, 427, 441
symmetric 3–5, 13, 83, 100, 128–130, 164, 165, –, rise 382, 384, 385, 401, 427, 441
184, 190, 210, 232, 233, 244, 251, 270, 271, 276, –, to failure 17, 20–24, 28
296, 298, 319, 346, 347, 483 tip line 153, 155, 508, 509
syncline tonalite 325
–, Beer Sheva 84, 141, 145, 171, 174, 217, topography 82, 130, 178–180, 234, 250, 261, 288,
251–281, 460, 461, 480, 481, 504–512, 390, 517, 522, 523
516–524 traction 3, 32, 33, 36, 49, 50, 52, 85, 155, 229–231
–, Lackawana 205 transport
–, Neqarot 474 –, of particles 524
–, Shephela 144, 171, 251, 253, 258, 261, 272, –, of solutes 521
273, 279, 459, 465, 473, 512 transverse section 327–329, 331
 Subject Index 569

triangularity 373, 374 W

Triassic 213, 321, 352
triple point 10, 12, 14 Wadi
tritium 518, 519 –, Naim 141–143, 253, 260, 262, 265–267,
turbidites 288 272–278, 462, 484, 486, 510, 511
Turonian 74, 429, 473, 474, 476, 479 –, Secher 135, 253, 258, 261–264, 270–279,
462, 508, 511
U Wales 187, 210, 214, 222, 230
water
undulation 9, 98–100, 121, 125, 127–130, 239, –, ground 266, 277, 520
240, 246, 250, 299, 305, 337, 338, 362 –, infusion 455
unilateral 146, 148, 316, 317, 340, 462 –, table 363, 364, 518
unsaturated 76, 79, 123, 519, 523 wave
unstable 85, 86, 343, 345, 365, 456, 520, 522 –, acoustic 46, 406, 426
updoming 288, 292 –, longitudinal 48
uplift 81, 141, 162, 176, 177, 179–181, 215, 234, –, Rayleigh 53, 113, 365, 384, 385, 395, 407,
235, 237, 239, 241–251, 258, 261, 264–269, 450, 458
272, 273–281, 284–287, 295, 296, 313, 316, –, transversal 46
324, 345, 351, 355, 363, 364, 444, 445, 459, 465, –, velocity 53, 112, 113, 385, 395, 450, 465, 466
466, 480–484, 495–497, 500, 504, 505, 510–512, wetting 520
516, 524, 526 whisker 6, 83
USA 168, 191, 234, 250, 286, 460, 524
Y
V
Yosemite National Park 84, 245, 246, 248, 286,
vadose zone 516, 519, 520, 522 317, 323, 324, 332, 334, 341–343, 347, 348, 493
Variscan 213, 285, 287, 292, 294, 295, 321 Young modulus 3–5, 8, 18, 28, 68, 395, 400, 403,
vein, calcite 213, 214 417, 438, 465, 466
velocity
–, final 44 Z
–, gap 57
–, jump 56, 129, 359 zigzag 104, 299, 305, 331
–, maximum 98 Zion National Park 126, 127, 248–250, 279–281,
–, oscillation 56, 365, 366 334, 341, 345, 347, 348
–, overshoot 356, 358, 359 zone 2, 59–61, 63, 84, 85, 98, 99, 105, 107, 109,
–, Rayleigh 400 112, 113, 115, 117, 120, 121, 123, 125, 133, 134,
–, terminal 53, 55, 56, 356, 358, 362, 365 144, 148, 149, 168, 182, 183, 186, 188, 190, 206,
–, threshold 55, 58 213, 215, 234, 235, 246, 248, 251, 264, 265, 270,
VLS (variable load sharing) 27, 29 277, 287, 288, 294, 295, 307, 308, 317, 318, 325,
void 3, 10, 16, 20, 84, 95, 150, 162, 271, 387, 463, 331, 334, 336, 338–344, 347, 354, 362, 363, 365,
520, 523 371–373, 391, 398, 399, 409, 413, 414, 417, 418,
volcanic 322, 380, 432, 433 422, 432–434, 438–450, 452–459, 469, 471–474,
volume, critical 84 480, 481, 494–503, 510, 516, 519, 521–523