The Science of The Perfect Swing

The Science of
the
Perfect Swing
The Science of the
Perfect Swing
Peter Dewhurst
1
3
Oxford University Press is a department of the University of
Oxford. It furthers the University’s objective of excellence in research,
scholarship, and education by publishing worldwide.
Oxford New York
Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto
With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam
Oxford is a registered trademark of Oxford University Press

in the UK and certain other countries.
Published in the United States of America by

Oxford University Press
198 Madison Avenue, New York, NY 10016
© Oxford University Press 2015
All rights reserved. No part of this publication may be reproduced, stored in

a retrieval system, or transmitted, in any form or by any means, without the prior
permission in writing of Oxford University Press, or as expressly permitted by law,
by license, or under terms agreed with the appropriate reproduction rights organization.
Inquiries concerning reproduction outside the scope of the above should be sent to the
Rights Department, Oxford University Press, at the address above.
You must not circulate this work in any other form

and you must impose this same condition on any acquirer.
Library of Congress Cataloging-in-Publication Data

Dewhurst, Peter, author.
The science of the perfect swing / Peter Dewhurst.
pages cm
Includes bibliographical references and index.
ISBN 978–0–19–938219–4
1. Physics. 2. Golf. 3. Force and energy. I. Title.
QC26.D49 2015
796.35201ʹ5316—dc23
2015020058
1 3 5 7 9 8 6 4 2
Printed in the United States of America
on acid-free paper
This book is dedicated to my wife Iris, who has been a source of support
and encouragement through my many struggles in the world of engineering
science and who always accepted, without complaint, my more than
occasional mental absences.
My children Nicholas and Claire have brought much joy and pride over the
years—expanding the family with Cynthia and Richard and later with our young
scholar athletes Harrison, Allison, and Jack who have provided much inspiration
with their love of the game.
CON T E N TS
Preface—important notes to the reader xi

Acknowledgments xiii
1. Introduction 1
Important developments in golf equipment over the past 50 years 5
A brief discussion of moment of inertia (MoI) in golf 6
Back to the history 7
2. Ball flight 15
The game without dimples 15
The golf ball as a projectile 16
The golf ball takes flight 18
The effects of atmospheric pressure and ball weight 24
The effects of head- and tailwinds 25
Curved ball flight 26
Effect of head- and tailwinds on hooks and slices 29
Effect of crosswinds 29
Optional Reading—The Supporting Physics 31
2.1. Aerodynamic forces 31
2.2. Lateral and drag force coefficients 32
2.3. Components of the drag and lateral forces
in 3-dimensional flight 35
2.4. Effect of winds on ball flight 38
3. Basic fundamentals of ball striking 39

Contact time between club and ball 39
Contact force between ball and club face 40
Ball speed relationships 42
(a) Club head speed factor 42
(b) Restitution Factor (1 + CofR) 44
(c) Mass factor = club head mass/(club head mass + ball mass) 45
The swing 47
Controlling the club loft 57
Effect of loft on ball launch angle 62
Effect of loft on ball speed 63
Estimating approximate values for CofR 67
Effective impact mass 68
vii
viii Contents
Effect of impact speed on CofR 72

Effect of impact speed on backspin 73
The power of the modern drive 75
3.1. Introduction to the science of ball striking 79
3.2. Dynamic loft 79
3.3. Wedge angle effects on ball speed 81
3.4. Determining dynamic loft from launch data 83
3.5. The effective impact mass of the golf club 86
3.6. Relationships between impact speed, impact force, contact
time, and CofR 93
4. The generation of ball spin 95

Types of golf balls and spin rates 95
Spinning and gripping 98
Ball cannon testing 103
Increased and decreased spin rates in “grassy” lies 106
The mechanism of overspin creation in golf striking 109
Application of the SCofR to ball striking 111
Gear effect spin 120
4.1. Wedge angle effects on spin rate and launch angle 125
4.2. Ball cannon testing 127
4.3. Effect of shaft mass on ball spin and launch angle 130
4.4. Gear effect on ball spin 131
4.5. The full complexities of spin generation in oblique impact 135
5. Analysis and formulation of rules for curved ball flight 147

Vertical and horizontal launch angles 147
Ball spin axis tilt 150
Lateral deviation of hooks and slices 151
Curved shots to defined targets 158
Further examples 159
5.1. A nalysis of the sideways components of launch
angle and spin rate 162
5.2. Estimating the tilt of the spin axis 165
5.3. Rules for curved ball flight to the target 166
6. A general model of driver impact and ball flight 173

Design parameters of a typical driver 173
The “straight” shot 174
Driver strikes offset from the face center 179
The Great Big Bertha II—a basis for comparing driver performance 179
Effects of increasing the MoI and the CofR 187
Iron strikes offset from the face center 189
The longest drives on the PGA Tour 192
ix Contents

6.1. The 3-dimensional ball striking model 194
6.2. Eleven equations of motion 196
6.3. Application of the equations of motion 198
7. Putting 201
Earlier investigations and other sources of information 201
Slipping and pure rolling 202
The “speed” of greens 205
The mechanics of launching the ball 208
Capturing the ball in the hole 214
Putting up or down the slope 217
Putting across sloping greens 219
Effect of imbalance of the golf ball (with David Marsh, 2013) 221
7.1. The science of ball striking with the putter 228
7.2. The roll of the golf ball on a level green 230
7.3. Putting straight up or down a sloping green 235
7.4. Putting across a sloping green 237
7.5. M
ore sophisticated models:Rolling with skidding
and slipping; velocity dependence on turf resistance 239
7.6. Conditions for capturing the ball 242
Appendix A: Energy transfer during club-ball collision 245

Rigid surface collisions 245
Driver ball impacts 250
Appendix B: Distance error propagation in putting 255
References 259
About the Author 265
Index 267
P R E FAC E — I M P O R TA N T N O T E S T O T H E R E A DE R
Every chapter in this book is divided into two separate parts. The first and main
body of each chapter is written in a descriptive style devoid of the jargon and abbre-
viations of science, with only simple step-by-step calculations where needed to
illustrate the effects of changes in the golf swing, the equipment, the launch condi-
tions of the ball, the “speed” and slope of the green, and so on. The descriptions,
results, and conclusions in these main-body sections occupy about two-thirds of the
book. They employ the units of golf as used in the United States, namely, miles per
hour, revolutions per minute, yards, feet, inches, and pounds. This is a mixed bag of
units to use in describing the science of the game. However, it leads to results whose
magnitudes relate directly to golf commentary, to the general golf literature, and to
every-day experience in a range of other activities.
In the latter parts of each chapter lie the more complex derivations of the sci-
entific models of the different aspects of the game. These sections are as concise
as possible and employ metric units of kilograms, meters, and seconds for direct
input to Newton’s laws, particularly as applied to impact mechanics. These sections
are present to provide the formal basis for the simplified descriptions and conclu-
sions that occupy the majority of the book, and to allow the scrutiny of the scien-
tific community. However, the simple descriptions are complete in themselves,
i ncluding rules, lookup charts, and step-by-step simple calculation steps, so that no
loss of understanding is involved in skipping through or jumping over the science
modeling sections.
The principles of the game are those found in the general theory of impact, the
science of flight, and the mechanics of motion including the bouncing and rolling
of balls. These are among the most fascinating of the physical sciences. The book
should be considered as a dual-path journey. On the one hand the basic sciences are
shown to explain even the finest details of the game. At the same time the different
aspects of the game from the swing to the conditions for capture in the hole, provide
wonderful examples for explaining scientific principles at work. Almost everything
about the game, of golf provides examples of perfection, when measured in terms
of the underlying scientific principles. This is particularly the case in The Science of
the Perfect Swing, where the examples and case studies are all taken from the per-
formance of PGA and LPGA Tour professionals. The intention of the book is to
describe the fundamental nature of the game in the simplest terms possible without
sacrificing rigorous checks on the accuracy of the work. The reader with even dis-
tant memories of high-school science, and maybe not even good memories, can fol-
low the general descriptions of the golf science, and in so doing reach a much deeper
and richer appreciation of the game.
xi
xii Preface
Physicists get upset about any misuse of their discipline. However it simplifies
many discussions if, for example, we go along with using pounds as mass or introduc-
ing centrifugal force into the swing rather than wrestling with centripetal force and
inertial resistance. So, I offer apologies in advance for such deliberate indiscretions.
Peter Dewhurst
June 2015
AC K N O W L E D G M E N T S
There is a rich literature on the science of golf evidenced by the 100 or so books and
articles that are referenced in this work. The reference list is not a suggestion for fur-
ther reading. Every one of the references adds a disconnected patch, which together
with other patches, some created by me and others discovered outside of the golf
world, provide a remarkable quilted story of the game.
As in any small branch of science, the larger and more significant patches
have been produced by a small number of individuals; these will become evident
throughout the book. Three of them require special note at the beginning.
Fredrik Tuxen is the inventor of the Doppler radar launch monitor and ball
tracking system and founder of Trackman A/S in Denmark that has had a game-
changing impact on player performance. Through his company Trackman A/S,
Fredrik Tuxen has published a series of articles based on the TrackmanTM monitor-
ing of PGA and LPGA professionals at Tour events. The data and empirical rules
in these articles provided a means for testing and refining some of the analytical
models developed for this book. The title of this book is based in part on its wide-
spread use of Trackman data on PGA and LPGA Tour professionals, which in the
aggregate represents the nearest thing to perfection in today’s game.
Prof. Rod Cross of the University of Sydney has carried out breakthrough
research into the bouncing of rubber balls and uncovered the mechanism by which
energy can be recovered in the bounce to generate extra spin. Adapting some of
his work in tennis and baseball to the golf club strike, particularly with respect to
the spin performance of the modern premium ball and the impact effect of the golf
shaft, has been a great pleasure.
Prof. Raymond Penner from Vancouver Island University has written on the
effects of the bulge of the driver face. Correspondence with him was very helpful
when I was building a 3-dimensional model of driver impact in order to understand
the cannon testing of a novel driver head that I had designed. The design was ruled
nonconforming by the United States Golf Association (USGA) while still in the
womb, but the modeling became the starting point for what eventually emerged as
this book. Penner’s work on the roll of the golf ball in putting also requires special
mention.
I also wish to mention personal friends who have made contributions to this
work. Tom Wishon, who became a personal friend during the writing of the book,
has followed an amazing career path, from drummer for the top-of-the-charts
Steve Miller Band and PGA professional, to Chief Technical Officer of GolfSmith
International and then owner of his own successful golf company (www.w ishongolf.
com). Tom is a fount of knowledge about everything in golf. Michael Apostal and
xiii
xiv Acknowledgments
Chuck Ritter, principals of JAR Associates (www.jar.com), are in the very top ech-
elon of complex dynamic finite element analysis. Working with them on golf equip-
ment design was always a very great pleasure and provided special insights into the
club-ball interaction. David Marsh, high school physical education teacher, golfer,
and outstanding bowling coach, developed a method for grading the imbalance of
golf balls. Testing of unbalanced balls, which we carried out together with very sur-
prising results, is reported in Chapter 7.
Much of this book was written while looking out from my office onto the 16th
fairway of Spruce Creek Golf Course. This was interspersed with rounds of golf and
occasional lessons with Mike Kelly, Head Professional and Director of the Mike
Kelly Golf Academy (www.mikekellygolf.com). His very insightful comments on
the golf swing have been greatly appreciated.
Spruce Creek, Volusia County, Florida
July 2015
The Science of the
Perfect Swing
1
I N T R O DU C T IO N
Golf is unique among games for the sheer perfection of its range of equipment. In the
wider category of sports, we might compare the perfection of golf clubs and golf balls
to the bows and arrows in modern archery, or even to the range of high-technology
bicycles used for the different stages of the Tour de France. The changes in golf
equipment are less obvious, mainly due the work of the governing bodies of golf in
keeping the game as close as possible to its traditional form. However, the science
underlying these changes, including ones from the beginning of the 20th century,
are equally profound. Frank Thomas titled his book, on the evolution of the golf
equipment rules, From Sticks and Stones (Frankly Golf Publications, 2011). The title
was clearly meant to imply an evolution from the simple clubs and balls used at the
time of the first code of golf rules, established in 1744 by the Honourable Company
of Edinburgh Golfers. However, the title is much more prescient than it appears.
The skills exhibited in golf are based on our unique physical ability to maintain
a highly stable upright stance while the arms and upper body perform powerful ath-
letic tasks. This allowed our earliest ancestors to throw stones and swing sticks with
great precision; and later to wield hammers, axes, and even golf clubs with precise
purpose. Children as young as four have an innate ability to perform these tasks.
Some with exceptional athletic ability can drive a golf ball after watching a skilled
golfer perform the task. Most can throw balls quite accurately underarm; by the age
of six, without instruction, they naturally apply a stiff-arm swing, thereby reducing
the number of degrees of freedom for increased repeatability when asked to throw
underarm at targets (Jacques et al. 1989). A precisely repeatable swing is of course
of little value without the stability of the stance; in fact, none of the primates with
much greater strength and agility, but without a stable upright stance, can accom-
plish any of these tasks. This is the essence of the golf swing and the putting swing;
that is, maintain the greatest stability to keep the rotation “center” of the swing as
stationary as possible. This issue will be addressed in Chapter 3.
Before embarking on the story of golf science, we will consider some compari-
sons of performance from the car and aerospace world to put golf performance in
some perspective. We will introduce some science to do this, which is not difficult to
follow. All of the book chapters have easy-to-follow explanations presented in a sim-
ilar form. So if you follow these initial arguments, you will be able to follow the sci-
ence of all aspects of the game from the club swing, through the impact-generated
launch conditions into ball flight, to the final roll of the ball on the green. The sci-
ence story of the game is truly remarkable.
The back of each chapter contains the full “Details of the Modeling,” which necessi-
tates some heavier physics. It is kept to the minimum possible but is needed to provide
1
2 Introduction
150 mph 1,500 lb
15 seconds
Figure 1A Average force from back wheels to go from 0 to 150 mph in 15 seconds.
the necessary support for some of the very surprising conclusions that are reached
throughout the text. These sections comprise about one-third of the total text and can
be skipped without any loss of understanding of the science principles of the game.
For others with a deeper science-oriented background, the physics sections will
hopefully stimulate discussion, and in some cases, further investigations.
Let’s start with a fantasy trip, to explain one of the most important science
aspects of golf. A 15-second period of this trip is illustrated in Figure 1A. Assume
you are heading out to the golf course in your Ferrari 730 hp F12berlinetta. Starting
through an intersection is an empty stretch of open road, leading to the course. You
step on the gas and go from zero to 150 miles per hour in 15 seconds; and coming
up over a rise and around a curve, you do a 2 g deceleration, which with much tire
squealing brings you back down to 30 miles per hour in 120 yards for a smooth turn
into the golf club. Don’t worry if that’s just a dream, because you are about to do
something even more amazing on the first tee. To appreciate this fact, we will use
the car ride to understand the principles involved:
Recall that Newton told us that

force equals mass times acceleration,
and during the car sprint, your average acceleration was 10 miles per hour per
second. So we can form a second relationship. The final speed is the acceleration
multiplied by the time, so we can multiply our relationship by time and get
force × time = mass × acceleration × time = mass × final speed.
Now it’s just a matter of using the language of impact:
force × time = impulse; mass × final speed = momentum.
More generally, impulse is always equal to change of momentum.

The mass of our Ferrari is 1,500 kilograms (3,300 pounds), and the final speed
is 67 meters per second (150 miles per hour). The conversions to metric units are
given to simplify the calculations. The problem here is mass, and strictly speaking
weight is not mass. Physicists get concerned about the difference because weight is
the force of gravity on an object that deflects the weighing scale. In space, an object
can be weightless, but it still retains its mass. Since all golf takes place on the surface
of the earth, it’s actually OK to talk about a driver head “mass” of 0.44 pounds, and a
ball “mass” of 0.10 pounds. But, if we use pounds with Newton’s laws, then we have
3 Introduction
150 mph 1,500 lb
0.00045 seconds
Figure 1B Average force from the club face impact for the ball to accelerate from 0 to 150 mph in
0.00045 seconds.
to cancel out gravity, which makes for messy calculations. If we work with kilograms
and meters per second, then we just plug the values straight in. So we have “force ×
15 seconds” equals “1,500 kilograms × 67 meters per second”; or force = (1,500 ×
67)/15, which gives 6,700 newtons of force. One pound is equal to approximately
4.5 newtons, and if you remember the tale about Newton putting all this together by
watching an apple fall from a tree, then it makes sense because we typically get about
4 or 5 apples in a pound. Therefore the average force from the back wheels of your
Ferrari, accelerating it forward, is 6,700/4.5 = 1,500 pounds.
Now you are on the first tee. It’s going to be a great morning, you just hit one
of your best drives, and with some bounce and roll, it looks like it’s out there about
250 yards. To do this you must have hit the ball around 100 miles per hour, and it
took off at approximately 150 miles per hour. For comparison with your Ferrari,
it did this, from a standing start, as shown in Figure 1B, in just 0.00045 seconds!
Surprisingly, the force you applied to the ball was just about the same as the driv-
ing force produced by the wheels of your 730 hp Ferrari. In this case, we have a
ball mass of 0.045 kilograms (0.1 pounds) instead of 1,500 kilograms and a time
of 0.00045 seconds instead of 15 seconds. So, the calculation for force changes to
“force × 0.00045” equals “0.045 × 67”; or force = (0.045 × 67)/0.00045, which gives
6,700 newtons or 1,500 pounds, exactly as before. In this case, the force starts from
zero, reaches a maximum at just over 0.0002 seconds, and then decreases back to
zero at 0.00045 seconds as the ball leaves contact. The maximum force is thus close
to 3,000 pounds; for a better feeling of this magnitude, let’s say a “ton and a half.” So
just laugh when you see those four extra miles per hour claims for “low friction” tees,
which magically add 80 pounds to the maximum impact force.
We can break contact time down into two parts. At just over 0.0002 seconds,
the ball was compressed to about a 1 inch diameter imprint on the face, and the
club head had slowed down to about 82 miles per hour, with the ball travelling
with it at the same speed. If you had hit a trick sticky golf ball filled with thick
molasses, instead of a real golf ball with a rubber core, the process would end at
this point, and the compressed ball would remain compressed and sticking to the
face. At this point, the mass at the end of the shaft would be both the 0.44 pound
head and the 0.1 pound ball. From Newton’s laws it can be shown that, slowed
down to 82 miles per hour, the energy of the moving mass at the end of the shaft
would still be 130 foot-pounds. As you get swept off your feet trying to slow down
4 Introduction
this handful of energy, be consoled that you have just experienced zero coefficient
of restitution.
Instead, a real golf ball starts to spring off the face shortly after 0.0002 seconds;
the action of restitution or recovery speeds the ball up to 150 miles per hour, while
the reaction slows the club head down further to 67 miles per hour. Therefore, com-
pared to the 100 miles per hour strike against a stationary ball, at the end of the
impact we have a 150 miles per hour ball and a 67 miles per hour club head with a
speed difference between the two of 83 miles per hour. So the golf ball, compared to
the molasses-filled ball, has recovered 83 percent of the impact speed. In proportion
terms, 0.83 of the impact speed has been recovered in speed away from the moving
club face; of course, this is the coefficient of the restitution for the impact, abbrevi-
ated throughout the book as CofR. The energy of the 0.44 pound head travelling at
67 miles per hour has been halved during restitution to 65 foot-pounds, which in
slowing down just provides a smooth follow-through, finishing in an elegant stance
with the belt buckle facing the target—nicely done!
However, the beautiful swing would have accomplished very little without the
dimples on the surface of the ball, first introduced in the early 1900s and now per-
fected in shape through exhaustive computer modeling and wind-tunnel testing.
Using a smooth round ball, even with 0.83 CofR, your drive would probably have
carried about 130 yards, with maybe bounce and roll taking it another 20 yards. We
will wait until the next shot to figure that one out.
The bounce and roll must be better than you thought because you are now only
115 yards out—just made for your pitching wedge shot. Your magic golf day con-
tinues. The ball launches around 40 degrees and just seems to keep on climbing, up
to somewhere around 80 feet. From its steep descent, it takes a single short bounce
about 10 feet from the hole, follows a smooth spiral trajectory as it rolls off the sloped
right side of the green, and has slowed down to about 1 mile per hour before crossing
just inside the edge of the hole. Any faster and it would have escaped, but instead it
rolls half way around the edge of the hole without touching the flag stick, runs out of
steam, and topples in sideways—maybe time to just go and celebrate.
The eagle approach shot happened because you managed to put around
8,000 revolutions per minute backspin on the ball, and in so doing changed it to a
little Harrier jump jet, which just kept on climbing as it appeared to do. The story of
this propulsion system, with its associated low-drag performance, must wait until
Chapter 2. The ball actually landed with most of this backspin still in place, which
caused it to check quickly and start its slow forward roll.
One final comparison with the Ferrari, and then we will move onward. The
amazing amount of backspin is produced because the ball sticks to the club face,
exactly like the rubber car wheels stuck to the road in the braking turn. If we could
put a microphone on the wedge, then amplify and slow down the signal, we would
likely detect very high-frequency squealing as the ball is forced to rotate while grip-
ping the face. This is exactly the same as the Ferrari tires gripping the road around
the bend while different points on the area of contact must travel at different speeds.
An even better comparison is the squealing of rubber soled athletic shoes as they
brake and accelerate during the step, all the while gripping the floor. Not coinciden-
tally, the latest high-spin golf balls have covers made from the same urethane poly-
mers as indoor running shoes. We will see later in the book how this produces such
5 Introduction
amazing amounts of backspin that the ball actually comes off the face skidding, just
as if a powerful internal spring had been wound up and then released—which is in
fact what actually happens.
It’s time to describe some of the important equipment developments that have
taken the game “from stick and stones” to the exciting game of today. We will
restrict the discussion to approximately the last half-century. The interested reader
may wish to consult Thomas (2008 and 2011) for a more detailed discussion of this
topic and an enjoyable discussion of the early history of the game.
I MPORTA NT DEV ELOPM ENTS IN GOLF

EQU IPM ENT OV ER THE PAST 50 Y E A R S
A primary source for understanding the fundamental aspects of golf is still the super-
lative study, titled Search for the Perfect Swing, by Cochran and Stobbs ([1968] 1999).
This study, which started in the 1960s and involved the collaboration of academics
from a number of research institutions in Britain, resulted in a wealth of information
that, 50 years on, still forms a solid foundation for the scientific study of the game.
The title of the present work is intended, in part, to pay tribute to many of the expla-
nations in Cochran and Stobbs, which were made before the existence of invaluable
experimental data from wind-tunnel testing, ultra high-speed photography, laser
monitoring, and Doppler radar tracking systems. However, the title has more to do
with the performance of PGA and LPGA Tour players, which has been captured by
the Trackman Company, and which over the aggregate represent as near to perfect
golf as it is possible to measure. The overall goal of this text has been to lay down, as
well as possible, the science necessary to describe the ball striking of this group of
elite players, and at the same time to try to establish some measure of the improve-
ment in performance of golf equipment in recent time, particularly for the modern
titanium driver and the current generation of premium golf balls.
Since the time of the “perfect swing” study, according to Thomas (Just Hit It,
2008), the Technical Director of the USGA from 1974 to 2000, only three equip-
ment innovations have made a profound change to the game of golf. These are
perimeter weighting of club heads, the graphite shaft, and the incorporation of
spring effect into the face of hollow titanium drivers. To this list should be added
the latest premium golf balls, designed to perform quite differently with the driver
and the grooved irons. But even this can be attributed to the thin-face titanium
driver, which from the necessity of surviving multiple impacts could not have a
grooved face.
The use of perimeter weighting, producing a “cavity back,” increases the “moment
of inertia” of the club head, and so reduces club head rotation for off-center hits.
This was first established in a putter in 1966 by Karsten Manufacturing Company.
Named the “Anser” putter, it became the most successful putter ever both in terms
of sales and tournaments won. By the 1980s, the inventor, Karsten Solheim, had
introduced an even more significant innovation by using investment casting to cre-
ate cavity-back irons using the trade name “Ping.” According to Thomas (2011), the
second generation of cavity-back irons, the Ping Eye2, was used by Bob Tway in win-
ning the 1986 PGA Tournament. Now investment cast cavity-back irons are used
by almost all amateur players. The original forged “blade” irons, with the weight
6 Introduction
concentration at the sole, are still used by a majority of professionals and highly
skilled amateur players. The reason typically advanced for this is that the blade irons
have better “feel” and provide better ball flight control. However, it seems much
more likely that these clubs are more suited for the downward angle of attack on the
ball and the resulting extra compression of the ball between the club face and the
inertial resistance of the ground.
Investment casting is just the modern term for the lost-wax process, with auto-
mation applied for low-cost manufacture. It allows parts to be cast with a very high
level of precision, including the very smallest features. This allowed Ping to cast
U-shaped grooves into their irons, which eventually brought the Ping develop-
ment in grooves to the attention of the USGA. As mentioned previously, golf balls
need to grip the face for high-spin generation, and they do this much better with
U grooves than the V grooves traditionally machined into forged irons. The main
issue became that if the U grooves are large enough with a very small profile radius
at the top edges, then the club face could grip the ball enough in deep rough to
still spin the ball well. This threatened to remove much of the penalty from errant
approach shots, which finally resulted in the USGA fighting a multimillion dol-
lar lawsuit against the Ping Corporation. One of the best lines in golf literature
is in Just Hit It (2008) by Frank Thomas in which he writes, “I came to think of
Karsten (Solheim, Founder of Ping) as a friend, even though he sued me personally
for $100 million.”
It was only a matter of time before investment casting would be used to cre-
ate hollow drivers with greatly increased moment of inertia. The first successful
introduction of this driver design, using investment cast stainless steel, was by the
Callaway Golf Company in the late 1980s. This led them to dominate the golf club
market for 10 years. By 1995 they had gone from $5 million to $500 million in sales.
At this point they introduced the Big Bertha, investment cast using high-strength
titanium alloy to produce even larger driver heads with consequently greater
moment of inertia. The high strength-to-weight ratio of the titanium alloy was
accompanied by the unparalleled lightweight spring quality of the material. This
produced additional performance improvements, which will be discussed follow-
ing a brief diversion.
A BR IEF DISCUSSION OF MOM ENT

OF INERTI A (MOI) IN GOLF
We will pause for a “moment of inertia” to describe what it is and why it is so impor-
tant in club head design. It is a consequence of Newton’s laws that the resistance
to twisting, or rotational inertia, of any item is proportional to its mass and to the
square of its distance from the center of the rotation. This is why MoI is expressed
in units of gram-centimeter squared. It is easiest to demonstrate this with a piece of
pipe being rotated about the axis through the center. If the tube has a thin wall, then
all of the mass can be considered to be at a distance equal to its radius from the rota-
tion axis. This sets us up for a simple experiment. We apply a torque to the tube and
measure its rotation speed. Next we get a piece of tube with double the radius and
half the wall thickness, which therefore weighs the same. We find that for the same
7 Introduction
torque, it spins at only one quarter of the speed; that is, the radius doubled makes
the resistance to rotation increase by 2 squared or 4.
So the reason for making driver heads large is to move the 0.44 pound typical
head weight as far away from the center of the club head as possible. The torque pro-
duced by off-center ball strikes then rotates the head backward less and so projects
the ball forward more. The same applies to off-center strikes with irons and put-
ters where perimeter weighting has pushed the mass outward from the center, thus
increasing the MoI.
BACK TO THE HISTORY

The carbon fiber, reinforced, light-weight shaft was developed in 1969 by the
Shakespeare Sporting Goods Company, for whom Frank Thomas (2011) was then
Chief Designer. Since that time, carbon-fiber shafts have undergone continuous
development, particularly with respect to improved stiffness per weight; shafts are
now available in the 50 to 60 gram range. Despite claims to the contrary, this has
resulted in only the tiniest of increases in swing speed, and likely no increase in ball
speed, when compared to even the heavier hollow steel shafts, which are now almost
never used in drivers. However, the fiber layup construction of shafts has allowed
the stiffness distribution along the shaft to be customized for different player abili-
ties. In particular, regions of increased flexibility are positioned toward the end of
the shaft to promote forward rotation of the club head for increased “dynamic” loft.
These issues will be discussed fully in Chapter 3.
As mentioned previously, in 1995 the Callaway Company introduced the first
high-strength titanium alloy driver with a thin diaphragm-spring face. It is inter-
esting that Callaway adopted titanium alloy as a means of further increasing the
volume, and therefore the moment of inertia, of their hollow Big BerthaTM range of
drivers while keeping the weight within the accepted 0.44 pound range. Popular
lore has it that the increased ball speed that resulted from the excellent spring qual-
ity of the titanium alloy, and allowed the thin driver face to act like a trampoline,
was an unexpected bonus. This may in fact be true, because Callaway did not file a
patent on the trampoline effect, with disastrous consequences for the company. By
1997, Callaway, after spectacular growth from a few million dollars to almost one
billion dollars in sales, was selling more of all golf products than almost all other
golf companies combined. However, by 1998, eighteen oversize titanium drivers,
from thirteen manufacturers, were competing for market share, and the Callaway
market share was in decline. It is interesting to conjecture that, with a robust patent,
Callaway would likely have dominated the entire golf market up to this time.
As Thomas (2011) reports, one of Callaway’s main competitors informed the
USGA that they were finding unusual performance results in their testing of the
titanium club. This performance improvement was the subject of a 2001 article
by Michal and Novak, the main focus of which was to demonstrate that even bet-
ter performance might be possible through the use of amorphous metals, often
called liquid metals, in driver head design. Michal and Novak used a measure of
trampoline, or more correctly diaphragm, spring quality (Dieter 1983, Ashby
2005, Dewhurst and Reynolds 1997) to compare different materials for club head
design. Michal and Novak showed that liquid metals, particularly VitreloyTM by
8 Introduction
Liquidmetal Technologies (2011), had the potential for superior golf club face per-
formance. However, attempts to produce a golf club with this material were unsuc-
cessful, apparently through a combination of fatigue-failure issues and difficulties of
bonding the faces to driver bodies. When the list of candidate materials is reduced
to those already applied successfully in club heads, titanium and beryllium copper
are shown to be equally best in class for driver diaphragm spring faces. However,
when the desire is to combine large head size with diaphragm spring quality, mate-
rial density must also be considered, which puts high-strength titanium alloy alone
at the top of the performance list for driver head design (see Chapter 2 of Design
for Manufacture and Assembly, Boothroyd, Dewhurst, and Knight 2011). It is inter-
esting to note that later on, the Callaway Corporation, apparently forgetting the
reason for the pioneering breakthrough they had made, developed a carbon-fiber
reinforced driver head and face for which diaphragm spring quality is far inferior to
high-strength titanium, which was unsuccessful in the market place.
Much more recently, golfer performance has been considerably enhanced
through the invention by Fredrik Tuxen (TrackMan A/S, U.S. Patents 2007,
2008, 2009) of a radar monitoring system of ball flight, and his founding of the
TrackmanTM Company in Denmark. Trackman performs precise measurements of
club head and ball velocities, launch angles, ball spin rates, and the tilt angles of the
ball spin axis, in addition to monitoring the complete ball trajectory. It has proved
to be a powerful training tool, allowing players to try different combinations of club
loft angle, shaft stiffness, and strike attack angle, along with variations of their swing
mechanics, to obtain optimum conditions for ball flight. This complemented the
development of new multilayer balls designed to allow high launch angles in driv-
ing without the generation of excessive backspin. Thomas (2008) envisioned these
golf ball developments and, in 1999, proposed an optimized overall distance rule.
The intention was to have a maximum overall distance for all balls but using the
particular optimum launch conditions for each manufacturer’s ball. This proposal
was never adopted, but a rule for overall distance under a standard set of ball flight
conditions was adopted in 2004 by both of the governing bodies of golf, the R&A
(Royal and Ancient Golf Club of St. Andrews) and the USGA, referred to with
slight abbreviation going forward as R&A/USGA.
Tuxen has published a number of articles (2007–2010) over recent years con-
taining comprehensive sets of data on ball launch conditions and ball flight, together
with a number of empirical rules relating fundamental aspects of ball striking to ball
flight. With the permission of Tuxen, these have provided invaluable data for this
work, in some cases providing validation of analytical predictions, for others allow-
ing estimations to be made of model input parameters. Reference will be made to
individual newsletters in later chapters.
For a given impact velocity with a given club head mass, the principal reasons
for loss of distance when driving a golf ball are the loss of energy in the golf ball
itself as well as the twisting of the club head resulting from off-center hits. A golf
ball is far from perfectly elastic. When deformed during impacts of approximately
100 miles per hour, with even the latest high-performance balls and titanium driv-
ers, 30 percent or more of the energy transferred to the ball in deformation is typi-
cally lost to internal friction in the ball polymer structure rather than recovered
in extra ball speed. The original developers of golf equipment, contending with
9 Introduction
much less efficient balls than those currently used, presumably discovered by trial
and error that greater distance could be obtained by striking the ball with a softer
club, which by sharing the deformation of impact would necessarily reduce the
ball’s deformation and consequently the amount of energy loss. The best material
for this purpose, with sufficient strength to sustain the impact loads, was found to
be hardwood, shaped so that the face cuts across the growth rings. This ensures
that the stiffness in the direction of impact is the lowest possible; in fact, less than
8 percent of the stiffness at right angles to the wood fibers, as, for example, that for
a baseball bat (Wood Handbook 1999). Compression tests carried out by me, on
specimens cut normal to the face of persimmon wood blanks from the Louisville
Golf Company, gave a stiffness modulus in the direction of impact of only 90,000
pounds per square inch. This is considerably less than the modulus value of 130,000
pounds per square inch used by Michal and Novak (2001), on the basis of which
they predicted that the maximum force from a 100 miles per hour impact, with
a wooden driver, would be 3,300 pounds; even less than their predicted value of
3,500 pounds for a diaphragm face modern titanium driver. However, Michal and
Novak used a “static” rather than “dynamic” solution, so we should only accept the
relative magnitudes of the two values.
The problem with the performance of hardwood drivers is that significant energy
is lost in internal friction between the wood fibers as well as in friction inside the
ball; with the modern ball, this may be a zero-sum tradeoff. Moreover, as discussed
previously, the distribution of the approximately 0.44 pounds of mass throughout
the bulk of the club head gives a very low moment of inertia compared to modern
hollow titanium drivers. Following the recognition of the increased performance of
the hollow titanium Callaway driver heads in the late 90s, the R&A/USGA estab-
lished a new equipment rule that the coefficient of restitution resulting from a 109
miles per hour impact with a golf ball should not exceed 0.83. To enforce this rule,
both the clubs and golf balls had to be subjected to independent tests. Both tests
use ball cannons equipped with ballistic screens to measure the rebounding ball
velocity after impact. For the ball test, a standard titanium alloy circular test plate
was adopted. The plate is 4 inches in diameter, with a 3-inch diameter inner region
0.115 inches thick, to form a spring diaphragm; and with an outer thick flange to
provide a total weight of 0.44 pounds, equal to the typical driver head weight. The
plate was made from the same titanium alloy and designed to give the same impact
performance as the Callaway titanium drivers already in the market. Balls exceed-
ing 0.83 CofR when fired at the plate at 109 miles per hour are deemed noncon-
forming. With this test in place, balls with the maximum 0.83 CofR value could
then be fired at the sweet spot (the point directly in front of the center of mass) of
driver heads to check that the CofR did not exceed 0.83. In addition to the CofR
rule, the R&A/USGA now restrict the volume of drivers to 460 cubic centimeters
and the moment of inertia component about the vertical axis (vertical MoI) to 5,700
gram-centimeter squared. Almost all manufacturers now supply their drivers with
the maximum volume. However, attempts to increase the vertical MoI into the
5,000+ gram-centimeter squared range, with rectangular-shaped driver head pro-
files, did not meet with success in the market; and the manufacturers seem to have
now settled on vertical MoI values in the area of 4,600 gram-centimeter squared for
460 cubic centimeter clubs.
10 Introduction
While much attention has been focused by the R&A/USGA on the impact
efficiency of clubs and balls, surprisingly nothing in the rules concerns the aero-
dynamic performance of golf balls, except the overall distance restriction. At the
time of the introduction of the CofR rule in 2002, the R&A/USGA overall dis-
tance standard (ODS) required that a ball struck by a conforming (0.83 CofR)
driver at 109 miles per hour should not exceed 297 yards in total distance, includ-
ing bounce and roll, under carefully controlled standard conditions. In 2004, to
reflect the increased driving distance of professional golfers, the test was extended
to the requirement that a total distance of 320 yards should not be exceeded with
an impact of 120 miles per hour. Currently, the average PGA Tour player drives the
ball for a carry distance of 269 yards, with an additional average bounce and roll
distance estimated in Chapter 3 to be 40 yards. This average overall distance of 309
yards is achieved at an average impact speed of 112 miles per hour, so it must be
assumed that the longest hitters are routinely driving further than the R&A/USGA
limit. The loop hole in the rules is that the ODS is for a defined launch angle and spin
rate of the ball. Therefore, manufacturers are free to design balls that may exceed the
ODS but with different optimum launch angles and spin rates than defined for the
ODS. If Thomas’s proposed Optimized ODS had been adopted, every new ball on
Tour would have been wind-tunnel tested to determine its optimum launch condi-
tions, which then would have been used to assess overall distance.
Chapter 2 is devoted to the aerodynamics of golf ball flight. However, it is fitting
to provide a short historical introduction to this work here.
From the substantial literature on golf ball flight, two papers are of particular
interest for the combination of careful experimentation and extensive databases
they provide. The first is an exhaustive experimental assessment of the perfor-
mance of golf balls, carried out in 1976 at Imperial College, London, by Bearman
and Harvey (B&H; 1976). They carried out a series of wind-tunnel tests of dimpled
balls, equipped with internal motors to provide varying spin rates. The authors mea-
sured lift and drag forces and produced tables of lift and drag data covering the range
of golf ball speeds and spin rates experienced in play. They tested two different 1976
Uniroyal brand balls: one with circular and one with hexagonal dimples. They dem-
onstrated close agreement between ball flight distances of the hexagonal-dimple
balls, as predicted using their wind-tunnel data, and measured sets of launch angle,
ball velocity, initial ball spin, and flight distance. Because of the distance validation,
this extensive data set provides a benchmark in time against which to compare the
aerodynamic performance of the modern golf ball. Two decades later, Smits and
Smith (S&S; 1994) used a higher-speed wind tunnel, mounted golf balls on slender
spindles, and measured lift and drag forces and the rate of spin decay, for a wide range
of air speed and spin rates applicable to the driver through short iron shots. Smits
and Smith obtained lift data that “in all respects are similar to the data obtained by
Bearman and Harvey,” although their values were higher than the hexagonally dim-
pled ball data of B&H by a constant increment of 0.04. S&S also obtained “broad
agreement” with the drag coefficient data of B&H, although their results indicated a
“stronger dependence on spin-rate.” This is to be expected; if golf ball designers had
managed to tweak the dimple spacing and profiles to achieve greater lift, then it is
likely this would have been accompanied by an added amount of induced drag. For
the present study, the B&H lift and drag data were incrementally adjusted by the
11 Introduction
40
(yards) 30
20
Callaway robot tests
10 B&H Hexagonal ball data
S&S ball driving data
0
0 50 100 150 200 250
(yards)
Figure 1C Trackman measurements of robot driving tests with predictions using the Bearman
and Harvey (1976) lift and drag coefficients and the Smits and Smith (1994) ball aerodynamic
drive model; data on the Great Big Bertha II courtesy of Callaway Corporation.
writer until best agreement was obtained over the range of PGA and LPGA Tour
player trajectories. For the drive, the S&S model was found to give better agreement
with current premium balls. The nature of the lift and drag forces acting on the ball
is explained in Chapter 2.
Figure 1C shows actual ball flight from robot testing, as monitored by the
Trackman radar system. The tests were carried out, as a demonstration for the
writer and one of his colleagues, by the Callaway Golf Corporation using a con-
forming golf ball and the Great Big Bertha II titanium driver. From an average of ten
100 miles per hour hits, the measured results of the drive were: initial ball
speed = 154.3 miles per hour; launch angle = 12.9 degrees; initial backspin = 3,106
revolutions per minute; maximum height = 38.3 yards; and carry distance = 245.1
yards. The dashed line (in Figure 1C) shows the predicted trajectory obtained
by the writer, using the original Bearman and Harvey lift and drag data for the
hexagonal-dimple, 1976 Uniroyal ball. Calculations were carried out as described
by Bearman and Harvey (1976), but using a spin decay rate as later determined
experimentally by Smits and Smith (1994).
With very minor adjustments, the S&S aerodynamic model simulated the
Trackman trajectory almost perfectly as shown by the square symbols in Figure 1C.
Before concluding this chapter, it is difficult to overlook the quite amazing per-
formance of the 1977 Uniroyal hexagonal-dimple ball. With all of the advances
in the understanding of fluid flows, and the development of sophisticated com-
putational fluid dynamics software systems for advanced aerospace design over
the last two decades, it would have been natural to expect a little more. However,
this comparison does not tell us anything about the ball striking behavior of
the Uniroyal ball compared to the modern premium ball. It would be wrong to
assume from these flight comparisons that the Uniroyal ball would perform as
well in actual play, and there is ample evidence to show that it certainly would not.
Likewise, the robot testing, with ten perfectly centered face strikes, gives no indi-
cation of its performance in actual play, particularly with a high-handicap player
whose ball strikes would be scattered quite widely from the face center. Dealing
with issues such as these, and of course different issues arising for every aspect of
the game, is the goal of this work.
As mentioned at the beginning of this Introduction, Trackman data on PGA and
LPGA Tour players will be used as the science “examples” throughout the text. This
Table 1A Average Trackman test results and modeling for PGA tour players
Club Attack Ball speed Vertical Spin rate Max Landing Carry
speed angle (mph) launch (rpm) height angle (deg) (yards)
(mph) (deg) (deg) (yards)
Driver 112.0 −1.3 165 11.2 2,685 31 39 269

3-wood 107.0 −2.9 158 9.2 3,655 30 43 243
5-wood 103.0 −3.3 152 9.4 4,350 31 47 230
Hybrid 100.0 −3.3 146 10.2 4,437 29 47 225
3-iron 98.0 −3.1 142 10.4 4,630 27 46 212
4-iron 96.0 −3.4 137 11.0 4,836 28 48 203
5-iron 94.0 −3.7 132 12.1 5,361 31 49 194
6-iron 92.0 −4.1 127 14.1 6,231 30 50 183
7-iron 90.0 −4.3 120 16.3 7,097 32 50 172
8-iron 87.0 −4.5 115 18.1 7,998 31 50 160
9-iron 85.0 −4.7 109 20.4 8,647 30 51 148
PW 83.0 −5.0 102 24.2 9,304 29 52 136
PW = pitching wedge.
Data reproduced with permission from Trackman A/C, Vedbaek, Denmark.
Table 1B Average Trackman test results and modeling for LPGA tour players
Club Attack Ball Vertical Spin rate Max Landing Carry

speed angle speed launch (rpm) height angle (deg) (yards)
(mph) (deg) (mph) (deg) (yards)
Driver 94.0 3.0 139 14.0 2,628 25 36 220

3-wood 90.0 −0.9 132 11.2 2,705 23 39 195
5-wood 88.0 −1.8 128 12.2 4,501 26 43 185
7-wood 85.0 −3.0 123 12.7 4,693 25 46 174
4-iron 80.0 −1.7 116 14.3 4,801 24 43 169
5-iron 79.0 −1.9 112 14.8 5,081 23 45 161
6-iron 78.0 −2.3 109 17.1 5,943 25 46 152
7-iron 76.0 −2.3 104 19.0 6,699 26 47 141
8-iron 74.0 −3.1 100 20.8 7,494 25 47 130
9-iron 72.0 −3.1 93 23.9 7,589 26 47 119
PW 70.0 −2.8 86 25.6 8,403 23 48 107
13 Introduction
data is presented here in Tables 1A and 1B as an easy to find reference for recurring
discussions of Tour player performance.
Ball flight or “carry” distances shown in the tables are not as large as might have
been expected. This suggests that for a player to be “on Tour,” the peak level of
achievement in the game, the major requirement is not ultra-long hitting, so it can
only be consistency of ball striking, coupled with accuracy on the putting surface
and of course a high tolerance for stress. However, we should certainly not underes-
timate the importance of distance combined with accuracy off the tee, for separat-
ing the very good from the very best.
On to the ball that actually does fly by creating its own propulsion system.
2
B A L L F L IG H T
As we know, golf ball velocities are high. They fly off the golf club face at speeds
ranging from about 50 miles per hour for partial wedge shots to 180 miles per hour
for the longest drives. They descend back to earth at speeds up to 70 miles per hour.
However, these high speeds do not explain for the amazing distances a golf ball can
be driven, the ease with which the same ball can be lofted over high trees, or the
manner in which they can be stopped on fast greens in a single bounce. A ball with-
out the elastic resilience of the modern ball, and particularly without its complex
dimple pattern, would do none of these things. How it is achieved is the subject of
this chapter and Chapter 3. Before explaining the science underlying the flight per-
formance of the modern ball, a preview of the game of golf without dimples might
be useful.
THE GA M E W ITHOUT DI MPLES

Figure 2A shows a side elevation view of the average trajectory of PGA Tour play-
ers. The initial ball speed is 165 miles per hour, and the initial backspin rate is
2,685 revolutions per minute. It is launched at an initial angle of 11.2 degrees and
continues to climb at that rate, defying gravity, for about the first 130 yards of its
flight. The flight path appears to be a straight line upward over this range, although
closer examination shows that it actually curves upward. This allows the ball to rise
to a height of 90 feet at around 170 yards, or at about two-thirds of the flight dis-
tance. From this height, the descent takes a further 100 yards for a carry distance of
269 yards. In contrast, the lower dashed curve is the flight distance of a smooth golf
ball given the same launch conditions, that is, struck by a driver with the same loft
and at the same impact speed of 112 miles per hour.
Of course if the game was played with smooth golf balls, then the launch condi-
tions to obtain maximum distance would be different. In fact, the likely range of
optimum launch conditions would be as shown in the upper dashed trajectories.
The lower curve is for a launch of 23 degrees with the same 165 miles per hour
speed and 2,685 revolutions per minute backspin. The upper one is for a launch at
27 degrees with the backspin rate decreased to only 1,000 revolutions per minute.
This one carries 167 yards, still 100 yards less than the average PGA Tour distance.
To take this comparison one historical step further, we might imagine what
the game might have become if the first golf balls had been smooth molded ones,
and the smooth ball had become the required standard. Without further regula-
tion of the ball, except the rule that the ball should be smooth, they would likely
have improved in springiness to get the higher ball speeds of today. However, to
achieve the maximum drive distance of approximately 170 yards, the ball would
15
16 Science of the Perfect Swing
40
smooth ball optimum launches
30
(yards)
20 average PGA drive

10
smooth ball PGA drive
0
0 50 100 150 200 250
(yards)
Figure 2A Average PGA Tour player performance with smooth and regular golf balls.
have to be teed up about 6 inches off the ground and struck outside of the front
foot to launch at a high angle using a driver having only a small loft to obtain a very
low backspin rate. The average player would drive the ball only about 125 yards,
somewhere around the distance of the average amateur 8- or 9-iron in the modern
game. I think we get the picture that the game would be a whole lot less exciting.
So how does the modern ball with its precisely optimized dimple pattern make
such a difference? To answer this question, we will first consider the golf ball as sim-
ply a passive, non-spinning projectile passing through the air. After this, the effect
of giving the ball backspin to become an active flier will be considered. The effect of
the latter is analogous to winding up the propeller on a rubber band powered model
aircraft.
THE GOLF BA LL AS A PROJECTILE

The fundamental effect of dimples is illustrated in Figure 2B—which gives the rela-
tive drag force acting on a smooth ball, a bullet-shaped projectile, and a golf ball at
speeds up to 250 miles per hour. This is strictly a comparison of the relative forces
for the different speeds and hides the fact that the drag force increases with the
square of the ball speed. So, if the speed through the air doubles from 75 miles per
hour to 150 miles per hour, the actual drag force on these objects increases four-
fold. This is the main reason why, with all of their athleticism and raw power, the
longest drivers of the ball only manage to carry the ball about twice as far as the
newest beginner of the game. The relative drag force of Figure 2B simply takes out
the effects of speed squared.
As seen in the top curve of Figure 2B, the relative drag force on a smooth ball has
an approximately constant value near 1.0 over the range of ball speeds involved in
most golf shots: approximately 50 to 200 miles per hour. In contrast, over the same
speed range, a dimpled golf ball is subjected to a relative drag force less than half of
that for the smooth ball. These differences are caused by the differences in the flow
of the air around the ball and particularly the manner in which it separates from the
ball as the ball passes by.
The upper illustration in Figure 2C shows how the air would flow around a
smooth golf ball at these speeds. As the ball pushes forward, the air is forced to
speed up sideways around the ball. The layer of air near to the ball on the leading
surface gains energy of motion, so-called kinetic energy, and so it must lose energy
in a different form to conform to the law of energy conservation. It suffers a net
loss of internal energy. Of interest here is the drop in the pressure of the moving
17 Ball Flight
1.0
smooth
ball
0.8
Relative Drag Force bullet

0.6
0.4
golf
ball
0.2
50 100 150 200

Ball Speed, mph
Figure 2B Relative drag of a golf ball, smooth sphere, and bullet with same diameter.
air. It is easy to imagine the molecules of air separating further as they speed up
around the outside of the ball, therefore becoming less dense with lowered pres-
sure. Maximum air speed and minimum pressure occur as the air reaches the full
diameter of the ball. Here the lower-pressure air must rejoin the higher-pressure
air behind the ball. This causes a chaotic mixing of the lower- and higher-pressure
air, resulting in turbulence and the formation of a wake.
The process thus dumps lower-pressure air behind the ball while the ball is push-
ing against higher-pressure air in front. This pressure difference between the front
and back is what is referred to as drag. So to reduce drag, we need the high-velocity
air passing over the ball to travel down the back surface of the ball. If it does this, it
will lose speed and by conservation of energy regain some of the pressure. This will
Figure 2C Separation of air flow behind smooth and dimpled non-spinning balls.

result in a smaller wake, with pressure nearer to the air ahead of the ball, therefore
reducing drag. As shown in the lower illustration of Figure 2C, the dimples do this
by “tripping” the layer of air passing over the ball, giving rise to a thin turbulent
layer. This layer mixes with adjacent thin layers of higher-pressure air and is able to
pass downward over the rear surface of the ball before separating in a much reduced
wake. This behavior, of the airflow remaining in a thin layer around to the back of the
ball, starts at a critical airspeed, which depends on the ball diameter and the surface
roughness. For the golf ball diameter, the critical speed with a dimpled surface can
be seen on the lower curve in Figure 2B to be approximately 40 miles per hour. For
a smooth sphere, following the upper curve to its smallest value in the lower-right
corner, it can be seen that the critical speed is approximately 225 miles per hour.
So now we can see that dimples do not universally lower drag. It would seem very
unusual that a rougher surface should always slip more easily through the air than a
smooth one. Instead, the dimples simply move the critical speed lower so that golf
is played in the so-called super critical region of ball flight. Notice finally that in the
supercritical region for a smooth ball, starting at 225 miles per hour, the relative
drag force is less than half of that for the dimpled ball, which puts things back into
common-sense order. For a game involving a more effective launch device, such as
a catapult firing balls at speeds above 225 miles per hour, smooth balls would fly
much further than golf balls of the same diameter and weight.
As for the bullet shape, the reduced-pressure layer formed around the nose can
pass smoothly along the sides, increasing in pressure before joining the air behind
the bullet. It is only included in the diagram (Figure 2B) to indicate the surpris-
ing efficiency of the dimpled golf ball. The effect of surface roughness on golf ball
flight is so dramatic that it is no surprise that the smooth, molded gutta-percha balls,
introduced in the 1850s, did not fly as far as the stitched leather case balls they were
intended to replace. However, the “guttie” balls were found to carry much further
when they became roughened in play. It was a small step from this realization to the
molding of marks into the ball, eventually becoming a pronounced dimple pattern,
and, in the past few decades, finely optimized dimple forms using sophisticated
fluid dynamics software.
THE GOLF BA LL TA K ES FLIGHT

The term “ball flight” was deliberately omitted in the last section. In this section, we
see that the golf ball, struck by a lofted club, really does fly, and by virtue of the same
principles that allow aircraft to fly. In the case of a golf ball, it results from backspin
and is enhanced by dimples. The discussion of drag in the last section did not take
account of the effect of the spinning ball. The intention was to explain in the simplest
way the basis for aerodynamic drag and the mechanism by which dimples reduce
drag. However, in all golf shots, the ball is launched with quite amazing amounts of
backspin. This section is concerned with the fact that backspin changes the ball into
a flying object, rather than just a passive projectile moving through the air.
For the average drive, the ball is spinning at about the speed of a car engine cruis-
ing at 55 miles per hour on the highway. In a full wedge shot, it’s spinning at the
rotational speed of a race-car engine at top speed on the track. This high backspin
is produced by the glancing blow of the lofted (wedge-shaped) club. It is absolutely
19 Ball Flight
lift
drag
down
draft
Figure 2D Rotating golf ball acting like a spherical fan.
central to the game of golf. It is the combination of backspin and dimples that cre-
ates the special excitement of the game. The next time you hear a golf commentator
say “he must have put a lot of topspin on that one,” just smile. An almost miss-hit,
glancing blow across the top of the ball is the only way to create topspin.
The backspin of the ball speeds up the layer of air flowing up and over the top
of the ball surface. The ball thus acts like a spherical fan, driving the layer of air,
through which it is passing, slightly downward. This action is greatly enhanced by
the presence of the dimples. Because by Newton’s laws, action and reaction are equal
and opposite, this downward push on the air must be accompanied by an equal and
opposite upward push on the ball. This is the aerodynamic lift that keeps the ball
airborne much longer than would be the case if it was merely a projectile. This pro-
cess is illustrated in Figure 2D. The source of this lifting force is easily defined from
our previous discussion. We know that the air must speed up as it is forced to move
around the ball. The rotating ball causes the air layer above the ball, assisted by the
moving upper surface, to have higher speed than the air layer below the ball, which
is resisted by the forward-moving lower surface. Since speed increase causes pres-
sure drop, the pressure on top of the ball must therefore be lower than the pressure
below the ball. The net force on the ball is thus upward, or more correctly, at right
angles to the ball’s direction of travel.
This effect is precisely analogous to the effect of an aircraft wing, which bulges
upward on the top surface, forcing the air to go faster over the top than across the
bottom. This causes a larger pressure drop on top and a net upward pressure on the
wing from the higher pressure below.
Data from wind-tunnel testing was used to illustrate the relationship between
the lifting force for a dimpled ball compared to a smooth ball. Just as for the drag
force, the lifting force varies with the speed squared. However, it also increases with
the ratio of the rotational surface speed of the ball to the forward speed. Because
we are considering a fixed ball size, we can represent this ratio in the golf units of
rpm/mph, as shown in Figure 2E. This ratio increases with club loft, with the high-
est ratio for loftiest wedge shots. From published data on PGA players, the average
1.1
0.9
Relative Lift Force all

0.7 fb
gol
0.5
all
othb
s mo
0.3
0.1
0 50 100 150 200 250
Backspin Rate/Ball Speed (rpm/mph)
Figure 2E Relative lift of a golf ball and same-diameter smooth ball.
value for a 7-iron shot as the ball leaves the face is 7,000 rpm/120 mph = 58. For
this case, the relative lift force from Figure 2E is approximately 0.6. In contrast, for
PGA players, the average drive ratio is 2,700/165 = 16, with a relative lift factor from
Figure 2E of approximately 0.3.
With a smooth ball, the 7-iron shot would produce a relative aerodynamic lift
of only about 0.15; for the drive, the “lifting” force can actually become negative
and pull the ball downward, considerably shortening the flight. The reason for this
unlikely behavior can be seen in Figure 2B, where the relative drag can be seen to
start decreasing slowly around the 165 miles per hour ball speed of the PGA Tour
average drive. With backspin, the bottom surface is travelling faster through the air
than the lower surface, so more favorable air flow conditions exist underneath the
ball. In particular, the airflow below can now separate later than above, the opposite
of that shown in Figure 2D.
Recall that the actual lift force increases with the ball speed squared; so, to com-
pare the lifting force for the drive and the 7-iron shots, we need to take the ratio
of 0.3 × (1652) to 0.6 × (1202), which gives 0.95. The actual lift force acting on the
ball as it launched in the PGA average drive can be calculated from wind-tunnel
testing data to be 0.153 pounds. For the PGA 7-iron shot, the calculated value is
0.166 pounds. The ratio of these two values is 0.92, in close agreement to our approx-
imate estimate from Figure 2E. The weight of a golf ball is 0.10 pounds, so these lift
forces are approximately 1.5 times the ball weight. Thus, in the initial phase of ball
flight, the ball is actually climbing at a progressively higher rate. This can be seen
very clearly in the long iron shots of powerful players when viewed from behind. The
ball can be seen to curve upward slightly in the first half or so of the flight.
It will be shown later in the chapter that the relative drag force, like the lift force,
is also very strongly dependent on the ball’s backspin/speed ratio. That is, a golf ball
travelling at 100 miles per hour with a backspin rate of 4,000 revolutions per minute
will experience approximately the same relative drag force as a ball at 50 miles per
hour with 2,000 revolutions per minute backspin. This is interesting from a physics
perspective, but for our present purposes, it allows us to present all of the necessary
21 Ball Flight
1.1
l
bal
oth
0.9 mo
Relative Drag Force

s
ll
0.7 f ba
gol
0.5
0.3
0 50 100 150 200 250
Backspin Rate/Ball Speed (rpm/mph)
Figure 2F Relative drag of a golf ball and same-diameter smooth ball.
drag data on a single curve. This is illustrated in Figure 2F, together with a short
section of the relative drag data for a smooth ball, which can be seen to be approxi-
mately twice the value as for the dimpled ball.
To complete our understanding of golf ball flight, we need to know how spin rate
decreases during ball flight. Spin rate is affected only by friction with the air. Unlike
lift and drag, which are determined by pressure variations around the ball and are
powerful forces, the braking torque applied by the air to the ball rotation is relatively
small. The change of spin rate during flight is thus also relatively small. The rate of
braking increases with the combined effect of both ball speed and spin rate and can
be represented by the approximate relationship, which was converted to golf units
from the work by Smits and Smith (1994):
Spin-rate loss/second = 0.0004 × (speed) × (spin rate)
For the PGA average 7-iron shot, the initial velocity and spin rate are 120 miles
per hour and 7,097 revolutions per minute. The initial rate of loss of spin rate is
therefore 0.0004 × 120 × 7,097 = 340 revolutions per minute per second.
Using this information, and the relative drag and lift data in Figures 2E and 2F,
it is relatively straightforward to construct the complete ball trajectory given the
launch conditions. The procedure for doing this is described in detail later in the
chapter using drag and lift data tables. In general terms, the process is carried out
using the following six steps:
1. From the launch conditions determine the initial values of drag force, lift force,
and rate of spin decrease. Drag and lift forces are obtained by multiplying the
relative drag and lift force values by the ball cross-sectional area, the density of
air, and the ball speed squared. The first two of these parameters were canceled
out of the relative force calculations. The rate of spin decrease is obtained from
the “spin-rate loss/second” formula.
2. Apply Newton’s laws to obtain the deceleration of the ball over a small time incre-
ment (0.001 second increments are sufficiently small) to obtain the changed ball
speed; of course, include gravity as well as lift and drag.
7000 130
6800 110
Ball Speed, mph

Spin Rate. rpm
6600 90
spin
6400 rate 70
ba
ll s
pe
ed
6200 50
6000 30
0 1 2 3 4 5 6 7
Flight Time, seconds
Figure 2G Changing ball speed and spin rate values during the ball flight of the average PGA
Tour player 7-iron shot.
3. Calculate the average of the ball speeds at the beginning and end of the time
increment. Multiply this average speed by the time increment to determine
where the ball has moved by the end of the time increment. The changed slope of
the ball trajectory over the time increment emerges from these calculations.
4. Multiply the rate of spin decrease by the time increment to obtain the decreased
spin rate value.
5. Determine the drag and lift forces for the changed velocity and spin rate.
6. Repeat steps 2, 3, 4, and 5 for the entire flight. Stop the process when the ball has
reached the ground!
The results of carrying out this process for the average PGA 7-iron shot are
shown in Figures 2G to 2I.
Figure 2G shows the changing ball speed and spin rate. The rate of decrease of
spin rate and speed are largest at the beginning of the flight. Starting from 120 miles
per hour, the ball speed reaches a minimum value of 40 miles per hour in 4 seconds
at the top of the trajectory. It then accelerates back to approximately 55 miles per
hour in the descent. From an initial backspin rate of 7,000 revolutions per minute,
the final spin rate of 6,100 revolutions per minute is a loss of only 13 percent. This is
the reason why it is possible to land and stop such shots on the green. If the spin rate
decayed substantially during flight, then all shots would have to be played to bump
and run from in front of the green. In contrast, the average PGA professional drive
launches with a spin rate of 2,700 revolutions per minute, which decays to 2,100
revolutions per minute over a flight time of 6.6 seconds.
The lift and drag forces acting on the ball during its flight are shown in
Figure 2H. The black lines correspond to the PGA average 7-iron shot, and the gray
lines are for the average drive on the PGA Tour. The forces in this plot are divided by
the ball mass (0.1 pounds), and so the scale is equivalent to the number of “g”s acting
on the ball. Note that the drag and lift forces are in line with and at right angles to
the trajectory throughout the flight, as shown in the inset ball figure.
It can be seen that for the drive, the drag force on the ball starts at 2.5 times
its weight and the lifting force starts at 1.6 times the weight. For the 7-iron shot,
23 Ball Flight
2.5
Lift and Drag forces/ball weight

drag forces lift
d
spee
1.5
spin
drag
weight
lift forces
driver
0.5
7-iron
0
0 1 2 3 4 5 6 7
Flight time, seconds
Figure 2H Changing lift and drag forces values during the ball flight of the average PGA Tour
player 7-iron shot and drive.
the drag and lift forces are almost identical through the flight and start at 1.5 times
the ball weight. These are substantial forces. They are proportional to the speed
squared; so as the speed drops rapidly in the first 2 seconds due to the high drag
force, the drag and lift forces decrease at a proportionally higher rate.
When following the preceding steps 1 through 6, we have also calculated the
positions of the ball throughout its flight. The resulting trajectories for the average
PGA drive and 7-iron shot are shown in Figure 2I.
The results in Figure 2I are in very good agreement with the trajectory data
published by the Trackman Company from thousands of measurements of PGA
Tour players on the practice grounds. The average heights and carry distances from
Trackman measurements are 31 and 267 yards for the drive and 32 and 172 yards
for the 7-iron, respectively. These agreements are not surprising. In developing the
model, lift and drag data from wind-tunnel testing was adjusted by the writer in
a small, systematic way to provide agreement with the average PGA and LPGA
Tour performance over the full range of clubs from driver to wedges. In essence, the
adjustments were made to model the average behavior of the range of “premium”
PGA 7-iron: Launch angle = 16.3, Ball speed = 120, Backspin = 7,079, Height = 31.4, Carry = 172
40
30
(yards)
20
10
0
0 50
150 100
200 250
(yards)
PGA drive: Launch angle = 11.2, Ball speed = 165, Backspin = 2,685, Height = 30.9, Carry = 268
Figure 2I Trajectories of the average PGA Tour player 7-iron shot and drive.
golf balls used on the PGA and LPGA Tour. These adjustments are described in
detail later in the chapter.
THE EFFECTS OF ATMOSPHER IC

PR ESSU R E A N D BA LL W EIGHT
The effects of lift and drag described in the previous section may be surprising to
anyone unfamiliar with aerodynamic forces. To put these in some context, the aver-
age automobile, under full braking, takes at least 120 feet to stop from 60 miles per
hour, which is a deceleration of approximately 1 “g.” The explanation, for decelera-
tions as high as 2.5 g acting on a driven golf ball, is simply the surprising density of
the air. To obtain the actual forces, the relative force values in Figures 2E and 2F
are multiplied by the cross-sectional area of the ball, the ball speed squared, and the
density of air. The density of air under standard atmospheric conditions at sea level
is 1.2 kilograms per cubic meter. In more familiar terms, a 3 × 3 × 3 foot cardboard
box contains 2 pounds of air; or in more appropriate terms, the weight of air that is
displaced by a 0.1 pound, 1.68 inch diameter golf ball, over the trajectory length of
a 269 yard drive, is 1.2 pounds. Two things are immediately obvious from these cal-
culations. First, a golf ball with the same launch conditions in thinner, less dense, air
will no doubt fly further. Second, a lighter ball of the same diameter will be subject
to the same aerodynamic forces and so will be influenced more by them.
To see the effects of atmospheric changes, the trajectory using the average PGA
Tour launch conditions was modeled with the same ball strike conditions but in the
mile-high city of Denver, Colorado. The air density there is 12 percent less on aver-
age than at golf courses much nearer to sea level. So all that is needed is to reduce the
lift and drag forces by 12 percent.
To test the effect of using a lighter ball, we assume the use of the floating balls at
some golf resorts for driving out over water. These have a weight of 39 grams com-
pared to 45 grams for the regular ball. They are also typical range balls, which have
hard rugged covers for long life and do not grip the face to provide the spin rates of
the premium golf balls. For this exercise, they are assumed to have the performance
of the modern premium ball. However, we do need to step ahead to some of the
results from Chapter 3. If a 39 gram ball is struck with the same club head speed
and club loft as a 45 gram ball, it will launch at a lower angle with a higher speed
and spin rate. The new launch values in this case will be: ball speed = 168.5 miles
per hour; spin rate = 2,973 revolutions per minute; and launch angle = 11.0 degrees.
The reason for these changes will be evident in the descriptions of ball striking in
Chapter 3. For now, all that is needed is a re-run of the six steps used to determine
ball trajectory, for the changed launch conditions and ball weight. The results of
these calculations are given in Figure 2J.
It is evident that the lighter ball is influenced to a significantly higher degree by
the lift and drag forces. The trajectory can be seen to curve slightly upward over the
first third or so of the carry distance. Also, the increased effect of the drag force,
combined with the higher climb, has resulted in the ball dropping 18 yards shorter
than the regular weight ball. The use of a lighter ball has been suggested in the past
as a means of limiting the distance of touring professional drives. These calculations
would seem to support the validity of that proposal. However, as discussed in a later
25 Ball Flight
Launch angle = 11.2, Ball speed = 165, Backspin = 2,685, Height = 27.9, Carry = 281
40
floating ball; standard conditions
30 average PGA tour
(yards)
20
10
standard ball; Denver conditions
0
0 50 100 150 200 250 300
(yards)
Launch angle = 11.0, Ball speed = 168.5, Backspin = 2,973, Height = 37.8, Carry = 250
Figure 2J Effects of an average PGA Tour player drive using a 39 g floating ball or playing a
standard ball in Denver, Colorado, standard atmospheric conditions.
section, hooks and slices also result from the aerodynamic forces, and a lighter ball
would make direction control considerably more difficult for the average golfer.
The 12 percent lower air density in Denver lowers both lift and drag. It can be
seen that as a result, the ball climbs about 9 feet less but still carries 13 yards further.
Teeing the ball up further to increase the launch angle and flight time would be a
good strategy there.
THE EFFECTS OF HE A D- A N D TA ILW IN DS

As we have discussed, a golf ball flying through the air is subjected only to gravity
and the aerodynamic forces of drag and lift. The latter two forces depend on the
square of the ball speed with respect to the air. So when subjected to a headwind,
the wind does not blow the ball in the everyday sense. It simply increases the ball
speed with respect to the air, as illustrated in Figure 2K. Because of this, it will be
subjected to increased lift and drag. The increased lift will make it fly higher, and the
increased drag will make it fall shorter.
Conversely, if the ball is struck into a tailwind, its speed with respect to the air is
decreased. Drag and lift will then be reduced, and the ball will fly lower but gener-
ally go further.
Consider a 7-iron shot, with a club head speed of 85 miles per hour and a ball
launch speed of 110 miles per hour, played into a 10 miles per hour headwind. At
launch, the ball speed with respect to the air is increased to 120 miles per hour.
Since the aerodynamic forces increase by the square of the velocity, they will
increase compared to still air by the factor (120 /110)2 = 1.19. So the lift and drag
will increase by 19 percent. Conversely, in the 10 miles per hour tailwind, the factor
will be (100 /110)2 = 0.83, so lift and drag will decrease by 17 percent.
headwind
ball groundspeed
ball airspeed
Figure 2K Effect of a headwind on ball flight.

Club speed = 77.3, Loft = 26.5, Ball speed = 105, Backspin = 4,531
40
30
(yards)
20
10
(yards)
0 20 40 60 80 100 120 140 160
Attack angle = –4.7, Launch angle = 21.6
Figure 2L Effect of 10 mph head- and tailwinds on a 145 yd., 7-iron shot; with headwind, 126
yd.; with tailwind, 162 yd.
A simulation of this 7-iron shot in still air, and with 10 miles per hour head- and
tailwinds is given in Figure 2L. In this case, the changes in carry are +17 yards for
the tailwind and −19 yards for the headwind. The middle trajectory is the ball path
in still air.
These differences are clearly of great significance when making approach shots
into a green. Since the differences in carry are caused by the changes in the aero-
dynamic forces, the time of flight is a large determining factor. For this reason, in
high winds a lower trajectory approach shot with a lower-lofted club, resulting in a
shorter flight time, is often a safer selection. Note that at the bottom of Figure 2L,
the “attack angle” value is given as −4.7 degrees. This means that the club has struck
the ball on a downward club path, inclined at −4.7 degrees to the fairway surface.
This negative attack of the ball is standard procedure for elite players. It allows them
to lower the trajectory of the ball for more penetrating shots while still getting a high
backspin rate. The follow-through of this, after the ball has flown, is the taking of a
divot. If the ball had been struck with a horizontal attack, the launch angle would
have been higher, the flight longer, and the effect of the head- and tailwinds even
more pronounced.
Figure 2M shows the trajectories for the same nominal distance with a 5-iron
using a reduced swing speed. The distances in this case are 145 yards in still air as
for the 7-iron shot, 132 yards with a 10 miles per hour headwind, and 154 yards
with a 10 miles per hour tailwind. The differences are +9 yards for the tailwind and
−13 yards for the headwind. This means that with only rough estimates of wind
speed available to the player, the low 5-iron shots have the potential for significantly
less error than the much higher, longer-flying-time, 7-iron ones. This is no doubt the
reason why the “bump and run” approach shots are so often seen on the windblown
links courses of Scotland and Ireland.
CU RV ED BA LL FLIGHT
Curved ball flight has just a single cause, and that is a tilt from the horizontal of the
ball spin axis. This in turn has only one major cause for almost all golf shots, and
that is the inclination of the club face to the club head direction at impact in the
horizontal plane. This gives a sideways glancing blow to the ball, which produces a
27 Ball Flight
40
30
(yards)
20
10
0 20 40 60 80 100 120 140 160

Attack angle = –4.7, Launch angle = 16.3
Figure 2M Short 5-iron shot with 10 mph head- and tailwind.
sideways component of spin, thereby tilting the spin axis. The direction in which the
club head is travelling at impact is usually referred to as the “club path.”
For drivers, a tilted spin axis is also caused by striking the ball at a position
on the face offset horizontally from the face center. This causes the club head to
rotate sideways during the impact, which in turn, like a pair of mating gears, causes
the ball to rotate sideways in the opposite direction; together with the backspin
caused by the club loft, this tilts the spin axis. The curved face of the driver from
heel to toe is designed to minimize this effect by starting the curved trajectory
away from the target line. This is discussed fully in Chapter 6.
If we address only the major cause of an angled face to the club path, then the
tilt of the spin axis is easy to estimate. If, for example, the club face is angled to the
club path by 3 degrees, and the club has 34 degrees of loft, then the spin axis tilt is
given approximately by tilt angle = 57 × (3/34) = 5.0 degrees. If the same face angle
error was made with a driver lofted at 10 degrees, the tilt of the spin axis becomes
57 × (3/10) = 17.0 degrees. For the latter, it would produce a large slice if the face
was 3 degrees open to the club path, or a large hook if it was closed. The multiplier
57, approximately equal to 180 divided by pi, comes from conversion of the scien-
tific radian measure of angle to degrees.
These calculations raise a very important point. With a low-lofted driver, quite
amazing precision is needed to keep the ball in the fairway. In this regard, most ama-
teur players use drivers with far too little loft, resulting in too much difficulty with
direction control. This issue is pursued fully in Chapter 5.
These simple relationships between the face angle and the spin axis tilt arise
from the fact that the ball will separate in a spinning condition from the face along
the direction angled across the face by the club path, which in the driver case is
3 degrees sideways and 10 degrees upward. The spin axis thus starts out parallel
to the face but angled with respect to the edge of the sole. Moreover, because of
the high spin rate, it flies through the air with the spin axis fixed in that orienta-
tion. This stems from the principle of gyroscopic stability, which allows us to ride
bicycles with ease and keeps the gyroscopic instruments in aircraft panels aligned
with the horizontal as the aircraft banks and turns. One other property of golf balls
that leads to gyroscopic stability is their very high level of spherical symmetry, suf-
ficient enough that they fly precisely the same from whichever orientation they are
placed on a tee.
Thus, while following a curved path, the ball slips through the air, exposing more
of the “side” of the ball toward the direction of flight as it turns.
*
axis tilt lift
spin
back
*
axis tilt
* fixed in space
drag
weight
Figure 2N Inclination of the spin axis and forces acting on the ball in flight.
In Chapter 4, we will learn that the curve in the ball flight of a drive will produce
approximately 3.5 yards of sideways movement, for each 5 degrees of spin axis tilt,
per 100 yards of carry. Thus, a 200 yard drive with the 10 degree lofted driver, hav-
ing the face 3 degrees open, would, because of the 17 degree inclination of the spin
axis, slice: to give a total sideways movement of approximately (17/5) × (200/100) ×
3.5 = 24 yards.
The stability of the spin axis in flight allows us to calculate the direction of the
forces throughout the flight, as the ball climbs then descends while continuing
on a curved sideways path. As illustrated in Figure 2N, the lift force always acts
at right angles to the spin axis and therefore is tilted to the vertical by exactly
the same amount. This provides a horizontal component of the lift force, which,
exactly as for a banked aircraft, produces a turn. Unlike a banked aircraft, how-
ever, which uses the coordinated effect of the rudder to keep the plane pointing in
the direction of the turn, the ball continues to face in the same direction in which
it left the face. In aircraft flight terms, the ball is “slipping” sideways through the
air and will strike the ground with the ball spinning backward at an angle to the
direction in which it landed. This can cause balls to spin backward to the side
of the landing direction, a condition that can be reduced or exaggerated if the
landing is on a side slope, depending on whether the tilted axis is aligned with or
against the slope.
Consider the effect of a 5-degree spin axis tilt on the average PGA professional
drive with a carry of 269 yards. The lifting force at the start of the flight is 1.5 times
the ball weight or 0.15 pounds. The horizontal component of this force is only
0.013 pounds; and during the flight, it decreases to 0.003 pounds. However,
whereas the vertical component of lift has to overcome gravity, the horizontal
component has very little resistance to contend with, and so such a tiny and
decreasing force can carry the ball almost 10 yards sideways. Moreover, for the
29 Ball Flight
Drive: Club speed = 90, Ball speed = 129, Backspin = 2,699, Launch = 14.3
(yards) 30
20
10
0
0 50 100 150 200
(yards)
Carry distances: in still air 200, with headwind 180, with tailwind 216
10
(yards)
0
10
20
0 50 100 150 200
(yards)
Side deviations: in still air 16.8, with headwind 18.9, with tailwind 13.6
Figure 2P Effect on slice of 17 mph tail- and headwinds.
average PGA Tour player drive, this is caused by having the face only 1 degree
open or closed!
EFFECT OF HE A D- A N D TA ILW IN DS

ON HOOKS A N D SLICES
Since in a headwind the ball airspeed increases, so the lift force increases, and a
given tilt of the ball spin axis will result in a larger hook or slice. As already dis-
cussed, the drag force will also increase and thus the ball drops short. Conversely,
tailwinds will reduce the lift force and the drag force and thus reduce hooks or slices
while increasing the carry distance.
Figure 2P shows the effects of 17 miles per hour head- and tailwinds on an aver-
age amateur player’s 200 yard drive. The face is 2 degrees open at impact, and with
10 degrees of loft, this produces a 17 yard slice in still air. For these results, the swing
path is assumed to be from outside to inside by just under 2 degrees. This would
start the ball down the centerline, so the lateral deviations are due entirely to the
curved ball path. If the club path was on the zero centerline, the 2 degree open face
would start the ball on a trajectory approximately 1.7 degrees right. This would
move all the landing points to the right by a further 6 yards. The effects of the head-
and tailwinds are given at the bottom of the figure.
EFFECT OF CROSS W IN DS
When a golf ball is subjected to a crosswind, it does not travel sideways with
the crosswind speed. This would produce much larger lateral deviations in the
ball f light than are actually experienced. Just as with the head- or tailwind, the
only things that can change the direction of a ball in f light are the aerodynamic
crosswind
ball airspeed
ball groundspeed
drag
Figure 2Q Effects of a crosswind on ball flight.
forces. In this case, as shown in Figure 2Q , the effect is to change the direction
of the ball speed with respect to the moving air mass. The relative effect, of air
travelling sideways toward the ball, is the same as the ball travelling forward at
the launch speed and sideways at the crosswind speed toward a stationary air
mass. This is indicated by the opposing (dashed) arrow facing the crosswind
on the right side and connecting the arrows representing groundspeed and air-
speed. Since the ball airspeed is angled toward the crosswind, the drag force,
which always opposes the forward movement through the air, is equally angled
away from the crosswind. Thus, the drag force, as well as slowing the ball in its
progress over the ground, will at the same time pull it sideways in the direction
of the crosswind.
Figure 2R shows the effect of a right or left 10 miles per hour crosswind on the
average PGA Tour drive. For the drives shown, the launch conditions are assumed
to produce a straight drive down the centerline in still air. The effect of the 10 miles
per hour crosswind is a 14 yard deviation in the wind direction. To put this into per-
spective, the PGA usually sets the fairway widths between 11 and 14 yards from the
centerline.
One last point: the crosswind speed of 10 miles per hour is equal to 5 yards per
second. So if the ball was carried sideways with the air mass for the 6.5 seconds
flight time, it would have travelled 5 × 6.5 = 33 yards. Fortunately it doesn’t work
that way.
Launch angle = 11.2, Ball speed = 165, Backspin = 2,685, Height = 30.9, Carry = 268
40
(yards)
30
20
10
0
0 50 100 150 200 250
(yards)
20
10
(yards)
0
10
20
Figure 2R Effect of 10 mph crosswinds on the average PGA Tour drive compared to the drive in
still air.
31 Ball Flight
OPTIONA L R E A DING—THE SU PPORTING PH YSICS*

2.1 Aerodynamic forces
It is difficult to start the discussion of any important topic in golf without first refer-
ring to Cochran and Stobbs ([1968] 1999) Search for the Perfect Swing. The research-
ers working with Cochran and Stobbs in the 1960s, a decade before the existence of
any reliable experimental data, described in some detail all of the principal character-
istics of golf ball flight. At this time, aerodynamic drag was of course very well under-
stood, as well as the reasons why spinning balls experience an aerodynamic “lifting”
force. The explanation of the latter is credited to the 19th-century scientist Gustav
Magnus and is referred to as the Magnus force. It is not clear whether the Cochran
and Stobbs researchers were aware of a limited set of wind-tunnel data, for a single
wind-tunnel speed, published by Davies (1949). In any case, the data was shown later
by Bearman and Harvey (1976) to be considerably in error. It will be interesting, later
in this section, to compare some of the quite profound conclusions reached in Search
for the Perfect Swing, with predictions based on later extensive wind-tunnel test data.
Figures 2.1 and 2.2 add appropriate labels to the earlier diagrams describing
spin axis tilt and the mechanism of aerodynamic lift.
FL
ω
αs
FD
Figure 2.1 Forces acting on a spinning golf ball in flight.
V density, ρ
ball cross-sectional area = s
Figure 2.2 Air flow disturbance over a spinning ball.
* bypass without any loss of understanding of the science principles of the game
For a more detailed discussion of the creation of both drag and Magnus
forces, with a wider focus on the aerodynamics of cricket, tennis, and baseball
as well as golf, the excellent review article by Mehta and Pallis (2001) is highly
recommended.
2.2 Lateral and drag force coefficients

The principle reference for the development of the ball flight model in this work
is Bearman and Harvey (1976). As mentioned in Chapter 1, these researchers car-
ried out wind-tunnel testing of hollow “golf ” balls, inside of which were mounted
motors to set required spin rates. The balls were held vertically by two wires, which
provided both support and power to the motors. Drag and Magnus forces were mea-
sured using a combination of strain guage and the standard wind-tunnel instrumen-
tation. B&H referred to the Magnus force as the lift force and used notation L to
describe it, as did other authors who followed. This did not cause confusion since
the interest was only “straight” ball flight. From this point, to avoid confusion from
all sources, we refer to the Magnus force as the lateral force with notation FL , and
represent drag force as FD .
B&H published the wind-tunnel data in the form of dimensionless drag and lat-
eral force coefficients (C D and C L , respectively) such that the actual forces are
FL = ( 12 ρ v 2 s)C L and FD = ( 12 ρ v 2 s)C D (2.1)

where
ρ = density of air = 1.204 kg/m3 for standard atmospheric conditions at sea level of

20°C and 101 kPa pressure (68°F and 29.92 in. of mercury);
v = ball velocity, m/s;
s = projected area of ball = π(0.02134)2 = 0.00143 m 2.
The form of Eq. 2.1 can be understood by considering the force, F, applied to a
rough blunt object passing through an air mass of density ρ at constant veloc-
ity v. Imagine that air molecules are being pushed ahead with velocity v as it
meets them. In time ∆t, the air mass accelerated to velocity v is ∆m = ρvs∆t ,
where s is the cross-sectional area of the object normal to the travel. The work
done by the force in time ∆t is Fv∆t , and the kinetic energy increase of the air is
(1 / 2)∆mv 2 = (1 / 2)(ρ vs∆t)v 2. Equating the work to the gain in kinetic energy gives
F = (1 / 2)ρv 2 s. This is of course the maximum possible force; any sideways deflec-
tion of the air mass must result in less drag. Hence the coefficients are less than 1.0.
B&H produced test data for two Uniroyal golf balls: one with normal circular
dimples and the other with hexagonal-shaped dimples. A more recent investiga-
tion by Smits and Smith (1994), using both wind-tunnel testing and R&A/USGA
data on driven ball trajectories, found that the B&H lateral force coefficients, for
the hexagonal dimpled balls, were only slightly low by a constant increment of 0.04.
For this work, I changed the data by B&H, presented in graphical form, to tabular
data using dividers, a magnifier, and a precision scale. The data is given in Table 2.1.
33 Ball Flight
Table 2.1 Bearman and Harvey force coefficients for the Uniroyal hexagonal

dimple ball
Ball velocity Ball spin rate (rpm)

(mph)
0 1000 2000 3000 4000 5000 6000 7000
Drag force coefficients
30.6 0.53 0.39 0.37 0.42 0.43 0.48 0.52 0.52

48.3 0.33 0.25 0.29 0.37 0.4 0.44 0.45 0.45
66.9 0.22 0.24 0.27 0.32 0.35 0.37 0.39 0.39
85.9 0.23 0.23 0.25 0.28 0.31 0.33 0.35 0.35
104.9 0.24 0.25 0.25 0.27 0.28 0.31 0.33 0.33
123.5 0.26 0.25 0.25 0.26 0.27 0.29 0.31 0.31
141.2 0.25 0.25 0.25 0.26 0.27 0.28 0.29 0.29
160.8 0.24 0.24 0.25 0.26 0.26 0.27 0.28 0.28
Lateral force coefficients
30.6 −0.15 −0.05 0.10 0.36 0.40 0.42 0.47 0.57

48.3 0.08 0.12 0.18 0.34 0.37 0.39 0.43 0.52
66.9 0.12 0.17 0.25 0.30 0.33 0.35 0.37 0.44
85.9 0.10 0.14 0.20 0.25 0.28 0.31 0.34 0.38
104.9 0.08 0.13 0.17 0.21 0.24 0.27 0.30 0.33
123.5 0.08 0.12 0.15 0.18 0.22 0.24 0.27 0.29
141.2 0.08 0.12 0.14 0.17 0.20 0.22 0.25 0.26
160.8 0.08 0.12 0.13 0.16 0.18 0.20 0.22 0.23
Unfortunately, B&H did not carry out any testing with spin rates above 7,000 rpm.
This places some uncertainty about the predictions made for 8-iron, 9-iron, and
wedge shots. In the absence of any better information, the lift and drag coefficients
are assumed constant above 7,000 rpm, as this gives good agreement for data on
PGA Tour professionals for 8-iron through pitching wedge shots.
Example
For the average PGA Tour player drive, the ball speed and spin rate off the tee are
165 mph (73.76 m/s) and 2,685 rpm (281.2 rad/s). From Table 2.1, the approximate
force coefficients are CD = 0.255 and CL = 0.145. From Eq. 2.1, the estimated drag
and lateral forces are
1
FL = 2 (1.204)(73.76)2 (0.00143)(0.145) = 0.68 N (newtons)
1
FD = 2 (1.204)(73.76)2 (0.00143)(0.255) = 1.19 N.
The gravitational force on the golf ball is FG = 0.045(9.81) = 0.44 N. So off the tee,
the lifting force for a straight drive is 55 percent greater than the ball weight, and the
drag force is 2.7 times the ball weight. In the Cochran and Stobbs investigations,
typical professional player drives are characterized as 134 mph ball speed with
3,600 rpm spin rates; the force coefficients for this can be interpolated from Table 2.1
as C D = 0.265 and C L = 0.19. With these coefficients, and the smaller British ball
radius of 0.0206 m, the lift and drag forces are FL = 0.67 N and FL = 0.76 N ; or in
British units, 2.4 and 2.7 oz., respectively, with a ball weight of 1.6 oz. Cochran and
Stobbs (1968) suggested that under these drive conditions, the initial lift would be 2 to
2.5 oz. and the drag 3 or 4 oz. It seems likely that these were estimated from the Davies
(1949) data, whose lift coefficient for this case is in approximate agreement with Smits
and Smith (1994), but drag coefficients are about one-third too high (Bearman and
Harvey 1976).
In contrast to the drive, the average PGA professional 7-iron launch parameters
are 120 mph (53.6 m/s) and 7,097 rpm (743.2 rad/s). From Table 2.1, the approxi-
mate force coefficients are CD = 0.31 and C L = 0.29. From Eq. 2.1, the estimated
drag and lateral forces are
1
FL = 2 (1.204)(53.6)2 (0.00143)(0.29) = 0.72 N
FD = 12 (1.204)(53.6)2 (0.00143)(0.31) = 0.77 N.
Almost two decades after the Bearman and Harvey study, the R&A/USGA spon-
sored further wind-tunnel investigations, which were published by Smits and Smith
(1994) in the World Scientific Congress of Golf (WSCG) II. These authors pre-
sented a restricted set of the data in the form of an empirical model, applicable only
to driver shots. For the wider range of ball speed and spin rates, they indicated gen-
eral agreement with Bearman and Harvey. In this work, I found that the best agree-
ment, with Trackman-produced data on the ball trajectories of PGA Tour players,
was achieved by decreasing the B&H drag coefficients by 0.015 and decreasing the
B&H lift coefficients by 0.005.
The Smits and Smith empirical relationships for the drag and lateral force
coefficients are
C D = Cd1 + Cd 2 W + Cd 3 sin(π(R # − A 1 )/ A 2 ) and C L = C l1 W . (2.2)

The coefficients in these equations were adjusted by Smits and Smith to provide
best agreement with seven different driver trajectories with 1994 balls. All of
these trajectory variations were covered by small variations in just the two coef-
ficients Cd1 and C l1 . The full set of coefficients for agreement with these driver
shots were
Cd1 = 0.21 ± 0.01; Cd 2 = 0.18; Cd 3 = 0.06; C l1 = 0.51 ± 0.02;

Cd 2 = 0.18; Cd3 = 0.06; A 1 = 90 , 000; A 2 = 200 , 000.
R # represents Reynolds Number, which is given by
R # = 2 rb v /(1.453 ×10 −5 ), (2.3)

35 Ball Flight
where rb is the ball radius (m), v is the ball velocity (m/s), and the constant
1.453 × 10−5 (m 2/s) is the air kinematic viscosity under standard atmospheric condi-
tions. Finally, parameter W is the dimensionless spin ratio rb ω / v .
Since Smits and Smith varied Cd1 and C l1 to provide agreement with 1994
driver trajectories, it seemed reasonable for me to vary these two coefficients in a
systematic way to obtain best fit to the 2010 average PGA Tour drives. The values
that were found to give the best fit are
Cd1 = 0.167; C l1 = 0.406.
These are quite radical reductions in both lift and drag. However, we should note
that at the time of the Smits and Smith investigation, the hollow stainless steel driv-
ers were new to the market, and the large volume titanium drivers were still a decade
away. These were to bring about major changes in optimum launch conditions, with
higher launch angles and much lower spin rates. It is not surprising that dimple pro-
files would be engineered to reduce drag at some sacrifice of lateral force.
The modified Smits and Smith parameters are used for all of the driver trajectory
predictions in this book. All other club strikes have trajectories modeled with the
modified Bearman and Harvey coefficients.
Smits and Smith (1994) also investigated the changes of ball spin in flight, and
their relationship for spin rate of decay is used in all of the flight modeling. This
relationship is defined by the single equation
ω = −0.00002ωv / rb , (2.4)

where ω (rad/s) and v (m/s) are the current angular and linear ball velocities,
respectively, and rb (m) is the ball radius.
2.3 Components of the drag and lateral

forces in 3-dimensional flight
The first step in predicting 3-dimensional ball flight, caused by a tilted spin axis,
is to separate the drag and lateral forces into components in a suitable co-ordinate
system, as illustrated in Figure 2.3. The simplest way to approach the problem is
to assign a co-ordinate system based on the club face at the moment of impact.
Specifically, one plane of the co-ordinate system passes through the bottom edge of
the face and is angled to the face by the amount of vertical loft, δ v. Two co-ordinate
axes, x and z, lie in this plane: x parallel to and z normal to the face lower edge. The
third axis, y, is normal to the x–z plane, pointing outward from the face. With respect
to this co-ordinate system, three important parameters are known. The initial ball
trajectory has azimuth and elevation defined by α hb and α vb with respect to the
y axis and the x–y plane. The spin axis, since it is established by rolling and slipping
or skidding on the face, lies on the plane that coincides with the x axis and is inclined
by amount δ v from the z axis. The spin axis tilt, α s , is defined with respect to the
x axis as shown in Figure 2.3. As discussed previously, the spin axis orientation can
be assumed to remain fixed in space. The amount by which the face is open or closed
at impact will be referred to as horizontal loft, with symbol δ h .
To make lightest work of the calculations to follow, it is best to use some rules
from vector mechanics. For this purpose, the angular velocity ω is represented
δv
FD
αvb
y
αhb
αs
ω
FL
club face
Figure 2.3 Co-ordinate system and parameters for flight calculation.
correctly as a vector along its spin axis. The spin direction is counterclockwise about
the spin axis, looking in the direction of the vector, which is in the negative × direc-
tion; and in this case, ω is negative.
The drag force FD by definition is the resistance in the direction of flight, and
so it is aligned with and opposite to the ball velocity vector v. The lateral force F L is
normal to the plane on which lie both the angular velocity ω and linear velocity v.
In vector mechanics, this direction is that of the cross product v × ω. To establish
this cross product, it is necessary first to determine the components of both ω and
v in the three co-ordinate directions.
From Figure 2.3, the components of ball angular velocity vector ω, with mag-
nitude ω , are
ω x = ω cos(α s )
ω y = ω sin(α s )sin(δ v ) (2.5)
ω z = − ω sin(α s )cos(δ v ).
Similarly, the components of ball linear velocity v are
v x = − v cos(α v )sin(α h )
v y = v cos(α v )cos(α h ) (2.6)

v z = v sin(α v ).
Since the drag force directly opposes the forward velocity v, its components are
37 Ball Flight
FDx = FD cos(α v )sin(α h )
FDy = − FD cos(α v )cos(α h ) (2.7)

FDz = − FD sin(α v ).
The components of lateral force FL in direction v × ω are taken directly from the
rule for vector cross product as
FLx = − FL [sin(α s )cos(δ)cos(α h )cos(α v ) + sin(α s )sin(δ)sin(α v )]
FLy = − FL [cos(α s )sin(α v ) − sin(α s )cos(δ)cos(α v )sin(α h )] (2.8)

FLz = + FL [sin(α s )sin(δ)sin(α h )cos(α v ) + cos(α s )cos(α v )cos(α h )].
The acceleration components acting on the ball during flight, including gravita-
tional acceleration, g, are
a x = (FDx + FLx )/ m b
a y = (FDy + FLy )/ m b (2.9)
a z = − g + (FDz + FLz )/ m b ,
which can be integrated numerically to predict velocities and displacements with

respect to the co-ordinate system in Figure 2.3.
The initial conditions—ball velocity v 1 ,angular velocity components ω x and
ω z , and launch angle components (δ h − α hb ) and (δ v − α vb )—are defined with
respect to the club face. In turn, the angles of the club face are defined with respect
to the club head velocity. The relationships are illustrated in Figure. 2.4.
ωx
δv–αvb v1
V0 rb
αvb
βv
δv
ωz
δh–αhb v1
V0 rb
αhb
βh
δh
Figure 2.4 Parameters used in flight trajectory simulation.

cross wind, magnitude vw
d, v*
αw ball air spee
vw
ball ground speed, v
FD
Figure 2.5 Aerodynamic velocity v* in the presence of a crosswind.
In the calculations, the horizontal club path β h is defined independently with

respect to a chosen target line, and the vertical club path β v is defined with respect to
the ground surface. A study of decreasing time step sizes found that a time step size of
0.001 s gives distance errors less than 0.03 yd. and flight time errors less than 0.01 s.
Flight paths modeled in this way were compared to Trackman-measured trajec-
tories in the first part of the chapter.
In the remainder of this chapter, we establish the changes needed to analyze the
effect on ball flight of head- and tailwinds and of crosswinds.
2.4 Effect of winds on ball flight

The situation of a direct headwind or tailwind is easy to handle. Assume the wind
velocity is defined by v w , which is positive for a headwind or negative for a tailwind.
From the moment the ball flies off the club face, its velocity with respect to the air is
changed to v + v w , whereas its spin rate ω is unaffected by the wind. As the ball
velocity and spin rate change in flight, the lift and drag coefficients will be changed
by relatively small amounts; see Table 2.1. However the calculation of the lateral
and drag forces will now change to
1 1
FL = ( 2 ρ( v + v w )2 s)C L and FD = ( 2 ρ( v + v w )2 s)C D . (2.10)

Because the velocity term is squared, the effect of head- and tailwinds is magni-
fied beyond its relative value. A 15 mph headwind acting on a 150 mph ball speed
drive will increase the air speed facing the ball by 10 percent from 150 to 160 mph.
It will then increase the aerodynamic forces by 1652 /150 2 = 1.21, or 21 percent. The
increased lift will cause it to rise higher, but the larger drag will make it drop shorter.
Applying Eq. 2.10 is a trivial change to the flight calculations.
A crosswind simply changes the direction of the air flowing over the ball to v*
as shown in Figure 2.5. This affects the direction of the drag force, which now has
a component FD sin(α w ), pulling the ball sideways in the direction of the cross-
wind. The Magnus force also changes direction to act in the vertical plane through
v* so that it has a component pulling the ball in the direction of the wind during
ascent and against the wind during descent. The changes of the Magnus force are of
secondary importance. However, all of the effects are easily modeled by simply
changing the azimuth angle α h in Eqs. 2.6 to 2.8 to (α h + α w ).
For any general wind direction, the separate components of head- or tailwind
and crosswind can of course be dealt with using both airspeed corrections just
mentioned.
3
B A S IC F U N DA M E N TA L S O F B A L L S T R I K I N G
This is perhaps the most important chapter in terms of understanding the funda-
mental principles of the game. Readers of this chapter will hopefully be able to rec-
ognize good advice when they hear it and ignore bad advice that does not accord
with fundamental principles. In addition they may come to appreciate even more
the almost unbelievable talent of the truly elite players. Chapters 4, 5, and 6 are
concerned with the development of backspin, with shaping curved shots, and with
the effects of off-center ball strikes off the tee or out on the fairway. These are based
on an understanding of the fundamentals laid down here.
CONTACT TI M E BET W EEN CLU B A N D BA LL

We start by expanding on this topic, which was discussed briefly at the start of
Chapter 1. For completeness, the discussion from there will first be reintroduced
in abbreviated form. Everything in golf results from the behavior of the golf balls
and clubs during the impact period, from first contact up to the point at which the
ball separates from the club face. Action and reaction between the two impacting
bodies during this period of time define the ball speed, launch angle, and spin rate
of the ball and the ensuing flight trajectory. The time period is amazingly small for
all shots as shown in Figure 3A.
Surprisingly, the harder the ball is struck, the shorter becomes the contact time.
This stems from the spherical shape being compressed. The volume of deforma-
tion increases much faster than the depth of compression. Consequently, the force
and resulting acceleration back to the original shape are proportionately greater
the larger the amount of compression. To put these times into some perspec-
tive, they range from about one-fiftieth of an eye blink for the tap in putt to about
one-hundredth of an eye blink for the drive.
Two consequences of these very short contact times are of great importance. The
first is that, of course, no adjustments to the launch of the ball are possible during
the impact. The player can only strike the ball. For this reason it is always neces-
sary to accelerate the club up to the impact point for a decisive strike and to focus
on the one goal of having the club facing and travelling in the required direction at
impact. As skills improve, the player may deliberately alter the “club path” through
the impact zone by small amounts to produce intentional curves in ball flight. But
for the higher-handicap player, the goal must be for the club head to pass as nearly as
possible back through the position the club had at the beginning of the backswing,
the so-called address position. It is worth noting that by the time the first vibrations
travelling up the shaft have reached the hands, the ball is taking flight. And an even
longer delay then occurs before the brain can start to react in any way to this delayed
39
Contact time, seconds

0.0008
0.0007
0.0006
0.0005
0.0004
Putt Chip Pitch Approach Drive
Figure 3A Contact times between club and ball.
transmission of the impact. So as the saying goes—just hit it—nothing more can
be done.
A second consequence is that provided the ball is struck first, what happens to
the club head during impact with the ground is of little importance. Even if the club
head has an average speed as high as 100 mph through the impact, the distance trav-
elled during the 0.0005 seconds of contact is still less than 1 inch. With a descend-
ing blow, rarely greater than 5 degrees, the club will be unlikely to encounter any
serious resistance while in contact with the ball unless the ball is on some hard sur-
face. Even on a hard surface, with an almost impossible 10 degree negative attack
angle, the club will still not reach the ground, with the ball still in contact, if the edge
of the club sole is more than about one-eighth of an inch from the ground at impact
with the ball. In contrast, if players try to sweep the balls off fairways, or particularly
out of short rough, then the club can certainly be deflected before reaching the ball,
with significant effects on the ball trajectory. Most amateur players can improve
their iron shots in no better way than striking the ball first with a downward angle of
attack. This simply requires the confidence that the ball will still fly upward off the
face, with in fact a higher ball speed, long before the club penetrates below the turf.
CONTACT FORCE BET W EEN BA LL A N D CLU B FACE

As discussed at the beginning of Chapter 2, a direct result of the short contact time
is that the contact forces between the club and ball are remarkably high. The easiest
way to determine this is with the impulse-momentum change relationship intro-
duced in Chapter 1.
“ force × time = mass × velocity change.”
For a 100 mph strike with a modern driver, the club head will be travelling
forward after impact at approximately 66 mph and the ball at 149 mph as shown
in Figure 3B. The 149 was rounded up to 150 mph in the Introduction for the car
comparison. Here we need to be more precise to demonstrate the validity of the
results. For an average driver head mass of 0.44 pounds, this gives a loss of momen-
tum in the club head of 0.44 × 34 = 15 pounds-mph. The gain in momentum of the
41 Basic Fundamentals of Ball Striking
action &
100mph 66 mph 149mph
reaction
approach velocity separation velocity

100 mph 149–66 = 83 mph
Figure 3B Force interaction between the club head and ball.
0.1 pound ball is 0.1 × 149 = 14.9 pounds-mph. The small difference is simply due to
speeds being expressed to the nearest mile per hour.
Before proceeding, it is important to define the coefficient of restitution more
formally as it applies to ball striking. It is defined with respect to the point of
impact as
CofR = (separation velocity)/(approach velocity).
For the driver impact example in Figure 3B, the approach velocity is 100 mph,
and the separation velocity is 149 – 66 = 83 mph. Thus the CofR is 83/100 = 0.83.
This is the maximum value allowed for competition and, as mentioned in the
Introduction, is only so high because of the spring trampoline face of the hollow
titanium driver—more on this later.
Recall also from the Introduction that we can take our last expression, “force ×
time = mass × velocity change,” and divide by the contact time to get the average
force. We can use either the change in club speed or the change in ball speed in this
calculation. We will choose the ball speed; and so we can write
Average contact force = (ball mass × ball velocity)/contact time
From this we obtained the average contact force for this drive to be 1,500 pounds.
The actual contact force rises from zero to a maximum value at mid impact (time
about 0.0002 seconds) and then reduces to zero at ball separation. The maximum
force at mid impact is thus approximately twice the average, that is,
maximum contact force = 3,000 pounds.
The force profile during the contact time is shown in Figure 3C. With short
irons, the ball speed may be around two-thirds the speed for a driver, with maxi-
mum impact force around 2,000 pounds.
We might pause to consider the enormity of these impact forces, which brings us to
a very important point. These forces greatly exceed any possible reaction forces from
the shaft. The role of the shaft is to bring the club head to the impact point with the
required speed, with the club close to its initial position with respect to the ball and
with the club path toward the intended target. The shaft then plays no active role dur-
ing the impact. This is an obvious fact since the shaft can be bent substantially with just
a few tens of pounds of force. No possible action from the grip during the impact can
Force, pounds
3000
2000
1000
0
0 0.1 0.2 0.3 0.4 0.5
Time, thousandths of a second
Figure 3C Force profile between driver face and ball.
have any significant effect when compared to the thousands of pounds of force interac-
tion with the ball. The use of terms such as shaft “kick point” by the golf manufacturers
is intended to disabuse you of this fact. As always in science there is an “exception.” In
this case it is except for the fact that the shaft is slowing down during the impact and
so losing momentum in the collision, and Newton’s laws do not allow for exclusions
of mass in an impact. Allowing for the mass of the shaft in the momentum transfer is
quite complex, and it has a very small effect on ball speed. It only matters when try-
ing to get exact values of coefficient of restitution for measurements of golfer perfor-
mance. The subject is discussed later in the chapter and ignored until then.
BA LL SPEED R ELATIONSHIPS
For a centered ball strike with a driver, and with the face “square” to the ball, the
speed of the ball depends on the club head speed, the club head mass, and CofR. For
more lofted clubs, the loft angle also has a significant effect on ball speed. We discuss
this in a later section. The relationship of ball speed to club head speed, CofR, and
club head mass can be expressed as follows:
Ball speed is proportional to

(a) Club head speed
(b) (1 + CofR)
(c) Club head mass/(Club head mass + Ball mass).
These three terms can be applied in any order as multiplying factors. We address
them in turn.
(a) Club head speed factor

The direct relationship between club head speed and ball speed is to be expected
and is easily recognized to satisfy both momentum conservation and the
action &
110mph 72.6 mph 163.9mph
reaction
Figure 3D Effect of a 10 percent driver speed increase.
CofR value for the impact. As shown in Figure 3D, if the club speed at impact
increases by 10 percent, then, all other things staying the same, both the club speed
after impact and the ball speed increase by 10 percent. Thus the momentum lost by
the club and gained by the ball both increase by 10 percent and are still equal and
opposite. Also, as a consequence of all the speeds increasing by the same propor-
tional amount, the division in the CofR calculation gives the same value; that is,
from Figure 3D we get CofR = (163.9 – 72.6)/110 = 0.83.
Because of this one-to-one relationship, club head speed is by far the most impor-
tant parameter in creating ball speed. If through physical training and range prac-
tice a golfer can increase his swing speed by say 10 percent, then the ball speed and
distance also increase by approximately 10 percent. For a mid-handicap player, a
200 yard drive becomes approximately a 220 yard drive, with maybe a 2-club shorter
shot into the green. The modifier “approximately” applies because the increased
impact speed deforms the ball a little more and slightly reduces the CofR. Also, the
proportional increase in distance is an approximate one. However, nothing else can
come close to increased swing speed for improving distance. This is shown in Figure 3E
where ball speeds and distances are compared to the average PGA Tour player drive
for decreasing club head speeds. For each 10 mph decrease in this case, the launch
angle is increased by 2 degrees to provide a more optimal trajectory for each shot.
The dashed-line trajectory is for the same 11.2 degree launch as the average PGA
drive but with an 80 mph impact speed. The very short carry in this case results from
the ball not having a long enough flight time, that is, reaching the ground with unused
potential for further flight. It lands at a low descent angle and consequently will make
up for some of the deficiency with a longer bounce and roll, but deceleration on the
fairway is very much greater than in the air, unless of course you are fortunate to land
on a significant downslope. So, for the low swing-speed player, the advice to tee the
ball high and hit it on the upswing for a higher launch is very sound.
Impact speeds: 80, 90, 100, 112 mph Ball speeds: 118, 132, 147, 165 mph
Launch angles: 17.2, 15.2, 13.2, 11.2 degrees Carry distances: 188, 215, 240, 268 yards
40
30
(yards)
20
10
0
0 50 100 150 200 250
(yards)
tour professional swing at 80 mph
Figure 3E Ball flights, with appropriate launch angles, for decreasing swing speeds.
For the three slower drives, the percentage decreases in impact speed from
112 mph are 11, 20, and 29, respectively. The percentage decreases in ball launch
speed are identical as calculated. The percentage decreases in flight distance are 10,
20, and 30, respectively; the modest increases in the launch angle keep them on trend
with the speed changes. These trajectory curves display the effect of aerodynamic
lift on ball flight. Recall from Chapter 2 that aerodynamic lift is proportional to ball
speed squared. So even for the same backspin rate, the aerodynamic lift at launch for
the professional grade 165 mph ball speed is over four times greater than for the 118
mph ball speed of the 80 mph ball striker. We can see in Figure 3E that for the former,
the ball is actually curving upward, that is, climbing at an increasing rate until about
100 yards out. For the latter, the rate of climb is decreasing from the start.
(b) Restitution Factor (1 + CofR)

This effect is disappointing. It might be expected that as CofR increases, then so does
ball speed. As drivers have progressed from the persimmon or laminated solid wood
heads, to the hollow titanium shell modern driver head with a diaphragm spring face,
the CofR has improved from approximately 0.74 to the present competition limit of
0.83. The expectation might have been that the ball speed would then increase by
factor 0.83/0.74 = 1.12, or by 12 percent. The reason this is not so is that the improved
spring effect to raise the CofR not only pushes the ball forward better but also pushes
the club head backward more. The club being pushed backward reduces ball speed,
while the ball being pushed forward increases ball speed. The latter wins, but by a
reduced margin. The changes in ball speed and distance because of changes in CofR
are actually proportional to (1 + CofR). This is not intuitively obvious but is a direct
consequence of momentum conservation. Thus the change in ball speed from wood
heads to titanium has been (1.83 /1.74) = 1.05, or an increase of only 5 percent, much
less than would be expected from all of the brouhaha that resulted from this innova-
tion. However, it should be noted that, with the much increased body and face size
of the modern driver, and the increased “sweet area” of the diaphragm face, there is
now less speed reduction from off-center hits. By reducing the penalty for off-center
hits, this has allowed players to swing faster with a smaller penalty risk, in this way
capturing much bigger gains in ball speed and distance.
The easiest way to demonstrate the validity of this relationship is by testing an
example. It seems that the laws of physics will not allow the CofR to be increased
above approximately 0.89 unless a springier non-conforming ball is used. Assume that
a non-conforming club with 0.88 CofR performance is available for non-competition
play. Our expression indicates that the ball velocity for a 100 mph ball strike will
increase from 149 mph to (1.88/1.83) × 149 = 153 mph as shown in Figure 3F. In this
case, the velocity of the club head after the impact must be 65 mph to give velocity of
separation (153 − 65) = 88 mph, or 0.88 CofR for the 100 mph impact.
The speeds must be correct if the loss of momentum by the head equals the gain
of momentum by the ball, that is, if they satisfy Newton’s law. In this case, the club
head of weight 0.44 pounds has decreased in speed by 35 mph, and so lost amount
(0.44) × (35) = 15.4 pounds-mph of momentum. The ball weighing 0.1 pounds has
gained (0.1) × (153) = 15.3 pounds-mph of its momentum. The difference, as before,
is simply due to the speeds being given to the nearest whole mile per hour.
0.88 Cof R
100mph 65 mph 153mph
Figure 3F Effect of increasing driver CofR to 0.88.
For simplicity, the loft angle of the club has been neglected in this discussion
of CofR; that is, all velocities have been taken as horizontal. We will see later that
only small errors are incurred for drivers in neglecting the effect of loft angle on
ball speed.
(c) Mass factor = club head mass/(club head mass + ball mass)

The validity of this expression can also be established by showing that it satisfies
both the correct CofR value and conservation of momentum. The majority of
driver heads weigh approximately 0.44 pounds. Assume we use a 25 percent heavier
driver, that is, increase the head weight from 0.44 to 0.55 pounds. The speed of the
 0.55   0.44 
0.1 pound ball would have a proportional increase of  / = 1.038, that
 0.65   0.54 
is, an increase of only 3.8 percent in ball speed for a 25 percent increase in club
weight. For a 100 mph ball strike with a 0.83 CofR driver, the ball speed will increase
from 149 mph to 154.7 mph. In this case, the velocity of the club head after the
impact must be 71.7 mph to give velocity of separation (154.7 − 71.7) = 83 mph, or
0.83 CofR for the 100 mph impact. The club head of mass of 0.55 pounds has
decreased in speed by 28.3 mph and so has lost amount (0.55) × (28.3) = 15.6
pounds-mph of its momentum. The ball weighing 0.1 pounds has gained momen-
tum of amount (0.1) × (154.7) = 15.5 pounds-mph, again with the small difference
due only to a decimal round-off error.
Of course, if this added weight slows the swing by more than 3.8 percent, then
all of the advantage is lost. Surprisingly, this is exactly what happens. Over a wide
range of head weights from 0.3 to 0.6 pounds, mounted on the same shaft, the
change in swing speed is almost exactly cancelled by the effect of the changed
head weight to give approximately the same ball speed. This was demonstrated in
the classic book by Daish on The Physics of Ball Games (English Universities Press,
1972). He carried out tests on four players with different swing speeds, using dif-
ferent head weights and measuring the speeds at the bottom of the swing with an
electronic timer. He found that with very small variations between the players, the
swing speed varied inversely with the 5th root of the head weight. Thus, going from
0.44 to 0.55 pounds, the swing speeds will be in the ratio 1 / 5 0.44 to 1 / 5 0.55 ,
equal to 1:0.956. This speed loss of 4.4 percent, combined with a speed gain of
3.8 percent with the increased head weight, leads to a predicted loss of 0.6 per-
cent in ball speed. The end result is a ball speed of 148 mph, a loss of 1 mph.
If instead we decrease the head weight to 0.33 pounds, then the swing speed
increases by a factor equal to the 5th root of (0.44/0.33), equal to 1.059, or from
160
Club head and ball speeds, mph

ball speeds
140
actual head weight values

120
club head speeds

100
80
0.2 0.3 0.4 0.5 0.6 0.7
Club head weight, pounds
Figure 3G Effect of driver head weight on “swing” speed and ball speed, for standard swing
speeds of 80, 90, and 100 mph.
100 to 105.9 mph. However, because of the impact with the lighter head, the ball
speed decreases by a factor (0.33/0.43)/(0.44/0.54) = 0.94. In consequence, the
ball speed changes to 149 × 1.059 × 0.94 = 148.3 mph, a loss of 0.6 mph. These
very small proportional changes, over the range 0.3 to 0.6 pounds head weight,
are the same for any swing speed; as noted by Daish, there is no evidence that a
powerful player should use heavier head weights or vice versa, even though this
is often done in club fitting.
Figure 3G shows the effect of different head weights on the ball speed and the
impact speed for impact speeds of 90, 100, and 110 mph with the standard driver
head weight. It can be seen that the optimum head weight of 0.44 pounds presents
only a tiny improvement over the range 0.33 to 0.55 pounds, and yet this precise
optimum value was long ago determined by trial and error.
From basic physics, there is no doubt that the swing speed is affected by the
swing resistance of the entire club and not just the weight of the head. The pos-
sible effect of shaft weight and length changes were excluded from Daish’s experi-
ments since the same shaft was used for the different club head weights. However,
it is worth noting the obvious fact that lighter shafts must bring about some small
increases in swing speed; unfortunately, this turns out to be very small.
Several studies reached approximately the same result from experiments to sup-
port this. The studies are Greenwald, Penna, and Crisco (2001); Smith, Broker, and
Nathan (2003); and Cross and Bower (2006). The common result is that the swing
speed increases approximately as the 4th root of the moment of inertia (MoI) of the
entire club decreases. The MoI in this case is the value for rotation about the grip
when the wrists are unhinged at the end of the swing.
So the effects of any small changes in club MoI will have very small changes
in swing speed. Very high prices are paid to go from 0.17 pounds (75 gram) driver
shafts to 0.11 pounds (50 gram) ones. The decrease in driver club weight with a
0.44 pound head and a lighter shaft, neglecting the grip, is 9 percent, down from
0.61 pounds to 0.55 pounds. This makes it seem worthwhile. However, the MoI
of the 0.44 pound driver head about the grip, at the end of a 44 inch shaft, is
0.44 × 442 = 852 pound-in.2 Because the mass of the shaft is distributed from
the head up to the grip, it turns out that its MoI about the grip is approximately
one-third of the shaft mass multiplied by the shaft length squared. Therefore, the
MoI of the heavier shaft is (1/3) × 0.17 × 442 = 110 pound-in.2 , while the value for
the lighter shaft is 71 pound-in. 2 The total MoI of the club has thus decreased from
852 + 110 = 962 pound-in. 2 to 852 + 71 = 923 pound-in. 2 , a proportional decrease of
0.96. However, the increase in swing speed is expected to be proportional inversely
to the 4th root of this value, which gives a proportional increase of 1/(0.96)1/4 = 1.01,
or a 1 percent increase! And even this is unlikely to produce higher ball speed
because of the small effect of the shaft mass during the impact—more on that later.
Of course, the lighter club will feel lighter when held and will also feel lighter
during the backswing when it is being raised at arm length. Unfortunately, it is very
unlikely to increase ball speed. This could easily be confirmed by Trackman, but
not surprisingly there seems to be an absence of any shaft manufacturers’ Trackman
data showing increased head or ball speed with lighter shafts. Tom Wishon (www.
wishongolf.com) is quite clear about the fact that the purchase of expensive light-
weight shafts is unnecessary.
These conclusions with respect to the driver apply equally well to the other clubs.
However, the head weights of the other clubs are selected following a long tradition
in club matching, to give the same “swing weight.” For this, a scale is used to ensure
that the clubs in a set are equally balanced, with the same holding force, about a
pivot point 14 inches from the end of the grip. Interestingly, when this is done the
MoI about the end of the grips varies by relatively small amounts. The average set of
irons has MoI values varying from 940 pound-in. 2, for the 3-iron to 880 pound-in.2 ,
for the pitching wedge, a steady decrease from the regular-shafted driver. According
to the fourth root rule, this would incur only plus or minus 1.5 percent swing speed
differences up and down the club range from the 5-iron. There has been some talk
in recent years about making sets of clubs with identical MoI values. It would seem
that such a change would produce only small value. However, if someone wishes
to try this, Cross and Nathan (2009) established a result to make it an easier task.
They showed that for any steady progression of club head weights, combined with
any constant length reduction between clubs, a pivot point distance from the grip
end can be chosen so that a constant MoI results from constant balance on the
“swing-weight” scale. Starting with the standard head weight of 0.528 pounds for
the 3-iron, they showed that moving the pivot point 18.5 inches from the grip end
would balance the MoI values exactly if the shaft lengths reduce progressively by
one-half inch while head weights increase progressively by 0.019 pounds at each
step rather than the standard 0.015 pounds. In the MoI balanced set, the pitching
wedge would weigh 0.661 pounds as opposed to the standard 0.633 pounds; that is
just over 4 percent heavier. So although the advantage may be small, there seems to
be no good reason for not having this scientifically correct!
So this brings us to the all-important swing.
THE SW ING
We know that nothing enhances ball speed and distance as much as increases in
impact speed. However, great precision is needed in ball striking. If a 100 mph
swing-speed drive is struck directly toward the target, but the face is angled only
2.5 degrees to the right or left of the club path, then it will curve in flight to land
approximately 30 feet wide of the target. A “good” golf swing, therefore, must com-
bine two attributes that do not easily come together. It should generate as much
impact speed as possible, and it should do this with a combined back- and down-
swing that ends precisely at the start point with the club head oriented in the start
position. To do this, the swing should revolve around a central pivot point, which
remains as still as possible, and with the simplest pivoting motions for precise
repeatability. The oft-repeated advice to the amateur player to “swing slowly” is sim-
ply because when a higher swing speed is attempted, the proportionately greater
inertial forces make it more difficult to maintain stability. Swinging fast, from the
top of the downswing, also invariably results in the highest club head speed being
reached too early—more on this later.
Reduced to its most simple form, the ideal golf swing can be represented by a
double pendulum as shown in the lower pane of Figure 3H. This is discussed in
detail in Cochran and Stobbs ([1968] 1999) Search for the Perfect Swing. High-speed
photography used in that study showed that the hands, or more precisely the center
of the left wrist, moved along an approximately circular trajectory with the center
of the circle just below the base of the neck. The main engine for the swing is then
a driving torque about this center, which in terms of the double pendulum model
is the main pivot. In addition to this main driving torque, there is a secondary and
much smaller torque from the wrists, which nevertheless can have a significant role
in developing club head speed.
The first move in the double pendulum model is illustrated in the upper pane.
This was sketched by the writer from video images of one of the world’s best play-
ers. It can be seen that this movement involves rotation of the hips, shoulders, and
arms acting almost as a single piece. It seems that this move may be best thought of
as a shoulder turn (Mike Kelly, personal communications, Fall 2014), although the
right leg is most likely providing some of the power. The fact that some of the power
of the swing must come from the legs is investigated later in a separate section on
“The power of the modern drive.” Some coaches recommend lifting the right heel at
the top of the backswing and then bringing it down as part of the execution of the
first move. The important point in this first move is that it does not move the pivot
sideways. Any lateral pivot movement at this point will result in a poor ball strike
(Mike Kelly, personal communications, Fall 2014).
The arrows on the pendulum model in Figure 3H are divided into three different
forms to avoid confusion. The ones with open arrow heads are the applied torques.
A continuous driving torque is required from the start of the downswing through
to the impact point. The first move should start smoothly and is overlapped in very
rapid succession with the shoulders driving the arms and the arms driving down-
ward and forward.
The mid-size arrows at the bottom swing position represent the motions of the
separate parts at impact. The small arrows define the angle of attack and the forward
hinge of the wrists. The sequence of the swing positions from the top represents
the downswing. However, for our purposes, the positions shown can also approxi-
mately represent the backswing. The upper arm of the double pendulum joining the
main pivot and the hands (or more correctly the wrists) starts the backswing and
a b
pivot
b
a
driving
torque
rotation speed
of arms
speed
hands
positive
wrist torque forward hinge
of wrists
rotation speed
of shaft
club head sp
attack angle eed
Figure 3H Initial move in downswing (upper pane) and double pendulum model (lower pane).
ends the downswing in a close to vertical position. At these points, the club shaft
(lower pendulum arm) is angled forward, away from the player, by an amount equal
to the club lie angle. Thus, the pendulum model in the lower pane of Figure 3H is
more appropriately viewed from in front and above the player, looking down at an
approximate 30 degree angle onto the “swing plane.” This angle involves two com-
ponents. The club is gripped when lying across the left palm at an angle of about
45 degrees. If the player stands with arms hanging straight down, the golf club
gripped correctly, and the wrists in the neutral position, then the club head should
be several inches off the ground. The wrists are then “cocked downward” about
15 degrees to bring the club into the address position behind the ball. Confusingly
in golf, this position is referred to as the “un-cocked” position because cocking refers
to the opposite motion, that is, raising the club upward by cocking the wrists. With
correctly fitted clubs, this position should be as far as the wrists will pivot backward;
that is, they have reached a backstop. It is important to note that the wrist cock is in
the plane of the hands. With the hands extended in front and the palms together, the
movement is from the little fingers angled downward to the thumbs angled upward.
This can be seen in Figure 3H, in the positions before impact, where the pivot is
caused by the wrist cock in the plane of the bottom edge of the club face. This is a
powerful pivot, with a fixed stop at the limit of the wrist cock. It can take the forces
involved in the reversal from the top of the backswing to the start of the downswing
without collapsing or “jackknifing.” In contrast, the wrist hinge, at right angles to
the wrist cock, plays no part in speed development. But it is essential that the wrist
hinge is fixed in its starting orientation throughout the swing so as to return the club
head along the correct swing path, ending in the correct path through the impact
zone. If this is accomplished correctly, then the back of the left hand should be trav-
elling toward the target through impact (Mike Kelly, personal communications,
Fall 2014).
From the address position, reaching the top of the backswing, as shown at left in
the upper pane of Figure 3H, involves four separate motions. The wrists are cocked
forward fully, the arms are lifted to the horizontal position and rotated backward,
and finally the shoulders and hips are twisted backward. Although these motions
overlap in the backswing, it is very instructive to do them in sequence, starting with
the wrist cocking. To do this, grip the club at the address position as firmly as pos-
sible with the middle, ring, and little fingers of the left hand. Add the relatively loose
grip of the index finger and a relatively loose right hand. With this grip, the average
player should be able to lift the club above the horizontal by just cocking the wrists.
Now try this with a harder grip including the index finger of the left hand. Because
of the extra tension of the tendons through the wrist, the amount of wrist cock is
now reduced; that is, the amount of rotation in the middle joint of our pendulum
model has been limited. This is why good golf instructors will advise a strong grip
of just the three fingers of the left hand. That does two things. It allows a substantial
wrist cock, which is needed to generate club head speed. The strong grip provides a
completely un-cocked stop, to which the hands can return during the downswing
so as to arrive back at the address position. With the wrists cocked, lift the arms to
the horizontal; and keeping the left arm straight, rotate the arms until the left arm is
lying across the chest and the passive right arm is bent at the elbow close to the body.
Now twist the wrists backward, and finally twist the right hip backward to extend
the swing and “coil” the body for the downswing.
Starting the downswing from this position, two of the ingredients needed to
bring the club head back to the ball with precision are keeping the left arm stiff
and maintaining the wrist hinge from the address position. The other ingredient is
maintaining a stable upper pivot position. This of course means maintaining a stable
athletic stance. The twisted-back wrists and the wrist cock should be held in posi-
tion until as late as possible in the downswing. This is illustrated in the pendulum
model in Figure 3H by the cocked wrist and edge of the club face remaining in the
swing plane until the final stage of the downswing. From this point a rapid uncock-
ing and roll back of the wrists and forearms increases the club head speed further
while closing the face at impact. Whoever said it was easy!
Additions of speed are aided by transfer of rotary momentum at each stage in

the downswing. The muscles that power the forward rotation of the shoulders and
torso, by reaction, slow down the hips. This is a mechanism for transferring momen-
tum from the hips to the shoulders. Next in line, the arms are powered separately
and receive a boost in rotational speed from the shoulders and torso being deceler-
ated. All of this is described in great detail by Cochran and Stobbs ([1968] 1999) in
their Search for the Perfect Swing. The system of interacting muscles does not of
course have the simplicity of a set of springs attaching hips to legs, shoulders to hips,
and arms to shoulders, which can be put into tension in the backswing and released
sequentially in the downswing. However, the principles are the same.
An important principle in the first stage of developing rotational speed is to
keep the golf club as close to the rotation axis as possible. We discussed earlier that
resistance to rotation is determined by the weight of objects and the square of their
distances from the axis of rotation. Specifically, a weight at half the distance from
the axis of rotation can be rotated at four times the speed for the same torque. Thus,
for a golf club, the important parameters are the distance of the golf club head and
the center of the shaft from the rotation axis. In Figure 3H, it can be seen in the
pendulum model that both of these distances, at the start of the swing, are about
one-quarter of the distances in the final stage just before impact. Thus the resistance
to rotation of the club is only one-sixteenth as much as it would be if the arms and
club were swung as a single unit. This is just one measure of the great efficiency of
the golf swing. The same distance comparison also applies to the swing resistance
of the arms, particularly the right arm, which are tucked in closer to the body at the
start of the downswing. As discussed in the Introduction, the sum of the separate
small weight elements multiplied by their distance squared from an axis of rotation
is the MoI. For the shaft, each element of weight along its length has to be allowed
for in a full MoI calculation. However, the MoI of the shaft about an axis of rotation
is least when the center of the shaft is nearest to the axis.
At around 70 percent into the downswing, the MoI of the club about the rotation
center is still less than half of that at impact. Up to this point, the skilled player has
retained a significant wrist cock in place. At the start of the downswing, this is not
difficult, since the inertia of the club holds the wrist cock against its fixed backstop.
However, holding the wrist cock for release later in the downswing requires care-
ful control by the player. It has been generally assumed in the literature that this is
achieved by applying a negative wrist torque. Recent work on swing mechanics by
Nesbit (2005), Nesbit and Serrano (2005), and Sharp (2009) suggests that this is
not the case. Instead, the strategy is to start the downswing slowly so as to reduce
the centrifugal force that acts to un-cock the wrists. This reserves sufficient wrist
cock for high acceleration late in the swing. Sometime after the shaft has reached the
horizontal in a good swing, the centrifugal force has reached a size where, with or
without small positive input from the wrist muscles, rapid un-cocking takes place.
Even without any positive wrist action, quite efficient transfer of rotational momen-
tum from the arms to the club will occur.
This is illustrated in Figure 3I where the current circular-arc trajectory of the
club head has been added to the swing position just a few hundredths of a second
before impact. The club head velocity is tangential to the downswing arc and so
creates a centrifugal force, at right angles to the swing arc as shown. This is just the
(–)
downswing arm rotation
arc slows
(+)
pivot pulled
e
lforc inwards
uga shaft rotation
trif head
cen speed speeds up
(exaggerated effects of centrifugal force in gray)
Figure 3I Increasing club head speed in the final stage of the downswing.
same as the force that would put tension into a rope if you were to swing it around
your head with a club head attached to the end. The difference in this case is that the
force is inclined to the inside of the two hinged members. The effect of this force on
the club head and the double pendulum is shown in gray. It is seen to move the club
head outward, so the wrist pivot point is forced to move inward, that is, automati-
cally un-cocking the wrists. This increases the counterclockwise rotation speed of
the shaft and at the same time decreases the counterclockwise rotation speed of the
upper arms as shown by the gray rotation speed change arrows labeled (+) and (−).
The upper arm is still being powered by the driving torque. It seems that this cancels
out the effect of the centrifugal force on the arms, and the hands typically move
through this final phase of the swing at constant speed. Thus, the driving torque
is going straight into speed increases of the club head. This final-stage transfer of
kinetic energy from the arms to the club head only occurs if significant wrist cock is
still in place at the beginning of this final swing stage.
We can go back to Search for the Perfect Swing for data that shows these relation-
ships quite clearly. The investigators in this study took high-speed photographs,
using strobe lighting, of three British Ryder Cup players: Bernard Hunt, Geoffrey
Hunt, and Geoffrey Wolstenholme. The one with the latest wrist cock release
of these three players was Geoffrey Hunt, and position-time data of his swing is
shown in Figure 3J. The first thing to be noticed is that the wrists do follow very
closely to the dashed-line curve, which is a circular arc. This illustrates the basic
validity of the double-pendulum model. The plots of the other two Ryder Cup
players also align very closely to the double pendulum. The first ring of numbers
around the outside of the swing are the times in seconds of the different positions,
starting from zero at the beginning of the downswing. Note that the first time step
is 0.05 seconds, the second one 0.04, the next three are 0.02, and the remainder
0.01. These changes in the time increments hide the smooth increase in the club
head speed through all of the time steps except for the last one-tenth of a second.
Extra dashed radial lines have been inserted at the beginning of the downswing to
illustrate the slow start.
The average speed between each of the time increments can be easily calculated. A
digital caliper was used in my measurements of the gaps between the successive club
head positions on the plot. The “club length” from head to wrist was also measured
49.2
55.4 40.1
0.15 0.13
61.5 0.16 0.11 27.8
0.17 0.09
68.9
18.5
0.18
0.05
80.0 6.2
0.00
0.19
+ increments
80.0 of 0.01 seconds
centrifugal 0.20
force
86.1
0.21
92.3
0.22
104.4 0.23
0.24
0.25
104.6
101.5
centrifugal
force
Figure 3J The swing of Ryder Cup Player Geoffrey Hunt superimposed on the double-pendulum
model from Search for the Perfect Swing (1968). Outer numbers are times in seconds and speeds in
miles per hour.
on the same plot. The scale of the plot was determined by assuming the club length to
be 44 inches. The average speed in each time increment is then simply the scaled gaps
divided by the time increment. These values are shown as the outer ring of numbers.
There is likely some significant level of error in the original recording process
and my measuring procedure. Nevertheless, with some hiccups around 0.18 sec-
onds, and a matter for discussion at the end of the swing, there is a steady speed
increase through the swing. The speed of the wrists also increases slowly until about
0.18 seconds and then slows to approximately constant speed as the wrist cock is
rapidly released during the last 0.05 seconds.
A plot of these club head speeds is shown in Figure 3K. The circles represent the
average time-interval speeds calculated in the way described. The smooth curve is
the best-fitting parabola. A parabola will fit to the data only if the acceleration is
increasing linearly with time. At the very end of the swing, in the last 0.025 seconds
it seems that the acceleration of the club head stalls and maybe even reverses in the
last time step. This speed decrease is validated to some degree by the fact that the
hands are seen to speed up in the last time step (see Figure 3J). However, it should
be noted that what has really been measured here is the swing speed of the shaft and
by extension the speed of the head in the swing plane, equivalently recorded on the
measured speeds speeds with continued trend

120
100
Club head speed, mph
80
60
40
20
0
0 0.05 0.1 0.15 0.2 0.25
Time, seconds
Figure 3K Club head speed during the downswing of Geoffrey Hunt. Circles indicate calculated
values from displacement-time plot. Square symbols are predicted speeds if the trend of
increasing acceleration had continued to impact.
plane of the photograph. In the last part of the swing, while club head speed in the
plane appears to slow down, the head is being rotated with increasing speed to close
the face. This is represented by the final square symbols.
Much more precise golfer swing data has been obtained by Nesbit (2005) in an
investigation supported by the National Science Foundation and the USGA. This
work was referenced previously with regard to the delay in wrist un-cocking. Nesbit
carried out a detailed and comprehensive experimental and modeling investigation
of the swings of golfers with a range of skill levels, from zero to 20 handicap. For
the experimental work, a multicamera motion analysis system was used to track
the position of 13 reflective markers attached to the players and the club. A total of
85 subjects were analyzed. In addition, a full body model of the player and the
club was built using a 3-dimensional modeling software system from Mechanical
Dynamics Incorporated. This can be described as a fast variant of the widely used
finite element modeling in which the entire volume is described by a fine mesh of
small rigid links attached by flexible connections. So we have gone from a two-link
model to one containing thousands of links. Both have an important role to play.
Even with all of this computer sophistication, the double pendulum is still of great
value in visualizing the principal elements of an efficient swing. The important part
about Nesbit’s work is that when the model has been fitted to the swings, much more
detail is then open to investigation.
Of particular relevance to this discussion, Nesbit found that the scratch golfer
(zero handicap), selected for detailed study, reached maximum swing speed con-
sistently 0.025 seconds before impact. The higher handicap players all reach maxi-
mum swing speed later, in the range of 0.01 seconds to 0.02 seconds after impact.
More importantly, the scratch golfer reached maximum club face speed at the
impact point while the higher handicap players reached maximum speed after pass-
ing through the impact zone. These results, shown in Figure 3L, have huge implica-
tions for improving the golf swing.
120
club face
scratch golfer
90 5 handicap
13 handicap
Speed, mph 18 handicap
60
30 wrists
0
–0.3 –0.2 –0.1 0.0 0.1
Time before impact
Figure 3L Speed of club face and wrists for a range of golfer skills; from Nesbit (2005).
First, considering the top group of curves in Figure 3L, all of the golfers except
the elite player reach maximum club face speed too late. Particularly the 5 and
18 handicap players would have reached substantially higher club face speeds if the
ball did not get in the way! It would seem that, as discussed previously, generat-
ing maximum swing speed too late, and not finishing the wrist roll soon enough,
is probably the culprit. Even more interesting is the speed of the wrists through
the swing. We can see from the lower group of curves that amazingly the maxi-
mum hand speed is the same for all the golfers. The club head speeds at about
−0.075 seconds (0.075 seconds before impact) are not so different: a spread of
about 8 mph. However, the accelerations at this point are quite different, with the
scratch golfer having about twice the acceleration of the 18 handicapper. So the
final difference in club head speeds at the impact point is all about the un-cocking
of the wrists. One last observation helps to explain this. The scratch golfer’s wrists
start to move appreciably more slowly than the higher handicap players’, and his
club head starts to move significantly more slowly. These can be seen clearly in
Figure 3L in the time increment −0.3 to −0.2 seconds before impact. The club head
speed differences might be explained by a jerking motion at the start of the back-
swing by the high handicapper, while the elite player starts the club with nothing
more than a small torque increasing smoothly from zero. And perhaps more impor-
tant, as discussed earlier, the slower club head of the elite player over the first half
of the downswing, to −0.015 seconds, would have reduced the amount of wrist
un-cocking up to that point.
A final link back to the Geoffrey Hunt swing position-time plot, in Figure 3J, is
worthwhile. The centrifugal force is always at right angles to the swing arc. It can
be seen that at 0.2 seconds, or 0.05 seconds before impact, the force is horizontally
outward and inclined as shown at a significant angle to the club shaft. This has a
very positive effect on increasing the rotational speed of the club. However, as the
club head approaches the impact point, the force is almost parallel to the shaft and
so much less effective, as also illustrated in Figure 3J. This is the point at which the
wrist rotation speed is peaking and thus can compensate for the slightly earlier peak
in swing speed.
Data published by the Trackman Company concerning speed variations across

the face of the driver, striking at 100 mph, suggests that a typical time to close the
face for a player with this swing speed is 0.035 seconds. In this time period, the
wrists have been rotated to close the face by 90 degrees or one-quarter of a full turn.
The average rotational speed is thus (1/4)/0.035 = 7.14 revolutions per second, or
428 rpm. At the same time, the primary swing of the arms continues to bring the
bottom edge of the club back to the horizontal plane. It is easy to visualize this sec-
ond rotation by adopting the position of the player in the upper right of Figure 3H.
Maintaining this position, un-cock the wrists and rotate the wrists and forearms.
The edge of the club head is now pointing downward, positioned above and behind
the ball. From here, just the rotation of the arm in the vertical plane will bring the
club head to the striking position. The speeds of these two rotations are similar
since they start at around the same point in the swing and both involve rotations of
about 90 degrees. These rotations produce significant speed variations across the
large face area of the driver. In mechanics, rotations can be moved to any point of
particular interest provided the velocity of that point is adjusted accordingly. In this
case, we can move the rotations to axes passing through the face center. The speed
of the center of the face is then the combined effect of all of the actions of the swing
leading up to the impact. We can think of it as the speed of the end of the shaft bur-
ied in the hosel plus the added effect of the complex set of rotations operating over
the offset distance from the end of the shaft to the center of the face. All of this can
be avoided by using a Trackman system and golf robot to simply increase the swing
speed incrementally until the face center is travelling at a target speed through the
bottom of the swing. Figure 3M shows the face speeds across a driver face for a
100 mph face center speed at the bottom of the swing.
These speed variations are consistent with a single rotation of 600 rpm about
an axis at 45 degrees across the center of the face from low heel to high toe. This
is the expected result of the combined separate rotations of approximately equal
speed: one caused by the rolling of the wrists and one by the swing of the arm.
In fact, if the resulting rotation is 600 rpm at 45 degrees, then separate rotations
424 rpm
600 rpm
98.2
100.9 95.5
0.5in
102.7 100 97.3

0.75in
424 rpm
104.5 101.8 99.1
Figure 3M Speed variations (in mph) on a driver face (adapted from Tuxen 2008).
must equal 424 rpm since rotation speeds are added geometrically so that 6002 =
4242 + 4242 , in agreement with our rough calculation in the previous paragraph.
The face speed map suggests that, from a ball speed perspective, the worst place
to strike the ball is at the high heel position. If the point is three-quarters of an inch
toward the heel and one-half of an inch high, the face speed would be 100 − 2.7 −
1.8 = 95.5 mph. Even worse, with the structural connection to the hosel, this is a
much stiffer portion of the face giving less spring effect and a lower value of CofR.
The vertical component of this rotation produces a small component of ball spin
about its vertical axis through a coupled gearing effect between the club head and
the ball during impact. This gives a slicing tendency to the driver, which only pro-
duces a straight shot if the face is slightly closed with respect to the club path—
much more on this in Chapter 6.
We now switch attention to divot-taking shots off the fairway.
CONTROLLING THE CLU B LOFT

As shown in Figure 3N, the effect of forward hinged wrists is to produce a negative
angle of attack on the ball, and as a consequence a divot is taken following the ball
impact. Because the hands have passed the lowest point prior to the impact, they are
now moving in a slightly upward direction while the club head is still descending.
This lowers the path followed by the club head so that the ball strike is in a direc-
tion below the direction normal to the shaft. The amount by which this is lowered
produces a corresponding reduction in the loft at impact. These relationships are
shown on the right of Figure 3N, where the club loft, reduced by the amount of
de-loft, is the actual loft applied to the ball in the impact. The actual loft will be
referred to as “dynamic loft,” as opposed to the static measurement of the angle
between the club face and the shaft axis. The latter will be referred to as “club loft.”
speed
hands
club loft
normal to
club speed shaft
face
normal to
club speed dynamic loft
Figure 3N Actual or dynamic loft resulting from a downward attach angle.

The amount of de-lofting of a club depends on the magnitude of the negative

attack angle and the speed of the hands through the impact. If the hands were
brought to a virtual stop at impact, then the club head speed would be along the line
normal to the shaft and there could be no de-lofting. Alternatively, to sweep the ball
off the fairway requires a horizontal club path or zero attack angle. For this to occur,
there can be no forward wrist hinge at impact, thus both the shaft normal and the
club speed are horizontal at impact, again with no de-lofting.
Determining the actual amount of de-lofting requires some fairly difficult
geometry and is reserved for later in the chapter. There it is shown that with
realistic hand and club head speeds, and negative attack angles between 3 and
5 degrees, the amount of de-lofting can range from 5 to 9 degrees: approximately
one to two club lofts less. This is exactly what elite players do. A comparison of
club lofts and the actual or dynamic lofts of the average PGA Tour player is given
in Table 3A together with the average attack angle. It can be seen that starting
with the 5-wood, the amount of de-lofting is in some cases equal to two club loft
differences. To provide data for later in the chapter, Table 3A also contains the
typical club head weights, average club head speeds, launch angles, ball launch
speeds, and backspin rates.
As shown in row 1 of Table 3A, in contrast with the other clubs, the average PGA
Tour player driver applies almost 5 degrees more dynamic loft than the actual aver-
age driver club loft. The loft change in this case is the result of an entirely different
mechanism related to the relative flexibility of the driver shafts. For the driver, the
center of mass of the head is at a significant distance behind the shaft. In the final
stage of the swing, as the rotational speed of the club increases, the large centrifugal
Table 3A Average data for PGA Tour players
Clubs Head Club Club Attack Launch Dynamic Ball Back-spin

weight loft speed angle angle loft* speed (rpm)
(pounds) (deg) (mph) (deg) (deg) (deg) (mph)
Driver 0.440 9.5 112 −1.3 11.2 14.4 165 2,685

3-wood 0.473 16.5 107 −2.9 9.2 14.8 158 3,655
5-wood 0.495 21.0 103 −3.3 9.4 16.0 152 4,350
Hybrid 0.517 22.5 100 −3.3 10.2 17.0 146 4,437
3-iron 0.528 20.5 98.0 −3.1 10.4 17.2 142 4,630
4-iron 0.543 23.5 96.0 −3.4 11.0 18.4 137 4,836
5-iron 0.559 26.5 94.0 −3.7 12.1 20.5 132 5,361
6-iron 0.574 30.5 92.0 −4.1 14.1 23.8 127 6,231
7-iron 0.590 34.5 90.0 −4.3 16.3 27.4 120 7,097
8-iron 0.605 38.5 87.0 −4.5 18.1 30.6 115 7,998
9-iron 0.620 42.5 85.0 −4.7 20.4 34.2 109 8,647
Wedge 0.636 48.5 83.0 −5.0 24.2 39.7 102 9,304
* Predicted values from Chapter 4.
force acts through the offset center of mass. This causes the slender tip of the shaft to
bend forward, which acts to increase the loft.
The loft is increased more if the shaft is made more flexible closer to the hosel.
This is illustrated in the right panel of Figure 3P where the flexible section is repre-
sented by a hinge. It can easily be seen that the closer the hinge is to the club head,
the more the loft will be increased as the centrifugal force moves toward alignment
with the shaft axis. Flexible regions near the tip of the shaft are often referred to as
“kick points” to give the impression that they add to the speed in some way. Their
main purpose is to increase the loft at impact and produce the high, soaring drives
that are the mark of the long-distance drivers of the ball. For PGA Tour players, they
produce the average dynamic loft of 14.4 degrees, a launch angle of 11.2 degrees to
the horizontal, and ball trajectories that rise to an average height of 90 feet.
This amount of shaft forward bending occurs because of the surprisingly large
magnitude of the centrifugal force. Centrifugal force is given by mass multiplied
by velocity squared and divided by the arc radius along which the object is trav-
elling. Using metric units for easy calculation, we can get an estimate of the cen-
trifugal force created by the average PGA Tour player by neglecting the forward
movement of the hands at impact. In this case, the final part of the swing can be
considered as a single pendulum of length equal to the shaft length: approximately
one meter. The mass of the head is 0.2 kilograms, and the average head speed at
impact is 50 meters per second (112 mph). The centrifugal force just before impact
is 0.2 × (502)/1.0 = 500 newtons = 112 pounds, a quite amazing 250 times the
static head weight. This force decreases with the square of the velocity, so the aver-
age male player, with a 87 mph swing speed, would produce a centrifugal force of
112 × (87/112)2 = 68 pounds. Moreover, if the wrist cock is released early, as is the
case with most low-skill players, then we have a single pendulum at impact, extend-
ing from the club head up through the arms to the main pivot—a length nearer to
2 meters. In this case, the centrifugal force becomes 68/2 = 34 pounds, insufficient
centrifugal force
head speed
dynamic loft
club loft
Figure 3P Forward bending of the driver head to increase dynamic loft.

Table 3B Approximate centrifugal force values
Swing speed (mph) Centrifugal force (pounds)

Wrist cock release
Early Late
80 29 57
90 36 73
100 45 90
110 54 109
to activate the so-called kick point by any useful amount with a regular flex shaft.
The approximate centrifugal force values for different swing speeds and early or late
wrist un-cocking is given in Table 3B.
We can note that the increase in driver loft is a matter of club head and shaft
selections and not typically of player control. However, players could deliberately
un-cock the wrists early, or never cock the wrists in the backswing. This would
decrease loft and would be accompanied by a decrease in impact speed. This can be a
useful strategy for higher-handicap players to employ in some over-long par-3 shots.
Further evidence of the forward shaft bending can be seen with reference to
Milne and Davis (1992). These authors determined the bent shape of a golf shaft at
different stages in the golf swing using the double-pendulum model. These results
were assembled into a complete picture of the downswing by McGinnis and Nesbit
(2010). The predicted shaft shapes, at incremental times in the downswing of an
elite player, are shown in Figure 3Q. It can be seen that the shaft is bent backward at
the top of the swing due to the inertia of the head as the hands come to a rest. This
backward bending increases, also due to the club head inertia, as the hands start to
accelerate on the downswing.
According to Milne and Davis, maximum backward bending occurs approxi-
mately 0.1 seconds into the downswing for a typical skilled golfer. By 0.19 sec-
onds, the centrifugal force from the offset center of mass of the head, has helped to
straighten the shaft into a shallow double curve. Then, during the last 90 degrees of
downswing, lasting only 0.035 seconds, the shaft takes up its forward bent shape
prior to impact. I have added the club head orientations to Figure 3Q to show that
the club face is rotated from 90 degrees open to square to the ball sometime in the
final 0.035 seconds.
The Milne and Davis predictions are supported by experimental studies using
strain gages fitted axially at positions along the club axis (Horwood (1994); Butler
and Winfield (1994)).
McGinnis and Nesbit (2010) utilized data from actual player downswings.
The data was gathered using a set of eight high-speed cameras, which provided the
velocity and acceleration inputs to an analytical model. The amount of forward head
tilt, predicted by this process for a “scratch” golfer, can be estimated from figures
in their article to be as high as 8 degrees. This seems surprisingly high. However,
we know that the average dynamic loft of PGA Tour player drives is approximately
5 degrees greater than the club loft.
102 ms
0
161 ms
190 ms
225 ms
Figure 3Q Model of a golf shaft during the downswing. Adapted from Milne and Davis (1992) as
in McGinnis and Nesbit (2010).
These conjectures about shaft bending are supported in an extensive investi-

gation of shaft custom fitting by Wishon (2013). He states categorically that the
“key swing characteristic to determine the correct tip stiffness design of the shaft
for the golfer is the point at which they un-cock their wrist on the downswing.”
For his development of shaft fitting methods, Wishon (2013) established a Shaft
Bend Profile standard test, which involves stiffness measurement at seven desig-
nated points along the shaft length. His company (Tom Wishon Golf Technology)
offers a database of shaft stiffness profiles that currently has test data on 2,000
commercial shafts. Within this database, it can be found that even regular stiffness
(R-designated) shafts from the same manufacturer can have huge overall differ-
ences in stiffness. The softest of this range are deemed by Wishon to be most suitable
for the 60 mph club head speed golfer, while the stiffest are suitable for a 100 mph
club head speed. With this wide degree of variation, there can obviously be no valid
relationships for swing speed versus inertial loft for the retail shaft stiffness desig-
nations, particularly when assembled to different heads and trimmed to different
lengths. This is where professional players gain much from custom fitting of shafts
to fit their particular swings and swing speeds.
It should be noted that with the shaft inclined away from the player by around
30 degrees in the last stage of the downswing, these forces have a vertical com-
ponent of approximately 85 percent of the total value and a horizontal compo-
nent approximately one-half of the total value. The vertical component is resisted
by the inertial mass of both the approximately vertical arms and, through the
shoulder joints, the inertial mass of the torso. In contrast, the horizontal com-
ponent can relatively easily cause the club head to be drawn outward on a wider
arc. A strong left-hand finger grip is required to tighten the tendons and resist
this outward arcing tendency, and of course firm arm muscles are a positive. The
often heard advice to grip the club like holding a small bird has no doubt worked
its way down from elite players with left-hand, steel-like grips who may get the
impression that they are gripping lightly (Mike Kelly, personal communications,
Fall 2014).
EFFECT OF LOFT ON BA LL LAU NCH A NGLE

The loft or wedge face angle of the golf club accomplishes two positive effects and a
negative one. Most importantly, it produces high amounts of backspin, which cre-
ates lift, enables high trajectories with steep descents, and together with the brak-
ing effect on the turf of the backspin, enables the player to stop the ball in short
distances. Loft also increases the launch angle, which allows obstacles from bun-
ker lips to high trees to be cleared with ease. The negative effect of loft is that it
decreases ball speed. We deal with these effects separately, starting with the launch
angle because it provides useful information about spin.
The two arrows in Figure 3R show the direction of the club head strike on the
left, referred to as the club path, and the resulting direction of the ball launch on
the right. The club path in this side view is defined by the “attack angle,” which is
measured from horizontal. In the case shown, we have a negative angle of attack.
Recall that the club path defines the dynamic loft, which is the angle between
the direction of the club head at impact and the line passing through the point of
impact at right angles to the face. We should also note, as discussed earlier, that for
an iron the center of mass of the head is close to the face, the shaft is stiffer, and so
the inertial force produces relatively little forward bending of the shaft compared
to the driver.
It is instructive to look at the ratio of launch angle to loft angle. To do this in
a way that leads to some fundamental relationships, it is necessary to eliminate
the attack angle. The only important measure is the launch angle with respect
to the club path. We will refer to this as the “normal” launch angle, which is
clearly the launch angle to the horizontal plus the attack angle magnitude for a
negative attack angle. Table 3C shows the normal launch to dynamic loft ratios
for the average PGA Tour player data.
back spin
attack angle
launch angle
dynamic
loft
Figure 3R Illustration of club direction before impact with ball direction after impact.
Table 3C Average data for PGA Tour players
Clubs Head Club loft Attack Dynamic Launch angle Normal

weight (deg) angle* loft (deg) launch/loft
To hori- Normal
(pounds) (deg)
zon* (deg) (deg)
Driver 0.440 9.5 −1.3 14.4 11.2 12.5 0.87

3-wood 0.473 16.5 −2.9 14.8 9.2 12.1 0.82
5-wood 0.495 21.0 −3.3 16.0 9.4 12.7 0.79
Hybrid 0.517 22.5 −3.3 17.0 10.2 13.5 0.79
3-iron 0.528 20.5 −3.1 17.2 10.4 13.5 0.78
4-iron 0.543 23.5 −3.4 18.4 11.0 14.4 0.78
5-iron 0.559 26.5 −3.7 20.5 12.1 15.8 0.77
6-iron 0.574 30.5 −4.1 23.8 14.1 18.2 0.76
7-iron 0.590 34.5 −4.3 27.4 16.3 20.6 0.75
8-iron 0.605 38.5 −4.5 30.6 18.1 22.6 0.74
9-iron 0.620 42.5 −4.7 34.2 20.4 25.1 0.73
Wedge 0.636 48.5 −5.0 39.7 24.2 29.2 0.74
* Data reproduced with permission from Trackman A/C, Vedbaek, Denmark.
It can be seen in column 8 that the normalized launch angle decreases as a per-
centage of the loft angle as the loft angle increases. This is because the increased
face friction, generated with higher wedge angle (loft) at impact, pushes the ball
relatively lower from the normal direction of the club face. These values are deter-
mined in the modeling section at the end of the chapter. In contrast, if the club face
was perfectly smooth (frictionless), the ball would slide up the face and bounce off
at right angles to the face, and column 8 would contain only 1.0 values.
The sudden increase of the launch angle to loft angle ratio for the driver, as
shown in Table 3C, is a result of the smooth face of the modern titanium driver.
This produces less friction between the face and ball, resulting in a launch closer to
a right angle with the face. For the same reason, it also produces less backspin on the
ball, which is almost certainly a disadvantage for the slower swing-speed amateur
players.
EFFECT OF LOFT ON BA LL SPEED

All other things being the same, the list of factors contributing to ball speed, which
included club head speed, CofR, and ball and club head mass, now must be extended
to include dynamic loft angle. The key to making the extension is that ball speed is
generated almost exclusively by the compression and restitution of the ball, and this
happens at right angles (normal) to the face. So the important parameter in the impact
is not the speed of the club but its component of speed normal to the face and the rela-
tionship between this velocity and the velocity of the ball at right angles to the face.
The CofR applies to these two velocity components and is more correctly referred to
as the “normal CofR.” The relationship between the actual club speed and its normal
component can be seen geometrically by inscribing them inside a circle on the club
face as shown in Figure 3S. The normal speed component is of course given by
normal club speed = club speed × cosine(dynamic loft angle).
The normal component of ball speed is then defined by the CofR acting at right
angles to the face and the requirement that momentum is conserved at right angles
to the face. This is a little more complex to deal with and is laid out fully in the mod-
eling section. The normal component of ball speed is related to the actual ball speed
by the cosine of (loft - normal launch angle), which is also illustrated in Figure 3S.
It can be seen in Table 3.1 that (loft - normal launch), column 5 minus column 7,
does not exceed 10.5 degrees over the full range of shots. Cosine(loft - launch) lies
between 1.000 and 0.983 over this range, and therefore can be neglected with maxi-
mum error of 1.7 percent for the wedge shot.
Thus, compared to the horizontal impact with negligible loft analyzed earlier,
both the club impact speed and the resulting ball speed are reduced by the same
factor cosine(loft). This means that both conservation of momentum and the CofR
relationship remain satisfied as with the application of the other factors.
For example, with reference to Figure 3D, it was shown that with 0.83 CofR
and 110 mph impact, the ball speed would be 163.9 mph if the loft angle effect was
neglected. If the dynamic loft was 14.4 degrees, as for the average PGA drive, then
the ball speed would reduce to 163.9 × cosine(14.4) = 158.8 mph. The normal club
head speed would be 110 × cosine(14.4) = 106.5 mph, and the normal club head
speed after impact would be 72.6 × cosine(14.4) = 70.3 mph. Since all three speeds
have decreased in the same proportion, they will still satisfy both CofR equal to
0.83 and conservation of momentum. Note that we are now concerned with con-
servation of momentum in the direction normal to the face. This is fine because
momentum conservation is always satisfied in any chosen direction.
So the list of factors influencing ball speed simply requires the addition of a
cosine factor to be complete. The first three factors are repeated for completeness.
Ball speed is proportional to
(a) Club head speed

(b) Restitution factor = (1 + CofR)
normal club speed
normal ball speed
ball speed
(loft-normal launch)
loft club speed
tangential club speed
Figure 3S Geometric relationships between club impact speed and ball speed.

(c) Club mass factor = Club head mass/(Club head mass + Ball mass)

(d) Wedge angle factor = Cosine(dynamic loft)
The effect of loft on ball speed over the range of dynamic loft angles is shown in
Figure 3T. The subranges indicated for the driver, 3-iron, 7-iron, and pitching wedge
(PW), extend from the average PGA Tour player dynamic loft up to the actual club
loft. For the high-handicap player, the difference between club loft and dynamic loft
is not usually significant; so the upper ends of the ranges apply. The other clubs have
been omitted to avoid confusing overlaps.
The club manufacturers have been forced to compete over this relationship for
some time. As one example, a modern 5-iron typically has club loft of 26.5 degrees,
and two decades ago, the 5-iron was more likely a 30.5 degree lofted club. The ball
speed losses for these two clubs, compared to a zero-loft face with the same weight
and same CofR, are 10.0 percent and 14 percent, respectively. Alternatively, on a
scale of performance from 100 for zero loft, the modern 5-iron is a 90 and the old
5-iron was an 86. So the modern 5-iron, which is simply a re-branded 4-iron, hits
balls 4 percent faster and about 4 percent further. In contrast, on a playability scale,
the average player who could never hit a consistent 3-iron shot now leaves the 4-iron
in the bag. This situation has been mitigated in large degree by the emergence of
hybrid clubs, which are more forgiving in ball striking. Of course this process of
loft reduction left a large 8-degree gap between the pitching wedge and the sand
wedge, for which the manufacturers could sell us another club appropriately named
the “gap” wedge!
Without any more reliable sources of information, we will assume that the pro-
portional change in swing speed between each club for the average PGA Tour player
can be applied generally to other players. From Table 3A then we get the relative
swing-speed values given in Table 3D. For example, the club head speeds for the
driver and fairway wood in Table 3A are 112 and 107 mph respectively. These are in
the ratio of 100 to 96 to the nearest mile per hour. Hence the entry of 96 in column 2
of Table 3D.
Before leaving this section, it might be instructive to use all of the previous
information to compare the ball speed produced by a particular player using a
driver and a pitching wedge, respectively. Assume the player has a swing speed
of 100 mph with a 12 degree dynamic loft drive. As previously determined, this
impact with zero loft would produce a ball speed of 149 mph with 0.83 CofR. With
a 12 degree loft, Figure 3T gives approximately 3 percent reduction of ball speed
to 144.5 mph.
Table 3D Relative swing speeds of the different clubs
Driver 3-wood 5-wood Hybrid Irons PW

3.0 4.0 5.0 6.0 7.0 8.0 9.0
100 96 92 89 88 86 84 82 80 78 76 74
Percentage decrease in ball speed 40

33% reduction
30
20% reduction
20
10 8% reduction
3% reduction
3-iron 7-iron p. wedge
0
0 10 20 30 40 50
driver Club loft angle, degrees
Figure 3T Percentage decrease in ball speed with club loft.
Moving now to the pitching wedge
(a) Because of the relative club head speed of 74 percent for a pitching wedge,
from Table 3D, the effect of club head speed, before considering other
factors, is a reduced ball speed to (0.74)149 = 110 mph, or a reduction of
26 percent.
(b) Without better information at this point, we will assume that the CofR of the
wedge is 0.80, compared to 0.83 for the driver. The restitution factor then
changes from 1.83 to 1.80, and the ball speed will therefore decrease further,
with respect to the driver, by the ratio (1.80/1.83); that is, a 1.8 percent decrease
from 110 mph to 108 mph.
(c) The mass factor for the wedge with a 0.1 pound ball, referring to Table 3A, is
0.636/(0.636 + 0.1) = 0.864, compared to 0.44/(0.44 + 0.1) = 0.815 for the
driver. This increases the ball speed to 108(0.864/0.815) = 114.5 mph, an
increase of 6 percent.
(d) Finally, we apply the wedge effect of the loft angle. If we assume the club is
de-lofted to the middle of the PW range in Figure 3T, then this gives an approxi-
mate decrease of 27 percent; or equivalently, a wedge angle factor of 0.73. The
final estimate of ball speed is therefore 0.73(114.5) = 84 mph.
We can apply these factors in any order, so the percentage change is the measure
of a factor’s relative importance. In this case, the changes in swing speed and loft
angle are equally important and have an effect that far outweighs the changes in
club head mass and CofR.
For a given swing speed, the only tactic available to strike the ball further with
the shorter irons is to decrease the loft angle. This is exactly the strategy used
by highly skilled players, as seen in the differences between columns 3 and 7 in
Table 3A.
ESTI M ATING A PPROX I M ATE VA LU ES FOR Cof R

We now have almost all the tools needed to predict ball speed for a particular ball
strike. The only thing missing is the ability to select a reasonable CofR value for a
particular club and swing speed. A sensible first step is to obtain estimates of typical
CofR values for the PGA players. We will calculate the value for the average drive
and just tabulate the remainder.
From Table 3A, the driver head speed is 112 mph and the ball speed is 165 mph.
Starting with the ball speed, we can simply divide repeatedly by the three other
ball-speed factors, so the end result will be (1 + CofR).
(a) Dividing by the club head speed gives 1.473, which is usually referred to as the
smash factor.
(b) The mass factor for the driver is 0.44/(0.44 + 0.1) = 0.815. Dividing this into
1.473 gives 1.808.
(c) Cosine(dynamic loft) = cosine(14.4) = 0.969. Dividing this into 1.808 gives
1.867 so the CofR is 0.867.
The CofR values calculated for all the average PGA player ball strikes are shown in
Table 3E. Note that for a 45 gram ball (0.099 pounds) the calculation above gives a
CofR value of 0.863. This is the ball mass used in all the science modeling sections,
which are the source of the calculated values in the tables.
Up to this point we have followed modeling procedures that are universally
accepted in the literature on golf science. However, they led to a glaring contradic-
tion between the calculated CofR values in Table 3E and accepted values in the golf
world. The central issue is the calculated CofR for the driver, which substantially
Table 3E Average data for PGA Tour players
Clubs Head weight Club loft Club speed* Dynamic Ball speed* Normal
(pounds) (deg) (mph) loft (deg) (mph) CofR
Driver 0.440 9.5 112 14.4 165 0.863

3-wood 0.473 16.5 107 14.8 158 0.847
5-wood 0.495 21.0 103 16.0 152 0.842
Hybrid 0.517 22.5 100 17.0 146 0.819
3-iron 0.528 20.5 98 17.2 142 0.802
4-iron 0.543 23.5 96 18.4 137 0.778
5-iron 0.559 26.5 94 20.5 132 0.764
6-iron 0.574 30.5 92 23.8 127 0.769
7-iron 0.590 34.5 90 27.4 120 0.754
8-iron 0.605 38.5 87 30.6 115 0.787
9-iron 0.620 42.5 85 34.2 109 0.798
Wedge 0.636 48.5 83 39.7 102 0.846
109 mph 36.64 mph 53.84 mph
pre-impact post-impact
Figure 3U Ball cannon test speed and maximum allowable rebound speed.
exceeds the maximum allowable CofR for competition use, namely, 0.830. This
CofR limit is scrupulously enforced on both manufacturers and competition play-
ers by the R&A/USGA. As described earlier, two fundamental tests are carried out
using ball cannons and ball speed measuring equipment. First, a standard circular
diaphragm plate having the weight of a driver head is used to test balls. The diaphragm
plate has the same diaphragm spring effect as a highest allowable performance tita-
nium driver head. Batches of new balls entering the market are fired against the test
plate to ensure that they do not rebound with speed greater than would be possible
with 0.83 CofR; any that rebound faster are deemed non-conforming and disal-
lowed for tournament play.
Club heads are then tested by firing conforming golf balls normal to the face of
a club head and ensuring the rebound speed does not exceed the 0.83 CofR limit.
The club head test procedure is illustrated in Figure 3U. The impact speed for the
test is 109 mph, and the maximum allowable ball rebound speed is 53.84 mph. The
club head is projected backward in the impact at a speed of 36.64 mph. The reader
may wish to check that these speeds satisfy 0.83 CofR and also that the momentum
values before and after the impact are the same. For the latter, it is important to note
that momentum has direction as well as magnitude. The original momentum, in
this case the ball momentum, is taken as positive from right to left. The momentum
of the rebounding ball is counted negative. It is easy to see that this must be the case.
If we were to compress a spring between the ball and the club face the momentum
would be zero. If we then release the pair, they would fly off in opposite directions
and with speeds that must make the momentum still zero. So obviously they would
possess equal and opposite momentum.
EFFECTI V E I MPACT M ASS

The important aspect of these two tests is that the only mass involved in the impact
is the 0.1 pound ball and the 0.44 pound driver head or test plate. In contrast to
this, when a 0.1 pound golf ball is struck by a driver, the impacting mass involves
a 0.44 pound driver head and a shaft with mass somewhere in the range 0.15 to
0.2 pounds. Since the shaft mass has considerable momentum going into the
impact, and at least some of it must slow down during the impact, it is inconceiv-
able that the ball will not gain some of the lost momentum. Newton did not allow
these sorts of exceptions.
The counterargument was first put forward in Search for the Perfect Swing
(Cochran and Stobbs [1968] 1999): the force between the club head and ball is so
large (the ton and a half from earlier) and the shaft so flexible that the shaft cannot
have any effect on the impact. An attempt was made in Search for the Perfect Swing to
support this assumption in an experiment that involved adding a hinge to the hosel
of a 2-wood so that the head could pivot freely during the impact. The club, without
the hinge, on average drove the ball 220 yards; in the word of the investigators, “only
5 yards further” than the one with the hinge. This result, directly opposed to their
assumption, was nevertheless discounted, and it seems that the argument has never
been challenged in the world of golf. We return to this example shortly.
In contrast, it is well proven in the science of tennis and baseball that the mass
of the shafts has a significant effect of the speed of a struck ball; see Cross and
Nathan (2009) and Cross’s Physics of Baseball & Softball (2011). In these sports,
the handles are bigger proportions of the bat and racquet total weights; but never-
theless, the basic principles of conservation of momentum apply just the same to
lighter-shafted clubs.
It is easier to imagine the effect of the shaft on the golf club impact by consider-
ing a possible CofR test based on a complete driver. This is shown schematically in
Figure 3V. When two bodies collide, in general they emerge from the collision both
moving along a straight trajectory and also rotating. The rotation is always about
their internal balance points, their so-called centers of mass (CoM). So the speed
after the impact is that of the CoM plus any rotation. Rotations are produced when-
ever the contact forces in the impact do not pass through the CoM.
Let’s apply this to the ball and club collision. For reasons we discuss later in the
chapter, we will ignore the club grip. A typical driver shaft is 44 inches from the cen-
ter of the face to the end of the shaft. If it’s balanced across a finger without a grip,
the balance point is about 8 inches from the club face center. This is the CoM of the
club and shaft. So, if a golf ball is fired at the face, with the shaft attached loosely at its
end, then the club CoM will be projected backward. At the same time, because the
contact force is 8 inches below the club CoM, the club will rotate clockwise about
its CoM. This is shown in Figure 3V. All this actually occurs because the impact
sends an impulsive wave up the shaft, which slows down sections of the shaft as it
passes by. It is mainly what is referred to as a shear wave, like the shear wave created
by layers of rock suddenly slipping in an earthquake, which travels upward, moving
layers above sideways and finally rocking buildings. This is what the player eventu-
ally feels, and it arrives at the hands at about the time the ball is leaving the face.
Things get complicated with shear waves in a flexible shaft because they excite a
44”
club
rotation
club
CoM
speed
8” ball rebound
speed speed
Figure 3V Proposed test for CofR of a golf club as opposed to just a golf head.
whole family of longer bending waves, which persist for some time after the impact.
This does not invalidate conservation of momentum in any way. Therefore we can
get a fair approximation of what happens by ignoring all this stuff and assuming that
the shaft is completely stiff.
Cross and Nathan (2009) posed the basic question, which for golf can be stated
as this: what would be the effective mass of an un-shafted club head if it was to offer
the same impact resistance as the club head/shaft combination? Once the question
has been posed, it is not difficult to answer, as shown by the authors. However, it
requires consideration of angular momentum, which is a topic for the later model-
ing section. The answer is quite amazingly simple for loft angles small enough so
that the cosine of the loft can be ignored, that is, for drivers and fairway woods. In
this case, the effective mass is the club head mass plus one-quarter of the shaft mass.
So we simply change the club head mass in the ball speed mass factor to the
effective mass for the club head and shaft as
effective impact mass = club head mass + shaft mass/4.
A typical high-performance driver shaft weighs 0.18 pounds (80 grams).

With the driver head mass of 0.44 pounds, this gives an effective impact mass of
0.44 + 0.18/4 = 0.485 pounds. The ball speed mass factor with a 0.1 pound ball is
thus 0.485/(0.485 + 0.1) = 0.829.
So if we redo the earlier CofR calculation for the average PGA drive, we now get
the following:
(a) Dividing ball speed by the club head speed gives 1.473.

(b) Dividing the new mass factor of 0.829 into 1.473 gives 1.776.
(c) Dividing Cosine(dynamic loft) = 0.969 into 1.776 gives 1.833, so the CofR is
0.833, and the golf world seems to have been put to rights!
Before proceeding to consider current Tour player data, we first address the
issue of the hinged 2-wood in Search for the Perfect Swing. The authors of that study
produced an empirical relationship for the carry distance of lower-lofted clubs.
Changed from speed units of feet per second to miles per hour for the ball launch
speed, their relationship becomes
Carry, yards = 2.2 × ball speed – 103
It is consistent to use this relationship because it was developed from ball driv-
ing tests carried out in the same study.
Applying the formula to the carry distance of 215 yards gives the estimated
ball speed as (215 + 103)/2.2 = 144.5 mph. This ball speed is based on the iso-
lated (hinged) weight of the 2-wood club head, which would have been close to
0.46 pounds. The club mass factor for this shot, using a 0.1 pound ball, would have
been 0.46/(0.46 + 0.1) = 0.821. A photograph of the club shows the shaft to have
been a steel one, and we will assume its weight to be 0.22 pounds. The effective
mass including the shaft, for the regular club without the hinge, is thus taken to be
0.46 + 0.22/4 = 0.515 pounds. The increased mass factor without the hinge becomes
0.515/(0.515 + 0.1) = 0.837. So we would expect the ball speed to increase by factor

0.837/0.821 = 1.02, or 2 percent. The 144.5 mph ball speed, with the hinged club,
should therefore be 144.5 × 1.02 = 147.4 mph with the regular club. The carry formula
then becomes
carry = 2.2 × 147.4 – 103 = 221 yards,
which agrees quite well with the 220 yards given in the study.
Calculating the effective mass for higher-lofted clubs is a little academic. CofR
is intended to allow collisions to be modeled as simply as possible. It is a metric that
has great validity since it also provides a measure of the energy lost in the colli-
sion. So as long as it is understood why the use of just club head mass gives too-high
CofRs, the rest is somewhat immaterial. The calculation of effective impact mass
for the more lofted clubs is explained later in the detailed science modeling section.
For completeness, the calculated values of effective mass and corresponding CofR
values for the full range of PGA Tour shots, are given in Table 3F.
In any calculation of golf shot performance we can use these effective mass val-
ues and corresponding CofR values. Alternatively, just club head weight can be used
with the artificially high Table 3E CofR values. Note that the CofR decreases from
the driver to the 5-iron and then starts to increase. The value of 0.77, for the 3-iron, is
approximately what would be expected from the lack of any spring effect in the club
face. This is all to do with energy storage and release and will be discussed later. The
further decrease down to the 5-iron is likely caused by the increasing sideways dis-
tortion of the ball, resulting from the increasing friction force from the higher loft.
This is a complex matter reserved for Chapter 4. Eventually this negative distortion
Table 3F Effective mass and CofR values for PGA Tour players
Clubs Head Effective Club Dynamic Ball Normal CofR

weight weight speed* loft speed*
Club mass Effective
(pounds) (pounds) (mph) (deg) (mph)
mass
Driver 0.440 0.491 112 14.4 165 0.863 0.828

3-wood 0.473 0.525 107 14.8 158 0.847 0.816
5-wood 0.495 0.561 103 16.0 152 0.842 0.807
Hybrid 0.517 0.584 100 17.0 146 0.819 0.786
3-iron 0.528 0.596 98 17.2 142 0.802 0.770
4-iron 0.543 0.612 96 18.4 137 0.778 0.748
5-iron 0.559 0.629 94 20.5 132 0.764 0.735
6-iron 0.574 0.650 92 23.8 127 0.769 0.740
7-iron 0.590 0.671 90 27.4 120 0.754 0.724
8-iron 0.605 0.692 87 30.6 115 0.787 0.756
9-iron 0.620 0.714 85 34.2 109 0.798 0.766
Wedge 0.636 0.741 83 39.7 102 0.846 0.811
effect is overcome by a positive lower impact speed effect, and we see increasing
CofR as we proceed down to the wedge. We discuss this next.
EFFECT OF I MPACT SPEED ON Cof R

The missing puzzle piece to apply the information in Table 3F to any golf shot is
the variation of the CofR with impact speed. One very careful investigation of this
variation was carried out by A. J. Cochran, one of the two authors of Search for the
Perfect Swing ([1968] 1999). This was published in Volume IV of Science and Golf,
the proceedings of a series of conferences of which Cochran was one of the found-
ers (Science and Golf IV: Proceedings of the World Scientific Congress of Golf; Thain
2002). Using equipment at the Callaway Spin/Impact laboratory, he measured the
CofR for golf balls impacting at right angles with a heavy block, with speeds from
4 to 112 mph.
The measured CofR values for the different test speeds are shown in Figure 3W,
which illustrates an approximate straight-line relationship between CofR and
impact speed. It can be interpreted as a rule that the CofR increases by 0.0125 for
every 10 mph decrease in impact speed. The reason for the decrease in rebound
efficiency as speed increases is that the resistance to deformation of the rubber-like
materials in golf balls increases as speed increases. The increased resistance is due
to internal friction between the large polymer molecules, and the energy used in
overcoming the internal friction is converted to heat. The faster they are com-
pressed, the greater is the proportion of energy lost as heat. In material science, we
say that the material is viscous, and the property is referred to as viscosity. “Thick”
liquids such as oils exhibit the same behavior. A common example is the oil-filled
dampers attached to automobile wheel spring suspension systems. The springs
need to be soft enough to cushion bumps at low speeds. However, they must also
absorb the shock of collisions with bumps at the highest cruising speeds, that is,
they must reduce the CofR. At the higher speeds, increased internal friction in
the oil flow through the damper absorbs the excess energy of the collision with the
bump, without which the car would become airborne. In just the same way, a “rub-
ber” ball bounced at higher speeds exhibits progressively less efficient rebounds.
0.90
0.86
Cof R
0.82
0.76
0.74
0 30 60 90 120
Impact speed, mph
Figure 3W Change of CofR for impacts at right angles to a heavy plate.

To apply this data to golf club impacts, it is necessary to account for the reduc-
tion in the ball speed caused by the loft angle and the reduced impact mass, com-
pared to the heavy block used in Cochran’s (2002) experiments. For the heavy
block, the total change in ball speed is determined directly from (1 + CofR). For a
club strike, the change in ball speed is determined additionally by the mass factor
and the cosine of the dynamic loft angle. So to make the CofR the same for both, we
simply multiply by the mass factor and the cosine of the dynamic loft angle.
For example, say a 7-iron shot with dynamic loft equal to 30 degrees and a swing
speed of 90 mph has a CofR value of 0.75. What will be the likely CofR for a swing
speed of 75 mph with the same dynamic loft?
We can start with 0.0125 increase for 10 mph decrease and therefore 0.0188
for a 15 mph decrease. The mass factor for the 7-iron using effective mass from
Table 3.6 is 0.671/0.771 = 0.87; coincidentally, the cosine of 30 degrees is 0.87. So
the CofR increase is likely to be 0.0188 × 0.87 × 0.87 = 0.014, that is, a corrected
value CofR = 0.764.
This calculation has been carried out for dynamic loft angles between 0 and 50,
using appropriate increases in club impact mass as the loft angle increases. The results
are shown in Figure 3X. From the graph, the change in CofR for a 7-iron is 0.0095 for
10 mph and therefore an increase of CofR by amount 0.014 for 15 mph as calculated.
Note that for the driver, Figure 3X gives the rule that
CofR increases by approximately 0.01 for every 10 mph decrease in impact speed.
The CofR value of 0.83 is for an impact speed of 109 mph. Therefore, for the
average male player with approximately 87 mph swing speed, the actual CofR of a
well-struck drive is approximately 0.85.
EFFECT OF I MPACT SPEED ON BACKSPIN

Any discussion of club ball contact would be incomplete without some reference
to the creation of backspin. With the modern golf ball, it is a somewhat complex
matter involving the introduction of a tangential or spin CofR. It will be dealt with
fully in Chapter 4. For a given club and dynamic loft, we show that the spin rate is
0.011
CofR Change per 10mph
0.010
0.009
0.008
0.007
0 20 30 40 50
Dynamic loft angle, degrees
Figure 3X Changes of CofR for 10 mph change in swing speed.

Table 3G Comparisons of average PGA Tour player with average male amateur player
7-Iron loft = 34.5 Club Club Attack Launch Dynamic CofR Ball Back-spin
loft speed angle angle loft (deg) speed (rpm)
(deg) (mph) (deg) (deg) (mph)
PGA player 34.5 90.0 −4.3 16.3 27.4 0.75 120 7,097
Average male A 34.5 69.6 0.0 25.2 34.5 0.80 88 6,755
Average male B 34.5 69.6 −4.3 16.3 27.4 0.77 94 5,488
directly proportional to the impact speed. This will allow us to carry out a compar-
ison of the average PGA Tour player with any other swing-speed golfer. We choose
to look at the 7-iron shot and compare the average Tour player with the average
male golfer.
According to Wishon (2013), the driver swing speed for the average amateur
player (AMP) is 87 mph. From the relative swing speeds of different clubs in
Table 3D, the expected AMP 7-iron swing speed is 87 × 0.8 = 69.6 mph. For the
angle of attack and the dynamic loft angle, two extreme cases will be taken. The
first is a horizontal attack angle and so no de-lofting. The second is the same angle
of attack and amount of de-lofting as the PGA player. Using the methods outlined
previously gives the launch angles, CofRs, ball speeds, and backspin rates as shown
in the second and third rows of Table 3G. Note that the launch angles have been
obtained from the launch/loft ratios in Table 3C. These ratios are approximately
constant with different swing speeds.
A comparison of the trajectories with these launch conditions is given in
Figure 3Y. The predicted trajectory, for the PGA player, has almost exactly the same
height and carry distance as the Trackman PGA Tour player data in Table 3A, which
provides some level of confidence in the amateur player predicted trajectories. The
main observation is of course the difference in flight distance between the AMP and
the professional. The average amateur does not hit down on the ball in an aggressive
way, so the chain-dotted line is most likely the performance. The predicted carry
distance in this case is only 114 yards compared to the average PGA distance of
170 yards. If the player adopted a negative attack by hitting the ball further back
in the stance, the shot could approach the dashed line trajectory with a carry of
7-iron shots:
30
(yards)
20
10
(yards)
0 20 40 60 80 100 120 140 160
average PGA tour average amateur, attack angle and de-loft = 0

average amateur attack with aggressive attack and de-loft
Figure 3Y Comparisons of average amateur male with the average PGA Tour player.
122 yards. The trajectory also has a much lower descent with a likely significant
bounce and roll distance.
THE POW ER OF THE MODER N DR I V E

A fundamental understanding of the performance of the hollow titanium driver is
best obtained by following the transfer of energy during a driver-golf-ball strike.
This is illustrated in Figure 3Z for a 0.83 CofR, 109 mph (48.7 meters per second)
driver strike. Recall that this is the test speed used by the R&A/USGA to check
conformance to the 0.83 CofR rule. To keep the argument as simple as possible, we
neglect the small effects of loft and the weight of the shaft.
The calculations are much easier using kilogram, meter, and second units so that
club head mass, and club head and ball speeds, can be plugged into the appropriate
expressions to get forces in newtons, energy in joules, and power in watts. The prob-
lem is the lack of feeling for these magnitudes when surrounded by everyday things
involving pounds, foot-pounds (ft-lbs), and horsepower, so we write the conversions
to everyday units as we go.
From initial contact, the ball continues to be compressed and the driver dia-
phragm face deflected until the ball and club head are travelling at the same speed.
This is referred to as the compression phase and lasts just under half of the total
contact time. At this mid-impact point, the speed of the ball and club head mov-
ing, as a pair, is 89.0 mph (39.8 meters per second). At separation, the club head has
slowed further to 72.4 mph (32.4 meters per second) and the ball has speeded up to
162.8 mph (72.8 meters per second). From our earlier discussion we can validate
these values by checking that the momentum at each stage is the same and that the
speeds satisfy the 0.83 CofR value.
A small discourse on energy is needed before we continue. The energy of motion,
so called kinetic energy, is one half of the mass of an object multiplied its speed
squared. In everyday units, this can be expressed in units of foot-pounds. This is
not obvious. To convert pounds force, which deflects the weighing scale, to pounds
mass, the amount of material, we need to divide by the acceleration that would
be applied to the ball if the scale was removed. This is of course the acceleration
of gravity. If we represent the acceleration in feet/second 2, then to obtain kinetic
energy we divide that into pounds force and multiply by speed squared with units of
(feet/second)2 . The result simplifies to just foot-pounds (ft-lbs). These are also
the units of mechanical energy. If we compress a 40 pound spring in an air rifle
175 ft-lb 143 ft-lb 165 ft-lb
109 mph 89.0 mph 89.0 mph 72.4 mph 162.8 mph
48.7 m/s 39.8 m/s 39.8 m/s 32.4 m/s 72.8 m/s
0.0000 seconds 0.00022 seconds 0.00045 seconds

compression restitution
–32 ft-lb +22 ft-lb
Figure 3Z Change of velocities and energy at stages in a 0.83 CofR impact.

by 3 inches, then we have stored 40 × (1/4) = 10 ft-lbs of energy. With some losses
due to friction, the pellet might emerge from the end of the barrel with 8 ft-lbs of
kinetic energy. This is not an off-topic example because the modern driver works by
storing spring energy in the trampoline spring face and releasing it to increase ball
speed. As for the conversions: 1 joule of energy equals 0.74 ft-lbs; 1 joule/second of
power = 1 watt = 0.00134 horsepower.
The kinetic energy at each stage is shown above the illustrations in Figure 3Z. At
the outset, the kinetic energy in the 0.2 kilogram club head is
initial kinetic energy = (1/2) × 0.2 × 48.72 = 237.4 joule = 175.7 ft-lbs.
At mid impact, the kinetic energy of the 0.2 kilogram head and the 0.045 kilogram
ball moving together is
kinetic energy at mid impact = (1/2) × 0.245 × 39.82 = 193.8 joule = 143.4 ft-lbs.
So we have temporarily “lost” 175.7 – 143.4 = 32.3 ft-lbs of energy, which has

been used in compressing the ball and deflecting the club face.
At separation, we have kinetic energy in both the club head and the ball, and
combined, this gives
final kinetic energy = (1/2) × 0.2 × 32.42 + (1/2) × 0.045 × 72.82

= 223.9 joule = 1 (value should be 165.7) 65.7 ft-lbs.
So we have recovered 165.7 – 143.4 = 22.3 ft-lbs of the temporarily “lost”

32.3 ft-lbs of energy at mid collision. Consequently we can state with confidence
that 22.3 ft-lbs of the “lost” energy at mid collision was stored as spring energy in the
ball and the trampoline club face. Most of the remaining 10 ft-lbs was lost in friction
within the ball material, while a small amount became locked into vibrations of the
club head.
There is a separate definition of CofR, called kinetic CofR, which is defined
for any collision as energy recovered/energy stored in the collision. It turns out,
surprisingly, that kinetic CofR = (normal CofR) squared. A proof of this is given
in the monograph on Impact Mechanics by Stronge (2000). For this case, we have
kinetic CofR = 22.3/32.3 = 0.69, which correctly equals 0.832 . So 69 percent of the
energy absorbed in the collision has been recovered; or equivalently, 31 percent of
the energy put into the club head and the ball during the impact has been lost. We
are now getting to the heart of the matter.
Recall that earlier it was shown that the average force during a 100 mph
driver strike with 0.83 CofR is 1,500 pounds. Since ball speed and contact force
increase with impact speed, the average force for the 109 mph collision would be
1,635 pounds. With a 109 mph impact, the face center of the titanium driver deflects
by approximately 2 mm (0.0066 feet) while the ball is compressed by about 5 mm.
Unfortunately, justification of this result involves a higher level of complexity than
is appropriate here, but it is included for completeness in an Appendix to the book.
Finite element modeling of the titanium driver head-ball collision by Apostal and
Ritter (JAR Associates, North Kingstown, RI) produced the same results.
From this work we can conclude that the amount of energy stored in the face
at mid impact is 1,635 × 0.0066 = 10.7 ft-lbs. We might assume that, with an effi-
cient diaphragm spring face, perhaps 90 percent, or 9.6 ft-lbs, is returned to kinetic
energy. So we are left with the conclusion that 32.3 – 10.7 = 21.6 ft-lbs was “stored”
in the ball and only 22.3 – 9.6 = 12.7 ft-lbs was recovered. Consequently, the kinetic
CofR for the ball is 12.7/21.6 = 0.59; or equivalently, 41 percent of the energy put
into the ball is lost. The normal CofR of the ball is the square root of 0.59 or 0.77.
This is roughly what we would expect if the driver face was rigid, although in that
case the contact forces would be higher, the ball compression greater, and so the
proportion of energy lost in the ball would be slightly greater. So we would expect
the CofR to be somewhat less than 0.77. Note that with kinetic CofR, we can obtain
the CofR of the club and ball separately, instead of just for the collision. For this
case, the CofR of the club has been assumed to equal 0.9 = 0.95.
It is now easy to see that if the driver face is made thinner, then it will deflect
more and so store more of the “lost” energy, which can then be returned to kinetic
energy efficiently and so increase the ball speed. There is a limit to how far this
improvement can go before the face does not respond quickly enough to return
available stored energy to the ball, and the CofR starts to decrease. The maximum
appears to be around 0.89 as shown by Cochran (2002).
This section was titled “The Power of the Modern Drive,” and we see that ball
driving power is increased by storage and return of spring energy. The actual driv-
ing power in this case gives an increase in kinetic energy of the ball from zero to
final ball kinetic energy = (1/2) × 0.045 × 72.82 = 119 joules = 88 ft-lbs
in 0.00045 seconds, or 88/0.00045 = 196,000 ft-lbs/sec = 355 horsepower. This is

not quite our v-12 Ferrari but an amazing amount for a manually driven instrument.
More interesting is the delivery of 176 ft-lbs of initial kinetic energy to the
impact in approximately 0.25 seconds of the downswing. This amounts to an aver-
age power input during the swing of 176/0.25 = 704 ft-lbs/sec = 1.3 horsepower.
This is just the power input to the club head. According to Daish (The Physics of
Ball Games, 1972), and making corrections for the reduced shaft weight since 1972,
the power input to the shaft is 0.2 horsepower. Daish also makes an estimate of the
kinetic energy still residing in the arms and body at impact to be 107 ft-lbs, which
required a power input of 107/0.25 = 428 ft-lbs/sec = 0.8 horsepower. This gives the
total power required to drive a ball at 109 mph as 1.3 + 0.2 + 0.8 = 2.3 horsepower.
Daish estimated the need for 3.3 horsepower based on a very heavy shaft and a very
fast downswing time of 0.2 seconds. In Search for the Perfect Swing, the investigators
estimated the need for 3 horsepower, also for a 0.2 second downswing. The obvi-
ous question then is how much muscle is required to produce this power. Daish
quoted two sources (A.V. Hill 1950 and D. R. Wilkie 1960) to estimate the amount
of power available from human muscle as 0.15 horsepower/pound. In Search for the
Perfect Swing, a value of 0.125 horsepower/pound was suggested without reference.
To put these values in some context, precise clinical studies of frogs (R. L. Marsh),
whose leg muscle power is certainly better than humans, estimate a value of
0.23 horsepower/pound; so watch out for competitors with webbed feet.
To estimate the amount of muscle involved in the golf swing, it is necessary

first to agree that in any one-sided act, we are only using half of our muscles: one
set contracting to produce power while an opposing set is extending in readiness
for the reverse movement. At a more simplistic level, right-handed golfers have
the muscles required to drive the ball left-handed, even if they lack the coordi-
nation to do so. So the amount of muscle required may be estimated at some-
where between 2 × 2.3/0.15 = 31 pounds and 2 × 3/0.125 = 48 pounds. A typical
180 pound golfer has approximately 75 pounds of muscle, and approximately
32 pounds are found above the waist. The most unlikely interpretation of these
values is that all of the power comes from the upper body. The more likely one is
that a significant part comes from the leg muscles. This could only come from the
initial rotation of the hips, body, and arms together, using the leg muscles to power
the first move. However, there has been a recent publication in Nature on “Elastic
Energy Storage in the Shoulder and the Evolution of High-Speed Throwing in
Homo,” which may overturn this accepted golf wisdom. Researcher N. T. Roach
and colleagues (2013) determined that the human body developed the capacity to
throw at a high speed by enabling the storage of energy in the tendons, ligaments,
and elastic portions of muscles across the shoulders. This gives a new meaning to
the value of coiling of the arms and shoulders around the torso at the end of the
backswing to put the system under strain. If this does result in a release of stored
energy at the start of the downswing, then the upper body alone may be capa-
ble of releasing the needed power. Mike Kelly (personal discussions, Fall 2014),
acknowledged at the beginning of the book, is of the firm opinion that a powerful
golf swing can be accomplished without input from the leg muscles.

3.1 Introduction to the science of ball striking
The goal of the remainder of this chapter is to provide a scientific basis for discus-
sion of ball striking with a lofted club. The golf swing has been discussed fully ear-
lier, and little more will be added here. There is a wealth of literature on the subject,
and the interested reader may wish to consult the articles that were referenced ear-
lier. In particular, the articles by Nesbit (2005) and, Nesbit and Serrano (2005) are
very insightful and contain an excellent, comprehensive reference list. The article
by Sharpe (2009), which adopts a triple pendulum model, is well worth studying.
The author shows that the model agrees quite well with the Ryder Cup players’
swings, which were documented in Search of the Perfect Swing. Unfortunately,
Sharpe seems to have been unaware of the paper by Nesbit (2005), which would
have provided a better basis for testing his model. Moreover, Sharpe’s conclu-
sions, that a slow swing start should be used to minimize early wrist un-cocking, is
strongly supported by the Nesbit and Nesbit and Serrano investigations.
Here we are concerned with the control of dynamic loft and the mechanics of
the ball impact. In particular the ball speed factors and effective impact mass values,
described and used earlier, are developed here.
3.2 Dynamic loft
The increase of loft of the driver, from the inertial forces passing through the offset cen-
ter of mass, needs no further attention. The second factor affecting the dynamic loft,
namely, the position of the golfer’s hands at impact and the relative speed of the hands
and the club head, does require some validation. This was discussed earlier in terms of
the forward hinge of the wrists at impact. To visualize this, it is helpful to exclude the
shaft bending effect. Figure 3.1 is a re-creation of the Milne and Davis (1992) plot with
a shorter, extra stiff, steel shaft assumed to result in insignificant shaft bending. I have
added the velocity vectors and created the velocity diagram on the right side. This gives
us a direct measure of the amount of de-lofting, labeled δ s−on the velocity diagram.
The angular velocity at impact of the upper part of the assumed double pendu-
lum is denoted ω h . The center of rotation is assumed to be fixed, so the velocity of
the hands at impact, denoted Vh , is normal to the upper pendulum arm. The hands
connection between the upper pendulum arm and the club shaft is denoted h.
The angle between the upper pendulum arm and the shaft is α sh ; this results
from the forward hinge of the wrists. The angular velocity of the shaft with
respect to the moving hands is denoted ω sh . The velocity of the club head at
impact, with respect to the hands, is labeled Vch , and this is normal to the shaft.
With these definitions, the velocity diagram on the right of Figure 3.1 is eas-
ily constructed. Starting from the origin at the left of the diagram, the hands have
velocity Vh = l u ω h at right angles to the upper (u) pendulum arm between ‘0’ and
‘h’, of length l u. The shaft rotational speed has been increasing during the final part
of the downswing, and the relative angular velocity of the shaft with respect to the
hands, ω sh , is positive. The club head therefore has velocity with respect to the
ωh
h Vh
αsh
ωsh
Vh αsh
Vch
Vch
Vc –
β δs
c
Figure 3.1 Model of golf shaft and head positions during the downswing; adapted from Milne
and Davis (1992); symbols and velocity diagram added by Dewhurst.
hands, Vch , equal to l s ω sh and at right angles to the shaft of length l s . Thus, velocity
Vch is at angle α sh to the velocity of the hands, Vh . It only remains to connect the
beginning of velocity vector Vh to the end of relative velocity Vch to establish the
actual club head velocity Vc . The angle between Vc and the horizontal is the angle
of attack and is represented by angle β in Figure 3.1. Finally, the angle between
normal to the shaft and the club path, shown as δ s− on the velocity diagram, is the
amount by which the club has been de-lofted. This is illustrated in Figure 3.2.
From the velocity diagram, we have sin(δ s− )/ Vh = sin(α sh )/ Vc . Using the
−
small angle approximation for sin(δ s ), this gives
normal to sh δ–
s
aft V
ch
Vc
face δ
normal to δc
Figure 3.2 Relationship of velocities to de-lofting of the club.

−
δ s = (Vh / Vc )sin(α sh ). (3.1)

This relationship includes three fairly obvious results. First, if the wrists are not
partially hinged at impact (α sh = 0), then there is no possible de-lofting of the club.
Second, if the hands are moving very slowly through the impact ( Vh ≈ 0), then there is
no de-lofting. In this case, the swing has become a single pendulum from the wrists at
impact; and the club head is moving at right angles to the shaft. Third, if the wrists are
hinged at impact (α sh > 0), then the faster the players hands move through the impact,
the greater will be the amount of de-lofting. The realistic maximum occurs when the
wrist forward hinge is held fixed through the impact. In this case, the swing becomes
that of a single bent pendulum, and the amount of de-lofting depends only on the hinge
angle, α sh , and pendulum arm lengths l u and l s. This is the shot used to keep the ball
trajectory close to the ground to recover from under low branches, or to play a very low
bump and run up to the green. For this condition, Eq. 3.1 simplifies to
 lu 
δ s− = sin(α sh )   ≈ sin(α sh )/ 2. (3.2)
 (l u + l s ) 
Figure 3.3 shows my sketch of a PGA player with one of the best short games
on Tour. The angle between the shaft and the upper pendulum arm (extending
from the center of the grip to below the chin), is approximately 35 degrees. Also,
the pendulum lengths are approximately the same. Assuming a club head speed
of 30 m/s, then for hand speed from 5 to 15 m/s, the amount of de-lofting, from
Eq. 3.1, will range from 5.5 to 16.5 degrees. The only hand impact-speed data
available in the literature is the approximately constant 9 m/s (20 mph) for a wide
range of drives; see Figure 3L. Thus the range of 5 to 15 m/s for a wedge shot
seems reasonable. The average amount of de-lofting of pitching wedges for the
PGA Tour player is 9 degrees, very nicely in the middle of this predicted range.
3.3 Wedge angle effects on ball speed

We consider first the simplest situation in which an “iron” club head is travelling
horizontally at impact with velocity V0 , and the ball is struck at the sweet spot on
the club face, that is, directly in front of the center of mass. In this case, the club head
is not subjected to any movements sufficient to produce significant rotation of the
club head. We will see later in the chapter, and in more detail in Chapter 6, how such
head rotations negatively affect ball speed.
The conditions at the point of separation of the club and ball, following this
sweet-spot impact, are illustrated in Figure 3.4. For this case, the ball angular veloc-
ity is labeled ω wedge , and the dynamic loft is δ. Because of the friction between the
ball and club face, the initial ball trajectory will be below normal to the club face as
shown in Figure 3.4; that is, the vertical launch angle α vb is less than the loft angle δ.
Note also that from conservation of momentum in the vertical direction, the upward
trajectory of the ball must be accompanied by a corresponding downward trajectory
of the club head, represented by α vc in Figure 3.4. Thus, we are faced with the solu-
tion of five unknowns. To simplify this problem, we assume some prior knowledge
ΩV
Figure 3.3 Aggressive wedge shot by an elite player.
about the initial ball launch angle α vb. From a very large number of experimental
observations made by Tuxen (October 2010), he formulated the rule that for irons,
the launch angle for a horizontal ball strike is approximately 75 percent of the loft
angle, that is, α vb ≈ 0.75δ.
ωwedge
v1
δ−α vb
rb αvb
V1 αvc
V0 δ
pre-impact
post-impact
Figure 3.4 Strike on the sweet spot of the club face with loft angle δ; V1, v1 are club head and ball
velocities at separation, respectively.
Later in this section, a full solution of the governing equations will show
that this parameter decreases slowly with loft, with an average value of 0.77
in good agreement with the Tuxen rule. This was shown in Table 3C. For the
moment, we will simply use this result to obtain some relatively simple but pow-
erful equations to predict ball velocity and CofR. The simplification occurs
through neglecting the term cos(δ −α vb ), which arises in the prediction of
ball velocity. If we restrict our attention to loft angles up to 50 degrees, which
includes all but the highest lofted wedges, then this cosine term lies in the range
1 < cos(δ − α vb ) ≈ cos(δ / 4) ≤ 0.976. We can therefore neglect this term, with the
acceptance of errors in our predictions increasing up to the order of 2.4 percent
for the higher-lofted clubs.
With the smooth face of the modern driver and the resulting lower friction
coefficient, the ball typically slides by some amount throughout the impact. This
necessarily moves the ball launch angle closer to the face normal. This conclusion
is supported by Tuxen’s (October 2010) modified rule for driver launch angles,
namely, α vb(driver ) ≈ 0.85δ.
From the definition of the CofR, e n , and conservation of linear momentum nor-
mal to the club face, we can obtain, with reference to Figure 3.4,
v 1 cos(δ − α vb ) − V1 cos(δ + α vc ) = e n V0 cos(δ) (3.3)

and
MV0 cos(δ) = MV1 cos(δ + α vc ) + mv 1 cos(δ − α vb ). (3.4)

Note that we use the symbol e n for CofR, with subscript n to denote that it is cor-
rectly named the normal CofR.
Eliminating V1 cos(δ + α vc ) between these equations, and neglecting the term
cos(δ − α vb ), as discussed previously, gives
V0 cos(δ)  1 + e n 
v1 = = V0 cos(δ)(1 + e n )[M /(m + M)]. (3.5)
cos(δ − α vb ) 1 + m / M 
Note that the right-hand expression is written as the product of the three ball-speed
factors used earlier.
Term cos(δ) is often omitted in the golf literature with significant errors for the
higher-lofted clubs.
Note that we refer to the “normal” CofR because it defines the relationship
between the normal component of the club and ball velocities. Equation 3.5 can be
inverted to give the expression for normal CofR as
e n = [v 1 / V0 cos(δ)](1 + m / M) − 1. (3.6)

3.4 Determining dynamic loft from launch data

The task here is to determine the actual club loft from measured values of ball speed,
launch angle, and spin rate. The first task is to determine the tangential club head
ω
v1n
Fn v1
δ
V1n δ−αvb
v1t αvb
Ft V1t
impact forces post-impact velocities
Figure 3.5 Velocity components before and after impact.
velocity component, V1t , after impact. Figure 3.5 illustrates, on the left side, the
force components acting on the ball during the impact. On the right it shows the
resulting velocity components after impact. It is assumed, as shown on the left, that
the normal contact force passes through the center of mass of the ball. We will see in
Chapter 4 that this is not strictly correct and that there is always some small amount
of rolling up the face during the deformation. This incurs rolling friction, which
moves the normal force slightly further up the face than the normal through the
center of mass. We will also see that this produces only a small error in the predic-
tion of dynamic loft and that the dynamic loft predicted in this section is, in fact, a
slight underestimate. This fact must await validation, but it is important for some of
the conclusions reached here.
From Figure 3.4, the momentum, tangential to the face before impact, is
MV0 sin(δ). Thus, from conservation of momentum tangential to the club face we
have, with reference to Figure 3.5,
MV0 t = MV0 sin(δ) = MV1t + mv 1 sin(δ − α vb ) (3.7)

or
V1t = [V0 sin(δ) − (m / M)v 1 sin(δ − α vb )]. (3.8)

Neglecting any center of mass offset, the tangential force integrated over the
contact time, which causes the tangential velocity of the ball, must also be solely
responsible for the ball angular velocity. Thus, over the contact time t c , the average
tangential force Ft produces the changes in both the tangential linear momentum
and the angular momentum of the ball; that is,
Ft t c = mv 1 sin(δ − α vb )
(3.9)
Ft t c rb = I bω,

where I b is the MoI of the ball.

Eliminating Ft t c between these two equations gives
ω = mv 1 sin(δ − α vb )rb / I b . (3.10)

Table 3.1 Trackman results for 7-iron shot off a tee
Club head Attack Spin Ball Launch Spin rate Dynamic

speed angle loft velocity angle (rpm) loft (deg)
(mph) (deg) (deg) (mph) (deg)
91.0 −4.3 24.5 126.0 14.1 6,481 24.3

91.2 −6.1 26.9 127.9 15.0 6,397 26.8
90.2 −4.7 27.3 127.9 16.5 6,725 27.2
90.4 −5.1 27.5 128.2 16.0 7,058 26.3
The MoI of the standard 45 g ball with uniform density throughout is

I b = (2 / 5)mrb2 = 82 g-cm2 . All of the polymers used in golf balls have densities less
than 1 g/cm2 , and to make the golf ball with volume 40.7 cm3 close to the allow-
able upper mass limit of 45.9 g, higher density materials must be added. Extra mass
can be added to the core to lower MoI below 82 g-cm 2 . This will increase spin rate
off the club face according to Eq. 3.10. This would be generally beneficial to the
lower-swing-speed golfer. Alternatively, mass can be added to the outer layers to raise
MoI, as claimed by one manufacturer, “to add distance in head winds”! Although
true, because the lower spin would be beneficial with increased air speed, the effect
is too small to be of any real value. In general, going forward, we represent the MoI as
I b = λmrb2 , (3.11)

where λ will be assigned value two-fifths unless stated otherwise.

Equation 3.10 is valid even if the final angular and linear velocities are indirectly
caused by storage and release of strain energy in the tangential direction. This result
will be needed again in Chapter 4, where the high tangential CofR of the modern
high-spin ball is investigated. The actual mechanism of spin generation does not
concern us here because we are simply using the measured values of the spin rate.
For the range of values of (δ − α vb ) up to the highest lofted wedges, the small
angle approximation sin(δ − α vb ) = (δ − α vb ) involves errors of less than 1 percent.
This can be validated using the Tuxen rule that the normal launch angle with grooved
“irons” is approximately 75 percent of the dynamic loft. Thus, for a 50 degree wedge,
the value of (δ − α vb ) is approximately 12.5 degrees. This value in radians is 0.2182,
and the sine of 12.5 degrees is 0.2164, a difference of less than 1 percent. Final pre-
dicted values of (δ − α vb ) will also be seen to be small enough for the approximation
to be insignificant. Substituting this and Eq. 3.11 into Eq. 3.10 gives
δ = α vb + λrbω / v1 . (3.12)

Note that provided we are using measured launch parameters, the mass of the
ball does not change this result. Fredrik Tuxen, CTO and Founder of Trackman
A/C, kindly carried out player tests with me to check the effect of hitting nominally
the same shots off a tee and off the fairway. The venue was a golf resort with floating
balls used on their driving range for hitting out over a lake. The results of two sets of
tests by an elite golfer with the 7-iron are given in Tables 3.1 and 3.2.
Table 3.2 Trackman test results for 7-iron shot off grass
Club head Attack Spin loft Ball Launch Spin rate Dynamic
speed angle (deg) velocity angle (rpm) loft (deg)
(mph) (deg) (mph) (deg)
88.8 −3.4 25.3 125.0 15.8 6,580 25.2

88.7 −3.1 21.8 126.0 13.1 5,997 21.7
90.9 −8.7 26.4 128.5 10.6 6,539 25.1
90.7 −4.0 21.8 124.5 11.9 6,308 21.7
90.9 −5.7 25.0 126.9 13.1 6,787 24.9
The main purpose of showing these results here is to test the validity of
Eq. 3.12. Predictions of dynamic loft using Eq. 3.12 are given in the last column.
Column 3 gives the loft angle predicted by the Trackman Company using proprie-
tary algorithms in their software system. It is reasonable to assume that these have
been tested thoroughly by the company, and we can see that, with one exception,
Eq. 3.12 is in agreement within two-tenths of a degree. This provides support for
the use of Eq. 3.12, the loft predictions from which we will continue to refer to as
“dynamic loft.”
As far as the effect of striking the ball and taking a divot is concerned, the results
in the two tables do show some significant differences. The average attack angle
is almost identical in the two tests: −4.9 and −5.0 degrees. The launch angle how-
ever is significantly different: 15.6 and 12.9 degrees, respectively. Also the average
dynamic loft for the divot-taking shots is 24.1 degrees, significantly less than the
26.5 degrees for the shots off the tee. Moreover, the 10 percent greater loft for the
tee shots only produces 1 percent greater spin rate on average. Although the ball has
flown before the club starts to take the divot, it does get pushed forward and slightly
downward during the impact, displacing some small mass of grass and top dressing.
It seems that this ground contact may increase the shearing of the ball against the
lofted face, resulting in increased spin. This is discussed fully in Chapter 4.
Equation 3.12 was applied to the average PGA Tour player data to give the
dynamic loft values in Table 3A. These loft values indicate very large amounts of
de-lofting, which are adequately explained by Eq. 3.1.
3.5 The effective impact mass of the golf club

Figure 3.6 illustrates the more general modeling of effective mass to include the
effect of club loft. The club is assumed to be suspended from the equivalent of the
compliant grip in the hands of the golfer. In the actual golf club impact, the reaction
of the golf club to the stress waves travelling upward is assumed to be unaffected by
the compliant grip or the small forces from the hands.
The effect of the normal force impulse, from the normal contact force Fn, will be
to produce a backward velocity, Vcm , of the center of mass of the club, situated some
distance b above the center of mass of the club head. At the same time, the shaft and
club head will be forced to rotate clockwise with angular velocity Ω about the com-
bined center of mass of the club head and shaft as shown. In Figure 3.6 M c defines
ls Ms
Ω
)
Co M M s(δ
Vcm bco
b v0
Mc Fn
Vch
Figure 3.6 Effect of the normal impact force on a freely supported club.
just the mass of the club head, M s defines the mass of the shaft, and M denotes the
sum of the two.
We start by assuming the shaft can be treated as a uniform hollow tube. The posi-
tion of the combined CofM can easily be shown to be given by
b = l s M s /(2 M), (3.13)

where l s is the shaft length.

As shown on the right of Figure 3.6, the velocity components of the club head
and the center of mass are related by
Vch = Vcm + bcos(δ)Ω. (3.14)

The normal impulse, Fn t c , applied during the impact, is responsible for both
the linear and angular momentum of the combined club and shaft. This gives
relationships
Fn t c = MVcm (3.15)

and
Fn t c bcos(δ) = I cm Ω , (3.16)

where I cm is the MoI about the CoM, and from use of the parallel axis theorem is
given by
I cm = M c b2 + M s (l s / 2 − b)2 + M s l 2s / 12. (3.17)

The effective mass of the club head for the normal impulse, M en , is now defined
so as to establish the same momentum exchange between the club and the ball but
without considering the angular momentum gain in Eq. 3.16. Thus, we have the
artificial relationship
Fn t c = M en Vch . (3.18)

Eqs. 3.13 to 3.18 can be manipulated to give the relationship
M
M en = . (3.19)
1 + Mb2 cos 2 (δ)/ I cm

For calculations of driver shots with δ < 15 degrees, we can write the approxi-
mate relationship
M (3.20)
M en(driver ) = = Mc + Ms / 4
1 + Mb2 / I cm
The simplicity of Eq. 3.20 is quite surprising. As discussed earlier, Cross and
Nathan (2009) included this rule as an aside in their detailed discussion of the effec-
tive mass in baseball, tennis, and golf. In their words, “if an implement has a light
handle and a heavy head” then during impact, “the effective mass at the head end
is M e = M c + M s /4, where M c is the mass of the club head and M s is the mass of
the shaft.
It can be noted from Eqs. 3.17 and 3.19 that only the mass of the shaft and the
head, and the ratio b / l s determine the effective mass. In the preceding calcula-
tions, the ratio was based on an assumed uniform weight distribution along the
shaft. This is not the case for graphite shafts, which are manufactured by laying
tapered strips of graphite fiber sheet material around a tapered forming spindle,
called a mandrel. The resulting product may have a uniform wall thickness. In con-
trast, steel shafts are manufactured by forming a tube blank into the tapered shape.
In this case, if the tube is simply reduced in diameter, then the wall thickness is
increased toward the tip end and the mass distribution along the length is nearer
to a uniform tube. We can test the effect of these differences by considering the
extreme case of a shaft being modeled as a hollow very-shallow cone of uniform
wall thickness, that is, tapering down from the butt to a very small diameter at the
tip. For this geometry, the center of mass is two-thirds of the length from the tip,
and Eq. 3.13 changes to
b = 2 l s M s /(3M). (3.21)

Also, the MoI of the tapered uniform-wall shaft about its center of mass is
I cm = M c b2 + M s (2 l s / 3 − b)2 + M s l 2s / 9. (3.22)

With these new values of b and I cm , Eq. 3.19 for effective mass still applies. The
reader may wish to confirm that the effect of these changes on the CofR values is
insignificant. The only changes to two decimal places in the normal CofR values,
from this different mass distribution in the shaft, are an increase from 0.83 to 0.84
for the driver and 0.80 to 0.81 for the pitching wedge.
Club specifications are assumed, for the average PGA Tour player, as shown in
Table 3.3.
From the Table 3.3 club data, we can calculate the dynamic loft and the effective
impact mass values for all of the average PGA Tour player shots. These can then be
used in Eq. 3.6 to recalculate the true normal CofR value for each impact. The val-
ues are shown in Table 3.4. For convenience, the details of each ball strike are also
included.
The corrected CofR values in the last column of Table 3.4 are seen to reach a
minimum value in the 5-iron to 7-iron range and then to increase into the wedges.
Two factors are at play here. As the loft angle increases with the corresponding
shorter shafts, the impact speed decreases, and at the same time the normal com-
ponent of the impact speed is reduced. We would expect this to give a progressive
improvement of CofR since the ball will be compressed less with a smaller pro-
portion of lost energy in the ball. This effect will be quantified in the next section.
The second factor concerns the effect of the increased friction force being
applied to the ball as the loft increases. This causes increasing “sideways” distor-
tion of the ball, which in turn causes the resultant normal force from the ball com-
pression to be misaligned from the center of mass of the ball. The result is that
the recovery contributes more to ball rotation and less to ball speed recovery.
This issue is addressed in Chapter 4 for the simplified situation of ball bouncing
against a fixed surface. From these tests, the effect of the offset normal force as a
braking torque during ball contact can be clearly established. In general terms, it
means there is some cross-coupling between ball spin and ball speed and the cor-
responding CofRs. However, any attempts to include this in a single model would
be impossibly complex; and by using actual ball striking data, we have subsumed
it into the normal CofR values. The true normal CofR values in column 10 first
decrease steadily from this effect, and then increase as the reduced impact speed
effect becomes dominant.
Before leaving this section, it is useful to consider just one further case study
using Trackman data for a club fitted with a heavier than normal shaft. The data in
Table 3.5 is the average of six shots using a fairway wood with an extra thin face,
which had been tested to produce exactly 0.83 for the CofR. It was fitted with a 125 g
shaft compared to the PGA average 80 g driver shaft. The head weight was 210 g and
the club loft was 15 degrees.
The calculated effective impact mass for this club, using Eq. 3.19, is 246 g. If only
the club head mass of 210 g is used in Eq. 3.6, then the normal CofR is predicted
to be 0.875, compared to the average PGA drive value of 0.863 based on the 200 g
head weight. When the effective mass of 246 g is used, the CofR becomes 0.826 as
expected.
Table 3.3 Specifications for average PGA Tour player’s clubs
Club Driver Fairway Woods Hybrid Irons PW

3 5 3 4 5 6 7 8 9
Head mass (kg) 0.200 0.215 0.225 0.235 0.240 0.247 0.254 0.261 0.268 0.275 0.282 0.289
Shaft mass (kg) 0.080 0.080 0.100 0.100 0.100 0.099 0.097 0.096 0.095 0.093 0.092 0.091
Shaft length (m) 1.105 1.080 1.055 1.030 0.991 0.978 0.965 0.952 0.939 0.926 0.913 0.900
Table 3.4 Average Trackman test results and modeling for PGA Tour players
Trackman data* Calculated values Effective True normal

impact mass CofR
Head mass Attack angle Club speed Ball speed Back-spin Dynamic Normal
(kg)
(kg) (deg) (mph) (mph) (rpm) loft CofR
(deg)
Driver/9.5 0.2 −1.3 112 165 2,685 14.4 0.863 0.223 0.828
3-wood/16.5 0.215 −2.9 107 158 3,655 14.8 0.847 0.238 0.816
5-wood/21.0 0.225 −3.3 103 152 4,350 16.0 0.842 0.254 0.807
Hybrid/22.5 0.235 −3.3 100 146 4,437 17.0 0.819 0.265 0.786
3-iron/20.5 0.240 −3.1 98.0 142 4,630 17.2 0.802 0.270 0.770
4-iron/23.5 0.247 −3.4 96.0 137 4,836 18.4 0.778 0.278 0.748
5-iron/26.5 0.254 −3.7 94.0 132 5,361 20.5 0.764 0.286 0.735
6-iron/30.5 0.261 −4.1 92.0 127 6,231 23.8 0.769 0.295 0.740
7-iron/34.5 0.268 −4.3 90.0 120 7,097 27.4 0.754 0.304 0.724
8-iron/38.5 0.275 −4.5 87.0 115 7,998 30.6 0.787 0.314 0.756
9-iron/42.5 0.282 −4.7 85.0 109 8,647 34.2 0.798 0.324 0.766
PW/48.5 0.289 −5.0 83.0 102 9,304 39.7 0.846 0.336 0.811
Table 3.5 Trackman test results and modeling for golf robot fairway wood shot
Fairway wood Trackman data Model

Model Head mass Attack angle Club speed Launch angle Ball speed Back- spin Spin loft Normal CofR
515GRT (kg) (deg) (mph) (mph) (rpm) (deg)
To horizon To club Head Effective
(deg) (deg) mass mass
Loft 15 deg 0.21 0.0 101.3 12.2 12.2 151.1 3,848 15.1 0.875 0.826
Data reproduced with permission from Tom Wishon Golf Technology, Boulder, CO, USA.
0.90
Normal coefficient of restitution

0.86
0.82
0.78
0.74
0 10 20 30 40 50
Impact speed, m/s
Figure 3.7 Measurements of CofR for varying impact speed; from Cochran (2002).
3.6 Relationships between impact speed,

impact force, contact time, and CofR
To make use of the PGA data for more general predictive modeling, it is necessary to
make corrections for the swing speed of different players. For this purpose, we can
use the results of a set of precise experimental measurements by Cochran (2002),
discussed earlier. Recall that he used a ball cannon to bounce golf balls at different
speeds off a heavy block. As well as measuring rebound speeds to determine CofR,
Cochran also used a force transducer mounted on the block to provide a force-time
history for each test. From this, time of contact could be obtained. Repeat tests with
and without the force transducer were taken to show that the transducer did not
affect the rebound velocities. The relationship between impact speed and normal
CofR from these tests is shown in Figure 3.7. This was shown earlier, and is pre-
sented here with speed expressed in meters per second. The straight line close to the
data points is the simplest fit to the data and is given by
e n = 0.9 − 0.003 v . (3.23)

Equation 3.23 can be written in a more useful form as
∆e n = −0.003∆v , (3.24)

where ∆v is any change in ball impact speed and ∆e n is the corresponding change
in normal CofR. To make this useful for golf club impacts, a criterion of equivalence
between the ball bounce tests and club striking is needed. It is reasonable to assume
that equivalence will be assured if the normal impulse is the same for both collisions;
or equivalently, the change in normal momentum is the same for both. This criterion
was used by Lieberman (1990a), expressed in terms of equivalent change in the ball
normal momentum.
For the impact with the heavy plate at ball velocity v , the momentum change
is mv(1+ e n ). For the impact with a club of mass M, with dynamic loft δ, at
velocity V, the ball momentum change normal to the face is given by Eq. 3.5 as
mV cos(δ)(1 + e n )/(1 + m / M). Using the criterion of equivalence, e n will be the
same in these two impact methods if
v = V cos(δ)/(1 + m / M). (3.25)

0.0008
0.0007
Contact time, sec

0.0006
0.0005
0.0004
0 10 20 30 40 50
Impact speed, m/s
Figure 3.8 Measurements of contact time for varying impact speed; from Cochran (2002).
So Eq. 3.24 can be rewritten for golf impacts as
∆e n = −0.003∆V cos(δ)/(1 + m / M). (3.26)

This must be considered an approximate estimator for higher loft clubs because
it takes no account of the effect on normal CofR of the shear deformation in the
oblique impact. However, the purpose of setting up this expression was to get an
estimate of relatively small differences in performance. For example, the average
male player has a driver swing speed of 87 mph (Wishon 2013). If we apply this
decrease of 25 mph (11.2 m/s) from the average 112 mph PGA driver impact speed
with CofR 112 = 0.828, then we get a predicted value of CofR 87 = 0.855. The change
in impact speed produces a proportional decrease in ball speed from 165 mph to
(87 /112)165 = 128 mph. The increase in CofR changes this result from 128 mph to
(1.855 /1.828)128 = 130 mph.
The relationship between contact time t c and the impact velocity v obtained
from Cochran’s experimental study is shown in Figure 3.8.
The curve that fits closely to this data in Figure 3.8 is given by
t c = 0.00078 v −0.145 . (3.27)

Just as for the CofR relationship, we can assume with some confidence that the
same change of ball momentum in any collision between a golf ball and a rigid strik-
ing surface will produce the same impact responses. So, for a golf club impact, we
can write the contact time as
t c = 0.00078(V cos(δ)/(1 + m / M))−0.145 . (3.28)

For the average PGA 5-iron shot in Table 3C, the expression in parentheses in
Eq. 3.28 gives 32.4 m/s, and the predicted contact time is 0.00047 seconds.
This topic is discussed much more fully in the Appendix. One result of impor-
tance to discussions of the spring-faced titanium driver in following chapters is
that the contact force is reduced, but the contact time is essentially unchanged at
0.00045 seconds. This result was also established by Cochran (2002).
4
T H E G E N E R A T IO N O F B A L L S P I N
The primary factor contributing to ball spin is the friction between the ball and club
face, without which the ball would depart at right angles to the club face with no
spin. However, particularly with the high-spin golf ball, spin is not produced directly
by the face friction. If that was the case, then the spin rate would reach a maximum
value equivalent to pure rolling up the club face. At this point, the face friction would
reverse to become rolling resistance, and the spin rate would decrease until separa-
tion from the face. Almost all of classical mechanics would define the process in this
way. However, we see that the bouncing of golf balls off the lofted club face follows
a surprisingly complex process. The behavior involves the storage of spring energy
in the ball across the area of contact with the club face in a manner that allows the
energy to be released for increased spin. This energy recovery into increased spin
generation can be defined just like the energy recovery into launch speed, that is, by
using a CofR. In this case, instead of the normal CofR for launch speed, we must
introduce a spin CofR for spin rate.
The behavior is exactly analogous to winding an internal clock spring, which
enables the ball to leave the face with “overspin”; that is, as the ball loses its grip on
the face toward the end of the contact period, it spins up to a higher level. This is a
very significant factor; particularly for golf balls with softer covers designed to bet-
ter grip the club face. We see later that although this overspin behavior is well under-
stood in the general science of the bouncing of balls, it has not been understood in
relation to golf. This is quite startling. As we see later, the average PGA Tour profes-
sional iron shot produces 25 percent overspin, and the average LPGA player pro-
duces even higher amounts. It seems that the reason this has escaped attention is
that a positive spin CofR value both increases spin and lowers the launch angle. It
does this in such a way that measured values of launch angle can be used to predict
spin rate, effectively hiding the true nature of the process. We will see that an under-
standing of the entire range of shots from tees, fairways, and rough, is considerably
enhanced when the role of the spin CofR (SCofR) is recognized.
Before addressing these topics in detail, we first add some historical perspective
and also consider spin rate requirements for different shots.
T Y PES OF GOLF BA LLS A N D SPIN R ATES

The modern ball, in the sense that it had all the characteristics of the balls cur-
rently being used, was invented in 1898 by Coburn Haskell of Cleveland, Ohio.
Haskell’s ball, appearing in the early 1900s, rapidly replaced the gutta percha
ball that had been molded from the dried rubber sap of the Malaysian sapodilla
tree. The Haskell ball had a highly resilient core produced by tightly winding fine
95
rubber strands over a small inner core. The core was covered by balata, which is a
natural rubber tapped from the Balata tree found in Central and South America.
We see later that for shots off the fairway, the balata-covered ball actually provides
more spin than the current urethane covered premium balls. The only problem
with the balata balls was that the cover was easily ruptured if the ball was struck
“thin,” that is, with the edge of the sole. It would also sustain surface damage
from bounces against stones or the hard surface of cart paths. In the 1960s, much
tougher covers were introduced from a class of polymers known as ionomers.
These were developed and marketed by the DuPont Company under the trade
name Surlyn. Combined with a molded rubber core, these became the most popu-
lar ball among amateur players because of their robustness. However, according
to Thomas (Just Hit It, 2008), “the professionals were not converts, because the
balls did not spin as much off wedges and other short irons as the soft balata balls.”
The reason is that balata grips the grooved club face sufficiently to allow energy
storage across the contact surface, and so generates overspin.
The new Surlyn balls came off the face of a driver with lower spin and slightly
higher speed than the balata ball. This produced greater driving distance. However,
the professionals were unwilling to trade greater distance for the better control
around the greens available with the balata ball. The Spalding Company was the
first to offer a solution to this issue when in 1984 they produced multilayer molded
rubber balls with soft covers. Following this, the big breakthrough came in late 2000
when Titleist brought out the Pro V1. These balls had a resilient core for speed and a
thin cover that would produce lower spin off the smooth driver face but still provide
sufficient spin rates off the grooved faces of the lofted irons. According to Thomas
(2008), within weeks of the formal introduction of the Pro V1 in early 2001, 54 per-
cent of Tour players were using the ball. Over this same time period the natural
balata rubber covers had been replaced by urethane polymers, which, although not
as rugged as Surlyn, are much tougher than balata. Urethanes, like balata, have very
high friction qualities and are used in applications such as skate board wheels where
maximum surface grip is desired. The different spin rates of the urethane and Surlyn
covered balls is seen very clearly in the next section.
Before continuing the more general spin-rate discussion, the desirability of
lower spin rates off the driver face should be addressed. The issue here, different
from that for all of the other golf shots, is that of maximizing the total distance
rather than just the flight or “carry” distance. According to the rules for compe-
tition by the R&A/USGA, a golf ball must satisfy an “overall distance standard”
under specified conditions on file with the R&A/USGA. As discussed previously,
the conditions include a defined launch angle and spin rate, no wind, and a land-
ing site on a carefully manicured flat fairway. The amount of bounce and roll is, of
course, a substantial part of overall driving distance. This makes a very significant
difference to the optimum launch conditions. We know that the steep descents with
very high backspin rates from wedge approach shots can sometimes result in the
ball spinning backward on greens. That is, the ball bounces and still retains some
of its flight backspin, and in consequence rolls backward. This never happens with
long approach shots or with the driver. With the lower descents, the backspin is con-
verted, at the first bounce, to a topspin consistent with pure rolling, and the ball
will take several rapidly diminishing bounces before rolling to a stop; Daish (1972),
97 Generation of Ball Spin
Haake (1989, 1991a, 1991b), and Penner (2002b). The distance of the bounce and
roll decreases with both increasing backspin and with steepening descent angles.
The research works just listed cast considerable light on the mechanism of spin
reversal and on the launch trajectory of the first bounce and subsequent bounces.
However, despite the great importance of bounce and roll distance in the drive,
there is no experimental data in the open literature on the CofR of the fairway
bounce under different grass conditions or on the rolling resistance of different fair-
ways. Tuxen (January 2010) provides two data points for an average dry flat fairway
for typical Tour player shots rising to a height of around 30 yards; namely, for a
45 degree descent angle, the ball will bounce and roll approximately 33 yards,
whereas with a 30 degree descent angle, the distance will be approximately 50
yards. We can write a very tentative rule from these two points, namely,
bounce and roll = 84 minus (17/15) times the decent angle from the horizontal.
For the average PGA Tour player drive with a 269 yard carry, the Trackman
measured descent angle is 39 degrees. This gives a predicted average bounce and
roll distance of 84 – (17/15) × 39 = 40 yards, for a total overall distance of 309 yards.
A tailwind will of course increase the carry distance, and a downward sloping fair-
way can make a very large difference to the bounce and roll distance. Record drives
into par 4 greens, or those leaving short iron shots into par 5s in tournaments, typi-
cally take advantage of one or both of these factors.
However, without trustworthy information on the relationship between launch
angle, spin rate, and bounce and roll distance, it is impossible to determine opti-
mum launch conditions in modeling. Thomas (2008) provides three launch con-
ditions as a general guide to optimum launch “assuming normal turf conditions.”
These are 12 degrees launch and 2,200 rpm with a 120 mph club head impact speed;
13 degrees and 2,400 rpm with a 100 mph impact speed; and 14 degrees, 2,600 to
3,000 rpm, for an 80 mph impact.
The average PGA Tour player drive with 112 mph club head impact speed has an
11.2 degree launch angle and 2,685 rpm backspin. Estimating between the Thomas
values as a guide would suggest the optimum conditions for a 112 mph strike to be
12.4 degrees and 2,280 rpm. These are not far removed from each other, particularly
when considered in the context of the different fairway conditions at the different
Tour events from which this data was gathered.
These launch conditions are considerably different than those used by profes-
sional players using wood driver heads and balata balls. Bearman and Harvey (1976),
whose work on ball aerodynamics was discussed in Chapter 2, used high-speed
photography to measure the drives of a group of British professional players. The
average professional drive was found to be 105 mph club head speed (calculated
using 0.78 CofR for a measured ball speed of 152 mph), 6.1 degrees launch, and
3,450 rpm backspin. With the grooved driver faces and soft balata covered balls, it is
likely that the higher lofts and launches used by current professional players would
have produced excessive spin rates, causing trajectories with too much height,
correspondingly steeper descents, and with considerable loss of overall distance.
Unfortunately, Bearman and Harvey did not supply sufficient information to deter-
mine the behavior of the ball as it launched from the driver face. However, we find
that the average PGA Tour drive launches the ball in a slipping condition, spinning
about 30 percent more slowly than if it was rolling up the face. This combined with
the much higher launch angles has produced the large increase in overall driving
distances we see today.
This can all be described by the spin CofR, which is approximately −0.3 for the
average PGA Tour drive, and in contrast +0.3 for the average PGA Tour pitching
wedge shot. These correspond quite accurately to 30 percent underspin from the
driver for lower drag and longer bounce and roll and 30 percent overspin from the
wedge for better stopping power on the green. These dramatic differences are of
much greater importance than the relatively small improvements in normal CofR,
over which there was so much alarm.
SPIN NING A N D GR IPPING

To avoid confusion with the discussion of restitution in Chapter 3, the CofR used
there will now be defined as the normal CofR, or NCofR. This is actually the cor-
rect name because the NCofR only acts on the velocity component at right angles,
or “normal,” to the face. The spin CofR will be abbreviated to SCofR.
SCofR has been introduced only once in published investigations of golf ball
impact. This work was carried out by B. B. Lieberman (1990a) for the USGA and
was published in 1990 in Science and Golf: Proceedings of the First World Scientific
Congress of Golf. Unfortunately, Lieberman’s focus was only on predicting spin rates
for investigations of the effect of “grassy lies,” that is, shots out of moderate to deep
rough. For this he eliminated SCofR from his equations, using measured launch
angles to predict spin rates. In consequence he missed the significance of this coeffi-
cient in his experimental results. It seems the SCofR in golf was to remain unrecog-
nized as an important parameter until this work, even though a complete theory of
the recovery of spring energy to increased spin had been established about 40 years
ago. This and recent experimental and theoretical investigations by the R&A and
USGA are discussed later.
To understand SCofR, and how the elite players manage to obtain such enor-
mous spin rates off lofted irons, it is best to first consider the simpler situation of
bouncing golf balls against angled plates. This has often been used in investigations
of golf ball performance, and in particular, it is used by the R&A/USGA in golf ball
testing. The results of firing a golf ball from a pneumatic cannon against a 30 degree
angled polished steel plate are shown in Figure 4A.
The authors of this study used ultra-high-speed photography to capture the
actual ball deformation at time increments during the impact. Figure 4A shows
some very revealing pictures, which were presented but not discussed in their
article.
The illustrations in the figure were drawn by me using a drafting program to
overlay the low-resolution photographs in the article, and to use line vectors through
the center of the brand name to measure the angular positions. The series of four
dots on the ball thus replace the manufacturer’s brand name by which the authors
measured the angular turn of the ball, although their interest was only in the total
amount of the angular rotation during the impact.
Orientation of marking on ball, degrees

4 6 18.5 27 32
(3,333 rpm) (14,900 rpm) (11,800 rpm) (6,800 rpm)
0 100 240 360 470

Time since contact, microseconds (average rotational speed between each time increment)
Figure 4A A 134 mph (60 meters per second) impact with a 30-degree “loft” stationary steel
plate; from Arakawa et al. (2006).
The numbers across the top of the figure represent the angular position of the
ball, as represented by the angle from the horizontal of the line of four dots. The
numbers under the separate figures are the times from the start of contact of the dif-
ferent picture captions, measured in millionths of a second (microseconds). From
these two corresponding sets of values, it is easy to calculate the average rotational
speed that has occurred between the separate pictures. For example, between pic-
ture frames two and three (from the left), the change in angle is 12.5 degrees, and
the time increment is 0.00014 seconds. The average rotational speed over this inter-
val is 12.5/0.00014 = 89,286 degrees per second, approximately 14,900 rpm. We
can conclude that at some time between 100 and 360 microseconds after impact,
the ball rotational speed reaches a maximum, which is greater than 14,900 rpm, and
yet it launches with a spin rate of only 6,800 rpm. This obviously leaves enormous
potential for higher spin rates, which is exploited very effectively with grooved clubs
and soft “gripping” ball covers.
Up until approximately 240 microseconds after contact, the ball is seen to be
sticking to the plate; that is, there is no perceivable sliding motion of the ball as
shown by the separation between the top of the ball and the upper dotted horizontal
line. After this point, the ball angular velocity decreases until separation as the ball
travels up the face in a skidding then slipping condition. Figure 4B shows the varia-
tion of average angular velocity for each time span between the high-speed pictures.
Note that these are the average spin rates for each interval, so they are assigned to
the center time of each interval.
The angular deceleration, starting at approximately mid impact, must be due to
one or both of two factors. One is just a reversal of friction as the ball loses its grip
on the face and starts to skid. The second is a forward separation occurring between
the resultant normal contact force and the center of mass of the ball such that the
normal force produces a braking torque. This braking torque always occurs when
a ball is rolling over a surface: from deformation of the surface, the ball, or both.
We see in Chapter 7 that this is what determines the “speed” of greens in putting.
Later in this chapter, we show conclusive evidence of friction reversals and braking
torques for more compliant rubber balls, which are easier to study at slower speeds
and with longer contact times.
We can make one important observation from these results. If the ball, illus-
trated in Figure 4A, had gripped the surface until later in the impact, then it would
16000
Average interval spin rate, rpm

12000
8000
4000
end of
contact
0
0 100 200 300 400 500
Time, microseconds
Figure 4B Variation of golf ball spin during oblique impact with a smooth rigid surface.
have launched with higher spin rate, that is, less braking time and maybe additional
spin generation.
We should fully realize what this test result teaches before continuing; namely,
the golf ball reached its highest spin rate while “sticking” to the surface. This is really
strange behavior. A consequence of this realization is that the cover of the ball is
being forced into the high pressure contact region between the ball and the plate, as
indicated by the curved arrow on the second illustration (from the left) in Figure 4A.
So, the actual behavior of the golf ball, especially with grooved club faces, is one
of gripping the face while rotating. Gripping and rotating may seem contradictory,
but it is exactly what happens in many everyday products of modern life that involve
gripping and moving. Perhaps the most common example is the curved gripping
soles of athletic shoes, which must allow rotation from the ball of the foot to the
toes while decelerating and accelerating through the contact, and all the while pro-
viding grip. This is accomplished by a process of micro slip whereby adjacent, tiny
subregions of the surface alternately grip and then slip. This is often accompanied by
squealing as the grip-slip process propagates across the contact surface. Automobile
tires, of course, do the same while accelerating or braking, and particularly during
turning when the outside and inside surfaces of the tires must be travelling at differ-
ent speeds while continuing to grip.
The approach then, as in the last chapter, must be to work around the need for
an analytical solution to the deformation problem by using a CofR but now in the
direction across the face from which spin is produced. Incorporating this into a
model for prediction of spin rate and launch angle is the subject of this chapter.
The easiest way to understand SCofR is through simple examples involving plate
“impact.” We can then understand the basic nature of some test results of “distance”
and “high spin” golf balls, which show surprisingly high spin-rate performance dif-
ferences. We first consider only the motion parallel with the surface. This is per-
fectly acceptable since the SCofR is based only on the speed parallel to the surface
before and after impact. It is then defined, exactly as for NCofR, as separation veloc-
ity divided by approach velocity with respect to the contact point.
Spin
Speedout
Speedin
Stopped
Figure 4C Simple mind experiment of transition from pure slipping to pure rolling.
The first example, given in Figure 4C, illustrates the condition of pure rolling
after impact. In this case, a ball with a notch on the underside is projected across
a smooth surface until it impacts a fixed stop. Before impact, the ball is in a state
of pure slipping, that is, with no rotation. On impact with the stop, the contact
surface in the notch stops without any rebound, so the ball continues forward in
the condition of pure rolling. We take this as just a mind experiment and do not
concern ourselves with the difficulty of getting the ball to pure slip on the edge of
the notch, or to stop dead with no rebound. Neither are we concerned about the
behavior after the impact, when the undercut will strike the back surface of the
stop. It is quite easy to show that for a uniform ball, the initial spin rate from this
experiment would satisfy
Spin = 1,000 × Speedin /7;
and the outbound speed after the collision would be
Speedout = 5 × Speedin/7,
where speeds are in miles per hour and spin rate is in revolutions per minute. These
expressions are determined, with suitable changes from scientific to golf units, from
equations developed in the modeling section of this chapter. Note that the spin rate
increases directly with increases of Speed in.
Because the notch stopped at impact, the separation velocity of the notch and
stop is zero. Thus, the SCofR, equal to “separation velocity” divided by “approach
velocity” (Speed in), equals zero. This is a general result. If the SCofR is zero in any
golf shot, then the ball separates from the club face in a condition of pure rolling.
Range balls with hard covers designed for long life can exhibit near zero SCofR.
The least possible value for SCofR occurs if the impacted face is frictionless.
A smooth club face, coated with a nonstick surface such as the PTFE used on non-
stick cooking utensils and knife blades, would come close to this condition. If we
remove the stop from the experiment in Figure 4C, then the sliding ball would just
pass through the “impact” zone with the speed unchanged. The separation veloc-
ity, which is measured in the reverse direction to the approach velocity (just as for
NCofR) is then equal to the negative Speed in. The SCofR is thus equal to −1, which
always represents 100 percent frictionless sliding. For impacts with the smooth
face of the modern titanium driver, the SCofR is typically less than zero, which
indicates some sliding through the impact. As mentioned previously, the average
value for a PGA Tour player drive SCofR is approximately −0.3.
Exactly as ball speed is proportional to (1 + NCofR), so spin rate is actually

directly proportional to factor (1 + SCofR). For example, an SCofR of −0.2 for a
driver shot gives 0.8 times the spin rate for pure rolling.
Before proceeding further, we discuss the limit of positive values for the SCofR.
Figure 4D shows another mind experiment, this time for an impact that could
provide high positive SCofR values.
In this case, a compression spring is assembled into the notch in a ball. The illus-
tration on the left shows initial contact between the spring and stop. The middle
illustration indicates the end of the compression phase of impact, where the contact
point inside the notch has been brought to a stop and the spring has reached the
maximum amount of compression. The ball would have rotated forward slightly by
this point. We choose to ignore this, so at this point the spin rate and forward speed
are exactly as for the impact in the Figure 4C experiment. However, we now have a
restitution phase, before separation, during which the spring re-extends. At separa-
tion, the notch can at most be moving back away from the stop with the same speed
as Speed in. The SCofR value in this case would be Speed in/Speedin = 1.0. The spin
rate in this case is increased by (1 + SCofR), or double the amount for pure rolling.
Thus we have
Spin = 2,000 × Speedin/7.
With the higher spin rate, the outbound speed after the collision is reduced:
Speedout = 3 × Speedin/7.
These results are again obtained from equations derived later in the chapter.
It can be shown that the energy of the “ball” with these values of Spin and
Speedout is exactly the energy of the “ball” before impact. Thus, the value of
SCofR cannot be greater than +1.0 because this would violate the principle of
conservation of energy; that is, it would require the creation of additional energy
with no possible source. This result was established in the scientific literature by
R. L. Garwin, an IBM researcher who published on the “Kinematics of an
Ultraelastic Rough Ball” (1969).
So the big open question is how does a golf ball store energy in the direction
parallel to the club face as in the spring example, and how does it release the energy
Spin
Speedout
Speedin
Stopped = Speedin
Contact End of compression End of restitution
Figure 4D Experiment to demonstrate an SCofR value of 1.0.

to increase spin rate? Before answering this question, we need to generalize the pre-
ceding relationships to cover the test firing of golf balls against angled plates.
BA LL CA N NON TESTING
The simple formula for spin rate with SCofR = 0, given previously for the example
in Figure 4C, is easily extended to the general angled impact ball test, illustrated in
Figure 4E. The ball spin in the Figure 4C case was proportional to the incoming ball
speed component parallel to the plate. For the angled impact, the parallel compo-
nent of the ball speed is obtained by multiplying the Speed in by the sine of the verti-
cal angle, labeled effective loft in Figure 4E. The term “effective” is used because
both the club loft and the impact speed that would produce the same spin rate under
the same conditions are slightly different because of the smaller weight involved in
the impact. The loft and speed differences will be resolved later.
Thus, the relationship of spin rate to the variables in ball cannon testing can be
expressed as follows:
Ball spin rate is proportional to

(a) Impact speed
(b) Restitution factor = (1 + SCofR)
(c) Wedge angle factor = Sine of effective loft.
Note the very pleasing similarity of these relationships to the ones for ball
speed in Chapter 3. NCofR and cosine in the former are simply replaced by SCofR
and sine.
Finally, the spin rate value in revolutions per minute is obtained by applying the
basic multiplying factor (1,000/7) from the Figure 4C example. This surprisingly
rounded multiplier arises from the conversion from metric to golf units. The value
actually turns out to be 1,000.085/7 to three decimal places. The divisor 7 is left
over from a ball mass and ball MoI factor, which reduces to 1/(1 + 2/5) = 5/7 in the
impact analysis. The value 2/5 comes from the MoI of a uniform ball used to repre-
sent that of the golf ball: that is, uniform ball MoI = (2/5) × (ball mass) × (ball radius
squared). If the ball is non-uniform and has a heavier core and lighter outer layers,
then the MoI will be reduced and the spin rate will be increased. Heavier outer lay-
ers will have the opposite effect. Low spin balls, intended to hook and slice slightly
Spin
Speedout
Launch
Effective angle
Speedin loft
Figure 4E Angled impact ball testing against a heavy grooved plate.

less, may be produced using both covers with less grip and with heavier outer layers.
Conversely high-spin balls may have heavier cores combined with high-grip covers.
Since the stiffness of the balls needs to be within a fairly narrow range, the ability
of the ball designers to change ball MoI is very limited. For example lightweight
foamed polymers cannot be used for the outer layers to allow a substantial propor-
tion of the mass to be positioned in the core, otherwise the coefficients of restitu-
tion would be greatly reduced. Later we determine the values of SCofR from the
averages of large numbers of Trackman test data and based on the assumption of a
uniform ball. Thus the application of the resulting models with the assumption of
a uniform ball will be perfectly consistent in predicting the average behavior with
a range of different balls. If a particular ball has MoI of say 10 percent less than the
norm, the factor 1/(1 + 2/5) = 0.714 becomes 1/(1 + 0.9(2/5)) = 0.735. This is an
increase in the basic factor of 3 percent and so will result in an increase in spin rate
of only 3 percent.
It still needs to be shown that different golf balls do actually create more spin
under the same ball-striking conditions. This has been demonstrated in the most
complete way in a set of tests carried out by the Technical Director of the R&A
with two co-investigators from academia: Cornish, Otto, and Strangwood (2007).
In these tests, balls of different make and type were fired at different angles against
a heavy plate into which standard grooves had been machined. Figure 4F shows
the results for two different balls with an impact speed of 67 mph (30 meters per
second).
With these measured spin rates in cannon testing, the preceding spin cal-
culation steps can be reversed to obtain the SCofR values. For example, for the
40 degree impact with the high-spin ball in Figure 4F, we have an impact speed of
67 mph and a spin rate of 8,400 rpm. Dividing the spin rate by the impact speed gives
125.4. Dividing next by wedge angle factor, which is the sine of 40 degrees equal
to 0.643, gives 195.0. Finally, multiplying by 7 and dividing by 1,000 gives 1.37.
This is (1 + SCofR), so the approximate SCofR value is 0.37. The more accu-
rate value using the precise mass factor, developed later, is 0.36 as given on
Figure 4F. Consequently, as the ball launches, its surface is skidding in overspin
approximately 36 percent faster than if it was rolling on the plate face at separa-
tion. Note that with a perfectly clean grooved surface, even the Surlyn (ionomer)
covered ball exhibits positive SCofR.
Further validation of the results in Figure 4F can be found in the results of can-
non testing eight different brands of balls against a 60-degree angled heavy plate by
the R&A/USGA; see the Interim Report: Study of Spin Generation (2006), which
can be downloaded from the R&A at http://www.randa.org. In these tests, the balls
were categorized as low-, to medium-, to high-spin. The plate contained U grooves,
which were at the limits of width, depth, and spacing allowed for competition.
The ball impact speed was 68 mph in all cases, only 1 mph more than the R&A
test speeds, so the results can be compared directly. The corresponding values of
SCofR, calculated in the same way as previously, had a very wide range: from −0.42
for the lowest spin ball to +0.30 for the highest spin ball. These ball cannon tests
results further validate the existence of substantial overspin in golf ball strikes using
the high-spin rate balls while allowing high amounts of sliding and underspin with
others. The corresponding spin rates from which the SCofR values were calculated
14000
High-Spin Ball
Polyurethane cover 0.44
12000
Distance Ball 0.44
Ionomer cover
10000
0.12
Backspin, rpm
0.36 0.18
8000
–0.17
0.14
6000
0.25
0.04 –0.47
4000
–0.2
2000
10 20 30 40 50 60 70 80
Effective loft, degrees
Figure 4F Calculated SCofR values for the firing of golf balls at 67 mph at different angles to a
heavy grooved plate; data from Cornish, Otto, and Strangwood (2007).
ranged from 5,000 rpm for the lowest spin ball to 11,000 rpm for the highest spin
rate ball.
Three general conclusions can be made from the test results shown in Figure 4F.
First and most obvious, with a clean grooved surface, the SCofR increases as
loft increases: to a maximum at 57 degrees for the “Distance” ball and at 60 degrees
for “High-Spin” one. The reason for the peak and decline of SCofR and spin rate is
quite clear: as the loft increases, so do the “shear” forces applied to the ball from the
grooved surface, while at the same time the normal forces and resulting ball speed
decrease. At some point, the shear forces reach a value that exceeds the limit of the
surface friction, and beyond that point slipping starts to occur earlier in the contact
period, or even throughout the entire contact.
The second conclusion is that the soft high-grip urethane cover has a pro-
found effect on spin rates. At 60 degrees of effective loft, the SCofR is 0.44 for
the high-spin ball compared to 0.12 for the harder surface of the ionomer covered
ball. The increase in spin because of the different cover is in the ratio 1.44/1.12; or
approximately 30 percent increase.
The third, and somewhat tentative, conclusion is that the low SCofR values for
driver and fairway wood shots may have as much to do with the small loft angle as
to the absence of aggressive grooves. All of these tests were carried out against a
plate that had the standard groove pattern used on irons, and yet SCofR has become
negative for both ball types below 20 degrees of loft. We should note, however, that
the impact speed of 67 mph is much less than that for drivers and fairway woods.
Together, with the small loft angle, this gives a speed component across the face
much lower than that encountered in actual golf shots, particularly with clubs of
20 degree loft or less. However, we see in a later section that the same trend of
decreasing SCofR with loft occurs in PGA and LPGA Tour player ball striking.
INCR E ASED A N D DECR E ASED SPIN

R ATES IN “GR ASSY” LIES
It is well known in golf that spin rates decrease dramatically when hitting approach
shots from the rough, which is defined as grass at least the height of the ball.
However it is not common knowledge that spin rates actually increase when hitting
longer to mid irons from so called “grassy lies.” This is counterintuitive. The juices
that are squeezed from grass blades trapped between the ball and the club face act as
a lubricant and serve to reduce the shear forces during the impact. This is the reason
for the so-called fliers out of the rough with short approach shots, that is, shots that
because of the reduced friction launch higher with reduced backspin and refuse to
stop on the green. An explanation for the increased backspin, with longer iron shots
out of grassy lies, will be offered later. For now, we introduce some results to explain
the great variation in spin rates found in ball cannon and player testing with “grassy”
conditions.
Lieberman (1990a), in his work with the USGA, was the first to recognize that
spin rates could increase because of grassy lies. He carried out ball cannon test-
ing against grooved plates and introduced surface treatments to simulate different
grassy lies. His no doubt painstaking investigation with film photography produced
sets of ball cannon results for different balls. Three sets of ball cannon results, for
different impact speeds, are shown in Figure 4G. At each speed, one set of tests was
carried out with the surfaces clean and dry. A second set of identical tests was car-
ried out with an artificial lubricant designed to replicate conditions in the rough.
The pairs of tests are shown by the black and gray lines, respectively. For example,
the solid lines are for an impact speed of 109 mph; the black one is for the dry condi-
tion, and the grey one is for the grassy condition.
15000
Impact speeds: ‘Dry’ conditions 0.29
109 mph Grassy lies
12000 81.8 mph
68.2 mph
Backspin, rpm
0.24
0.19
9000
0.17
0.17
6000
0.04 0.10
experimental data points
3000
with calculated SCofR values
0.09
0
15 20 25 30 35 40 45 50 55
Loft, degrees
Figure 4G Spin rates produced by ball cannon shots at inclined plate angles with varying speeds
and plate angles. Black curves represent dry impact conditions, and gray curves simulate a given
level of “grassy” lie; adapted from Lieberman (1990a).
There are three aspects of great interest in this work. First, it provides further
support to the existence of positive SCofR values; see the calculated values for
actual experimental data points in the figure. As mentioned earlier, this was not
reported by the author, whose main interest was on the effect of grassy lies on spin
rates. He used the measured values of launch angle to predict the spin rates, which
gave the false impression that the deformation and recovery of the ball was not a
factor in the resulting ball spin. However, as mentioned earlier, SCofR affects both
spin rate and launch angle in a manner that renders its effect to be insignificant on
the relationship between them. This point is explained fully later in the chapter.
Figure 4G shows very clear evidence that with face lubrication produced by
simulated grass juices, the loft at which maximum spin rate occurs is smaller; also,
the subsequent rates of spin decline are greater. It also shows the very surprising
result that, for lofts up to the 6-iron range, the spin rate can actually increase in the
presence of face lubrication. A likely explanation of this phenomenon will be given
later. Note the very high level of sensitivity of the results to surface friction at high
loft angles. With the highest speed at 55 degrees of loft, the grass lubrication has
reduced the spin rate from 14,000 to 7,000 rpm.
A more recent ball cannon investigation of “grassy” lies was presented in the same
R&A/USGA Interim Report: Study of Spin Generation (2006) discussed earlier. More
so than in the Lieberman results, the R&A/USGA experimental study shows higher
spin rates occurring from simulated rough lies in ball cannon testing. In this case,
higher spin rates were obtained with simulated light rough conditions than with a
clean grooved-plate surface, from 25 to 45 degrees of loft. For example, for 35 degrees
of loft, a spin rate of 6,000 rpm off a clean plate with standard grooves became 8,000
rpm when lubrication was added to the plate to simulate light rough.
Even more interesting in the 2006 R&A/USGA report are the results of pro-
fessional player shots from light rough and from the fairway. Descriptions of these
tests and the results are given in Appendix A of the report. In this case, a group of
six professional players were asked to make 5-iron, 8-iron, and sand wedge shots from
fairway lies and from light rough in which the grass was about half of the ball height.
The investigation compared the performance of a pre-1990 club/ball combination
with a modern club/ball combination. The difference between the clubs was simply
the V grooves milled into the face of the pre-1990 clubs and the more aggressive U
grooves in the modern clubs. For the latter, the U grooves were machined to the
limit of the current R&A and USGA conforming standard. The selected pre-1990
ball was the Titleist Tour Balata with a natural balata rubber cover. The selected
modern ball was the Titleist Pro V1 392, which has a three-layer construction and a
urethane cover. In total, 600 player shots were taken by six professional players, and
the results were averaged for the six conditions of three clubs and two different lies.
The results for the “dry” fairway conditions were remarkable in that the pre-1990
club/ball combination produced almost identical ball speeds but higher spin rates
than the modern club/ball combination. The spin rates from the pre-1990 equip-
ment averaged 25 percent higher for the 5-irons, 27 percent higher for the 8-iron
shots, and almost identical spin rates for the sand wedge shots. From our earlier
discussion, this tells us that the natural balata rubber cover grips the V grooves bet-
ter than the urethane cover grips the modern U grooves under “dry” conditions for
the 5-iron and 8-iron shots. The grip is about the same for the two combinations of
sand wedge shots. Although results for the other possible combinations were not
reported, it seems likely that the Titleist Tour Balata would give even higher spin
rates with the U-grooved clubs, including the sand wedge.
So it is clear that Tour players are not playing the modern urethane covered balls
for their performance off the fairway. We know that one reason the modern ball is
preferred is that it grips the smooth face of the driver less, which lowers spin rate
and consequently lowers drag, and the lower backspin also helps to increase bounce
and roll distances.
However, the main advantage of the modern club/ball combination is its per-
formance out of the rough. This advantage can be seen very clearly in Figure 4H.
This shows launch spin rates divided by launch ball speeds, referred to as the nor-
malized spin rates. The V symbols are for the pre-1990 club/ball combination, and
the black U symbols are for the modern combination. Both sets of shots are played
from the light rough. It can be seen that the difference between the pre-1990 per-
formance and that of today is profound. Since ball speeds are almost exactly the
same for the pairs of combinations, the relationship between the normalized spin
rates is the same as the relationship between the actual spin rates. These show that
the spin rates, out of the light rough with the modern equipment compared to the
pre-1990 equipment, are 84 percent higher with the 5-iron, 100 percent higher with
the 8-iron, and 38 percent higher with the sand wedge. Of course, if wider, deeper,
and sharper U grooves were allowed by the R&A and USGA, then the approximate
doubling of spin rates from the rough could likely extend to the sand wedges.
For comparison, the gray U symbols are the average data for PGA Tour players
from Trackman (Tuxen, January 2010). This data was all obtained at the practice
ranges for Tour events, so it represents shots off the fairway. As also shown by the
ball cannon testing in Figure 4G, for the longer irons, higher spin rates are obtained
from the light rough than for the fairway shots. For the 5-iron shots, for example,
70
U fairway U
U
U light rough
V n
der
mo
Spin/Ball speed, rpm/mph
U
55
U
our
V
AT
PG
40 U
0
99
e-1
V V
pr
25
5-iron 8-iron SW
Figure 4H Spin rates divided by ball launch speeds for professional player iron shots. The black
U and V symbols are for shots from partial rough from the R&A/USGA Interim Report: Study of
Spin Generation (2006). V symbols are for pre-1990 equipment and U symbols represent modern
equipment. Gray U symbols are the average 2010 PGA Tour player data for fairway shots.
the normalized spin rate is 40 from the fairway and 52 from the light rough, a dif-
ference of 30 percent. Moreover, compared to the pre-1990 5-iron shots from the
light rough, with a normalized spin rate of 31, the modern 5-iron from the rough has
68 percent more spin. We discuss the likely source of these differences in the next
section.
It would be improper to leave the topic of ball bouncing without mentioning
important scientific work that addresses the topic of energy release into the spin.
An analytical solution for the bouncing of perfectly elastic balls against angled sur-
faces was established by Maw (1976) and improved as a predictive model by Maw,
Barber, and Fawcett (1976). It has been shown by R&A and USGA researchers to
predict the form of behavior in Figure 4F. This work is discussed in Appendix H of
the R&A/USGA Interim Report: Study of Spin Generation (2006). It was also shown
in the study to predict spin rates “somewhat higher” than the limit of pure rolling.
The work has a high level of complexity, and it is difficult to imagine its extension to
the large deformation impact behavior of golf balls with appropriate rubber proper-
ties. But this is exactly why CofRs are introduced into the study of impacts—to
eliminate the need for often impossibly difficult analysis of the deformation itself.
On that note, we continue with a discussion of how storing and releasing energy in
golf ball striking can occur without the steps, stops, and springs of our earlier mind
experiments. We can then go on to determine SCofR values from Trackman data of
actual ball striking.
THE M ECH A NISM OF OV ER SPIN

CR E ATION IN GOLF STR IK ING
The main purpose of reviewing the ball cannon test results was to demonstrate con-
clusively that the SCofR is a major factor in golf, one that can make a big difference
in both flight trajectories and in the ability to stop longer approach shots on greens.
From our simple example in Figure 4D, it is clear that the primary requirements
for getting a substantial positive SCofR are that the ball should possess spring qual-
ity parallel to the face. In addition, under dry conditions, a large velocity parallel
to the striking surface, as provided by higher lofted clubs, will better compress the
spring material lying across the striking surface and thus more effectively store
spring energy to be recovered as increased spin rate. The cores of all golf balls are
made from synthetic rubber, which is the ultimate spring quality material. Its abil-
ity to store spring energy in a given weight of material is second to none. That is the
reason, for example, why nothing emerging from modern material science has been
able to outperform rubber bands in wind-up model aircraft where spring energy per
weight is the single most important property. The synthetic rubber having among
the very highest spring qualities is polybutadiene rubber or PBR. This is the mate-
rial of so-called “superball” toys, which have quite amazing bounce characteristics.
Not surprisingly, PBR is also the preferred core material in premium performance
golf balls. So, for its small mass, the golf ball is able to store large amounts of spring
energy in any direction during the bounce. The release of spring energy normal to
the face of course gives the high NCofR values and resulting high ball speeds.
Having shown that SCofR values greater than zero occur in golf-ball collisions,
the question still remains as to the mechanism by which this occurs. First, to store
material
Larger enters
pressure
and friction compressed
zone
Smaller
pressure material
and friction leaves more slowly
Figure 4I Mechanism of the storage and release of energy in spin restitution.
spring energy in the direction parallel to the surface, or face of the club, it is neces-
sary for the ball to be stopped from slipping across the surface; that is, the ball must
grip the surface. That brings us to the interaction between the ball cover and the
grooved club face.
Figure 4I shows a golf ball gripping a club face while turning, just as in the
early stages of the plate collision shown in Figure 4A. Its behavior during the
impact can be represented by the two internal compression springs as shown on
the right. These are assumed to behave independently of one another. We show
later in the chapter that the action of spring energy storage parallel to the face
(represented by the black spring) tends to offset the spring effect at right angles
to the face toward the top of the ball. This is a secondary effect, which produces
some reduction in the spin rate. But the effect is simply reflected in a small reduc-
tion of the SCofR.
The spring effect at right angles to the face, represented by the gray spring,
acts to provide the high separation speed normal to the face. The spring effect
parallel to the face acts to increase the separation speed of the ball parallel to the
face as described in the Figure 4D mind experiment. The key to this action is the
movement of the ball surface across the contact region. We can visualize the pro-
cess as follows: the torque applied to the ball, by the shear force from the club
face, offset from the ball center, imposes rotational motion to the ball. This in
turn forces insertion of the surface of the ball under the contact surface at the
highest contact point, as noted by the figure caption “material enters.” Under the
surface the movement is greatly restricted, so the material across the width of
the contact surface is being subjected to increasing compression. This causes the
storage of spring energy across the area of contact as signified by the black spring.
The second part of the puzzle is solved by the realization that the pressure under
the ball must be higher where material is being inserted than at the opposite side,
or lowest point of the contact surface, where material is locked or emerging more
slowly. The non-uniform distribution of pressure “under” the ball, represented by
the parallel arrows, illustrates this.
Thus, as the ball recovers from compression, and the pressure under the ball
decreases, the surface will slip first at this lowest contact point where the pressure
is least. This will be accompanied by release of the energy from the black spring,
putting the ball into a skidding condition as the entire ball spins up to distribute the
internal rotational momentum uniformly. My understanding of this process came
about through personal discussions with Dr. Rod Cross of the University of Sydney.
Some of his groundbreaking research on the mechanics of bouncing balls will be
described after the next section.
The time during the contact at which the ball first starts to slip at its lowest point
of contact is clearly of the greatest importance. As the ball recovers from maximum
normal compression, while sticking to the face, some of the parallel spring compres-
sion will push out equally on both sides of the ball. If the ball never starts to slip,
then all of the parallel strain will push out equally on both sides and none will be
converted to increased spin. Pursuing this conjecture further, the SCofR may be
low for smaller loft angles, as in the Cornish, Otto, and Strangwood study, because
the much higher normal pressure retards the slip and parallel energy release until
too late. If this is the case, then a decrease in friction from grass juices allowing
earlier slipping at the lower edge of contact is a likely explanation for the higher spin
rates from the light rough with the lower-lofted iron. With deep rough, the friction
is no doubt reduced too much and slipping occurs too early, allowing little or no
buildup of strain energy.
A PPLICATION OF THE SCof R TO BA LL STR IK ING

Surprisingly, the steps for determining the spin rate in actual golf ball striking
require only a very simple change to the steps used for cannon testing. That is
the multiplication by a “ball mass-MoI factor,” which represents the difference
between the heavy plate in cannon testing and the weight of the club head. The
factor in this case contains both the ball mass, the club head mass, and the ball
MoI. It reduces to 1/[1 + 2/5(1 + ball mass/club head mass)]. For a ball mass of
0.1 pounds and the range of club head mass from 0.44 pounds for the driver to
0.64 pounds for a wedge, the factor varies over a very narrow range from 0.67 to
0.68. So within the level of accuracy of the modeling, it can be represented by
factor (2/3) for all golf shots. In the calculations that follow, we use the closer
approximation of 0.675 to demonstrate equivalence to more complex calculations
used in the second part of the chapter. It can be seen that if the effective “club head
mass” is very high, as for the plate in the cannon testing by Cornish, Otto, and
Strangwood, which weighed 44 pounds, then ball mass/club head mass is negli-
gibly small and the mass-MoI factor becomes the value 5/7 used in the previous
section. So, switching 0.675 for 5/7, the basic multiplier 1,000/7 used to convert
speed to spin for a golf shot becomes 135 to the nearest integer. Note that 1,000/7
equals 143 to the nearest integer, so even a 44 pound “club head” only produces
5.6 percent more spin.
Finally then, to obtain the ball spin rate from impact speed, club loft, and SCofR
value, we can use the following simple calculation steps:
(a) Multiply the impact speed by 135

(b) Multiply result (a) by the sine of the loft
(c) Multiply result (b) by (1 + SCofR)
Spin coefficient of restitution values

140 PGA: driver, 3-wood, 5-wood, hybrid, 3-, 4-, 5-, 6-, 7-, 8-, 9-, PW 0.4
LPGA: driver, 3-wood, 5-wood, 7-wood, 4-, 5-, 6-, 7-, 8-, 9-, PW
0.3
Spin rate/ Impact speed, (rpm/ mph)
120
0.2
0.1
100
0.0
80
60
40
20
10 15 20 25 30 35 40 45
Spin loft, degrees
Figure 4J Chart relating normalized spin rate to loft and SCofR values.
These calculations can be graphed to provide a lookup chart of spin rate divided
by impact speed, or normalized spin rate, for different values of loft and SCofR val-
ues. This is shown in Figure 4J.
For example, the average PGA Tour player 5-iron shot has an impact speed of
94 mph, a dynamic loft value of 20.5 degrees, and an SCofR value of 0.2. Reading
upward from just past the 20-degree vertical line to the third diagonal curve, the
0.2 SCofR curve, brings us to approximately 57 on the vertical scale. This point is
marked by one of the small circles, which from left to right represent the average
PGA Tour player shots from driver to pitching wedge. The predicted spin rate, in
this case for the 5-iron, is 57 times the impact speed of 94 mph, which almost exactly
equals the Trackman measured value of 5,361 rpm.
We can reverse the calculation steps of course to obtain SCofR values from
Trackman data giving impact speed, spin rate, and loft. For example, the aver-
age 6-iron PGA Tour player shot has an impact speed of 92 mph, a dynamic loft
of 23.8 degrees, and a spin rate of 6,231 rpm. Dividing 6,231 by 135 gives 46.2,
which divided by 92 gives 0.502. The sine of 23.8 degrees is 0.404, which divided
into 0.502 gives 1.24. This is the estimated value of (1 + SCofR), so the estimated
SCofR for the shot is 0.24. Using the exact equations developed later, instead of the
approximate 135 divisor, gives SCofR = 0.232.
Figure 4J also includes the calculated SCofR values over the full range of shots
of average PGA and LPGA Tour player shots published by the Trackman Company.
The caption across the top of the figure lists the clubs in increasing loft order. We can
see that the LPGA players generate higher normalized spin rates and SCofR values
on average than the PGA players. The PGA Tour players reach a maximum SCofR
of 0.32 for the 8- and 9-iron shots. In contrast, the LPGA Tour players reach a maxi-
mum of 0.38 SCofR for the 8-iron shots. The differences are most likely a result of
different ball preferences. Note that in both cases the SCofR increases with loft, as
for the ball cannon test results.
An additional coefficient is very helpful in visualizing the effect of increasing

SCofR values. This is used instead of SCofR in Chapter 6, where we deal with the
complications of offset strikes on the curved driver face. This is referred to as the
“slip/skid” parameter. It measures the amount by which the spinning ball surface
travels slower than the club face surface in a slipping mode, or faster than the club
surface in a skidding mode. The skidding mode is overspin. This parameter brings
out more clearly the basic nature of the results, that is, that the ball in most cases
jumps off the face while spinning up to a higher value. It is calculated from the
change of momentum equations later in the chapter and it turns out to be related
with good approximation to the SCofR by
slip/skid = 1 + SCofR.
We can now turn our attention back to the details of the performance of Tour
professionals. Table 4A shows, on the left-hand side, the Trackman data for the
average PGA Tour player, which was analyzed for NCofR in Chapter 3. The cal-
culated values of the slip/skid parameter, and the SCofR, are shown the right side.
The relationship between the slip/skid parameter and the SCofR is as expected.
The predicted launch to loft ratios were discussed in Chapter 3 and are included to
demonstrate the consistency of the results. The values in Table 4A were determined
from a modeling expression derived later, which shows that the loft/launch ratio
changes according to the ratio of (1 + SCofR)/(1 + NCofR). This is a very pleasing
result that demonstrates the relationship between launch angle and overspin. The
ratios are identical to the values in Chapter 3, which were obtained directly from the
measured launch angles.
The SCofR appears to reach a maximum at a loft angle somewhere between the
8- and 9-iron shots. However, only one data point, namely for the pitching wedge
shots, suggests that the maximum has been reached. We first discuss the LPGA data
and then show some data on partial wedge shots that further strengthens the con-
clusion that maximum SCofR is reached somewhere around 35 degrees of dynamic
loft. The decline starts at a significantly smaller loft angle than for the ball cannon
tests, which showed continued SCofR increase up to approximately 60 degrees of
loft. We will consider more ball striking data before attempting any explanation
for this.
Table 4B shows Trackman data from the same source but for LPGA Tour pro-
fessionals. The trends are very much the same. There is less face slip with the driver
and more overspin with the irons by the female Tour players. As already mentioned,
this is possibly due to their ball selections; perhaps a higher number favoring the
softest cover balls. The amount of de-lofting is similar to that for the PGA players.
Women players tend to use clubs with slightly higher lofts, and the launch angles
for 4- through 9-iron, taking due account of the different attack angles, are seen to
be around 2 degrees higher for the female players. Also it should be noted that for
these players, the SCofR values increase up to a loft of 39.7 degrees, excluding the
anomalous result for the 9-iron.
More recently the Trackman Company (Tuxen 2014) has released data on pitch
shots with wedges ranging from 47-degree pitching wedge shots through 60-degree
lob wedges; see Table 4C. The data is for pitch shots in the 50 to 80 yards carry
Table 4A Average Trackman test results and modeling for PGA Tour players
Trackman data* Model

Head mass Attack angle Club speed Launch angle Ball speed Back-spin Dynamic loft Slip/skid Spin CofR Launch/loft
(kg) (deg) (mph) (mph) (rpm) (deg)
To horizon To club
(deg) (deg)
Driver 0.2 −1.3 112 11.2 12.5 165 2,685 14.4 0.71 −0.28 0.87
3-wood 0.22 −2.9 107 9.2 12.1 158 3,655 14.8 0.99 0.00 0.82
5-wood 0.23 −3.3 103 9.4 12.7 152 4,350 16.0 1.14 0.14 0.79
Hybrid 0.24 −3.3 100 10.2 13.5 146 4,437 17.0 1.13 0.12 0.79
3-iron 0.240 −3.1 98.0 10.4 13.5 142 4,630 17.2 1.19 0.18 0.78
4-iron 0.247 −3.4 96.0 11.0 14.4 137 4,836 18.4 1.18 0.17 0.78
5-iron 0.254 −3.7 94.0 12.1 15.8 132 5,361 20.5 1.21 0.20 0.77
6-iron 0.261 −4.1 92.0 14.1 18.2 127 6,231 23.8 1.25 0.23 0.76
7-iron 0.268 −4.3 90.0 16.3 20.6 120 7,097 27.4 1.27 0.26 0.75
8-iron 0.275 −4.5 87.0 18.1 22.6 115 7,998 30.6 1.34 0.32 0.74
9-iron 0.282 −4.7 85.0 20.4 25.1 109 8,647 34.2 1.34 0.32 0.73
PW 0.289 −5.0 83.0 24.2 29.2 102 9,304 39.7 1.30 0.28 0.74
Table 4B Average Trackman test results and modeling for LPGA Tour players
Trackman data* Model

Head mass Attack angle Club speed Launch angle Ball speed Back- spin Dynamic loft Slip/skid Spin CofR Launch/
(kg) (deg) (mph) (mph) (rpm) (deg) loft
To horizon To club
(deg) (deg)
Driver 0.2 3.0 94.0 14.0 11.0 139 2,628 13.2 0.91 −0.09 0.84
3-wood 0.22 −0.9 90.0 11.2 12.1 132 2,705 14.4 0.89 −0.11 0.84
5-wood 0.23 −1.8 88.0 12.2 14.0 128 4,501 18.0 1.24 0.22 0.78
7-wood 0.24 −3.0 85.0 12.7 15.7 123 4,693 20.1 1.20 0.19 0.78
4-iron 0.247 −1.7 80.0 14.3 16.0 116 4,801 20.7 1.26 0.25 0.77
5-iron 0.254 −1.9 79.0 14.8 16.7 112 5,081 21.9 1.29 0.27 0.76
6-iron 0.261 −2.3 78.0 17.1 19.4 109 5,943 25.7 1.31 0.29 0.76
7-iron 0.268 −2.3 76.0 19.0 21.3 104 6,699 28.7 1.37 0.35 0.74
8-iron 0.275 −3.1 74.0 20.8 23.9 100 7,494 32.5 1.41 0.38 0.73
9-iron 0.282 −3.1 72.0 23.9 27.0 93 7,589 36.4 1.32 0.30 0.74
PW 0.289 −2.8 70.0 25.6 28.4 86 8,403 39.7 1.40 0.38 0.72
Table 4C Trackman test results and modeling for partial wedge shots
Wedges Trackman data* Model

Head mass Attack angle Club speed Launch angle Ball speed Back-spin Dynamic loft Slip/skid Spin CofR Launch/
(kg) (deg) (mph) (mph) (rpm) (deg) loft
To horizon To club
(deg) (deg)
47 0.29 −4.5 50 25.6 30.1 60.8 5,642 40.8 1.28 0.26 0.74
50 0.29 −4.5 50 27.6 32.1 58.4 6,002 44.0 1.28 0.27 0.73
52 0.29 −4.5 50 28.9 33.4 56.7 6,232 46.1 1.28 0.27 0.72
56 0.268 −4.0 64 29.8 33.8 68.5 8,252 47.7 1.30 0.28 0.71
54 0.29 −4.5 50 30.5 35.0 54.9 6,236 48.1 1.24 0.23 0.73
55 0.29 −4.5 50 30.9 35.4 54.3 6,178 48.6 1.22 0.21 0.73
56 0.29 −4.5 50 31.5 36.0 53.9 6,129 49.1 1.20 0.19 0.73
58 0.29 −4.5 50 32.4 36.9 52.9 6,023 50.1 1.16 0.15 0.74
60 0.29 −4.5 50 33.4 37.9 51.9 5,919 51.1 1.12 0.12 0.74
distance range using club head speeds of 50 mph for all but one of the shots. These
partial shots provide a valuable contrast to the full shots in Tables 4A and 4B. Of
particular interest in Table 4C is the relationship between spin rate and loft. The
right-side of the table shows the predictions from the preceding equations. The most
interesting aspect of the data is the gradual reduction of the SCofR and the slip/skid
parameter as the dynamic loft increases above 48 degrees. Note the data comprises
a set of wedge shots with a −4.5 degree angle of attack. However, row 4 contains
the results of a single shot taken with an angle of attack of −4.0 degrees and with a
significantly higher club head speed. This shot included more aggressive de-lofting
of the club, so the loft and associated slip/skid parameter do not follow the trend of
the other loft values.
Neglecting this off-trend result, a virtually constant spin rate is achieved above
45 degrees as the dynamic loft increases and SCofR decreases. It indicates that the
extra spin rate, which would have been produced by the extra loft, has been can-
celled through increasingly early release of the shear strain energy prior to final
ball separation. This makes complete sense since the loft increase is associated with
decreased normal impact force; therefore, the maximum possible shear force with-
out slipping decreases.
The lower impact speed in the partial wedge tests, than found in the PGA
and LPGA data, suggests some possible effects of impact speed on the maximum
spin-rate loft value. The loft angle at which maximum spin rate occurs in the PGA
and LPGA data is at 32.5 and 40 degrees with corresponding impact speeds of
86 and 73 mph, respectively. However, this does not explain in any way the increase
in spin rate all the way up to 60 degrees of loft in the Cornish et al. (2007) ball
cannon tests. The most likely candidate for the continued increases of spin rate
beyond 45 degrees in the ball cannon tests is better control of the friction condi-
tions. Laboratory experiments where a clean new ball is used for every test, and the
grooved surface is perfectly clean, are not likely to replicate the club face condi-
tions out on the course or on the practice grounds. This is clear from the simulated
“grassy” lie testing of Lieberman (1990a) and the R&A/USGA. From those results,
it can be assumed with confidence that even small amounts of grass getting into the
club-ball interface will diminish spin rates. And, of course, this becomes much more
likely as both lofts and angles of attack increase. This suggests that a more effective
face cleaning method than just a wipe with a towel after each shot might be worth-
while, particularly after shots from the rough. Trials with a cleaning agent such as
denatured alcohol would certainly be worthwhile. This may seem excessive, but the
wider literature on deformation of materials illustrates wide variations in friction
conditions caused by the slightest contamination of otherwise clean surfaces.
Ball manufacturers will almost certainly further increase the SCofR of golf
balls. The rubber materials used in golf ball cores are so-called viscoelastic materi-
als. That is, they exhibit viscous behavior, as, for example, in thick liquids such as
molasses. A gob of molasses fired at a plate will of course just collapse into a larger
blob stuck to the plate. In doing this it has absorbed all of the energy in internal
friction between the molasses molecules. The golf ball has a significant amount of
this behavior, as well as elastic, or spring-like behavior. As discussed in Chapter 3,
the recovery of energy in a viscoelastic collision is approximately equal to the CofR
squared. Thus, for the driver shot with 0.83 NCofR, the amount of energy recovered
from the normal compression against the face is approximately 0.83 squared, equal
to 0.69. Equivalently we can say that 31 percent of the energy has been lost in inter-
nal friction between the polymer molecules. So the elastic behavior dominates.
If we turn our attention to the SCofR, the very best energy recovery perfor-
mance in the average PGA and LPGA Tour shots is 0.38 for the LPGA 9-iron and
wedge shots. For these cases, the proportion of energy recovered from the shearing
of the ball is 0.38 squared or 0.14. This equality of SCofR squared to the propor-
tion of energy recovered has been proved by Cross (2014). Conversely, for this shot,
86 percent of the energy put into the ball, by the shearing effect of the lofted face
strike, has been lost in friction. In this case, however, it can be a combination of both
internal friction between the polymer molecules and friction between the ball and
the face as the ball slips during the impact.
One obvious way to improve SCofR is through the development of new ball cover
materials that better grip the club face. This is no doubt a major area of research by
the golf ball manufacturers, particularly with regard to the loftiest wedges and the
effects of grass juice lubrication when hitting out of rough. Also, if even thinner cov-
ers than on the Pro V1 can be made to withstand the impact conditions, then even
better gripping will result. It is easy to visualize the inner softer rubber core pushing
a thinner cover even deeper into the U-shaped grooves. In addition, the manufac-
turers are almost certainly investigating new core materials that lose less energy
in restitution when recovering from shear deformation. The viscoelastic material
property that defines the energy loss, when recovering from shear, is named the
“shear loss” modulus. The reader, with particular interest in this topic, may wish
to search Google Patents with the key words “golf shear modulus” to see the large
number of patents filed and get a measure of the very large amount of research that
is ongoing in this area in the golf ball R&D departments.
This is certainly an area of great interest to the R&A/USGA. It is the reason
why these governing bodies added new and precise limitations to their equipment
rules on the depth, width, spacing, and edge sharpness of grooves. It is easy to get
a feel for the stakes involved by simply assuming that more aggressive grooves
would allow the urethane covered premium ball to grip the face with the loftiest lob
wedges. In that case, we would expect the SCofR values in Table 4C to continue to
rise rather than decline as the loft increases. From the trends in Figure 4J, we might
expect an SCofR value of 0.4 with the 60 degree lob wedge. Even when de-lofted to
51.1 degrees as shown on the last row of Table 4C, with an actual SCofR equal
to 0.12, this would give a spin rate increase from 5,919 rpm to 5,919 × 1.4/1.12 =
7,399 rpm for the 50 mph impact. With the PGA average wedge strike speed of 83
mph, this would give a spin rate of 7,399 × 83/50 = 12,282 rpm, sufficient to con-
trol ball movement on the most difficult of greens. Toward the other end of the loft
range, a decrease in the shear energy loss in the golf ball would allow elite players to
hit even longer approach shots with lower lofts, still with sufficient braking power to
aim directly at the flag.
We can get some measure of the possibilities for further increasing overspin in
ball impacts by looking at test results of high SCofR rubber balls (Cross 2014). In this
case, one of the test balls was molded PBR of a similar size to golf balls. Recall that
PBR is the core material of choice for premium golf ball cores. In these tests Cross
simply hand projected the balls onto a granite slab at speeds between 9 and 16 mph.
The slower speeds provide clear images with digital video for precise data analysis.
With hand launching, the balls possessed a small amount of spin going into the col-
lision; therefore, the evaluation of SCofR is based on the difference in spin before
and after.
The entire data set from the PBR ball investigations was kindly given to me by
Cross. The PBR balls have a diameter of 1.8 inches compared to 1.68 inches for a golf
ball and a mass just slightly larger than a golf ball. The spin rate from angled impact,
all other things being equal, reduces proportionately with increasing diameter. So
instead of the basic multiplier in our spin rate calculation being 1,000/7, it changes
to (1.68/1.8) × (1,000/7) = 133. Using this value gives approximate estimates of the
calculated SCofR values as shown in Figure 4K. These SCofR values are in almost
precise agreement with the values calculated without approximation in Cross’s 2014
article.
The first thing to be noted about these results is that they were obtained from
bounces on a smooth granite slab. Even so, grip does not start to fail and SCofR
decrease until about 45 degrees of loft. The second observation is that SCofR remains
mostly above 0.6 until 50 degrees of loft. The property of viscoelastic materials is
that they lose a bigger proportion of stored energy the more they are deformed. So
these values may be unattainable at the impact speeds of golf. However, knowing
the large number of graduate polymer chemists in the employ of the large ball man-
ufacturing companies, and the patent proliferation in this area, we should certainly
expect further improvements in spin performance.
The main value of the Cross data is that it includes the rebound speed and the
launch angle. This allows some assessment of the full complexity of the bounce.
The principle finding, from analysis of this ball data bounce later in the chapter,
is that the loft values calculated in Chapter 3 and used in this chapter underestimate
0.8
0.7
0.6
Spin coefficient of restitution
0.5
0.4
0.3
0.2
0.1
Cross, superball experiments, 2014
0.0
–0.1
–0.2
5 15 25 35 45 55 65 75
Equivalent loft, degrees
Figure 4K Results of “golf ball core” bounce testing; Cross (2014).

the true dynamic loft by an approximately constant value of -6 percent. For example,
from the average PGA Tour player 5-iron shot in Table 4.1, the dynamic loft was esti-
mated to be 20.5 degrees. An improved estimate would be 20.5/0.94 = 21.8 degrees.
This calculation is presented at the end of the chapter without the approximation to
be 21.9 degrees.
This result also means that the predicted SCofR values are in error by some
small amount. However, the predicted SCofR loft values are consistent with pre-
dicted dynamic loft values and thus provide a consistent basis for modeling. If the
slightly better estimate of loft is used, then the SCofR values turn out to be slightly
larger. The reason is that we must then include rolling resistance in a more complex
model, which, with a higher SCofR value, will bring the predicted spin rate back to
the Trackman measured value. Since the error is small, it seems preferable to use the
spin rate calculated loft values and the correspondingly much simpler model.
GE A R EFFECT SPIN
When we shift our attention to the driver, which has its center of mass significantly
behind the face, a second mechanism of ball spin becomes important. This is known
as the gear effect and has been recognized as creating sidespin from heel or toe
strikes since the earliest scientific studies of the game; see Cochran and Stobbs
([1968] 1999). Its primary importance is for off-center driver hits, but it also has a
small effect on the center hit. What is generally not recognized is that, because of
the gear effect, ball strikes high and low on a driver face can have a profound effect
on backspin. We deal with this situation first.
Gear effect is relatively easy to evaluate if we make direct use of the slip/skid
factor. Since the angled face applies a shear force to the ball to create backspin, the
equal and opposite tangential impulse acting on the club face must of course pro-
duce an opposite rotation of the club head. This is shown in Figure 4L, where the
club head rotation is shown to act about the sideways axis through the club center
of mass. The value of the rotation speed can be determined by the condition that
the shear force that causes the ball to both spin and be launched lower than the loft
angle, by reaction from the ball, also causes the club head to be rotated backward.
This is a relatively small effect for a center hit.
For example, for the average LPGA driver strike in Table 4B where impact
speed = 94 mph, dynamic loft = 13.2 degrees, and launch angle = 11 degrees, the
gear (3)
gear (2) (1) launch angle

force
s
(1)
wedge (2)
Figure 4L Force and spin interactions for center ball strike with a flat-face driver.
shear force from the ball to the face causes a backward head rotation speed of
117 rpm. Keep in mind that all of this happens in 0.00045 seconds, so while this is
represented as a rotation of 117 rpm for ease of assessing the relative magnitudes, it
is more appropriately thought of as an angular kick backward.
The head rotating in one direction causes the ball to rotate in the opposite direc-
tion just like a pair of gears. The head rotates about its center of mass, which in the
modern driver is usually further back from the face than the radius of the ball. This
is the same as a large gear driving a smaller one. The interactions between the ball
and club are labeled in Figure 4L. The shear force from the angled face gives a direct
force (1) to produce wedge-effect spin (2) on the ball. The reaction shear force
from the ball, given by dashed arrow (1), produces backspin of the head (2), and
through the gear effect a forward spin component back to the ball (3). In this case,
for a typical driver, the ball will rotate forward with an additional speed of 176 rpm.
Thus, without this gear effect, the average backspin for the LPGA drive would be
2,804 rpm rather than 2,628 rpm.
The driver head, in Figure 4L, is shown with a flat face from sole to crown. This is
deliberate because we have not yet justified why it should be otherwise. For that we
need to explore the gear effect further.
The gear effect becomes much more significant for off-center hits. Figure 4M
shows the ball separating from the face following an off-center impact, at a distance
offset upward from the center of mass of a vertically flat-face driver. The club head
is now forced to rotate backward by the very much larger normal force reaction
between the ball and the club face. The interactions between the contact forces and
the rotations are again shown in numbered sequences. Consider again the average
LPGA drive in Table 4B. For a high hit, 0.5 inches above the club center of mass,
and a typical 460cc driver head, the backward head rotation speed is the very much
larger value of 1,230 rpm, and the ball will rotate forward with added speed of
1,833 rpm. Thus, with a flat driver face, the average backspin for an LPGA 0.5 inch
high ball strike with a flat-face driver would be 2,628 – 1,833 = 795 rpm.
Moreover, because of the backward rotation of the head, the club face at the point
of impact would lose forward velocity during the impact by a further amount equal
to 3.6 mph. If the NCofR was unchanged, then the ball velocity will decrease by the
same 3.6 mph. However, it is more likely that the NCofR has decreased to around
0.8 at this position on the face. This will lead to a further fractional loss in forward
ball speed of 1.8/1.83 = 0.98, or an approximate further ball speed reduction in this
gear (3)
(1) launch angle

forces
(1)
wedge (2)
gear (2)
Figure 4M High ball strike on a flat-face driver.

gear (3)
increased launch
(1)
(1) extra loft
wedge (2)
gear (2)
Figure 4N High ball strike with a vertically curved face (roll).
case of 2.7 mph. All of these changes, particularly the very low backspin, are det-
rimental to driving distance. The values given here are calculated precisely using
3-dimensional modeling of the driver strike in Chapter 6.
Conversely, if the hit is low on the face, 0.5 inches below the center of mass,
then the club head will rotate in the opposite direction. In this case, the gear effect
will add further backspin to the ball; and for this example, the backspin would be
approximately 2,628 + 1,833 = 4,461 rpm. It will also produce a ball speed decrease
as for the high hit.
Both of these conditions would produce poor drives. The low face hit, with
excessive backspin, will rise too high and drop short. Conversely, the high face hit,
with anemic backspin rate, will give a low trajectory with short flight time and low
carry distance.
We can now see the task facing the club designer. To obtain less variation in
backspin for hits high or low on the face, we require more wedge effect (greater loft)
for high face hits and less wedge effect (smaller loft) for low face hits. This is the
reason for the vertical face curvature from sole to crown; referred to as “roll”; see
Figure 4N.
With face roll, two changes occur: for hits high on the face, the effective loft angle
increases and wedge effect increases; whereas for low hits, the opposite occurs. Thus
roll is used to decrease the variation in backspin from high to low hits on the club
face. Unfortunately it changes the launch angle between high and low face hits, so it
could be argued that low hits should create at least some more backspin, providing
more aerodynamic lift for the lower trajectory. This is not the usual design practice.
Typically, on the modern driver, the roll is a constant arc with radius in the range
12 to 13 inches. This gives a loft increase between 2.2 and 2.4 degrees for hits
0.5 inches high on the face, and the identical loft decrease for hits 0.5 inches low. We
discuss this further in Chapter 6.
The design of a driver face to minimize the negative effect of sideways off-center
hits follows the same rule of balancing the wedge and gear effects. Figure 4P shows
the instant of separation for a ball strike offset toward the toe for a flat face driver
in the left pane and one with curvature across the face, called “bulge,” in the right
pane. Without face curvature, only gear effect spin occurs, which would cause a suf-
ficient sidespin component to produce a severe hook, curving across the player to
the opposing side of the fairway.
wedge(2)
gear(3)
gear(3)
(1)
(1) (1)
gear(2) gear(2)
Figure 4P Toe hit: (left pane) severe hook from gear effect sidespin with sideways flat face; and
(right pane) sideways curved face (“bulge”) providing wedge effect, which gives sideways launch
away from target and reduces gear effect spin to produce gentle draw back to the target.
Bulge across the face of the club provides a sideways wedge effect for toe and
heel strikes, which produces smaller and opposite wedge effect spin to diminish the
gear effect and also launches the ball away from the player. The resulting sidespin
then causes the ball to curve back to the center of the fairway. A heel strike reverses
all the sidespin directions and produces the same effect, starting the ball sideways
toward the rear of the player and then curving it back to the center. These beautiful
mechanics of self-correction produced by face bulge are described in the text by
Cochran and Stobbs ([1968] 1999), and elegantly explained in the work of Penner
(2001a). In contrast, the need for face roll to control backspin variation, for hits high
or low on the face, does not seem to have been described in the open literature and
is widely misunderstood.
Figure 4Q shows the predicted trajectories for a center hit and heel and toe
strikes for the Callaway Great Big Bertha II (GBBII). The heel and toe strikes are
40
center
30
(yards)
heel & toe

20 with bulge
10
0
20
10 0.75 in. heel
(yards)
0
–10 0.75 in. toe
without face bulge
–20
0 50 100 (yards) 150 200 250
Figure 4Q Ball flight for 104 mph impact with the Great Big Bertha II; center, heel, and toe
strikes with a flat face or 11.8 in. bulge radius (gray lines for flat face).
0.75 inches from the face center along the face centerline. This is an interesting
case study because it is an almost perfectly designed club head and was the first
to combine 0.83 NCofR with a volume greater than 400 cc. I measured the GBBII
driver head and found it to have a bulge radius of 11.8 inches and a roll radius of
10.6 inches. These values were used in analytical modeling of offset impacts with
the GBBII, which is described fully in Chapter 6.
Figure 4Q shows the predicted trajectories from the modeling. The solid line
is for a center hit with an impact speed of 104 mph. This was the speed of robot
test results of the club described in Chapter 6. The 0.75 inches offset heel and toe
strikes were modeled with the measured bulge radius of 11.8 inches and also with
a flat face. The modeling predictions with the face bulge agree well with the robot
test. These are shown by the chain-dotted and dashed lines in the plan view of
Figure 4Q , both of which return to the centerline. The carry predictions of these
two strikes are predicted to be approximately 230 yards for both offset shots com-
pared to a center hit distance of 243 yards. All three shots are less than 1 yard to
the right side of the centerline. These are quite amazingly precise for trajectories
that start out 4 degrees to the left and right and would end up 16 yards to the left
and right without the corrective sidespin. With the flat face, the carry is reduced
to 218 yards, and the side movement is approximately 36 yards left and right, land-
ing way outside of the fairway. These are illustrated by the outer gray trajectories
in Figure 4Q.
It is worth mentioning that as drivers have become larger in volume, the MoI
about the vertical axis has increased significantly. The effect is a large reduction
in the gear effect sidespin for toe and heel strikes. This gives rise to a need for less
severe bulge across the face. It seems that the most common design for 460 cc driver
heads is with bulge and roll both having a 13-inch radius.

4.1 Wedge angle effects on spin rate and launch angle
It seems that the use of a tangential CofR was first introduced, and included in a gen-
eral theory of impact, by Horak (1937). Goldsmith (1960) refers to this work in his
monograph on impact and to the fact that the classical treatment of impact neglects
the effect of any tangential storage of energy during oblique impact, “even though
such storage always occurs and under certain circumstances may be released dur-
ing restitution to significantly increase both tangential velocities and spin rates.”
Goldsmith used the term Spin CofR, which seemed an appropriate name to use in
the earlier discussion that was focused on the main issue of spin rate. Lieberman
(1990a) was also focused on spin rates, and he referred to the “CofR of Spin” in his
investigation. In addition to his 2014 article discussed earlier, there is a large body of
research by Cross (1999, 2002a, 2002b, 2005) on the bounce of rubber balls, which
confirms the existence of substantial amounts of overspin when the ball “grips” the
surface during the impact. Therefore, we now proceed to analyze golf ball striking
in the certain knowledge that “overspin” will be involved.
Despite the use of the term SCofR in earlier studies, it is more appropriately
titled “tangential” CofR since it is based on the total tangential velocity of the ball
and not just the rotation component. Since e n was used to denote normal CofR in
Chapter 3, we use e t to denote the more general tangential CofR here.
Using the symbols in Figure 4.1, the tangential CofR can be written as
e t = (v 1t + rbω − V1t )/ V0 t . (4.1)

As discussed in simple terms earlier, the tangential CofR can take any value
between plus and minus one.
We can see that if e t is zero, then the numerator of Eq. 4.1 must be zero, which
gives v 1t + rb ω = V1t . This means that the points of contact on the two surfaces at
separation are travelling at the same speed; so with respect to the club face, the ball
is in the general condition of pure rolling.
If e t equals −1, then from Eq. 4.1 we have v 1t + rb ω = V1t − V0 t . This can only
result from the face of the club being completely frictionless, in which case the club
face simply slides across the ball without imparting any spin to it. It would still be
compressed in the impact of course, and would fly off the face at right angles. Since
ω
v1n
V0n
δ v1
V1n v1t δ–αvb
V0 αvb
V0t V1t
v1t + rbω
Figure 4.1 Velocity components before and after impact.
the ball is given no tangential velocity component, by conservation of momentum

the tangential velocity of the club must be unchanged. Thus, we have V1t − V0 t = 0 ,
v 1t = 0 , and ω = 0, which trivially satisfies Eq. 4.1 for e t = −1.
When e t = +1, all energy stored in components of strain tangential to the
impacting surface is recovered. Garwin (1969) established an elegant proof that if
both the NCofR and tangential CofR values are 1.0, then all of the kinetic energy
of any collision is completely recovered from the collision. More recently, Cross
(2014) has shown that provided the resultant normal force from the face passes
through the ball center of mass, the proportion of energy recovered in the tan-
gential direction is equal to e 2t , as for the case of energy recovery in the normal
direction. These results greatly authenticate the introduction of e t into the general
theory of impact.
Without exception, all of the analytical studies in the literature on golf ball
impact have assumed that the resultant normal force from the face, acting on the
ball, does pass through the ball center of mass. For now we also work with this
assumption. We show later that the very high tangential forces cause shear distor-
tion of the ball and deflect the center of mass in the tangential direction by distances
of the order of 1 mm from the resultant force, and that the errors in neglecting it
are manageably small. Offsets of this order of magnitude can also be determined
from experiments by Cross (2002b) on the low-speed bounce of plain rubber balls.
However, any changes in ball spin rate from the resultant normal force being off-
set from, and producing a moment about, the ball center of mass will be reflected
in the value of the tangential CofR. So for now, e t should be viewed as measur-
ing the consequence of both recovery of tangential strain energy and the applica-
tion of any braking torque to the ball from the non-uniform pressure distribution
between the ball and the impacted surface. Thus, SCofR values are being slightly
underestimated.
Rearranging Eq. 4.1, and including the expressions for V0 t and v 1t from
Figure 4.1, gives
rbω = e t V0 sin(δ) + V1t − v 1 sin(δ − α vb ). (4.2)

The relationship between angular velocity and the other launch parameters was
established in Chapter 3 as
ω = v 1 sin(δ − α vb )/(λrb ), (4.3)

where λmrb2 is the ball MoI.

Eliminating v 1 sin(δ − α vb ) and V1t between Eq. 3.8 in Chapter 3 and Eqs. 4.2
and 4.3 gives
rbω = V0 sin(δ)(1 + e t )/[1 + λ(1 + m / M)]. (4.4)

For the range of club head masses from 0.2 kg for the driver to 0.29 kg for
a wedge, the inverse of [1 + λ(1 + m / M)] with λ = (2 / 5) varies between 0.67
and 0.68. Thus, we can write the surprisingly simple expression for ball angular
velocity:
ω = (2 / 3)V0 sin(δ)(1 + e t )/ rb . (4.5)

Recall from Chapter 3 that for the range of values of (δ − α vb ) up to the high-
est lofted wedges, the small angle approximation sin(δ − α vb ) = (δ − α vb ) involves
errors less than 1 percent. Substituting for rb ω from Eq. 4.3 into 4.4, and substitut-
ing for v 1 from Eq. 3.5 in Chapter 3, gives (after some manipulation)
λ tan(δ) (1 + e t )
α vb = δ − . (4.6)
[λ + 1 /(1 + m / M)] (1 + e n )
For the range of club head masses, the expression λ /[λ + 1 /(1 + m / M)]
with λ = (2 / 5) varies between 0.33 and 0.32. Thus even more surprisingly we
can write
(1 + e t )
α vb = δ − (1 / 3)tan(δ) . (4.7)
(1 + e n )
Equations 3.5, 4.4, and 4.6 allow us to predict the ball speed, launch angle, and
spin rate if we know the mass of the club head, the mass properties of the golf ball,
the impact velocity, the dynamic loft, and the appropriate CofR values. Our next
step, therefore, is to explore typical values of tangential CofR for the range of clubs
from driver to wedge. Inverting Eq. 4.4 gives
e t = [rbω /(V0 sin(δ))][1 + λ(1 + m / M)] − 1. (4.8)

Equation 4.8 was used to calculate the tangential CofR for the LPGA and PGA
Tour players as presented earlier. From Eq. 4.5, tangential CofR can be given almost
precisely by e t = (3 / 2)[rb ω /( V0 sin(δ))] − 1.
4.2 Ball cannon testing

For ball cannon testing, as illustrated in Figure 4.2, Eqs. 4.4 and 4.8 remain valid,
provided of course the impact velocity V0 for the club head is changed to v 0 for the
ball. If the plate is sufficiently large so that m/M is negligibly small, we can write
rbω = v 0 sin(δ)(1 + e t )/(1 + λ), (4.9)

ω1 v1
δ θ
v0
Figure 4.2 Experimental setup for testing oblique impact of golf balls.

and conversely,
e t = rbω(1 + λ)/ v 0 sin(δ) − 1. (4.10)

Using λ = 2 / 5, the expression for the angular velocity becomes
rbω = (5 / 7)v 0 sin(δ)(1 + e t ). (4.9)ʹ

Equation 4.9 can readily be validated directly from Figure 4.2 by the same
assumption as for the golf ball strike, namely, that the tangential impulse is respon-
sible for both the change in tangential momentum m b ( v 0 sin(δ) − v 1 sin(θ))
and the ball angular momentum λm b rb2 ω. This gives the “direct relationship”
rb ω = ( v 0 sin(δ) − v 1 sin(θ))/ λ. Lieberman (1994) used this relationship to predict
the spin rates from the measured launch angles of his ball cannon tests. Unfortunately,
on this basis he concluded that tangential CofR played only a very minor role in the
spin rate generation. This only appears to be the case because the tangential CofR
has the opposite effect on the launch angle; that is, a positive SCofR decreases the
launch angle further from the vertical while increasing the spin rate, and vice versa
for a negative NCofR. In fact, if we eliminate ω between the tangential CofR defi-
nition, rb ω = v 1 sin(θ) + e t v 0 sin(δ), and Eq. 4.9′, we see that the second term in the
expression used by Lieberman is equal to v1 sin(θ) = v 0 sin(δ)(1 − λe t )/(1 + λ).
We can, in fact, calculate the tangential CofR using the “direct relationship”
used by Lieberman. For example, for the 35 degree impact of the Surlyn ball at
48.77 m/s (160 ft./s) with the grooved plate as shown in Figure 4G, Lieberman
provides the measured result ω = 1118.6rad/s (178 revolutions per second in the
paper). From the “direct relationship” this gives v1 sin(θ) = 18.42 m/s. The periph-
eral speed of the ball is rb ω = (0.2134)(1118.6) = 23.87 m/s. Finally the initial tan-
gential velocity is v 0 sin(δ) = 48.77 sin(35) = 27.97 m/s. By definition the tangential
CofR for this impact equals (23.87 − 18.42)/27.97 = 0.19. This is exactly the value
given in Figure 4G, where the tangential CofR values were calculated from the
approximate relationship of Eq. 4.9′.
From Eqs. 4.5 and 4.9′, the difference between the spin rate for a cannon test
against a heavy plate and that from a golf strike, with the same impact speed and
effective loft, is an increase by factor (15/14).
A more complete presentation, than given earlier of the 30 m/s ball-cannon tests
carried out by Cornish et al. (2007), is given in Figure 4.3. The amount of variation
in the Cornish et al. repeated tests for each particular ball was very large. In particu-
lar, three of the results fail to follow the smooth trends suggested by the entire set
of results. These have been shifted to follow the second order Bezier curves, which
I fitted to the data. The shifted values are shown as the solid black dots, and their
tangential CofR values are given in italics. These adjusted smooth trends were used
in the earlier discussion.
Lieberman (1990a) analyzed the changes that should be made in the impact
speed and the loft to ensure the same ball spin rate for both test configurations.
The results of his analytical equivalence over the range of club head lofts and cor-
responding weights for the 30 m/s cannon tests are shown in Figure 4.4. The circles
in the figure are the calculated values of equivalent club loft for the plate and loft
14000
High-Spin Ball
Polyurethane cover 0.45
12000
Distance Ball 0.43
Ionomer cover
10000 Adjusted to 0.36
0.12
Backspin, rpm
smooth trend 0.18
0.38 0.13
8000
–0.16
0.14
6000 0.33
0.25
0.08 –0.45
4000
–0.1
2000
10 20 30 40 50 60 70 80
Figure 4.3 Results of firing non-spinning golf balls at 30 m/s at different angles to a heavy
grooved plate; adapted from Cornish, Otto, and Strangwood (2007).
angles up to 60 degrees. The mass of the club head is taken for each case to follow
the standard trend of club lofts. The percentage difference in equivalent loft is larg-
est across the middle of the range, with a maximum difference just over 10 per-
cent. The equivalent speed of the club head to the 30 m/s ball cannon impact speed
decreases from 38 m/s to 32 m/s as loft and club head mass increase. None of these
differences would suggest a change in the basic nature of the relationship between
loft and tangential CofR.
60
Equivalent club loft, deg/Velocity, m/s
50 approx. 10%
40 Equivalent clu
b velocity
30
Ball cannon velocity
20
Equivalent club loft
10
0
0 10 20 30 40 50 60
Effective plate loft, degrees
Figure 4.4 Club loft and club velocity values for the same spin rate as 30 m/s oblique heavy plate
impacts; after Lieberman (1990a).
A discussed earlier, it is instructive to define a “slip/skid” parameter k s as the

ratio at separation of the ball surface speed to the club face tangential speed. From
previous calculations, this is easily written as
v 1t + rbω wedge v 1 sin(δ − α vb ) + rbω

ks = = . (4.11)
V1t V0 sin(δ) − (m / M)v 1 sin(δ − α vb )
Eliminating ω between Eqs. 4.4 and 4.11 gives
et
ks = 1+ ≈ 1 + et . (4.12)
1 − (m / M)(v 1 / V0 )sin(δ − α vb )/ sin(δ)
The approximation k s = 1 + e t is quite accurate since, over the range of shots
from driver to wedges, the denominator in Eq. 4.12 ranges between 0.95 and 0.96.
In Chapter 6, it is necessary to analyze ball striking in three dimensions. The use
of the slip/skid parameter simplifies this extension of the modeling.
4.3 Effect of shaft mass on ball spin and launch angle

Before moving on from the consideration of wedge-effect spin, we should consider
the effect of the shaft mass on the tangential CofR. This turns out to be even less sig-
nificant than for the NCofR due to the mass-MoI factor of [1 + (2 / 5)(1 + m / M)]
in the expression for e t , compared to [1+ m / M] in the expression for e n . For this
reason it was omitted from the earlier discussion.
The approach here is exactly as in Chapter 3 except the focus now is on the
momentum changes from the tangential impulse. This changes the moment
arm of the impulse acting about the center of mass of the club, as shown in
Figure 4.5.
Thus the only changes in the calculations, from Chapter 3, are simply to replace
cos(δ) for the normal impulse with sin(δ) for the tangential impulse. This gives the
expression for effective tangential impact mass as
M
M et = . (4.14)
1 + Mb2 sin 2 (δ)/ I cm

ls (δ) Ms
bsin
Ω
CoM M
Vcm
b v0 Mc
Vch
Ft
Figure 4.5 Effect of the tangential impact force on a freely supported club.

As before, M = M c + M s , where M c is the club head mass and M s is the mass of

the shaft.
For calculations of driver shots with δ < 15 degrees, we can write the approxi-
mate relationship
M et(driver ) = M c + M s . (4.15)

Equation 4.15 is easily perceived to be true. If the face is near vertical, then the
tangential impulse from the ball to the club face will cause a compressive impulse
wave to pass very quickly up the shaft slowing sections of it as it passes by.
Tables 4.1 and 4.2 show the effect of tangential effective impact mass on the
average LPGA and PGA ball striking data, respectively, published by Trackman
(Tuxen January 2010). For comparison, in columns 6 and 7, the tangential and
NCofR values are given based only on the club head mass values. It can be seen that
the use of the higher tangential effective impact mass has a much smaller effect on
the tangential CofR values than was the case for the normal CofR values, for the
reason discussed previously.
4.4 Gear effect on ball spin

The results produced by the following analyses were all presented earlier in the
chapter.
In Figure 4.6, the club head angular velocity is labeled Ω s to denote rotation
about the sideways axis through the club center of mass. The value of Ω s can be
determined by the condition that the moment of the tangential ball impulse about
the center of mass of the club head must equal its angular momentum at the end of
the impact; that is
I s Ωs = − mv 1 sin(δ − α vb )s t , (4.16)

where s t is the normal distance from the face to the center of mass and I s is the
sideways MoI component of the club head. Because of club head rotational speed,
the tangential velocity of the club face after impact, given by Eq. 3.8 in Chapter 3,
must decrease by amount s t Ωs . Thus the relationship between the ball and club face
tangential velocities at separation is given by
k s [V0 sin(δ) − (m / M)v 1 sin(δ − α vb ) − s t Ωs ] = v 1 sin(δ − α vb ) + rbω , (4.17)

where ω, the resulting backspin, is reduced from the value of ω wedge by amount
ω gear = −(k s s t / rb )Ωs . (4.18)

Note that the relationship in Eq. 4.18 only applies for a center hit, where both
the ball and club head rotations can reasonably be assumed to result solely from the
tangential impulse.
Consider the average LPGA driver strike in Table 4.1 where M = 0.200 kg,
δ s = 13.2 degrees, s t = 0.035 m, I s = 0.0003 kg-m 2 , v 1 = 62.14 m/s, α vb = 11 degrees,
Table 4.1 Trackman test results and effective-mass modeling for LPGA Tour players
Club Trackman data* Head mass model Effective-mass model

Head mass Club speed Ball speed Back-spin CofR Effective mass (kg) CofR
(kg) (mph) (mph) (rpm)
Tangential Normal Tangential Normal Tangential Normal
Driver 0.2 94.0 139 2628 −0.09 0.86 0.218 0.262 −0.10 0.83
3-wood 0.215 90.0 132 2705 −0.11 0.83 0.234 0.276 −0.12 0.81
5-wood 0.225 88.0 128 4501 0.22 0.84 0.245 0.284 0.21 0.81
7-wood 0.235 85.0 123 4693 0.19 0.84 0.267 0.324 0.17 0.80
4-iron 0.247 80.0 116 4801 0.25 0.83 0.286 0.353 0.23 0.79
5-iron 0.254 79.0 112 5081 0.27 0.80 0.294 0.358 0.25 0.76
6-iron 0.261 78.0 109 5943 0.29 0.82 0.304 0.360 0.28 0.78
7-iron 0.268 76.0 104 6699 0.35 0.82 0.315 0.363 0.33 0.78
8-iron 0.275 74.0 100 7494 0.38 0.86 0.326 0.364 0.36 0.82
9-iron 0.282 72.0 93 7589 0.30 0.86 0.339 0.365 0.29 0.82
PW 0.289 70.0 86 8403 0.37 0.84 0.350 0.367 0.36 0.80
* Data reproduced with permission from Trackman A/C, Vedbaek, Denmark
Table 4.2 Trackman test results and effective-mass modeling for PGA Tour players
Club Trackman data* Head mass model Effective-mass model

Head mass Club speed Ball speed Back-spin CofR Effective mass (kg) CofR
(kg) (mph) (mph) (rpm)
Tangential Normal Tangential Normal Tangential Normal
Driver 0.200 112 165 2,685 −0.28 0.86 0.219 0.261 −0.29 0.83
3-wood 0.215 107 158 3,655 0.00 0.85 0.234 0.276 −0.02 0.82
5-wood 0.225 103 152 4,350 0.14 0.84 0.244 0.286 0.12 0.82
Hybrid 0.235 100 146 4,437 0.12 0.82 0.265 0.327 0.11 0.79
3-iron 0.240 98 142 4,630 0.18 0.80 0.276 0.350 0.16 0.76
4-iron 0.247 96 137 4,836 0.17 0.78 0.284 0.355 0.15 0.74
5-iron 0.254 94 132 5,361 0.20 0.76 0.293 0.360 0.18 0.73
6-iron 0.261 92 127 6,231 0.23 0.77 0.303 0.363 0.22 0.73
7-iron 0.268 90 120 7,097 0.26 0.75 0.313 0.365 0.24 0.72
8-iron 0.275 87 115 7,998 0.32 0.79 0.324 0.367 0.31 0.75
9-iron 0.282 85 109 8,647 0.33 0.80 0.336 0.368 0.31 0.76
PW 0.289 83 102 9,304 0.28 0.85 0.350 0.367 0.27 0.80
ωgear
ωwedge
V1
rb δ−α vb
st αvb
V1
Ωs
Figure 4.6 Strike on the sweet spot of a flat-face driver with loft angle δ.
and k s = 0.91 . From Eq. 4.16, Ωs = 12.33 rad/s (84 rpm); and from (4.18),
ω gear = −18.4 rad/s (−176 rpm). Thus, if the club head did not rotate backward from
the tangential impulse, we would predict that the backspin rate would be 2,628 +
176 = 2,804 rpm.
Figure 4.7 shows the ball separating from the face following an off-center impact,
at a distance s n , from the center of mass of a vertically flat-face driver. The angular
momentum of the club head is now determined by the moments about the center of
mass of both the normal and tangential impulses; that is
I s Ωs = mv 1 (cos(δ − α vb )s n + sin(δ − α vb )s t ), (4.19)

where s n and s t are the moment arms for the normal and tangential contact forces,
respectively. It can readily be seen that this can produce much higher club head
rotational speeds because for driver loft values, cos(δ − α vb ) >> sin(δ − α vb ).
Consider again the average LPGA drive in Table 4.1, but with the face flat and
the strike 12.5 mm above the center of mass; s n = 12.5 mm. Substituting the same
values as for the center hit into Eq. 4.19 gives Ω gear = 129 rad/s; and from Eq. 4.18,
ω gear = −192 rad/s (−1,833 rpm).
With face roll, as shown in Figure 4.8, the normal distance, s n , from the club
center of mass to the normal through the impact point, decreases for a given face
ωgear
ωwedge
V1
sn
rb δ−α vb
st αvb
V1
Ωs
Figure 4.7 High ball strike on a flat-face driver.

ωgear
ωwedge
V1
δ−α vb
αvb
st
sn
V1
δ
Ωs
Figure 4.8 High ball strike with a vertically curved face (roll).
position because the normal is now an inclined radial line, and this decreases the
magnitude of the head rotation. This decrease is small but by no means negligible.
Equation 4.18 still applies with appropriate changes to both moment arms s n and
s t and to the loft, which now depends on the impact position. Analysis of this situa-
tion must wait until Chapter 6.
4.5 The full complexities of spin generation in oblique impact

For oblique impacts of golf balls, only five articles in the literature were found
to contain credible analyses or experimental results (Cornish et al. 2009; R&A/
USGA 2006, 2007; Arakawa et al. 2006; Tanaka et al. 2006; Monk et al. 2005;
Lieberman 1990a). None of these papers deals with the complex nature of the
deformation itself, although the work by Arakawa et al. (2006) shows some very
revealing high-speed photographs, as discussed at the beginning of the chapter.
Rodney Cross, of the University of Sydney, whose 2014 article was discussed
earlier, has published widely on ball sports, with particular emphasis on tennis and
baseball. A series of experimental and theoretical papers by him (2002a, 2002b,
2005, 2010) provide a very complete explanation of the bizarre bounce behavior
of elastic balls that undergo significant deformation during the impact. Some of his
focus has been on the so-called “superballs,” which can have normal CofR values as
high as 0.9 and tangential CofR values as high as 0.75. As mentioned earlier, super-
balls are manufactured from polybutadiene, a synthetic rubber, which now is used
for the cores of many of the premium high-spin golf balls.
Courtesy of Cross (2014), Figure 4.9 shows the time-interval average rotational
speeds of a hollow Spalding High-Bounce rubber ball striking obliquely onto a pol-
ished granite slab. The form of the spin buildup during the compression phase, and
reduction during the restitution phase, is seen to be identical to the Arakawa et al.
golf ball test results in Figure 4B, although the reduction in spin rate from the maxi-
mum value is less in the Cross experiment. This may be due to the higher MoI of
the hollow ball, when compared to a solid one. Moreover, for similar amounts of
deformation, the hollow ball will likely be subjected to lower rolling resistance than
a solid one.
400
300
∆θ/∆t (rad/s) 200
100
contact time
0
–5 0 5 10
Time (ms)
Figure 4.9 Experimental results from Cross (2014) of an oblique impact, showing the variation
of angular velocity during the 6 ms of contact time.
Before proceeding further, it may be useful first to define the relationship

between these surface bounce tests and golf ball striking. This is illustrated in
Figure 4.10.
In the left pane the surface represents a club face that is moving with velocity v 0
toward a stationary ball. The normal golf club face orientation has been changed to
the horizontal for easier comparison to the ball bounce off a stationary surface in the
right pane. In both panes, symbol δ represents the effective loft angle. For this case,
we assume that the striker on the left has the same large mass as the plate on the
right so that the deformation of the ball will be identical in both cases. The purpose
here is not to make direct comparisons with golf club strikes but to understand the
complexities of oblique impact more, and also to assess the likely errors in using the
calculated values of dynamic loft in our simplified model.
On the left of Figure 4.10, symbol α is the launch angle with respect to the club
path; but for simplified presentation the subscripts have been omitted. The chain
dotted arrows represent the trajectories that would occur following impact if the
surfaces were frictionless. In the left pane this is vertical since the ball is stationary
v1
y
ω1 ω0 ω1 v1
α δ θ
v0
v0
δ
Figure 4.10 Relationships between ball striking on the left and the bounce of a ball off a
stationary surface on the right.
50 50
δ = 69° normal δ = 45.5°
40 v = 2.7 m/s tangential 40 v = 2.7 m/s
30 30
Force, newtons
Force, newtons
20 20
10 10
0 0
–10 –10
–20 –20
–30 –30
–1 0 1 2 3 4 5 6 7 –1 0 1 2 3 4 5 6 7
Time (ms) Time (ms)
Figure 4.11 Measurements of normal and tangential contact forces during the bounce of a
hollow rubber ball; from Cross (2014).
prior to impact and no horizontal velocity component can be produced by the fric-
tionless surface. In the right pane the angle of the trajectory off a frictionless surface
would be inclined at an angle further from the vertical than the loft angle since the
vertical velocity component would decrease while the horizontal one would stay
the same. In all bounce studies, the reflected angle θ is smaller than δ because the
velocity loss horizontally through friction is typically greater than the vertical loss
through normal restitution. The situation on the right allows the incident ball to
be rotating, even though this of course cannot apply to the moving striker and sta-
tionary ball on the left. Some of the bounce experiments we evaluate have the ball
projected at the surface with spin.
Cross (2014) also measured the normal and frictional forces during the bounce
of the hollow rubber ball. To obtain these measurements, he bounced balls off a
plate, which had piezo-electric transducers attached onto the top and end surface.
The plate was mounted on rollers to ensure accurate measurements of the tangential
forces as well as the normal ones. Figure 4.11 shows the transducer outputs for two
bounces of the hollow rubber balls at different incident angles, landing on the upper
surface piezo plate and driving the roller-mounted carriage back and forth with the
friction reversals.
The force outputs, on the left, show a complete reversal of the friction force
approximately in the middle of the impact. This reversal point corresponds to the
point of maximum angular velocity and equivalently to the start of the spin decel-
eration. It can be seen that the friction coefficient is greater than unity for the
first 2 milliseconds of the impact. This may seem strange but does not violate any
principles of mechanics. The only limitation on the tangential stress itself is that
it cannot exceed the shear strength of the material surface. From examination of
the high-speed video pictures taken of the bounce, Cross (2014) noted that these
friction reversals coincide with tangential oscillations of the ball during impact.
Because of this oscillation mechanism more than one friction reversal may occur.
This is seen in the result on the right side of Figure 4.11. These tangential oscilla-
tions occur when the ball grips the surface, and so no doubt occur during a golf ball
ω1
ω0
Fy
δ θ v1
d
.
v0
r*
Fx
Figure 4.12 Realistic representation of the impact of a compliant ball.
strike with a grooved face. So to maximize spin rate, it may be necessary to move
the reversal point further toward the end of the contact time. However, moving it
too far may release too much of the tangential strain energy equally on the two sides
of the ball. So to increase spin, the ball design considerations must include varying
the ratio of the tangential to the normal natural frequency. The objective would be
to reverse the shear force at a point that allows the maximum possible difference
between strain energy released to increased spin and spin reduction through the
braking effect of the reversed shear force itself.
One other important aspect of the bounce of compliant elastic balls has been
demonstrated by Cross (2002b, 2014), namely, that the normal force resultant does
not pass through the center of mass of the ball. It may in fact lag the position of
the ball center of mass at an earlier phase of the impact, but it certainly leads on
average and produces some spin reduction prior to separation. This situation is
shown in Figure 4.12, which also includes some of the other complications facing
the analyst. The moment arm of the shear force with respect to the center of mass,
denoted r*, is affected by both the compression of the lower surface of the ball and
the resulting upward deflection of the center of mass. The latter effect will clearly
be smaller than the former one, so we know that on average through the impact,
r * will be less than the ball radius rb . In addition, the flattened lower surface will
result in a reduction of the ball MoI, so that we should use a value I *b < (2 / 5)mrb2
as the average value for the MoI during the impact. We should note that they have
an opposing effect; the smaller moment arm will decrease spin rate while the lower
MoI will increase it. We will return to this point later. These factors are unlikely to
form part of a viable predictive model. However, we can determine an approximate
value of offset d for any particular ball bounce.
Because the bounce data from Cross (2014) includes some spin of the ball
prior to impact, we need to include this in our analysis. Referring to the right
pane of Figure 4.10, we can write the normal and tangential impulses acting on
the ball as
p x = Fx t c = m(v 0 sin(δ) − v 1 sin(θ)), (4.20)

p y = Fy t c = m(v 0 cos(δ) + v 1 cos(θ)), (4.21)

Table 4.3 Oblique impact ball cannon test results
δ (deg) ν 0 (m/s) θ (deg) ν1 (m/s) ω1 rad/s d (mm)
Ball A 10 35 7.0 29.0 183 0.33

20 35 15.5 28.0 325 0.61
30 35 25.0 27.5 492 0.65
Ball B 10 35 8.0 27.5 189 0.22
20 35 17.0 26.5 325 0.53
30 35 24.5 28.0 492 0.64
From Tanaka et al. (2006).
where Fx and Fy are the average tangential and normal forces during the impact
(respectively), and t c is the contact time.
With reference to Figure 4.12, but neglecting the variations in moment arm and
MoI, we also have the relationship
p x rb − p y d = I b (ω1 − ω 0 ). (4.22)

Rearranging Eq. 4.22, substituting for p x and p y from Eqs. 4.20 and 4.21, and
using the value I b = (2 / 5)mrb2 for a uniform solid ball, gives
rb (v 0 sin(δ) − v 1 sin(θ)) − (2 / 5)rb2 (ω1 − ω 0 )

d= . (4.23)
v 0 cos(δ) + v 1 cos(θ)
We apply this formula first to the Tanaka et al. (2006) golf ball cannon tests
discussed earlier. These authors fired golf balls from an air gun, in a “strain free and
non-rotating” condition, at the end face of a steel bar of mass 20 kg. Tests were car-
ried out with balls incident at 10, 20, and 30 degrees to the bar axis. The inbound
velocity and outbound velocity and spin rate were obtained using a 20,000 frames
per second video camera. Tests were carried out with two different balls, identified
as Ball A and Ball B. These were two different brands of 3-piece balls, both compris-
ing an ionomer cover with an inner and outer core of polybutadiene.
The results for the different incident angles with the two ball brands are shown
in Table 4.3. The test results were obtained as carefully as possible from graphed
results in the Tanaka et al. article. Both inbound and outbound angles are with
respect to the axis of the bar. The “d” offset values given by Eq. 4.23 are shown in
the last column.
The Cross (2014) experimental results for rubber ball bounces, with effective
loft angles less than 70 degrees, are shown in Table 4.4, again with the calculated
d values in the last column. The relationship between effective loft and the “d” offset
for these two sets of bounce experiments are shown in Figure 4.13. For comparison
with the Tanaka et al. data, only the Cross (2014) values for effective loft less than
40 degrees are shown.
Table 4.4 Experimental results for superball bounces
δ (deg) ν 0 (m/s) ω 0 (rad/s) θ (deg) ν1 (m/s) ω1 (rad/s) d (mm)
12.2 4.40 5.6 6.8 3.86 40.7 0.42

18.6 4.07 7.9 11.4 3.54 62.1 0.32
21.1 4.67 10.2 13.6 3.95 72.1 0.51
27.2 5.05 12.5 16.2 4.23 109.2 0.65
30.4 5.19 17.0 19.9 4.20 121.2 0.66
30.9 4.80 11.4 19.1 3.91 110.5 0.82
31.3 5.03 14.5 19.2 4.07 122.7 0.78
31.5 4.92 14.4 19.7 4.04 121.6 0.64
32.7 5.55 15.5 19.8 4.40 139.6 0.96
33.2 5.20 15.4 20.1 4.12 132.9 0.98
37.2 5.13 15.2 22.7 4.02 145.3 1.04
37.3 5.69 6.2 21.2 4.31 166.9 1.11
40.1 4.75 20.3 25.1 3.80 139.8 1.14
40.8 5.32 15.5 25.3 4.09 166.9 1.00
42.8 5.95 7.0 27.0 4.39 180.7 1.27
45.8 5.80 8.9 28.8 4.15 202.8 1.13
47.5 5.73 25.4 32.0 4.17 201.1 1.22
50.8 6.03 14.9 34.8 4.17 223.4 1.18
51.8 6.50 7.2 36.8 4.61 209.7 1.45
53.3 6.04 13.7 37.6 4.10 218.0 1.54
61.2 5.85 −4.3 48.5 3.92 212.0 0.85
63.5 5.79 0.0 55.0 4.04 166.7 1.58
67.3 5.80 8.5 57.6 3.98 200.6 1.17
67.9 5.86 7.8 59.6 4.17 180.2 1.30
69.2 5.79 7.0 64.2 4.39 137.5 1.51
70.0 5.96 9.0 65.0 4.49 155.5 1.06
From Cross (2014).
The results of Figure 4.13 provide conclusive evidence that friction between the
impacting surface and a ball is not the sole determinant of the resulting ball spin.
More specifically, the results show that the product of the normal force and the off-
set between the normal force and the CofM, integrated over the contact time, pro-
duces a reduction of the spin rate. An unfortunate corollary is that the parameter
that we defined as “dynamic loft” earlier cannot be the actual dynamic loft. It is an
artificial value of loft that correctly relates ball spin and launch angle based on the
assumption that offset d = 0 and that the friction moment arm and ball MoI can be
assumed constant throughout the impact.
So, with respect to our simplified model based on d = 0, together with coef-
ficients e t and k s , we have arrived at one of our main interests in investigating
bounces off fixed surfaces. The question is, what predicted value of dynamic loft
will allow for a correct prediction of spin rate for these tests, and by how much will
1.2
Cross
1.0 Tanaka, Ball A
Tanaka, Ball B
Center of Mass offset, mm 0.8
0.6
0.4
0.2
0
0 10 20 30 40
Figure 4.13 Comparisons between normal force offset values for ball cannon oblique impacts
and superball low-speed ball bounces.
it differ from the actual loft values? The basis for assessing these values is shown
in Figure 4.14. As shown, the predicted dynamic loft value, for ball bounces, is
established by assuming that the surface onto which the ball is projected slopes
upward by some small amount ε. This will decrease the effective loft to δ − ε and
lower the launch angle to δ + ε.Thus, the tangential impulse will decrease dispro-
portionately to m( v o sin(δ − ε) − v 1 sin(θ + ε)), which will have a magnified effect
on spin rate for a small angle change ε .
Thus we can write
tangential impulse , p x = m( v o sin(δ − ε) − v 1 sin(θ + ε)) (4.24)

and
tangential impulse moment, p x rb = I b (ω1 − ω 0 ). (4.25)

(a) (b)
ω0 ω1 v1 ω0 ω1 v1
δ θ δ−ε θ+ε
v0 v0
actual loft spin loft
Figure 4.14 Relationship between actual loft and predicted dynamic loft.

70
Cross
60 Tanaka, BallA
Spin loft, degrees Tanaka, Ball B
50
–5.65%
40
30
20
10
10 20 30 40 50 60 70
Dynamic loft, degrees
Figure 4.15 Relationship between actual loft and spin-rate calculated loft for golf ball cannon
and superball bounce tests.
Eliminating p x between Eqs. 4.24 and 4.25 and substituting I b = (2 / 5)mrb2 for
a solid uniform sphere gives
(2 / 5)rb (ω1 − ω 0 ) = v 0 sin(δ − ε) − v 1 sin(θ + ε). (4.26)

If we assume ε to be small, then we can make the following simplifications:
sin(δ − ε) = sin(δ) − ε cos(δ) and sin(θ + ε) = sin(θ) + ε cos(θ).
Making these small angle substitutions into Eq. 4.26 and rearranging gives
( v 0 sin(δ) − v 1 sin(θ) − (2 / 5)rb (ω1 − ω 0 )

ε= . (4.27)
v 0 cos(δ) + v 1 cos(θ)
Equation 4.27 was applied to the data in Tables 4.3 and 4.4 to obtain estimates
of the dynamic loft, equal to δ − ε. The results are plotted in Figure 4.15. Quite sur-
prisingly, the results for the low-speed superball impacts and the high-speed golf
ball impacts are almost perfectly aligned. Clearly more golf ball cannon test data is
needed to validate this equivalence. However, it is clear that only small differences
between the actual loft and the predicted dynamic loft are needed to account for the
torsional braking effect in the latter part of the impact. From Figure 4.15, it can be
determined that the difference is an almost constant 5.65 percent. Part of the reason
for this relatively small error may be, as mentioned earlier, that we also neglect the
changes in moment arm and ball MoI, which have a cancelling effect on the errors in
assuming d = 0.
0.8
Tanaka, Ball A
0.6 Tanaka, Ball B
Center of Mass offset, mm
0.4 4
= 2.4
pe
slo
0.2
offset (mm) = 2.44 × impulse(N-s)
0
0 0.1 0.2 0.3
Tangential impulse, N-seconds
Figure 4.16 Relationship between tangential impulse and CofM offset for oblique impact
cannon testing.
It is reasonable to assume that the main influence, on the CofM offset, is the tan-
gential impulse, p x , defined by Eq. 4.20. Figure 4.16 illustrates the relationship for
the Tanaka et al. (2006) ball cannon tests. The line fitting closely to both sets has
slope 2.44. We will assume that this approximate relationship can be applied to golf
ball striking.
To test this assumption, Eq. 3.9 in Chapter 3, relating tangential impulse to tan-
gential momentum and spin rate, must now be expanded to include the effect of the
CofM offset. This gives
Ft t c = mv 1 sin(δ − α vb )
(4.28)
Ft t c rb − Fn t cd = λmrb2 ω ,

where Fn t c is the normal impulse that produces the change of momentum normal to
the face, and δ is the improved loft estimate. Using the small angle approximation
as before, this is
Fn t c = mv 1 cos(δ − α vb ) = mv 1 . (4.29)

Eliminating Ft t c and Fn t c between these equations, and using the small angle
approximation sin(δ − α vb ) = δ − α vb as before, gives
δ = α vb + (λωrb + v 1d / rb )/ v 1 . (4.30)

Spin-rate calculated loft, denoted δ s, satisfies
δ s = α vb + (λωrb )/ v 1 . (4.31)

0.9
Trackman PGA & LPGA driver and 3-wood, 2010
0.8 Trackman partial wedge shots, 2014
Cross, superball experiments, 2014
Tanaka et al., Balls A & B
0.7
Trackman PGA data, 2010
0.6 Trackman LPGA data, 2010
rbω/V0
0.5
0.4
?
0.3
0.2
0.1
0 15 25 35 45 55 65 75
Spin loft, degrees
Figure 4.17 Relationship between spin loft and normalized spin rate for golf PGA and LPGA
average shots, ball cannon tests, and superball bounce tests.
Subtracting Eq. 4.31 from Eq. 4.30 gives the elegant result
δ − δ s = d / rb . (4.32)

As an example, consider the case of the average PGA 5-iron shot in Table
4.2. Using the δ s value of 0.357 rad (20.5 degrees) established in Chapter 3,
together with the measured values α vb = 0.276 rad (15.8 degrees) and v1 = 59.0
m/s (132 mph), gives tangential impulse 0.216 N-s. Assuming the relation-
ship in Figure 4.16 gives CofM offset d = 0.000526 m. Equation 4.32 gives
δ − δ s = 0.526 / 21.34 = 0.0246 rad = 1.4 degrees. The improved loft estimate is
21.9 degrees. In this case, the error in using the approximation for dynamic loft is
6.4 percent, in reasonable agreement with the ball bounce predictions. The aver-
age percentage loft overestimate, for the entire PGA player shots in Table 4.2, is
6.2 percent, with a minimum of 4.7 and a maximum of 6.7.
Tanaka et al. (2006) gave no information about the golf balls being used in their
tests. Hopefully more exhaustive oblique impact testing will become available to
make a more complete assessment of the accuracy of loft approximations for dif-
ferent ball types. Spin rate estimated loft will continue to be used in the following
chapters, with the understanding that it is a close underestimate of the actual value,
and importantly it can be used in modeling without introducing an unknown “d”
offset.
Finally we can compare the effect of loft on spin rates, across the entire set of
ball-cannon and Trackman data, as shown in Figure 4.17.
It can be seen that the results fall on three separate curves. The upper curve is for
the polybutadiene “superballs,” which exhibit extremely high friction coefficients,
greater than unity in some cases as shown on the left pane of Figure 4.11. These
balls do not start to slide through the impact until the effective loft angle is greater
than 55 degrees, at which point the dimensionless spin rate starts to fall. The middle
curve passes through the PGA and LPGA data and the Tuxen (January 2014) opti-
mized wedge data.
Recall that the Tanaka et al. (2006) data experiments, through which the bot-
tom curve passes, were carried out with a polished steel surface. The lowered fric-
tion limits in this case are the reason for the lower normalized spin rates. There is
little doubt that if these tests had been carried out with a higher series of effective
loft angles, then the maximum would be reached at some loft angle considerably less
than the 47 degrees for the groove clubs. The question mark on Figure 4.17 indicates
this assumption.
5
A N A LY S I S A N D F O R M U L A T IO N O F R U L E S
F O R C U R V E D B A L L F L IG H T
In Chapter 3 we considered the straight golf shot, where the club path before impact
is aligned with the target, and the leading edge of the club face is normal to the club
path. In this case, for an iron strike, the ball will be launched at right angles to the
leading edge of the face, the ball will spin backward about a horizontal axis, and the
ball will fly over a straight line along the ground. For a driver strike under the same
conditions, the closing of the club face at impact will produce a small gear effect
sidespin component inducing a small slice or fade. This initial rotation effect will be
addressed in Chapter 6. In this chapter, we extend the results of Chapters 3 and 4
for the situation where a deliberate curved shot is to be made. The ability to do that
separates the elite players from the majority whose main concern is eliminating the
occasional badly curved shot. We see this all the time on the Tour where drives will
be “shaped” to avoid fairway bunkers, or approach shots will be executed to start
toward and curve away from hazards on one side or another of the green. The player
who can make such controlled shots to the left (draws) or to the right (fades) with
confidence, has a great advantage in navigating safely around a golf course. Also of
course, deliberate more highly curved shots can be used to steer the ball around
obstacles or to follow tight curves around difficult fairways.
Most of us know the general rules to execute such shots. Playing the normal
swing, but with the club face slightly closed or open at impact will produce a draw
or a fade respectively. Alternatively, the same result can be achieved by having the
club face square to the ball with respect to the player’s stance at impact, but having
the club path move slightly away from or toward the player through the impact zone.
This marks the extremes, and of course any combination of club path and club face
angle may be used.
Defining the precise conditions needed for any particular curved shot is the
subject of this chapter.
V ERTICA L A N D HOR IZONTA L LAU NCH A NGLES

In Figure 5A, the vertical approach angle and the horizontal approach angle define
the “club path” toward the ball. The effect of the vertical approach angle, or attack
angle, in lowering or raising the vertical launch of the ball was discussed in previous
chapters. In the case shown, we have a downward vertical approach angle and an
inside to outside horizontal approach angle.
147
back
spin
vertical approach angle
(attack angle)
(dynamic loft) vertical

launch angle
vertical face angle
side
spin
horizontal
approach angle
horizontal horizontal
face angle launch angle
Figure 5A Illustration of club direction before impact and ball direction after impact.
The horizontal face angle, identified in the lower pane, defines the amount by
which the face is closed or open with respect to the club path. In the case shown,
the face is closed. Note that in the upper pane of Figure 5A, loft is also defined
as the vertical face angle. This simply gives some consistency in presenting the
results. We can equally refer to the vertical and horizontal face angles as vertical
loft and horizontal loft respectively. Because we have been conditioned to refer
to the vertical face angle as “loft,” we will continue to do so where appropriate,
and in some cases use simply “face angle” where it clearly denotes horizontal
face angle.
Modeling this face strike simply means setting up the same relationships that
were used in Chapters 3 and 4 for the straight shot. In this case, however, the task
is performed twice: once for the horizontal plane and again for the vertical plane.
The two sets of relationships allow us to predict the horizontal and vertical launch
angles of the ball and the components of ball spin in the vertical and horizontal
planes. The latter results define the tilt of the spin axis. As we shall see, some simple
elegant relationships emerge from this 3-dimensional modeling.
The first items of importance are the vertical and horizontal launch angles. The
relationships of the horizontal and vertical launch angles to the horizontal and
vertical face angles are almost the same. These are expressed as percentage val-
ues in columns 4 and 5 of Table 5A. It can be seen that they only start to diverge
slightly beyond the loft of a 6-iron. The horizontal launch angle percentage values
are unchanged between 10 degrees open and 10 degrees closed, with only small
errors outside of this range. This covers the vast majority of intentionally shaped
curved shots.
A complication arises, in applying this data, from the fact that loft angle and hor-
izontal face angle are not independent. The face can only be closed or opened, with
respect to the club path, by twisting the head about the shaft axis, or equivalently
149 Rules for Curved Ball Flight
Table 5A Relationships between face angles and launch angles
Club Lie angle Dynamic Launch percentage Percent change

loft angle of face angle of loft due to
face angle
Vertical Horizontal
Driver 54.0 14.0 87 87 73

3-wood 56.0 17.0 82 82 67
3-iron 58.0 22.5 78 78 62
4-iron 59.0 25.5 78 78 60
5-iron 60.0 28.5 77 77 58
6-iron 61.0 32.5 76 76 55
7-iron 62.0 36.5 75 74 53
8-iron 63.0 40.5 74 73 51
9-iron 64.0 44.5 74 72 49
PW 65.0 49.0 74 71 47
turning the stance while keeping the club-face orientation unchanged. In either
case, the relationship between the club loft and the face angle is established through
the shaft. It is easier to think in terms of the shaft being twisted. If, for example, the
shaft is twisted clockwise from above to open the face, it is clear that the top edge
of the face will move backward and the loft will increase. The amount of twisting,
about the vertical and horizontal, are in the ratio of the sine to the cosine of the
“lie” angle, which is the term used for the angle of the shaft to the horizontal at the
address position. The effect of closing or opening the face on the loft angle is given
in the last column in Table 5A. For example, from row 1, if the face of the driver is
closed by 5 degrees, then the dynamic loft will decrease by 73 percent of 5 degrees or
3.65 degrees, to a value of 10.35 degrees. Opening the face by 5 degrees would
increase the dynamic loft by 3.65 degrees to 17.65 degrees. This of course explains
why closing the face, to produce a hook, gives a much lower trajectory than does
opening the face to produce a slice.
The percentage values, in columns 4 and 5 of Table 5A, for estimating launch
angles, all indicate that the direction of the face (more correctly the direction of
the line normal to the face) is the biggest determinant of the initial direction of the
ball. It is most important to recognize this fact for horizontal launch angles. If a ball
is launched in an undesired direction, it is only because the club face was pointing
in that general direction at impact. The dominance of the face direction is particu-
larly obvious to a player when making high chip shots around the green with an
open-faced wedge for increased loft, as shown in Figure 5B. In this case, the wedge is
open by 30 degrees. From column 4 in Table 5A the predicted ball direction will be
approximately 0.71(30) = 21 degrees to the right of the club path, or only 9 degrees
to the face normal as shown. To land at the desired target, the club head should be
directed 21 degrees to the left. Note that with reduced friction in the rough, the ball
would be projected even closer than 9 degrees to the club face normal.
9°
21°
30°
Figure 5B Illustration of dominance of face direction on ball direction.
BA LL SPIN A X IS TILT
Vertical and horizontal face angles produce horizontal and vertical components of
ball spin, respectively. In combination, this results in a tilt of the spin axis, as shown
in Figure 5C. Here the ball is viewed from behind, travelling away from the club
face, and the axis is tilted by amount labeled “tilt.” As discussed in Chapter 2, the
aerodynamic lifting force on the ball is inclined at the same tilt angle to the vertical.
The horizontal component of the force pulls the ball sideways, producing the curved
flight.
The “physics modeling procedures” section later in the chapter shows that, pro-
vided the horizontal face angle is less than 40 percent of the loft angle, the spin tilt
angle can be given by the ratio of face angle divided by the loft angle. The ratio must
be multiplied by 180 / pi, approximately equal to 57, to convert from scientific units
of radians to degrees. So the approximate rule is
tilt angle = 57 × ( face angle / loft angle ).
horizontal
component
lift vertical
component
back spin
tilt
Figure 5C Angled spin axis produces both vertical and horizontal aerodynamic lift.
face square to centerline

10° spin loft spin axis tilt = –5.7°
1 degree
8 yards
center line
club path
200 yards
Figure 5D Effect of 1.0 degree closed face on spin axis tilt and ball flight; Tuxen (October 2010).
The loft angle used in this calculation must be the actual value, changed by the
open or closed face angle.
Yet again we can find examples from the extensive experimental investigations
by Tuxen (October 2010) to assess the validity of our results. For a driver strike with
10 degrees of “spin loft,” Tuxen provides the results shown in Figure 5D. Applying
our relationship to this drive gives
axis tilt = 57 × (1/10) = 5.7 degrees
exactly in agreement with the published result. Note that in this case, the “spin”
or dynamic loft is given for the shot and so needs no correction. We can perhaps
conjecture from this example that the Trackman Company discovered these same
relationships. Nevertheless, we can consider the result to provide validation since
Trackman has all the experimental capabilities needed to test their proposed rules.
LATER A L DEV I ATION OF HOOKS A N D SLICES

The preceding simple formula explains the main reason why it is so much easier
to hit accurate shots with a high-lofted club compared to a driver. With the face
unintentionally open by 5 degrees at impact, the vertical loft of the driver would
increase by 0.73(5) = 3.7 degrees, while the vertical loft of the 8-iron would increase
by 0.51(5) = 2.6 degrees; see the last column of Table 5A. The dynamic lofts of the
driver and 8-iron in Table 5A would thus increase to 17.7 and 43.1 degrees, respec-
tively. The tilt of the spin axis would be 57 × (5/17.7) = 16.1 degrees for the driver.
For the 8-iron, it would be 57 × (5/43.1) = 6.6 degrees. The driver shot in this case
would slice 11 yards to the right for every 100 yards, while the 8-iron shot would go
6 yards to the right for 100 yards. For the average player, the drive would be about
200 yards and 22 yards right, while the 8-iron shot would be around 130 yards and
8 yards right. Of course, in addition to this face angle comparison, we should not
neglect the greater difficulty of controlling the face angle at impact with the longer
driver shaft. These lateral deviation estimates were obtained with an approximate
rule developed by Tuxen (October 2010), which predicts
“A ball will move sideways by 3.5 yards for every 5 degrees of spin axis tilt, for every
100 yards of carry.”
+1 degree 30° spin loft

club path spin axis tilt = –1.8° 3 yards
face angle zero

150 yards
Figure 5E Effect of 1.0 degree closed face on spin axis tilt and ball flight; Tuxen (October 2010).
For the drive example in Figure 5D, this gives a predicted side movement of
3.5 × (5.7/5) × (200/100) = 8 yards as shown.
We will see that this rule agrees with predictions of the ball flight model for
driving the ball. As must be expected, it underestimates the sideways movement
for soaring slices with relatively long flight times, or overestimates it for low severe
hooks with short flight times. It also underestimates the side movement for fair-
way woods and irons, which produce higher relative backspin rates than with the
modern driver. It will be shown that in these cases, the sideways movement per
100 yards, per 5 degrees of axis tilt, is more appropriately 4.5 yards.
A second example in Tuxen (October 2010) is given for a 30 degree loft iron with
again a 1 degree close face.
The estimated spin axis tilt in this case is
axis tilt = 57 × (1 / 30) = 1.9 degrees ,
in almost precise agreement with the Trackman Company value given in the fig-
ure. The normalized side movement in this case is 3 × (5/1.0) × (100/150) = 5.3
yards per 100 yards per 5 degrees of axis tilt. Increasing the loft by approximately
0.56 degrees for the 1 degree open face makes little difference in this case.
In Chapter 3, we discussed the value to the average amateur player of using a
much higher-lofted driver. The claim was made that not only would this produce
a better ball trajectory and carry distance, but that it would also reduce slices sig-
nificantly. The example used was an increase of the dynamic driver loft from 14.7 to
21 degrees. Assume that, as previously, the face is 5 degrees open, giving an
additional loft of 3.7 degrees. The spin axis tilts for these two cases would be
57(5 /18.4) = 15.5 degrees for the 14.7 degree loft and 57(5 / 24.7) = 11.5 for the
21 degree loft, with a corresponding reduction of 25 percent in the slice. This
reduction of slice, with increased loft, applies of course to all clubs. This is why the
decreasing of lofts by the manufacturers, in an unavoidable race with others claim-
ing increased speed, has been to the detriment of the average player. Wishon (2008)
quotes the “24/38 rule, which basically says that the average golfer—male or
female—cannot even come close to hitting an iron that has less than 24 degrees of
loft or is more than 38 inches in length.” He continues by noting that a few decades
ago “the 24/38 line fell to the far side of a 3-iron but has now moved into the ter-
ritory of the 5-iron—making the 3-, and 4-iron un-hittable for most players.” As
mentioned earlier, this has been mitigated by the introduction of the hybrid clubs.
40
(yards) 30
20
10
0
0 50 100 150 200 250
(yards)
20
10
(yards)
0
initial ball trajectory
–10
–20
0 50 100 150 200 250
(yards)
Horizontal launch = 0.26° , Spin axis tilt = –8.9°, Side carry = 14.2 (total 15.4)
Figure 5F Predicted trajectory for average PGA drive with face closed 2.0 degrees in error.
To consider the effect of the Tuxen rule, we will start with the average PGA Tour
player drive with 14.4 degrees of dynamic loft (Tuxen, January 2010). For a 2-degree
closed face, with the average driver lie angle of 54 degrees, Table 5A gives a loft
decrease of 1.45 degrees, that is, from 14.4 to 12.95 degrees. The effect of a 2-degree
closed face with this loft is an 8.9 degree tilt to the spin axis. Also because of the
decreased loft, according to the cosine of the loft angle (as discussed in Chapter 3),
the ball speed will increase from 165 mph to 166.0 mph. Associated with these
changes, from Chapter 4 results, would be a decrease in backspin to 2,435 rpm. The
resulting modeled trajectory is shown in Figure 5F.
The side movement past the centerline in Figure 5F is given by the model to be
14.2 yards. However, the horizontal club path was taken to be aligned 2 degrees
inside the target line with the face square to the target line. From row 1, column 5
of Table 5A, this gives a horizontal launch angle with respect to the club path of
0.87(2) = 1.74 degrees, or 0.26 degrees right of the target line. If the ball travelled
straight, it would land 1.2 yards right of the target. The lateral movement of the
ball along its trajectory is thus 14.2 + 1.2 = 15.4 yards. This corresponds to (15.4) ×
(100/261) × (5/8.9) = 3.3 yards per 100 yards per 5 degrees of axis tilt, in very good
agreement with Tuxen’s approximate rule.
In contrast, for a 2-degree open face, the dynamic loft will increase to
15.9 degrees, the ball speed will decrease to 164 mph, and the backspin will increase
to 2,968 rpm. The trajectory in this case is shown in Figure 5G. The side movement
from the target line is predicted by the model to be 14.2 yards. The ball launch is
now 0.26 degrees left of the target line, so the total amount of side movement is
14.2 + 1.45 = 15.65 yards. However, the spin axis tilt, for the 2-degree open face with
the 15.9 degree loft, is 7.3 degrees. Thus side movement corresponds to (15.65) ×
(100/272) × (5/7.3) = 3.9 yards per 100 yards per 5 degrees of axis tilt. The reasons
for this increase in the normalized side movement are the larger spin rate and pro-
portionately longer flight time. The total flight time in this case is predicted to be
40
30
(yards)
20
10
0
0 50 100 150 200 250
(yards)
20
10
(yards)
0
–10
–20
0 50 100 150 200 250
(yards)
Horizontal launch = –0.26°, Spin axis tilt = 7.3°, Side carry = 14.2 (total 15.4)
Figure 5G Predicted trajectory for average PGA drive with face opened 2.0 degrees in error; by
coincidence the side carry is the same as for the draw.
6.8 seconds or 2.5 seconds per 100 yards. For the draw, the flight time was
6.0 seconds or 2.3 seconds per 100 yards. With increasingly open face angles, the
normalized side movement continues to increase with the flight time. With the face
8 degrees open, the trajectory rises to a height of 50 yards, and the ball falls short for
a carry distance of only 230 yards. The flight time is 8 seconds, and the normalized
side movement has increased to 4.5 yards per 5 degrees per 100 yards.
A comparison of the main parameters of these drives with the straight shot is
given in Table 5B. Table 5B also includes an approximate estimate of the bounce
and roll distances using the very approximate rule formulated in Chapter 4, namely,
bounce and roll in yards is approximately 84 minus (17/15) times the descent angle
from the horizontal.
With this rather crude approximation of roll distances, we can predict the total
drive distances for the hook, straight shot, and slice to be 310, 314, and 313 yards,
respectively. Because of the change to pure rolling conditions in the bounce, the
ball will, in the absence of any fairway side slope, continue along the final horizontal
trajectory. Under these conditions, the total side displacement of the slice and hook
shot are estimated to be 18.5 and 16.8 yards, respectively.
In this section, we have simply added some magnitudes to what is clearly true
from watching the most athletically talented players in the game; namely, driving
long distances and landing in the fairway is extremely difficult. The latest PGA sta-
tistics show that of the top 200 players in driving accuracy, the average percent of
fairway misses is 40 percent; the best is 27 percent, and the worst is 53 percent. Since
the fairway widths of the landing areas of tournament courses are typically around
30 yards, our results indicate that the ball must be struck with the face within the
tiny interval of plus or minus 2 degrees from square to achieve a fairway landing.
And we know that this must be achieved when swinging a driver head through more
than a full circular arc, at the highest controlled speed possible, and rotating the
club face from 90 degrees open in the last few hundredths of a second.
Table 5B Average PGA Tour drive with +/− 2.0 face angle for draw and fade with club head speed = 112 mph, attack
angle = −1.3 deg, club path =0
Loft Ball speed Normal Spin rate Max. height Landing Flight distance Bounce and Flight time
(mph) launch angle (rpm) (yards) angle (deg) roll (yards) (sec)
Vert Horiz Length Side
(deg)
(yards) (yards)
Vert Horiz
14.4 0 165.0 11.2 0 2,685 31.0 34.0 269 0.0 44.7 6.49
12.9 −2 166.0 10.0 +0.26 2,435 26.3 30.9 261 −14.2 48.8 6.04
15.9 +2 163.9 12.4 −0.26 2,968 35.2 36.6 272 +14.2 41.2 6.81
Vert = vertical; Horiz = horizontal.
40
30
(yards)
20
10
0
0 50 100 150 200 250
(yards)
20
club path
10
(yards)
0
–10
–20
0 50 100 150 200 250
(yards)
Horizontal launch = –3.3°, Spin axis tilt = 7.3°, Side carry = 0.96 (total 15.4)
Figure 5H Predicted trajectory for average PGA drive with face opened 2.0 degrees to produce a
fade back to the center
Of course, if a player can consistently strike the ball with the face 2 degrees open,
then the controlled fade takes on a whole different appearance. This is shown in
Figure 5H, where the ball is started left 3 degrees, probably toward the outside of the
bunkers on the left side of the green.
With this result, we can proceed with some confidence to apply an axis tilt to
the full range of average PGA player shots. The reason for doing this is to formulate
rules for curved shot making, which is the topic of the next section. In all cases,
the launch parameters of the straight shot are used, and a 5-degree spin axis tilt is
applied in the flight model. The results, shown in Table 5C, can be considered to
approximately represent either a slice or a hook. Note from the previous example
that this likely gives a small underestimate of the side movement for a fade and a
small overestimate for a draw.
The normalized side movement for the drive in column 9 of Table 5C is
3.8 yards. The average of the normalized side movements, for a 3-wood through
pitching wedge, is 4.6 yards. However, with a reasonable expectation of the preci-
sion in these results, and allowing for the probably extensive Trackman testing, we
might offer the modified rule as follows:
(a) When using a driver, a ball will move sideways by approximately 3.5 yards per
100 yards for every 5 degrees of spin axis tilt.
(b) When using other clubs, a ball will move sideways by approximately 4.5 yards per
100 yards for every 5 degrees of spin axis tilt.
Table 5C includes the carry distances, maximum trajectory heights, and flight
times for the different shots. The flight times are obtained from the modeling. For
the carry distances and maximum heights, both the Trackman measured values
and the values predicted by the model are shown. To the nearest yard, the latter are
identical with or without the 5 degree spin axis tilt. This is to be expected since the
Table 5C Flight parameters for average PGA Tour shot with 5-degree spin axis tilt
Loft (deg) Carry (yards) Side (yards) Maximum height (yards) Side/100 Flight time
(sec)
Vert Horiz Trackman Model Trackman Model
Driver 14.4 1.25 269 269 10.2 31 31 3.8 6.5

3-wood 14.8 1.28 243 248 11.0 30 29 4.5 6.8
5-wood 16.0 1.38 230 234 11.1 31 30 4.8 6.8
Hybrid 17.0 1.47 225 224 10.5 29 30 4.7 6.6
3-iron 17.2 1.48 212 215 10.0 27 29 4.7 6.5
4-iron 18.4 1.58 203 205 9.4 28 28 4.6 6.4
5-iron 20.5 1.76 194 193 9.0 31 29 4.6 6.3
6-iron 23.8 2.02 183 181 9.0 30 31 4.9 6.4
7-iron 27.4 2.31 172 170 8.2 32 31 4.8 6.3
8-iron 30.6 2.55 160 161 7.5 31 31 4.7 6.2
9-iron 34.2 2.82 148 151 6.7 30 31 4.5 6.1
PW 39.7 3.20 136 138 5.8 29 32 4.3 6.0
Vert = vertical, Horiz = horizontal.
s
yard
200
25 ya
3.95°
rds
7.1°
5°
Figure 5I Desired trajectory of a slice to a predefined target.
spin axis tilt was assumed to have no effect on the initial ball speed, launch angle, or
spin rate; that is, it is an approximate representation of either a hook or a slice. The
good agreement between the modeling and the full range of trajectories provides
some confidence in the side movement predictions. The step up in flight time from
the driver to the fairway woods is a little problematical but may be due to the switch
from the Smits and Smith lateral and drag coefficients for the driver to the modified
Bearman and Harvey coefficients for the other clubs. The reason for this switch was
discussed in Chapter 2. It seems that neither applies so well to the fairway woods.
The larger discrepancy between the actual and modeled carry distances for these
two shots reinforces this point.
CU RV ED SHOTS TO DEFINED TA RGETS

We apply these relationships, of face angles to spin axis tilt to sideways displace-
ments, to model the execution of desired trajectories.
For example, assume it is desired to make the shot illustrated in Figure 5I. The
small circle represents the intended target, but we assume it is necessary to launch
the ball toward a point offset 25 yards from the target at a distance of 200 yards,
that is, 7 degrees to the left. For this shot 12.5 yards of side movement for each
100 yards is required. Assume for this shot an LPGA player decides to use a 3-wood
with 17 degrees of dynamic loft. Using the modified Tuxen rule of 4.5 yards per
100, per 5 degrees of spin axis tilt, she would need to impose a spin-axis tilt angle
on the ball of (12.5/4.5) × 5 = 13.9 degrees. When the face is opened by angle A, the
loft will increase by (0.67)A as given in the last column of Table 5A. Thus the shot
requires the ratio 57A/(loft) = 57A/(17 + 0.67A) = 13.9, which is satisfied with face
angle A = 5.0 degrees. So we would expect this shot to be executed with the 3-wood
face opened by 5 degrees and so increasing the loft to 20.4 degrees. Note finally that
with the face open 5 degrees with respect to the club path, the difference between
the club path and the ball launch angle will be 0.78 × 5, or approximately 4 degrees;
see Table 5A, row 3, column 5. Thus, for a ball to be launched toward offset distance
25 yards, at 200 yards, we need the club path a further 4 degrees left, or 11 degrees
left of the target to an offset distance of approximately 40 yards at 200 yards. If
we got all this right, then the ball should be launched 7 degrees left as shown in
Figure 5I.
A simulation of this shot, using equations developed in Chapters 3 and 4
to predict the launch conditions of ball speed and ball spin rate, is shown in
Figure 5J. In particular, the club head speed to achieve this result is 90 mph. We
3-Wood, Loft = 20.4°, Ball speed = 129.7 mph, Flight time = 6.75 sec
40
30
(yards)
20
10
0
0 50 100 150 200
Launch angle = 16.3°, Backspin = 4,116 rpm, Height = 35.5 yds, Carry = 200 yds
30
20 club path ectory
initial traj
(yards)
10
0
–10
0 50 100 150 200
Horizontal launch = –7.1° , Spin axis tilt = 13.9° , Side carry = 24 yds
Figure 5J Predicted ball flight for the example in Figure 5I.
can see that the predicted horizontal launch, spin axis tilt, and trajectory back to
the target are almost exactly as planned, the ball being predicted to land 1 yard
short of the target line.
It should be noted that the application of the curved shot rule depends on appro-
priate club selection, with attention to the change in loft produced by face opening or
closure. If this shot were attempted with a hook instead of a slice, then the loft would
be reduced to 14.5 degrees, resulting in a low trajectory with a flight time of only
5.6 seconds. In this case, the side movement turns out to be 20 yards, leaving the
carry 5 yards right of the target line. These issues are discussed in the next example.
To summarize these steps
(a) We first select the club to reach the target, making possible adjustments in selection
for the likely amount of loft change.
(b) We next determine the value of the spin axis tilt that will cause the desired amount of
hook or slice to the target.
(c) Next we estimate the amount by which the face needs to be opened or closed.
(d) Finally, we determine the required club path that will start the ball with the desired
horizontal launch.
Of course these estimations could not possibly be carried out on the course!
However, it certainly could be possible to make some alignment mats for use in
practicing such shots on the driving range.
FU RTHER EX A MPLES
Assume the average PGA Tour player wishes to perform a full shot of 150 yards
carry distance with a sideways movement from 20 yards left of the target. We use
40
30
(yards)
20
10
0 20 40 60 80 100 120 140 160

(yards)
Launch angle = 20.4, Ball speed = 109, Backspin = 8647, Height = 30.4, Carry = 150
20
ath
10 club p
(yards)
–10
–20
0 20 40 60 80 100 120 140 160
(yards)
Horizontal launch angle = –7.6, Spin axis tilt = 15, Side carry = 17
Figure 5K Predicted parameters for a 150 yard fade with an 8-iron.
the data directly from Table 3.3 in Chapter 3. We can see there that the normal club
selection for a straight shot of 150 yards would be a 9-iron. Since this is a controlled
slice, the open face will increase the loft and shorten the carry distance. A sensi-
ble choice in this case would be to step up to an 8-iron with normal dynamic loft
30.6 degrees.
The reader may wish to step through the approximate calculations, as in the last
example, to validate the following predicted conditions for this shot:
Face angle = +9 degrees; dynamic loft = 35.2 degrees; club path = 14 degrees left.
The loft is approximately that for the straight 9-iron shot, so a full shot can be
taken to give the 9-iron launch conditions but with a spin-axis tilt, which can be
shown to be 15.2 degrees. The launch angle in this case is 20.4 degrees and the
backspin rate equals 8,647 rpm. The modeled trajectory for this shot is shown in
Figure 5K.
If these calculations are repeated for a controlled hook from the opposite side of
the target (now stepping down to a pitching wedge to allow for the loft decrease), the
following results are obtained:
Face angle = 7.6 degrees; dynamic loft = 35.2 degrees; club path = 14 degrees right
The launch conditions are again approximately that of the straight 9-iron shot.
The modeled trajectory is given in Figure 5L.
40
30
(yards)
20
10
0 20 40 60 80 100 120 140 160

(yards)
40
30
(yards)
20
10
0 20 40 60 80 100 120 140 160

Horizontal launch angle = 7.6, Spin axis tilt = –15, Side carry = 3 (total 17)
Figure 5L Predicted parameters for a 150 yard draw with a pitching wedge.
The perfect symmetry of these two results was completely unexpected when the
problem was first formulated. It raises the possibility that the skilled player could
have a go-to substantial draw and fade for every club. In each case, the shot would
give a known carry distance, which would be approximately the straight shot dis-
tance for the next club up or down, that is, one number lower for the fade and one
higher for the draw.
The full calculations for this example are given at the end of the chapter.

5.1 Analysis of the sideways components
of launch angle and spin rate
Figure 5.1 combines the velocity and geometrical configuration of the club head at
impact with the launch conditions of the ball after impact. In the case illustrated,
the club is being swung from inside to outside of the target line, as defined by β h in
the lower pane. The club head is also being swung downward at the ball, as defined
by negative attack angle β v in the upper pane. In this chapter, dynamic loft from
Chapters 3 and 4 are redesignated as δ v to indicate that it is the angle of the club face
to the club head velocity in the vertical plane normal to the face. The horizontal club
face angle δ h is defined as the angle between the club head velocity and the face nor-
mal in the horizontal plane. The illustration in the lower pane of Figure 5.1 shows a
closed face, for which δ h is taken to be negative. As discussed earlier, the two angles,
δ v and δ h , are not independent. The face can only be closed or opened, with respect
to the club path, by the connection with the shaft, which is angled to the horizontal
according to the club lie angle γ. Therefore, the amount of twisting about the vertical
and horizontal will be in the ratio sin(γ ):cos(γ ). Thus, for face opening by magnitude
δ h , the normal dynamic loft of the club, δ, will increase by amount δ h /tan(γ ); and
for a closed face, the loft will decrease by the same amount, that is
δ v = δ ± δ h/tan(γ ), (5.1)

where ± corresponds to an open or closed face, respectively.
ωv
V0
rb δ v −αvb v1
αvb
βv
δv
ωh
V0
rb δ h−α hb v1
αhb
βh
δh
Figure 5.1 Initial club head velocity, resulting ball velocity, and resulting spin rate components;
upper pane showing negative angle of attack β v and lower pane illustrating positive (inside to
out) approach angle β h with negative (closed) face angle δ h.
With reference to Figure 5.1, the initial velocity of the club head normal to the
face is now given by V0 cos(δ v )cos(δ h ); therefore, from Eq. 3.5, the ball velocity is
now given by
 1+ e 
v 1 = V0 cos(δ v )cos(δ h )  . (5.2)
1 + m / M 
Equations 3.7 through 3.10 in Chapter 3 still apply, but they now predict the spin
rate and launch angle components in the vertical direction. (Note that the spin rate
predicted by Eq. 3.10 should actually be considered to be horizontal since angular
velocity is defined as a vector by the direction of the axis of spin. For now the equa-
tions are easier to follow if we define all variables in the upper pane of Figure 5.1
with suffix v, and all those in the lower pane with suffix h.)
As in Chapters 3 and 4, we neglect any ball center of mass offset to relate the
horizontal and vertical loft and launch angles to the components of the spin rate.
Applying Eq. 3.10 in the horizontal and vertical directions gives
ω v = mv 1 sin(δ v − α vb )rb / I b (5.3)

and
ω h = mv 1 sin(δ h − α hb )rb / I b . (5.4)

Applying conservation of momentum in both the vertical and horizontal direc-
tions on the club face as in Eq. 3.7 in Chapter 3, and neglecting the term cos(δ h ) ≈1
in Eq. 5.2, gives
MV0 sin(δ v ) = MV1 vt + mv 1 sin(δ v − α vb ) (5.5)

and
MV0 sin(δ h ) = MV1ht + mv 1 sin(δ h − α hb ), (5.6)

where V1 vt and V1ht are the vertical and horizontal components of the tangential
velocity of the club face after impact.
As described in Chapter 4, we can express the relationship that the ball separates
from the club face in a condition of skidding or slipping by relating the tangential
components of velocity of the club face with the components of the ball peripheral
velocity through the slip/skid factor k s . This gives
k s V1 vt = v 1 cos(δ h − α hb )sin(δ v − α vb ) + rbω h = v 1 sin(δ v − α vb ) + rbω h (5.7)

and
k s V1ht = v 1 cos(δ v − α vb )sin(δ h − α hb ) + rbω h = v 1 sin(δ h − α hb ) + rbω h . (5.8)

Note that we are assuming that the same value of the slip/skid factor applies to
both the horizontal and vertical tangential directions on the face. There is no funda-
mental justification for this. However, it leads to some simple expressions that agree
well with experimental results.
Eliminating V1 vt and V1ht between Eqs. 5.5 and 5.7, and between Eqs. 5.6 and
5.8, gives the kinematic relationships for the sidespin components as
ω v = [k s V0 sin(δ v ) − (1 + k s m / M)v 1 sin(δ v − α vb )]/ rb (5.9)

and
ω h = [k s V0 sin(δ h ) − (1 + k s m / M)v 1 sin(δ h − α hb )]/ rb . (5.10)

Equating the right-hand sides of Eqs. 5.3 and 5.9, and Eqs. 5.4 and 5.10, writing
I b = λmrb2 , and rearranging, gives
k s V0  1 k m
sin(δ v − α vb ) = sin(δ v )/ 1 + + s  (5.11)
v1  λ M 
and
k s V0  1 k m
sin(δ h − α hb ) = sin(δ h )/ 1 + + s  . (5.12)
v1  λ M 
As before, for the small angles, sin(δ v − α vb ) ≈ δ v − α vb and

sin(δ h − α hb ) ≈ δ h − α hb with negligible error; therefore
k V  1 λm  
α vb = δ v −  s 0 sin(δ v )/  1 + + (5.13)
v
 1  λ M  

and
k V  1 λm  
α hb = δ h −  s 0 sin(δ h )/  1 + + . (5.14)
v
 1  λ M  

Dividing Eq. 5.11 by Eq. 5.12 gives an additional elegant relationship:
sin(δ h − α hb ) sin(δ v − α vb )
= . (5.15)
sin(δ h ) sin(δ v )
Results of experimental studies of launch angles were reported by Tuxen (October

2010). Figure 5.2 illustrates one of the reported cases where δ v = 20 o , δ h = 8o,
α vb = 15o, and α hb = 6 o . This gives sin(δ h − α hb ) = 0.251 and sin(δ v − α vb ) = 0.255,
sin(δ h ) sin(δ v )
in good agreement with Eq. 5.15.
–5 degrees 15 degrees
unch angle
attack angle normal la
20 degrees
spin loft
+6 degree
club path 6 degrees
normal launch angle
–8 degree
face angle
Figure 5.2 Corresponding vertical and horizontal launch angles; Tuxen (October 2010).
5.2 Estimating the tilt of the spin axis

Figure 5.3 shows the components of the spin in the horizontal and vertical planes.
These spin rate components are of course vector quantities (treated for simplicity as
scalars in this section). Therefore, with reference to Eqs. 5.3 and 5.4, the angle of the
spin axis can be represented by
ω   sin(δ h − α hb ) 
α spin = tan −1  h  = tan −1  ; (5.16)
ω
 v  sin(δ v − α vb ) 
FL
ωh
ωv αspin
Figure 5.3 Resultant spin rate ω with horizontal and vertical components ω h and ω v .
Resulting lift force F L acts normal to the spin axis.
Table 5.1 Average PGA Tour shots with face 3 degrees open: predicted vertical and
horizontal loft and launch angles, and spin axis tilt
Club Lie angle Loft Normal launch Ratio of Spin

angles launch/loft* axis
tilt
Vert Horiz Vert Horiz Vert Horiz
Driver 54.0 16.6 3.0 14.4 2.6 0.87 0.87 10.4

3-wood 55.0 16.9 3.0 13.9 2.5 0.82 0.82 10.2
5-wood 56.0 18.0 3.0 14.3 2.4 0.80 0.79 9.6
Hybrid 57.0 18.9 3.0 15.1 2.4 0.80 0.79 9.2
3-iron 58.0 19.1 3.0 14.9 2.3 0.78 0.78 9.1
4-iron 59.0 20.2 3.0 15.8 2.3 0.78 0.78 8.6
5-iron 60.0 22.2 3.0 17.2 2.3 0.77 0.77 7.9
6-iron 61.0 25.5 3.0 19.4 2.3 0.76 0.76 6.9
7-iron 62.0 29.0 3.0 21.9 2.2 0.75 0.74 6.2
8-iron 63.0 32.1 3.0 23.8 2.2 0.74 0.73 5.6
9-iron 64.0 35.7 3.0 26.2 2.2 0.74 0.72 5.1
PW 65.0 41.1 3.0 30.3 2.1 0.74 0.71 4.6
* Within +/− 0.01 for face angles in the range −10 to +10 degrees. Vert = vertical, Horiz = horizontal.
and from Eq. 5.15, this can be further simplified to the truly elegant result
 sin(δ h ) 
α spin = tan −1  . (5.17)
 sin(δ v ) 
Table 5.1 shows the results of applying Eqs. 5.13, 5.14, and 5.17 to the average
PGA Tour data in Table 4A of Chapter 4 and using the loft and slip/skid parameter
values presented there. For the values shown, the face is 3 degrees open. The launch
to loft ratios given in columns 7 and 8 vary only by ± 0.01 for face angles in the range
−10 to +10 degrees. Thus the ratio values are generally useful for modeling curved
ball flight.
5.3 Rules for curved ball flight to the target

In this section, we combine the analytical relationships between spin rate compo-
nents and face angles with the rule for side movement presented in the earlier dis-
cussions. This will establish the simple rules used earlier for executing curved ball
flight. The central question to be answered is what combination of club path β h ,
club selection, and face angle δ h will result in the ball landing at a predefined point.
If the club path is along the target line, straight toward the predefined landing
point, and the face angle is zero, then the problem is simply one of club selection and
swing speed. If, however, the initial ball flight must be angled away from the target
line, then a satisfactory result involves manipulation of all four variables. We can
define the problem as shown in Figure 5.4 where an initial ball trajectory, inclined
at angle α t to the desired target line, results in a curved flight to the desired landing
V0 βh
αt
δh
αhb
lc ls
Figure 5.4 Parameters of curved shot problem to a pre-defined target.
spot. To achieve the desired hook, the club face must be closed by amount δ h ; so
the club path must be at angle α t + α hb to the target line. The required side move-
ment is l s , and the nominal carry distance is equal to l c . The latter distance is used
in club selection and swing speed. The first step, in arriving at a satisfactory rule
 sin(δ h ) 
for this problem, is to investigate the relationship α spin = tan −1   between
 sin(δ v ) 
spin axis tilt and the combination of face and loft angle. From the form of this
relationship, it seems likely that the same ratio of face angle to loft angle for dif-
ferent clubs will produce very similar spin axis tilts. For example, if we have a face
angle of 5 degrees with a 30 degree loft iron, then the spin axis tilt is predicted to
 sin(5.0) 
be α spin = tan −1   = 9.9 degrees. For comparison, a 2.5 degree face angle
 sin(30.0) 
with a 15 degree driver will give α spin = 9.6 degrees. To eliminate the sin and arc-
tan functions for more transparent rule formulation, we can use the result by Lyons
(2011) that arctan can be approximated with excellent precision by
 1 
tan −1 ( x ) =   , (5.18)
 1 / x + cx
where c = 0.2813. We can also note that a / b is a very close approximation to

sin(a)/ sin(b) for a , b ≤ 45 degrees, which includes all of the clubs with which
“shaped” ball trajectories are made. Thus we could anticipate that if x in Eq. 5.15
represents (δ h /δ v ), then it might be possible to select a value of c close to 0.28 to
obtain a single approximate expression for α spin covering all loft angles. From trial
and error, c = 1 / 4 satisfies this requirement. Substituting (δ h /δ v ) for x, rearrang-
ing, and converting from radians to degrees, gives
 (180 / π) 
α spin =  . (5.19)
δ /
 v h δ + 0.25 δ / δ v 
h

A plot of Eq. 5.19, together with plots of Eq. 5.17 for different loft angle clubs, is
shown in Figure 5.5.
One further simplification step is possible. It can be seen from Figure 5.5 that
up to (δ h /δ v ) = 0.5 the plot is almost perfectly linear. The linear relationship in
Lyons approximation
40 αspin = (180/π)(δv/δh)
δv = 45°
Spin axis tilt, degrees 30 δv = 15°
20
10
0
0 0.2 0.4 0.6 0.8 1.0
Face angle/vertical loft angle
Figure 5.5 Plots of α spin against δ /δ v .

h
Figure 5.5 is simply Eq. 5.19 without the additional term 0.25δ h/ δ v in the denomi-
nator. The relationship can thus be approximated as
α spin = (180 / π)(δ h / δ v ) degrees for δ h /δ v ≤ 0.5, (5.20)

a range that applies to the majority of intentionally shaped shots played in golf.
This of course corresponds to the rule that, expressed in radians, has the agree-
ably simple form α spin = δ h /δ v . Note that δ v can take large values, so we could not
have inferred this directly from Eq. 5.17.
Tuxen (October 2010) provides one further specific example of a curved shot,
but with a 30-degree loft iron as shown in Figure 5.6. The lower pane of this figure
was discussed earlier. To achieve the spin axis tilt of −1.8 degrees, our analytical
prediction from Eq. 5.17 of the required face angle is
δ h = sin −1 (tan(1.8)sin(30)) = 0.90° ,
in excellent agreement with the experimental value of 1.0. Note that the normalized
side movement in this case is (3/1.5)(5/1.8) = 5.6 yards per 100 yards, per 5 degrees
of tilt. Because the side movement in Figure 5.6 is given to the nearest yard, the side
movement could be taken to be as low as 2.6 yards, with a normalized side move-
ment of 4.8 yards. These values are not inconsistent with our conclusions regarding
grooved clubs.
Equation 5.20 agrees very well with the two experimentally determined
data points by Tuxen (October 2010) in Figures 5E and 5.6. For the 10 degree
loft driver, with 1.0 degree face closed, the spin axis tilt is predicted to be
α spin = 180(1)/10 π = 5.7 o ; and for the same face closure with a 30 degree loft
7-iron, the tilt is α spin = 180(1)/ 30 π = 1.9o: that two experimentally validated
golf ball spin axis tilt values could be almost exactly 18/ π and 6 / π is a very
pleasing result.
es
23 degre angle
la h
u n c
normal
horizontal
attack
30 degrees
spin loft
30° spin loft

+1 degree
club path spin axis tilt = –1.8° 3 yards
face angle zero

150 yards
Figure 5.6 Tuxen example of curved iron shot (October 2010).
We can now extend this, to solve the shaped shot problem, by applying the
modified Tuxen (October 2010) rule of 3.5 yards of side movement per 100 yards
of carry, for every 5 degrees of spin axis tilt for the driver; and 4.5 yards for the
other clubs.
Starting with the driver, we can write the rule for side movement, l c , as
α spin l c
ls = (3.5) = 0.007α spin l c yards. (5.21)
5 100
With reference to Figure 5.4 this can be expressed in terms of the target angle
α t as
ls
tan(α t ) = = 0.007α spin . (5.22)
lc
Substituting for α spin from Eq. 5.17 gives

ls
tan(α t ) = = 0.40(δ h /δ v ). (5.23)
lc
For δ h /δ v ≤ 0.4 , target angle α t is restricted to a range less than 10 degrees, or
less than 20 yards of side movement per 100 yards. Within this range we can use
the small angle approximation for tangent, which on conversion from radians to
degrees gives
α t � 23(δ h/δ v ) for the driver (5.24)

Using 4.5 yards of normalized side movement for the other clubs gives the equiva-
lent expressions as
ls
= 0.52(δ h/δ v ) (5.25)
lc
and
α t � 30(δ h/δ v ). (5.26)

In setting up for a particular shot, we need first to determine club selection and
face angle.
Example
This example is the average PGA tour player 8-iron shot, but executed to give a fade
from 20 yards right of the target. This was discussed earlier. The resulting trajectory
is repeated here for convenience. The calculations of the launch conditions for this
shot are given next.
The initial launch angle away from the target must be α t = tan −1 (20 /150) = 7.6
degrees. For the required side movement, the ratio of horizontal to vertical loft from
Eq. 5.25 is
δ h/δ v = (20 / 150)/ 0.52 = 0.256.
With the face open, and an 8-iron lie angle of 63 degrees, the vertical loft
becomes
δ v = 30.6 + δ h /cos(63) = 30.6 + 0.51δ h .
Hence we require
0.256 = δ h/(30.6 + 0.51δ h ) or δ h = 9.0 degrees,
and
δ v = 9 / 0.256 = 35.2 degrees.
The ball speed for the straight 8-iron shot is 115 mph. With the increased loft, we
would expect this to decrease to 115cos(35.2)/cos(30.6) = 109 mph; coincidentally
this is the average ball speed for a 9-iron shot; see Table 3A in Chapter 3. The loft has
become very close to that of the 9-iron, so we could also select the 9-iron spin rate of
8,647 rpm and launch angle of 20.4 degrees.
Finally, the spin axis tilt is
α s = tan −1 (sin(9)/ sin(35.2)) = 15.2 degrees.

40
30
(yards)
20
10
0 20 40 60 80 100 120 140 160

(yards)
20
ath
10 club p
(yards)
–10
–20
0 20 40 60 80 100 120 140 160
(yards)
Horizontal launch angle = –7.6, Spin axis tilt = 15, Side carry = 17
Figure 5.7 Predicted launch parameters and trajectory for a 150 yard fade with an 8-iron.
The modeled ball trajectory for these launch conditions is shown in Figure 5.7. It
can be seen that the ball is predicted to return 17 of the required 20 yards.
Finally, note, with reference to column 5 for the 8-iron in Table 5A, that the club
path must be in the direction as illustrated in Figure 5.7, where
β h = α t + α hb = 7.64 + 0.73(9.0) = 14.2 o.
In Chapter 6, we return to the driver and investigate off-center strikes with the
curved driver face.
6
A G E N E R A L M O DE L O F DR I V E R
I M PAC T A N D B A L L F L IG H T
In this chapter we are concerned mainly with the performance of the modern
driver. For this purpose we describe the results of applying a 3-dimensional
model to the driver impact. This allows us to discover the effects of off-center hits,
variations in face angle, variations in club path, and variations in the basic club
geometry. The variations in the club geometry may include changes in the attach-
ment position of the hosel, changes in the vertical and horizontal face curvature,
further increases in the moment of inertia (MoI) values, and changes in the posi-
tion of the center of mass (CoM).
The “forgiveness” of the first truly modern driver, the Callaway Great Big
Bertha II (GBBII) is assessed as a benchmark against which subsequent improve-
ments can be judged. This is compared with one of the best available drivers in
2015, assumed to be further improved to the maximum performance allowable
within the R&A/USGA equipment rules for competition.
Rules for speed loss in off-center hits are formulated from the modeling results.
These allow estimates to be made of ball speed loss for poor face hits using forgiving
cavity back irons.
Finally we will show why the longest drivers of the ball on the PGA Tour must
combine greater power with a disproportionate improvement in the precision of
ball striking.
DESIGN PA R A M ETER S OF A T Y PICA L DR I V ER

The first step in applying the model is to define the characteristics of the typical
latest-generation driver. These are:
Head weight = 0.44 pounds (0.20 kg)

MoI components = (vertical 4,600 g-cm2), (sideways 3,000 g-cm 2)
Center of mass (CoM) distance behind face center = 1.34 inches (3.4 cm)
Bulge radius = 13 inches (33 cm)
Roll radius = 13 inches (33 cm)
CofR at center = 0.83 maximum
Dynamic loft = 14.4 degrees (for average PGA Tour player)
Position of hosel connection behind face center = 0.55 inches (1.4 cm)
Units of gram-centimeter squared are used for MoI because they are expressed in
these units by the golf industry and the R&A/USGA.
173
THE “STR A IGHT” SHOT

We will first use this club head to check the model against the average PGA Tour
player drive. For this drive, the principal impact parameters are
Swing speed (measured at the face center) = 112 mph

Rotation of the head at impact, about vertical and horizontal axes = 475 rpm
The rotational speed of the head was determined from the estimates made by
Tuxen (Trackman Newsletter, May 2008) for a 100 mph driver strike; see Figure 3M
in Chapter 3. The speed of rotation is assumed to be proportional to the swing
speed, so the value there was increased by factor 1.12. The resulting face speed map
is shown in Figure 6A.
In this case, for reasons that will become clear as we continue, the face map is based
on the speed at the end of the shaft extended to the horizontal plane cutting through
the center of the face. Choosing this point as the datum provides a clearer understand-
ing of the effect of the club geometry on the ball flight. We can transfer the club head
rotations to this artificial transition point from the shaft to the head. We refer to the
point as the “primary rotation point”, as indicated in Figure 6A. This allows the swing
to be thought of as bringing the extended end of the shaft to the correct location for
impact and independently applying the out-of-plane face closing rotation.
Figure 6B illustrates two representations of the impact conditions for the average
PGA Tour player drive. The left illustration shows the speed of the center of the face,
while the right one shows the speed of the club CoM. Both are based on the speed at the
end of the shaft, extended to the rotation point, and the rotation in the horizontal plane
about the rotation point. The illustration of the head represents the section of the club
head viewed on this center plane. A typical driver head has a face width of 3.5 inches
across the center plane, while the top of the face is usually around 4.25 inches wide.
The effect of the 475 rpm rotation, over the 1.75 inch distance to the face center, is an
additional 7.3 mph. Consequently, the swing speed at the rotation point is 104.7 mph,
to give the required 112 mph center face impact speed. The rotation, acting over the
distance of 0.55 inches from the rotation point to the face center, as shown in the
475 rpm
primary
rotation
point
face speeds, mph
113 110 107

0.5 in
115 112 109 105
0.75 in
475 rpm
117 114 111
Figure 6A Face speed map for the average PGA Tour player.
175 Model of Driver Impact and Ball Flight
left illustration, produces a sideways movement of the face toward the heel at speed
2.2 mph. Note that it is assumed that the rotation center is travelling directly toward
the target. This corresponds to the advice of good teaching professionals to have the
back of the left hand moving toward the target at impact (Mike Kelly, personal com-
munication, Fall 2014).
If instead of focusing on the center of the face, our attention is transferred to the
CoM of the club head, the result is as shown on the right side of Figure 6B. Here the
rotation point is 0.79 inches in front of the CoM, which produces a sideways move-
ment toward the toe of 3.2 mph. The net result is that at impact, the CoM is moving
1.63 degrees to the right, while the face center is moving 1.15 degrees to the left.
These angles are of course greatly exaggerated in the figure. Because of the small
angles, the speed of the CoM and the center of the face are the same 112 mph to the
nearest 0.1 mph.
The CoM moving 1.63 degrees to the inside of the swing plane is equivalent to
the face being closed by 1.63 degrees, so it will produce a counterclockwise sidespin
component on the ball; and if acting alone, it would produce a hook. At the same
time, the counterclockwise rotation of the club head, and the resulting sideways
movement of the face toward the heel, will produce a clockwise sidespin compo-
nent on the ball through the gear effect, as discussed in Chapter 4. Acting alone this
would produce a slice. We can easily isolate this gear effect by having the club path
directed straight at the target. To do this, the extended end of the shaft would be
directed 1.63 degrees to the left of the target, thus directing the CoM at the target.
For this condition, the resulting trajectory, as modeled in Figure 6C, would be a
slice landing 14 yards right of the centerline.
The ball sidespin component in this case is 368 rpm, so the total backspin has
increased to 2,710 rpm. The normalized side movement per 100 yards, per 5 degrees
of spin axis tilt, is 3.35 yards, in good agreement with the Tuxen (October 2010) rule
discussed in Chapter 5.
If instead the head rotation speed was zero and the club path was 1.63 degrees
from inside the target line as in Figure 6B, a hook would result, giving 11.4 yards
movement to the left. These effects are not entirely independent and not strictly
additive, since for each case the spin rate, launch angle, and ball speed are slightly
different. However, this turns out to be very nearly the case. The precise result is
112.0 mph
1.15°
104.7 mph 3.2 mph 0.79 in

2.2 mph
0.55 in 676
rpm
676
rpm
112 mph
1.75 in
1.63°
Figure 6B Impact velocities of the average PGA Tour player (angles greatly exaggerated).
40
30
(yards)
20
10
0
0 50 100 150 200 250
(yards)
Launch angle = 11.2, Ball speed = 165, Backspin = 2710, Height = 31.0, Carry = 267.8
20
10
(yards)
0
–10
–20
0 50 100 150 200 250
(yards)
Horizontal launch = 0.26° , Spin axis tilt = 7.8° , Side carry = 14.0
Figure 6C Effect of the club head rotation with club CoM directed straight to the target.
very close to the difference between these two effects, that is, a slight fade 14 – 11.4 =
2.6 yards right of the target line. However, before we put these separate effects
together into the model, we must also consider the horizontal component of the
club head rotation speed. Because the horizontal head rotation rate is approximately
the same as the vertical one, and the distances from the face center to the rotation
point and the CoM are the same, the corrections to the impact conditions are the
same. This is shown in Figure 6D where the extended shaft end, the rotation point,
is assumed for easier illustration to be travelling horizontally toward the target at
impact. It is easy to visualize that this rotation, transmitted directly from the swing
arc, must produce a slightly downward trajectory of the CoM and a slightly upward
trajectory of the face center.
The effect of these complete impact conditions on the ball trajectory are shown
in Figure 6E, a very small fade.
These are very satisfactory results, to be expected for equipment that has been
subjected to exhaustive careful design iterations to achieve best performance, start-
ing with forgotten early craftsmen working with wood. There is in fact no better
place to position the hosel. Assume the hosel was moved forward in Figure 6D so
that the rotation point is aligned with the center of the face, as shown in the left
illustration of Figure 6F.
rotation center
angles greatly exaggerated

1.63°
2.2 mph
112.0
mph 676 1.15°
rpm
Figure 6D Impact velocities of the average PGA Tour player.

40
30
(yards)
20
10
0
0 50 100 150 200 250
(yards)
Launch angle = 11.3, Ball speed = 165, Backspin = 2612, Height = 31.0, Carry = 268.9
20
10
(yards)
0
–10
–20
0 50 100 150 200 250
(yards)
Horizontal launch = 0.04°, Spin axis tilt = 1.3°, Side carry = 2.8
Figure 6E Average PGA Tour drive with face square and hosel directed at the target.
In this case, the vertical rotation has a larger sideways effect on the speed
of the CoM, and the club path has moved further inside to 2.76 degrees. The
sideways movement of the face, shown in Figure 6B for the typical driver design,
has now been eliminated. The ball sidespin now results only from the face being
closed by 2.76 degrees with respect to the club path. With the average PGA Tour
swing, the resulting trajectory is a hook 18 yards to the left of the target line.
The flight distance is reduced to 260 yards because of the lower launch from the
closed face.
The appearance of a fully forward hosel design is likely unappealing. However,
forward hosels to reduce slice have been introduced into the market including the
Ping G10 DrawTM and Cobra BafflerTM. Any reduction from the typical 0.55 inches
rearward placement of the hosel will reduce slicing for amateur players who have a
tendency to leave the club face open at impact. The effect is proportional to the rear-
ward position amount, diminishing to zero when the hosel is aligned with the face
center. It should be mentioned that the alternative way, of promoting a draw with the
driver, is to move the CoM of the driver head toward the heel of the club. The impact
‘rotation point’
676
rpm
1.34 in 1.34 in
5.4 mph
676
rpm
112 mph
2.76° 104.7 mph
Figure 6F Driver designs with forward and centered hosel positions.

40
30
(yards)
20
10
0
20
10 forward hosel
(yards)
0
–10 body-center hosel
–20
0 50 100 150 200 250
(yards)
Figure 6G Flight trajectories for the average PGA Tour player swing with forward and
body-centered hosel locations.
force for a center-face hit, acting outside of the CoM toward the toe, produces a
clockwise rotation of the head, which through the gear effect gives a counterclock-
wise draw sidespin to the ball. This effect has been known since the earliest times of
custom shaping wood driver heads. We see in the next section that it was a signifi-
cant attribute in the design of the ground breaking Callaway Great Big Bertha II.
More recently, moveable weights have been introduced to allow the player to adjust
CoM position from the center toward the heel or toe, for varying draw or fade effects
respectively.
The other extreme, to the full-forward, heel-positioned hosel, would be to con-
nect the hosel so that the rotation point is at the CoM. In this case, the impact speed
does not increase through the action of closing the face. Also if the club path is
directed at the target, the rotating face will cause the ball to slice to 13 yards right.
This is reduced from the amount in Figure 6E simply due to the reduced club head
speed and the corresponding reduction in flight distance. The trajectories for these
two hosel configurations are illustrated in Figure 6G.
The forward hosel location is a helpful equipment change for the habitual slicer
of the ball. It also seems to be a commonly accepted design for game-improvement
irons, although the CoM of irons is so close to the face that the effect is likely to be
small. The body-centered hosel example was included simply because the heel place-
ment of shafts to golf clubs is not obviously the optimal design solution, although it
turns out to be so.
The trajectory in Figure 6E can be considered the “straight shot,” which an elite
player achieves through trial and error with direct feedback on the practice range.
A final adjustment to launch the ball 0.5 degrees to the left would land the ball on
the centerline.
To obtain agreement with the Trackman PGA average drive data in Figure 6E,
the slip/skid factor had to be increased from 0.71, used in Chapter 4, to 0.725. This is
not surprising. The analysis in Chapter 4, from which the value 0.71 was determined,
neglected the horizontal rotation of the club head, which reduces the amount of
backspin. Thus the slightly higher slip/skid factor was required to still obtain the
required estimate of the 2,685 rpm measured average spin rate.
gea rotat
r -w ion
edg
e
driving
gea
gear
radius e
rd
driv
gear
rive
head head wedge-gear
speed speed rotation
Figure 6H High toe driver strike.
DR I V ER STR IK ES OFFSET FROM THE FACE CENTER

To illustrate the main variables involved in applying the model, Figure 6H shows
the case of a high toe strike where the club path is directed straight toward the tar-
get and the angle of attack is zero. For this strike position, the normal reaction from
the ball to the club head will cause the head to rotate as shown in both the vertical
and horizontal planes; these rotations are shown by the curved arrows inside the
club body acting about the CoM. They are labeled “gear drive”; and the radius of
the “driving gear” is illustrated by the open-tipped arrow from the CoM to the face.
Because of the face curvature, the driving gear radius is slightly larger than the dis-
tance from the CoM to the face center, but this difference can be neglected. Because
the ball grips the face during some of the contact time, the rotating club head causes
the ball to rotate in the opposite direction to the club head in both the vertical and
horizontal planes. These effects were discussed in Chapter 4.
THE GR E AT BIG BERTH A II—A BASIS

FOR COMPA R ING DR I V ER PER FOR M A NCE
A good starting point to illustrate the effect of offset ball strikes is in the perfor-
mance of the Callaway Great Big Bertha II driver (GBBII), which was introduced to
the market in 2004. This was the second iteration of truly higher performing driv-
ers than all their predecessors. It had broken through into the 400cc volume range
with a volume of 415cc, and this opened the door for even larger drivers until the
R&A/USGA imposed the 460cc limit. It also had a high spring-effect trampoline
face providing 0.83 CofR for center hits and giving high CofR for off-center ball
strikes. The GBB II was manufactured from lightweight, high-strength titanium
alloys, which, as discussed earlier, allowed the Callaway designers to increase the
head size to achieve higher MoIs. We will see that this reduces the ball speed loss
for off-center miss-hits. Also, of course, the material choice enabled a large flexible
diaphragm face.
Callaway later introduced the GBBII+, which had an even more flexible face, pro-
viding approximately 0.85 CofR. This caused some concern within the USGA about
the limits of the trampoline face technology, and they decided to impose a limit of
0.83 CofR on driver impact performance. This was later adopted by the R&A.
The Callaway Corporation generously gave me a demonstration of their robot

testing facility, carrying out a full set of off-center ball strikes using the GBBII. The
results of this testing are shown in Table 6A. The results are for a 104 mph face cen-
ter speed and are the average of 10 hits at each face position. There was a light cross-
wind from left to right during the tests. We will see that this was the cause of the
slice bias in the results, shown by the negative side movement in column 8 for all of
the results. The effect of the crosswind will be accounted for later in the modeling.
The offset face positions are 0.75 inches each side of the center. These are labeled
“Center heel” and “Center toe.” The “High center” and “Low center” positions are
0.5 inches up and down from the center. At the heel and toe, the high and low posi-
tions are 0.4 inches up and down from the “Center heel” and “Center toe,” respec-
tively. The Trackman trajectories for the center, center toe, and center heel positions
are shown in Figure 6I.
This shows the self-correcting ball flight for heel and toe strikes as discussed in
Chapter 4, together with the center-strike trajectory. The best fit for the model for all
of the cases was found to be with a face closing speed of 500 rpm at impact; equiva-
lent to 354 rpm in both the horizontal and vertical planes. To model the GBBII ball
strikes, it is necessary to know the club head geometry and the mass properties of
the head. For the GBBII these are
Head weight = 0.45 pounds (0.206 kg)

Effective impact mass = 0.49 pounds (0.223 kg)
MoI components = (vertical 3,535 g-cm2), (sideways 2,283 g-cm2)
CoM position measured from the face center:
distance behind face center = 1.22 inches (3.1 cm)
offset toward heel from face center = 0.24 inches (0.6 cm)
offset toward crown from face center = 0.24 inches (0.6 cm)
Bulge radius = 11.8 inches (30 cm)
Roll radius = 10.6 inches (27 cm)
CofR at center = 0.82
Club loft = 11 degrees
Dynamic loft = 14.6 degrees (at the 104 mph robot test speed)
Position of hosel connection behind face center = 0.55 inches (1.4 cm)
The head mass includes a hosel tube, which passes through the head and is
attached to the sole plate. The MoI values were measured by the writer on a Mass
Properties Measurement System Model KSR 330-60 at Space Electronics LLC in
Berlin, Connecticut. The one surprise in these values was the fact that the CoM is
both above and toward the heel from the face center. This is easily checked using
just a string attached to a small suction cup. Simply adjust the position of the string
attachment to the crown until the club head hangs with the face vertical (completely
de-lofted) and the score lines horizontal. Next, repeat the process with the suction
cup attached to the side wall at the toe or heel so that the score lines are vertical to
get the CoM offset high or low on the face. The setup for the side offset is shown in
Figure 6J.
Table 6A Callaway Great Big Bertha II robot testing with Trackman test results
Face position Ball speed Launch angle (deg) Spin rate Maximum height Carry (yards) Side Flight
(mph) (rpm) (yards) (yards) time sec
Vertical Horizontal
Center 152.8 12.9 0.8 3,075 37.9 243.3 −5.0 7.1

Center toe 150.0 12.7 −3.2 3,562 35.7 227.6 −5.4 7.1
Center heel 152.0 12.7 −5.1 3,265 36.7 239.3 −6.7 7.1
High center 148.2 16.0 1.0 2,488 38.8 238.1 −7.8 7.0
Low center 150.7 9.7 0.8 4,202 33.5 224.9 −6.3 6.9
High toe 144.7 15.4 −3.3 2,993 36.6 228.7 −9.6 6.9
Low toe 147.0 9.8 −3.7 4,127 30.6 215.6 −10.2 6.6
High heel 145.3 15.3 4.9 2,588 36.2 235.7 −6.4 6.9
Low heel 148.3 10.3 5.1 3,834 32.4 219.8 −5.1 6.8
GBBII
40
center
30
(yards)
heel & toe

20
10
0
20
10 0.75 in. heel
(yards)
0
–10 0.75 in. toe
–20
0 50 100 150 200 250
(yards)
Figure 6I Trajectories for center, heel, and toe strikes with the Great Big Bertha II; center face
speed 104 mph.
The figure also shows the face speed map for a 104 mph center-face speed with
the face closing speed of 354 rpm about the vertical and horizontal axes.
Moving the center of gravity nearer the heel helps to eliminate the slice bias of
the driver head. It also serves to increase the ball speed for strikes toward the heel,
and decrease the ball speed for strikes toward the toe, and so compensates for the
decrease in face speed from the toe to the heel. Likewise, the high CoM will com-
pensate for the decrease in face speed from sole to crown.
The first step in modeling the GBBII face strikes is to find the strike parameters
that give good agreement with the center hit. These strike parameters are then held
constant for all of the ball strikes. This represents the way in which the testing was
carried out. That is, the robot was not adjusted during the tests except for reposi-
tioning to give the required face striking positions.
It turns out that with the offset CoM position, the GBBII is perfectly designed to
provide a “straight” shot. To achieve the correct launch angle and backspin rate, the
dynamic loft was found to be 14.6 degrees with a slightly upward (positive) angle of
string
support
354 rpm
Center
face speeds, mph of mass
102.5
105.0 0.5 in 100.6
354 rpm 106.2 104 101.8

0.4 in 0.75 in
107.4 103.0
105.5
Figure 6J Face speed map from robot testing of the Callaway Great Big Bertha II.
GBBII
40
center
30
(yards)
heel & toe
20
10
0
20
10 0.75 in. heel
(yards)
0
–10 0.75 in. toe
–20
0 50 100 150 200 250
(yards)
Figure 6K Modeled trajectories for center, heel, and toe strikes with the GBBII; center face
speed 104 mph, club path zero, face angle zero, attack angle +0.6 degrees.
attack of 0.6 degrees. The loft value was obtained by applying the method developed
in Chapter 4. Both the club path and face angle were set to zero. The predicted tra-
jectory with these strike conditions is shown by the solid lines in Figure 6K. With
these strike conditions, the club gives an almost perfectly straight center shot and
perfect self-centering trajectories for toe and heel strikes. The latter are shown by
the dashed line and dot-dash line, respectively, in the lower pane of the figure. In the
vertical plane, the heel- and toe-strike trajectories are predicted to be almost identi-
cal and are represented by the double-dot dash line in the upper pane. These results
attest to the excellence of the design. The parameters of these three trajectories are
given in Table 6B.
The next task is to determine the crosswind conditions and club path that would
provide the initial trajectory and landing offsets in agreement with the Trackman
data in Table 6A. The best fit to all of the data set was found to be with a left to right
crosswind of 5 mph and an initial launch 0.8 degrees to the left; the latter value was
taken from the Trackman data for the center hit in Table 6A.
With these conditions, the modeled trajectories for the center and toe and heel
strikes are shown in Figure 6L. The side landing positions are in close agreement
with the measured values in Table 6A.
The results across the range of shots, as shown in Table 6C, are in reasonable
agreement with the Trackman measured values. With only two exceptions, they are
within the range of variation of the 10-shot results for each face position; and all are
within the 90 percent confidence intervals for all of the results.
Nevertheless, it is clear that the model generally underestimates the ball speed
for all off-center hits by between 2 and 4 mph, the highest underestimation being for
heel strikes. In my view, this is almost certainly due to the effect of the shaft.
Recall from Chapter 3 that the effect of the shaft is represented by the use of an
effective mass value. However, when offset ball strikes are considered, it is clear that
the loss of the momentum of the shaft will vary with strike position. This is easy to
see with respect to the high toe and low heel positions as shown in Figure 6M. For
the toe strike, the forward speed loss of the shaft will be reduced by the club head
rotation, while the opposite will be true for the heel strike. Since the momentum
Table 6B Model predictions of the Callaway Great Big Bertha II for “straight” strike at center, toe, and heel positions
Face position Ball speed Launch angle (deg) Spin rate Maximum Carry Side Flight time
(mph) (rpm) height (yards) (yards) (sec)
Vertical Horizontal
(yards)
Center 153.1 12.9 0.0 3,110 37.8 243.0 −0.9 7.1

0.75 toe 148.0 12.5 −4.1 3,498 34.9 230.1 −0.7 6.9
0.75 heel 148.1 12.5 4.1 3,559 35.0 229.4 −0.7 6.9
40 center
30
(yards) heel & toe
20
10
0
20
10 0.75 in. heel
(yards)
0
–10 0.75 in. toe
–20
0 50 100 150 200 250
(yards)
Figure 6L Modeled trajectories with 5 mph crosswind for center, heel, and toe strikes with the
GBBII; center face speed 104 mph, club path 0.6 degrees inside, face angle zero, attack angle +0.6
degrees.
loss of the shaft for the heel strike will be greater than for the toe strike, this will
result in higher ball speed for the heel strike—as shown by the Trackman results.
This is a very complex problem that may be beyond reasonable modeling methods.
There is a further important point to be understood about off-center driver
strikes, and this can be found in the modeling results. Table 6D shows the velocities
and rotational speeds of the club head and ball before and after impact for the GBBII
heel strike.
The CofR for this shot would be calculated by the R&A/USGA, and in the
general golf literature, as separation speed/approach speed = (148.0 – 74.7)/104 =
0.70. However, although this is a measure of the effectiveness of the ball strike, it is
entirely specific to this club and does not represent the efficiency of the impact in
any general terms. I refer to this measure as the CofR-MoI index. To understand
low
high heel
toe strike
strike
105.0
103.0
Figure 6M Opposite effects of heel and toe strikes on the shaft speed after impact.
Table 6C Model predictions of the Callaway Great Big Bertha II face strikes with 5 mph crosswind
Face position Ball speed Launch angle (deg) Spin rate Maximum height Carry Side Flight time
(mph) (rpm) (yards) (yards) (yards) (sec)
Vertical Horizontal
Center 153.1 12.9 0.8 3,110 37.7 243.1 −5.1 7.1

0.75 toe 148.0 12.5 −3.3 3,498 34.8 229.2 −4.9 6.9
0.75 heel 148.0 12.5 4.9 3,559 34.9 230.4 −5.0 6.9
High center 146.3 16.0 0.8 2,478 38.5 235.0 −5.8 7.0
Low center 148.8 9.2 0.8 4,300 30.6 224.1 −5.1 6.8
High toe 143.1 15.1 −3.2 2,892 36.0 226.4 −5.6 6.8
Low toe 145.2 9.8 −3.3 4,081 29.0 218.3 −6.2 6.6
High heel 143.0 15.0 5.0 2,978 36.1 227.6 −5.0 6.9
Low heel 145.1 9.8 4.9 4,124 29.1 219.5 −3.3 6.6
Table 6D Conditions before and after impact for 0.75 inch offset heel strike with Great

Big Bertha II
0.75 Heel strike Club head Ball speed Club head rotational Ball rotational speed
speed (mph) speed
(mph)
Horizontal Vertical Horizontal Vertical
(rpm) (rpm) (rpm) (rpm)
Before impact 104.0 0.0 354.0 354.0 0.0 0.0

After impact 74.7 148.0 1,391.0 48.0 532.0 3,519.0
why this is the appropriate name, we need to calculate the correct CofR value, which
is defined as the separation speed/approach speed, but both measured at the impact
point. From Figure 6J, the approach velocity of the face at the heel position is 101.8
mph. As shown in Table 6D, after impact, the club head is rotating counterclockwise
in the horizontal plane at 1,391 rpm. This rotation, over the distance of 0.75 inches
from the center to the heel position, produces a backward face speed of 6.2 mph.
Thus at separation, the face contact point at the heel is moving forward at speed
74.6 – 6.2 = 68.4 mph. Finally, the face velocity must be multiplied by the cosine
of the loft angle to give the speed at right angles to the face. This value is 98.5 mph.
Thus, the correct measure of CofR is (148.3 – 68.4)/98.5 = 0.81. This value makes
much more sense since even 0.75 inches toward the heel the ball is still striking a
thin face with some trampoline spring effect. However, the impact is now nearer to
the trampoline support of the club head side wall, and so the face is stiffer there with
reduced spring effect. If the GBBII had greater MoI values with the same face, the
ball speed would increase and thus the CofR-MoI index would increase, but the true
CofR would be the same 0.81. This is because the true CofR is fundamentally based
on the energy loss between the ball and the striking surface, which are unchanged
by changes in MoI.
The simulated trajectory for this heel strike, using the predicted launch param-
eters in row 3 of Table 6C, is shown by the solid lines in Figure 6N. The dot-dash line
shows the Trackman measured trajectory.
EFFECTS OF INCR E ASING THE MOI A N D THE Cof R

Having established some confidence in the validity of the model, it is now pos-
sible to estimate how far driver club development can improve and still remain
conforming to the R&A/USGA rules for competition. To do this, we simply
increase the MoI components of the club head and set the true Cof R to 0.83
at any position on the playable area of the face. The rules of driver head design
limit the volume to 460cc, the width of the head to 5 inches, the depth of the
head to less than the face width, and the height of the head from sole to crown
to 2.8 inches. In addition, the MoI should not exceed 5,900 g-cm 2 about the
vertical axis through the club head when the club is in the horizontal address
position. This MoI limit is very difficult to reach. If we look at simple geometry
Club speed = 104; Ball speed = 148.0; Flight time = 6.9

40
30 Predicted
(yards)
Trackman
20
10
0
Attack angle = 0.6; Vertical launch = 12.5; Spin = 3559; Carry = 230.4; Height = 34.9
20
10 0.75 in. heel
(yards)
0
Crosswind = 5 mph
–10
–20
0 50 100 150 200 250
(yards)
Club path = –0.8; Face angle = 0; Horizontal launch = 5.1; Spin axis tilt = 8.6; Side = 5.0
Figure 6N Predicted launch parameters and ball trajectory for GBBII toe strike; with
comparison to Trackman test results courtesy of Callaway Corporation.
analogs, half of the 200 g mass of the club head would need to be spread equally
around the club perimeter to achieve this value; that is, the “vertical” walls
would need to form an approximate ring 5 inches in diameter with the CoM in
the middle of the ring. For this configuration, if the sole and crown use the other
half of the 200 g, then the vertical axis MoI is 6,150, just over the maximum.
However, the MoI component, sideways through this body, is only 2,370 g-cm 2 .
Unfortunately, the face needs to be heavier than the average wall thickness to
accept the impact loads, and it needs to have less curvature. This brings the
CoM toward the heavier mass of the face and reduces the MoI value. Therefore,
more of the total mass needs to move to the sides of the perimeter to get back
up to the 6,000 g-cm 2 level. This logically moves the design toward the “square”
profile club heads such as the Nike SasQuatch, which were introduced into the
market several years ago.
So a starting point for the design should be some sensible discussion of the
desirable relative values of the MoI components about the vertical and hori-
zontal axes through the club head. The basis for such a discussion is shown in
Figure 6P.
vertical
MoI
component
horizontal
MoI
component
Figure 6P Typical error strike pattern for a high handicap amateur player.
The circles in the figure represent the distribution of ball strikes for the aver-
age weekend golfer. The ellipse is the best ellipse fit through the center of the ball
impressions on the face, neglecting occasional very poor strikes. The horizontal
length across the ellipse is approximately 1.5 times the distance from the lowest to
highest point of the ellipse.
The shape of the error pattern is no doubt due to the fact that the length of the
swing pendulum is much easier to control than the plane of the swing. This is why
the ellipse is inclined across the swing arc. In turn, this is presumably why the width
of driver heads became greater than the face depth.
This brings us to a sensible rule for the relative MoI values. The amount of club
head rotation during the strike is proportional to the offset distance from the center,
that is, to the torque applied by the impact force about the CoM. However, the amount
of rotation is inversely proportional to the MoI. So if the impact force is on average
1.5 times further from the vertical axis than the horizontal one, then it is best to
have the vertical MoI component 1.5 times greater than the horizontal one.
Therefore, being truly simplistic, instead of approximately sharing 6,000 and
2,500 g-cm 2 for these two values, it is much better to share the total in a propor-
tion of 3 to 2, or 5,100 and 3,400 g-cm 2 . Typical values, for the most forgiving
460 cc drivers, approach 4,600 and 3,000 g-cm 2, which is in the 3 to 2 ratio.
The outstanding 919 THI driver head designed and sold by Tom Wishon (www.
wishongolf.com), for custom fitting at centers all over the United States, has MoI
values 4,759 and 3,220 g-cm2, which are quite close to the suggested maximum.
It is not unreasonable to consider future creative designs reaching the 5,100 and
3,400 g-cm2 target. If, in addition, we assign 0.83 to the entire playable area of the
face, then the modeled face performance is as shown in Table 6E.
The average carry distance for the eight off-center hit positions is 236.4 yards
compared to 226.3 yards for the GBBII. These gains are not huge in percentage
terms. However, for club head comparison, the launch conditions were taken to be
as near as possible to those in the GBBII testing. These launch conditions are likely
far from optimal, and custom fitting of the 919 THI, as intended by Wishon, would
without doubt make a much bigger difference to player performance; see Wishon
2006, 2008.
IRON STR IK ES OFFSET FROM THE FACE CENTER

We can formulate a simple rule for speed loss for offset hits. If a ball is launched at
say 100 mph then the change in momentum of the ball is 100 times the ball weight of
0.1 pounds, or 10 pound-mph. Recall that this is equivalent to the impulse through
Newton’s laws, and that an equal and opposite impulse is applied from the ball to
the club face. The impulse acting through the offset on the face produces an impulse
torque equal to the impulse multiplied by the amount of the offset. Through a more
complex extension of Newton’s laws it can be shown that the impulse torque pro-
duces a change in rotational speed that decreases with increasing MoI. So in general
terms we can write
Club head rotational speed = impulse torque/MoI = ball speed × ball mass × offset/

MoI.
Table 6E Predictions for highest performance driver with 0.83 CofR over the playable area of the face
Face position Ball speed Launch angle (deg) Spin rate Maximum height Carry Side Flight time
(mph) (rpm) (yards) (yards) (yards) (sec)
Vertical Horizontal
Center 154.3 13.0 1.1 2,946 38.0 245.4 −1.4 7.1

0.75 toe 151.8 12.7 −2.2 3,297 37.2 239.1 2.0 7.1
0.75 heel 152.0 12.6 4.5 3,386 35.8 239.4 0.1 7.0
High center 150.4 15.4 1.1 2,443 39.3 241.6 −3.0 7.1
Low center 153.7 10.2 1.1 3,997 34.6 235.1 0.5 7.1
High toe 149.1 14.7 −2.2 2,806 38.6 238.2 −3.2 7.1
Low toe 151.7 10.7 −2.3 3,795 34.0 233.4 1.4 7.0
High heel 147.3 14.6 4.5 2,917 37.2 233.8 −1.2 7.0
Low heel 150.0 10.6 4.5 3,864 32.7 229.3 0.9 6.9
Note that impulse torque/MoI for rotational speed changes is the same as
impulse/mass for regular (linear) speed changes.
Next we can recognize that when the club head is rotating backward from the
offset impulse, the loss of speed of the face at the impact point, because of the back-
ward rotation, is the rotation speed multiplied by the offset. Consequently
Face rotation speed loss = ball speed × ball mass × (offset squared)/MoI.
If we were using units of kilograms, meters, and seconds we would just plug the
values directly into this expression. With our mixed bag of units including pounds,
miles per hour, and grams-centimeter squared for MoI, we simply need to convert
the units in turn and so end up with a multiplier which is the product of all the
conversions. This gives a multiplier of 290 including the ball weight. Finally, we
can write
Face rotation speed loss = 290 × launch speed × (offset squared)/MoI,
where speed is expressed in miles per hour, offset is in inches, and MoI is in the
industry units of gram-centimeter squared.
So, consider strikes 0.5 inches up or down from the center of the GBBII driver
with sideways MoI of 2,283 g-cm 2 . There is a complication for the driver curved
face on which the ball reaction force is inclined inward toward the CoM. The offset
distance of the force passing the CoM is thus reduced (see Figure 3N). In this case,
the offset decreases from 0.5 inches to 0.44 inches. Plugging these values into our
expression gives
Face rotation speed loss = 290 × 154.3 × 0.442/2,283 = 3.8 mph,
where 154.3 mph is the center strike ball speed, which is good enough for an approx-
imate estimate.
We can compare this to the Trackman measured values. Since the CoM is
also offset from the face center in the GBBII, we compare the average of both the
high and low strikes. From Table 6A the average high and low speed is 149.5 mph.
Compared to the center ball speed of 152.8 mph this is a loss of 5.6 mph. The extra
1.8 mph can be attributed to the decrease in the true CofR at these locations.
These estimates neglect the effect of the friction force acting from the ball about
the club CoM which is significantly behind the face. This friction effect, the reduc-
tion in the offset, and the change in CofR all disappear when the flat, shallow, and
non-flexible face of irons are considered. The MoI values of “forgiving” cavity-back
irons have not changed since the Ping Eye2s, and are approximately 2,600 and
1,500 g-cm 2 about the vertical and horizontal axes through the face center respec-
tively. With shorter shafts the face striking errors are less than with the driver for all
players. However, because of the smaller MoI values the penalties are larger. A badly
off-center ball strike 0.5 inches toward the toe or heel, for a 100 mph ball speed
with a center hit, would give an approximate ball speed loss of 290 × 100 × 0.52/
2600 = 2.8 mph. A 0.5 inch offset high or low ball strike would, with the lower MoI,
cause a loss of 4.8 mph. If the strike is both 0.5 inches sideways and 0.5 inches high
or low then the rotations about both the vertical and horizontal axes occur and the
ball speed loss would be 2.8 + 4.8 = 7.6 mph.
These effects are not so damaging to distance. The main problem with iron strikes
for amateur players is hitting the ball “thin,” that is, very low on the face so that the
ball wraps over the edge of the sole as it deforms against the face. The club rotation is
then of secondary importance compared to the non-uniform deformation of the ball
giving very poor restitution. Alternatively, hitting the ground first, or hitting “fat,”
decelerates the club head before impact with equal loss of club head and ball speed.
Even though the CoM is much closer to the face than for a driver, heel and toe
hits will still produce some gear effect fades and draws respectively since there is
no longer any cancelling wedge effect from a bulged face. According to Tuxen (July
2009), offset hits sideways on the face of a 6-iron will produce about one-third of the
spin axis tilt as for the same offset on the driver face. This is probably little different
for the other irons.
THE LONGEST DR I V ES ON THE PGA TOU R

Except for the later valuable innovations in player customization, such as the
ability to position the CoM for draw or fade bias, the ease of switching shafts, and
the adjustment of loft, the peak in driver ball-speed performance from the princi-
pal equipment manufacturers had arrived by 2006. In that and the following year
the “hot” drivers included the Callaway Big Bertha 460, Cobra HS9, Nike SQ
Sumo, Ping G5, and the Taylormade r7. In 2006 the longest driver of the ball on
the PGA Tour had also just arrived. His average drive in PGA competitions that
year had a carry distance of 319.6 yards; see www.pgatour.com. His average driv-
ing distance over the following 9 years put him in first place on Tour for average
driving distance 5 times, and second place twice. Let’s put a Green Jacket on him
and give him a title - Mr. Bubba Watson. I had the pleasure to follow him around
the course at the CVS Classic several years ago. The most remarkable aspect of
his drives, apart from huge carry distances and souring heights sometimes more
than 50 percent higher than the average PGA Tour drive, is the accuracy of his
ball striking with average swing speeds around 125 mph. By the 2014 season he
had become the first player to average 310 yards of carry while hitting 60 percent
of fairways. Knowing the speed and distance penalties for off-center hits, it is
clear that his striking must be consistently very close to the center of the face.
However, the much bigger problem, resulting from the increased flight times
and, more significantly, from the launch conditions, is hitting the fairway with
such drives. The control difficulty is proportionately very much larger than the
distance increase.
From Chapter 4, we know the likely optimum launch conditions needed to
achieve extra long drives. These were provided by Thomas (2008) from his work
on the Overall Distance Standard at the USGA; namely for 120 mph club head
speeds, the launch angle should be 12 degrees and the backspin rate should be
2,250 rpm. However, Thomas does not discuss how such launch conditions are
achieved. Recall that the average launch angle and backspin rate on the PGA
Tour are 11.2 degrees and 2,685 rpm with an impact speed of 112 mph, which
are achieved with an attack angle of −1.3 degrees and a dynamic loft angle of
14.4 degrees. So the task is to strike the ball faster and launch it higher, while
decreasing the spin rate. The only way to achieve the reduced spin rate is to
reduce the dynamic loft. This will lower the launch angle for the same swing, so
the attack angle must be adjusted upward to increase the launch angle. Trackman
monitoring of Bubba Watson, reported by Tuxen (January 2008), showed that
he achieved his long drives with an average launch angle of 13.5 degrees, an aver-
age spin rate of 2,250 rpm, and an average attack angle of +5 degrees. The swing
speed and ball speed required to drive 310 yards with these launch conditions are
125 mph and 185 mph respectively. With these values the techniques established
in Chapters 3 and 4 give normal and spin CofR values, using the effective mass
values, of 0.81 and -0.23 respectively, and a dynamic loft of 9.9 degrees. As a reality
check, from the variation of NCofR with impact speed discussed in Chapter 3, we
would expect a decrease of 0.016 for the 16 mph increase above the 109 mph test
speed for a 0.814 CofR value. We should also note that to achieve a dynamic loft of
9.9 degrees, even with an ultra-stiff shaft, requires a small club loft; probably
somewhere in the 6.5 to 7.5 degree range.
This brings us to the important point of this section. To remain in the fairway
with a longer drive requires even more precision in ball striking than would appear
to be the case. This is because two factors are in play. First, the longer drive will
have a longer flight time, and so all other things remaining the same, a given tilt
of the ball spin axis will carry the ball further sideways. Second, and more signif-
icant, is that the longest drives require less dynamic loft with an increased angle
of attack. The smaller loft means that any given error in face angle will produce a
higher tilt of the spin axis. For example the average PGA drive has a dynamic loft of
14.4 degrees; see Table 4A. So for a 2-degree angled face with respect to the club
path, the spin axis tilt is approximately 57 × (2/14.4) = 7.9 degrees. In Chapter 5
this was shown to carry the ball sideways approximately 15.5 yards from the initial
launch direction in either a hook or a slice, the typical distance from centerline to
edge of a Tour fairway landing area. On the other hand, a 2-degree face angle error
with one of Bubba Watson’s long drives will result in a spin axis tilt of 57 × (2/9.9) =
11.5 degrees. This will increase the sideways movement per 100 yards of carry by
multiplier 11.5/7.9 = 1.46, or 46 percent. To give just the same sideways movement
the maximum allowable face error is, instead of 2 degrees, 2/1.46 = 1.4 degrees.
With the increased carry and consequent increase in flight time, even with this
smaller face angle error, the ball is likely to move sideways by 15.5 × 310/269 = 18
yards. This calculation is an approximate one. However, it is sufficient for us to rec-
ognize the amazing skill required to hit 8 or 9 of the fairways, out of the 12 par 4s or
5s on a PGA Tournament course, with such monster drives.
In Chapter 7, we consider the other extreme in ball striking, namely, the putt.

6.1 The 3-dimensional ball striking model
This work was undertaken after reading the 2-dimensional analysis by Raymond
Penner on the effects of driver face bulge on toe and heel strikes; see Penner, 2001a.
Figure 6.1 shows the global coordinate system (X, Y, Z), with its origin at the
CoM, and the Y axis normal to the face. The point where the Y axis passes through
the face defines the “sweet spot.” The sweet spot should ideally also coincide with a
region of highest spring effect, and so be the strike spot for highest ball speed. The
point of impact is defined by offset s γ , starting at the sweet spot on a trajectory at
angle γ to the X axis and passing over the curved face surface. The sweet spot is
distance d 0 from the CoM in the direction of the Y axis. As discussed earlier, the
“sweet spot” is often compromised by moving the CoM away from the face center
for easier trajectory control, or to equalize the penalties for off-center hits in differ-
ent directions on the face. This is easily accomplished in the model by moving the
origin of the global coordinate system to the desired position.
A local coordinate system (x, y, z) is defined at the impact point. The y axis is nor-
mal to the face and the x axis is normal to the global Z axis. In Figure 6.2, the view is
tangential to the face through the contact point, in the direction of the local z axis.
The ball is shown as it is about to separate from the club face following a strike on the
club heel with velocity V0 at the club head CoM. The ball is now moving forward
with velocity v 1 and rotating with angular velocity ω. The club head has slowed to
velocity V1 and is rotating with angular velocity Ω.
The curved face of a driver can be treated as a toroidal surface and any offset, s γ ,
from the face center will in general follow a complex helical path. This makes the strike
positions on the face difficult to establish analytically with respect to the coordinate
system. However, to carry out our impact analysis, we only need to establish the direc-
tion of the normal contact force and the magnitude of moment arms of the contact
force components about the CoM of the club head. Figure 6.3 illustrates the two sec-
tion views, normal to the point of impact. The moment arms about the CoM of the
Z
sγ
γ
X
impact
point
center
of mass
X
do
Figure 6.1 Center of mass (CoM) location, impact point, and principal axes.
V1
R
do
y
sh
x
rb
v1
Figure 6.2 Section through impact location at separation of the ball and club face.
club, associated with the x, y, and z components of the contact force are denoted sx, sy,
and sz, respectively. These can be represented approximately by the equations
s x = s h − d 0 sin(α h )
s y = d 0 cos(δ + α v ) (6.1)
s z = s v − d 0 sin(δ + α v ),

where d 0 = distance to CoM from center of face; δ = loft angle at the face center;
α h = s h / R b ; α v = s v / R r ; R b = bulge radius; and R r = roll radius. Angles α h and
α v are the azimuth and elevation of the normal to the club face at the impact point
with respect to the direction of the normal to the face center. The moment arms
will be used to define the changes in angular momentum of the club head about
pz
Rb
sy sx
Rr py
do sy
αv sv py
sz αh
δ px
sh
Figure 6.3 Side and top views of oblique impact illustrating offsets from the sweet spot and
moment arms about CoM.
coordinate axes, which are parallel to the local coordinate axes, but with the origin
shifted to the CoM.
These moment arm values are based on the CoM lying directly behind the center
of the face. If the effect of moving the CoM from this position, by small amounts
∆X and ∆Z, is required, then sh and sv will be replaced, with sufficient accuracy, by
s h − ∆X and s v − ∆Z ,respectively.
6.2 Eleven equations of motion

To simplify the expressions that follow, the impulse resulting from the contact force
will be defined by symbol p; that is, p = Fav t c , where Fav is the average value of the
contact force acting during the contact time t c . The components of p in the local
coordinate directions are denoted (p x , p y , p z ). Figure 6.4 has the same sectional
views of the impact point as in Figure 6.3, but shows the impulse components,
and the resulting velocity components, of the ball and club head at the instant of
separation. To assist understanding in this figure, the linear and angular velocity
components are shown in the directions that are likely to result from the indicated
strike position. Some of these components will have negative values with respect
to the local coordinate system. Suffix 1 is used to denote the final club head veloc-
ity components. The initial velocity components of the club head are denoted
(V0 , Ω x 0 , Ω z 0 ), where V0 is the impact velocity of the club CoM, and Ω x 0 , Ωz 0 are
the angular velocity components of the club head translated from the hosel to the
CoM. Since the ball is initially at rest, only suffixes x, y, and z will be used to denote
final ball velocities. In the expressions following, I xx , I yy , I zz are the components of
the MoI of the club head with respect to the local coordinate system but with origin
at the CoM; and I b denotes the MoI of the ball.
Referring to Figures 6.3 and 6.4, the relationships between the impulse and
impulse-moment components and the changes in linear and angular momentum
components of the ball and club can be expressed as follows:
Ω1z
pz vz
V1x
vy V1y py
py ωx
px
Ω1x sv
sh v1x
ωz
V1y δ
V1z
v1y
Figure 6.4 Side and top views of oblique impact illustrating impulse, velocity components, and
local coordinates.
The change in linear momentum of the club head is given by
p x = M c (Vx1 − Vx 0 ); p y = M c (Vy1 − Vy 0 ); p z = M c (Vz1 − Vz 0 ). (6.2)

The change in linear momentum of the ball is given by
p x = − m b v x 1 ; p y = − m b v y 1 ; p z = − m b v z1 . (6.3)

The change in angular momentum of the club head is given by
p y s z + p z s y = I xx (Ω x1 − Ω x 0 ) (6.4)

p x s y + p y s x = − I zz (Ω z1 − Ω z 0 ) (6.5)

p z s x − p x s z = I yy Ω y 1 . (6.6)

The change in angular momentum of the ball is given by
p z rb = I bω x (6.7)

p x rb = − I bω z . (6.8)

Substitution between Eqs. 6.2 through 6.8 to eliminate the impulse components
gives
m b v x + M c Vx 1 = M c Vx 0 ; m b v y + M c Vy 1 = M c Vy 0 ; m b v z + M c Vz1 = M c Vz 0 (6.9)

m b v y s z + m b v z s y + I xx Ω x1 = I xx Ω x 0
(6.10)
m b v x s y + m b v y s x − I zz Ω z1 = − I zz Ω z 0 (6.11)

m b v x s z − m b v z s x − I yy Ω y 1 = 0 (6.12)

m b v x rb − I bω z = 0 (6.13)

m b v z rb + I bω x = 0. (6.14)

In addition to the eight independent conservation of momentum equations

(6.9 – 6.14), there are three additional independent relationships between the ball
and club face velocities. Two of these are determined by the condition of slipping
or skidding between the ball and club face at the time of separation, as discussed
in Chapter 4. For the smooth-faced driver, we know that slippage on the face is
to be expected. For this we use the kinematic slip/skid coefficient k s . The rela-
tionship between tangential ball and club face velocities in the x and z directions,
which must include velocity components plus the effects of all rotations, can be
written
v x + rbω z − k s (Vx1 − s y Ω z1 − s z Ω y1 ) = 0 (6.15)

v z − rbω x − k s ( Vz1 + s x Ω y1 + s y Ω x1 ) = 0. (6.16)

Finally, strain energy lost due to ball hysteresis, and to residual vibrations in the
club head, is accounted through the normal CofR, en, relating approach and separa-
tion velocities in the y direction normal to the impact; that is,
v y − Vy1 − s z Ω x1 + s x Ω z1 = e n ( Vyo − s x Ω z 0 + s z Ω x 0 ). (6.17)

Note that in the golf industry, CofR is often defined erroneously as
e n = ( v y − Vy 1 )/ Vy 0 ; that is, using the velocity of separation of the CoM and ignor-
ing the club head rotation. The metric is easy to obtain experimentally and is a valid
assessment of club head performance in terms of ball speed. However, it does not in
general compare the impact efficiency of different club heads; and except for sweet
spot impact, it has no fundamental meaning in mechanics. We refer to the metric as
the CofR-MoI index.
It will be assumed that the deformation between the ball and club face is suf-
ficiently rotationally symmetric about the y axis, so the rotation about the y axis
can be assumed negligible. Equations 6.9 to 6.17 thus comprise the 11 independent
linear simultaneous relationships, which can be solved numerically to obtain the
11 unknown components of ball and club head velocity after the impact. These
equations can be expressed in matrix form as given in Figure 6.5.
6.3 Application of the equations of motion

The remaining step in application of these equations is to transform the MoI compo-
nents about the global axes to the local ones. Golf driver heads have a high degree of
symmetry. The asymmetry of the hosel is usually counterbalanced by a slight bulg-
ing of the head behind the toe of the club. For the GBBII discussed earlier, the MoI
measurements included the value at 45 degrees diagonally in the plane of the face
and through the CoM. This diagonal value was determined to be I 45 = 2 , 736 g-cm 2 .
This indicates that the MoI components about the vertical axis and the horizontal
axis sideways across the face are close to the principal components. If they were
principal components, then the average of the two would be equal to I 45 . Referring
back to the data given earlier, we can see that the average of the two components is
(2,283 + 3,535)/2 = 2,909, which is only 6 percent larger than I 45 . Since the club
is most symmetrical about the horizontal axis from front to back, we treat the three
measured values as principal components, labeled I XX , I YY, and I ZZ . The transforma-
tion to local coordinates is simply
I xx = I XX cos2 (α h ) + I YY sin 2 (α h )cos2 (δ v + α v ) + I ZZ sin 2 (α h )sin 2 (δ v + α v )

I yy = I XX sin 2 (δ v + α h ) + I YY cos2 (α h )cos2 (δ v + α v ) + I ZZ cos2 (α h )sin 2 (δ v + α v ) (6.18)
I zz = I YY cos2 (α h ) + I ZZ cos2 (δ v + α v ).

A program written in Matlab (MathWorks Inc., Natick, Massachusetts) was
used to solve the set of simultaneous equations. The inputs to the program include
the ball and golf club head dimensions and mass properties, as well as the linear and
angular velocity components of the club head at impact expressed in terms of the
global coordinate system. Figure 6.6 illustrates a driver strike with the club path
from the inside to the outside of the player (angle βcz ), the attack angle upward
m 0 0 M 0 0 0 0 0 0 0 vx MVox
0 m 0 0 M 0 0 0 0 0 0 vy MVoy
0 0 m 0 0 M 0 0 0 0 0 vz MVoz
0 msz msy 0 0 0 0 0 Ixx 0 Ixz Vlx IxxΩx0 – IxzΩz0
0 0 mrb 0 0 0 Ib 0 0 0 0 Vly 0
0 0 1 0 0 – ks – rb 0 – kssy – kssx 0 = Vlz 0
msy msx 0 0 0 0 0 0 Ixz 0 – Izz ωxl IxzΩx0 – IzzΩz0
mrb 0 0 0 0 0 0 – Ib 0 0 0 ωzl 0
1 0 0 – ks 0 0 0 rb 0 kssz kssy Ωxl 0
msz 0 – msx 0 0 0 0 0 0 – Iyy 0 Ωyl – IyyΩy0
0 1 0 0 –1 0 0 0 – sz 0 sx Ωzl en(Vy – sxΩz0 + szΩx0)
Figure 6.5 Matrix form of the governing equations.

αh y
Xh
z y
δh
x αv
Ωx0
Ωz0
δv
V0 βcx
βcz
V0
Figure 6.6 Club path, face angles, and impact velocities for driver strike.
(angle βcx ), the face closed with respect to the club path (angle δ h ), and the strike
at the upper toe position (angles α h , α v ).
To carry out the impact analysis, it is necessary to impose the velocity compo-
nents of the CoM in the directions of the local coordinate system. These are
V0 x = V0 sin(δ h − α h )
V0 y = V0 cos(δ h − α h )cos(δ v + α v ) (6.21)
V0 z = V0 cos(δ h − α h )sin(δ v + α v ).

The program output for the high toe strike in row 6 of Table 6C is given in
Table 6.1.
Table 6.1 Calculated values for high-toe ball strike in Table 6E
Co-ordinates where applicable x y z

Initial club velocities, m/s 2.490 −44.574 −12.981
Initial club angular velocities, rad/s 52.397 0.000 −52.397
Final ball velocities, m/s −0.131 −66.610 −2.505
Final club velocities, m/s 2.516 −979.000 −12.469
Final ball angular velocities, rad/s 293.469 0.000 −15.320
Final club angular velocities, rad/s 94.827 −8.733 81.344
Vertical and horizontal launch, deg 14.683 −2.182
Moment arms, (sx, sy, sz), m −0.023 0.037 0.003
Mass, (m_b, M_c), kg 0.045 0.220
Ball props, (r_b, I_b), m, kg-sq.m 0.02134 0.0000082
Club MoI, (I_xx, I_yy, I_zz), kg-sq.m 0.00034 0.000297 0.00051
7
PU TT I NG
In this chapter, we look at ball trajectories on the green. The main intention is to
explore the physical factors that influence accuracy and distance control in putting.
Wherever possible the discussion is reduced to distances. Good golfers have devel-
oped the feel for putting desired distances on greens of a given “speed” and slope.
Even the consideration of the amount of curve of the ball on a sloped green, needed
to arrive at a particular hole position, is reduced to considering the position of a dis-
placed target and its distance sideways from the hole. This provides the target line.
The second consideration is the slope of the target line and the equivalent distance
of the putt if it was on level ground. With this information, the well-practiced player
can draw on his or her muscle memory to execute the stroke.
So whereas the physics is concerned with velocities off the putter face, the transi-
tion from slipping across the green to pure rolling, the rolling resistance of the turf,
the likely random deviations along the way, and the chance of capturing the ball if it
does happen to cross over the edge of the hole, the results all need to be in terms of
distances. That is the goal of this chapter.
E A R LIER IN V ESTIGATIONS A N D OTHER

SOU RCES OF IN FOR M ATION
Cochran and Stobbs ([1968] 1999) devoted only 14 pages out of 240 to their chapter
titled “Science on the Greens.” However, they still managed to impart considerable
light on a part of the game that had never been subjected to scientific enquiry. They
measured the time of contact between the putter face and the ball by making and
breaking an electrical contact between the two. Their measurements of 0.0006 s
for a 10 mph strike, and 0.00075 s for a tap-in putt, agree with more recent experi-
ments and modeling described in Chapter 3. They discussed the initial slipping
phase through the grass and the transition to pure rolling. They opined that with
the putter, the player could never impart enough topspin to significantly decrease
the slipping distance of the ball. They were clear that the slipping portion of the putt
is only part of the ball path to be affected by side spin. These issues and others are
discussed in detail in this chapter.
The Cochran and Stobbs investigators also considered the effect of off-center
ball strikes. They discussed the shortcomings of the putter designs of the day and
recommended moving much of the putter mass to the heel and the toe to increase
the vertical-axis MoI. They estimated that the loss of distance, with toe and heel
strikes, could be halved with a redistribution of half of the putter mass to the toe and
heel. They noted that very few putters of the day had this type of design feature and
suggested that “those that do, do so by accident.” One of the examples they gave was
201
an early Ping putter named the “Cushin,” which we know from the Chapter 1 was
no accident. The inventor, Karston Solheim, was a talented engineer working for the
General Electric Corporation. A few years later, according to Thomas (2011), the
next putter to be released by Solheim was the “Anser,” which became the most suc-
cessful putter ever in terms of sales and number of tournaments won.
Since the publication of Cochran and Stobbs, three articles have added consider-
ably to the fundamental understanding of the science and underlying principles of
this important part of the game. Holmes (1986, 1991) published detailed analyses of
the measurement of “speeds” of greens and of the necessary conditions for capturing
the ball in the hole. Penner (2002a) analyzed the rolling resistance of the golf ball
and the trajectories across sloped greens. There are also many texts and articles writ-
ten by teachers of the game and playing professionals. Two texts stand out for their
depth of coverage and the substantial knowledge and experience of the authors.
The Fundamentals of Putting, by Thomas and Melvin (2012), must be taken
very seriously. Thomas has a deep understanding of the game from his 26 years as
Technical Director of the USGA, and coauthor Valerie Melvin is a trained sports
psychologist who played international golf for Scotland. A visit to www.franklygolf.
com is worthwhile, both for general advice on equipment and information on the
putting instruction.
Similarly, Dave Pelz’s Putting Bible by Pelz (2000a) stands out because of his
detailed and exhaustive experimental investigations and the wealth of informa-
tion on which he has based his recommendations and conclusions. Pelz has some
interesting insights on the “unseen” factors influencing ball trajectory on the green.
These are discussed later in regard to the imbalance of the golf ball and the basic
instability of rolling “through” the grass.
Here the focus is solely on conclusions that can be reached from the fundamen-
tal physics of striking the ball and the resulting roll over the green.
SLIPPING A N D PU R E ROLLING
When a golf ball is struck on the putting surface, it always slips for an initial stage of
its travel. It starts in a condition of almost pure slipping: the absence of any signifi-
cant rolling. In this condition, the lower surface of the ball is travelling at approxi-
mately the same speed as the center of the ball. Friction between the grass and the
lower ball surface rapidly reduces the speed of the lower surface, while the speed of
rotation increases until the lowest point in contact with the ground stops. This is of
course the condition of pure rolling, which was discussed in the generation of spin
on the club face. Each stationary contact point in turn acts as the pivot for continu-
ation of the roll. It is quite difficult to visualize and presents countless students with
great difficulty in applying Newton’s laws to rolling balls and wheels. It has, how-
ever, large consequences for the behavior of putts.
The proportion of the slipping distance is approximately constant irrespective
of the putt distance. From examples in Search for the Perfect Swing (Cochran and
Stobbs), it appears that the decreasing slipping phase is typically about one-seventh
of the putt distance. This is illustrated in Figure 7A.
By definition, when a ball is in a condition of pure rolling along the ground,
energy is not being lost in sliding friction. Energy is, however, being lost in the
203 Putting
blade and
structure 1/7 of total travel
putter
impact
discretionary
weight
pure slip pure roll
Figure 7A Transition from slipping to rolling over approximately one-seventh of total travel.
contact deformation of the ball and/or the ground. With a golf ball rolling on the
green, this energy is lost in the deflection of blades of grass over which the ball is
passing. With very short grass, this allows the golf ball to roll for the surprisingly
large distances. Also due to this low resistance, gravity can cause large deviations of
the ball path on even the smallest of slopes.
When the ball is sliding, grass leaves are being deformed in the same manner,
but energy is also lost in friction between the sliding surfaces. So if a ball is started
along the ground in pure rolling with a given velocity, it will travel further than one
that starts at the same velocity in a slipping mode. This presents a small problem in
relating the way greens are tested using the “stimpmeter” to the speed and distance
of an actual putt.
There is some advantage to be gained if it was possible to strike a golf ball so that
it starts in the condition of pure rolling. A rolling ball, like a bicycle, has much more
stability than one that is slipping. For this reason, deviations of its track, resulting
from imperfections in the surface of the green, are diminished when the ball reaches
the point of pure rolling. This stems from the principle of gyroscopic stability pre-
viously discussed with respect to ball flight. Even with this stability, the deviation
of the ball due to minor, invisible imperfections in the green has a major effect on
the success rates of putts. As the ball slows toward the end of the putt, gyroscopic
stability decreases and the ball is influenced much more by turf imperfections or
small amounts of side slope. Pelz (2000a) concluded that PGA Tour players’ putting
strokes are so good that they should be expected to make 90 percent of 6 foot putts.
Instead, because of invisible imperfections in the grass, they only make 50 percent
of them. I believe that we can add to this the influence of very small amounts of
imbalance in the ball itself. This is discussed in a later section.
Sadly, analyses carried out in this chapter indicate that, despite claims in equip-
ment advertising and in the golf literature, it is impossible to apply any significant
topspin to a golf ball with the putting strike. Putters are usually designed with
between two and three degrees of positive loft, to strike the ball cleanly out of the
shallow nest its weight has compressed in the grass. This generates a small amount
of backspin, between 1.5 and 2 percent of the amount required for pure rolling, but
in the wrong direction! So any attempts at topspin generation have to first cancel
out the wedge-angle generated backspin. Unlike a pool ball, which can be struck
on the upper half to generate a large amount of topspin, attempts to do that with
the edge of the putter simply drive the ball into the turf. This is seen when a ball is
“topped” with an iron shot, and is knocked down into the turf, but springs out with
topspin and often rolls a surprising distance.
The player can attempt to strike the ball with an upward angle of attack to cre-
ate some topspin. However, it is clear that very little upward attack is possible for
a ball sitting on the ground. Cochran and Stobbs ([1968] 1999) determined, from
high-speed photographs of players using a loftless putter with the maximum pos-
sible 5-degree upward angle of attack, that a topspin of 2.2 revolutions per second
could be achieved for a 17 feet/s (12 mph) ball speed. However at 12 mph, the top-
spin rate in pure rolling is 39 revolutions per second, so this only creates 6 percent
of the topspin needed.
Another possibility is to use a putter with a CoM lower than the center of the
ball. This would give the putter a rotationally backward kick, which in turn through
the gear effect would produce topspin in the ball. Moreover, as we discussed in
earlier chapters with respect to the driver, the gear effect would be increased if the
center of mass was moved backward. A rear CoM is commonly the case in putter
designs. It is achieved by moving any discretionary weight back from the face as
shown in Figure 7A. Analyses, carried out later in the chapter, show that approxi-
mately 6 percent of the topspin required for pure rolling could be generated by a put-
ter in this way. However, to do this, the CoM would need to be halfway between the
ball center and the ground. This could conceivably be achieved using small heavy
discretionary weights of say tungsten carbide inserted on the underside of the rear.
Striking up on the ball would decrease the gear effect, so the two are not completely
additive effects.
The only published study of topspin in putting since Cochran and Stobbs is the
study of the “C-Groove” putter by Hurrion and Hurrion (2002). In comparison
with an unknown alternative (Brand X) putter, these authors showed an additional
32 degrees of ball “top” rotation in the first 0.04 seconds. This would be the result of
an additional topspin rate following impact of approximately (32 / 360)/ 0.04 = 2.2
revolutions per second; co-incidentally the Cochran and Stobbs value. These mea-
surements were taken using nominal 20 foot putts. We see in the next section that
for an average “speed” green (stimpmeter rating = 8), this would have involved
an initial ball velocity of 6.7 mph. The topspin rate for pure rolling at this speed
is 22 revolutions per second. So the C-Groove putter test produced approximately
10 percent of the pure rolling rate. From the preceding discussion, this is about the
maximum that could be expected from a combined upward stroke and gear effect.
It should be noted that the C-Groove putter has quite an aggressive upward facing
groove to provide the grip required for gear effect topspin. The effect, of even this
relatively small amount of topspin, on stability off the putter face should not be dis-
counted, nor should the small amount be considered a blemish on the excellence in
design or manufacture of the C-groove putters.
If the ball is hit off center on the putter blade, or with the putter face angled to
the club path, then a sidespin component will be imparted to the ball. This will only
cause the ball to swerve in its path when it is in the slipping phase. Moreover, even
though the ball will suffer an initial launch away from the intended target line, the
ground friction will quickly force the tilted spin axis to be level with the surface. In
other words, it will not curve increasingly further from the target line as for a slice
or hook, but it will roll true soon after the initial launch deviation.
205 Putting
Just as with all the other clubs, the major determinant of the launch direction
is the direction in which the face is pointing. The relationship is the same as for the
smooth faced driver, that is, between 80 and 85 percent of the face angle determines
the launch direction. For example, if the club path is directly toward the target, but
the face is angled from the club path by 5 degrees to the club path, the ball will
launch 4 degrees or slightly more away from the target. Therefore control of the face
angle toward the intended target must be a primary concern.
THE “SPEED” OF GR EENS

The “speed,” or more correctly the “rolling resistance,” of greens is measured using a
“stimpmeter,” named for its inventor Edward Stimpson. The stimpmeter is a grooved
plastic beam which has an indent, 30 inches from the end, and from which a ball is
released automatically when it is lifted to an angle 20 degrees from the horizontal.
The distance the ball travels in feet on a level part of the green is the “stimp rating.”
Stimp ratings typically range from as low as 4 (feet) for some municipal courses, to
as high as 12 (feet) for premier courses. Major championship greens are a special
case and may often exceed a rating of 12.
Instructions for using a stimpmeter can be found on the USGA website (www.
usga.org), which includes a metric for using upslope and downslope measurements
to determine an equivalent horizontal roll distance on sloped greens. The use of the
stimpmeter seems to be a matter of some controversy in course management. An
interesting article by Brevard (2010) discusses the problems of green speed compe-
tition between course superintendents and the concern of focusing on green speed
rather than the health of the turf.
The ball leaving the bottom of the stimpmeter, as shown in Figure 7B, transi-
tions immediately from the impact with the turf to a state of continued pure roll-
ing. Its speed at that point is 4 mph. We can use this information to calculate the
rolling resistance and the effect of different putting speeds. At the simplest level, we
know that a golf ball rolling at a speed of 4 mph on a level green will roll 4 feet on
a 4-stimp green, 6 feet on a 6-stimp green, and so on. Although the values will be
slightly in error because of the initial slipping phase, we can nevertheless use these
speeds and others derived from them to get a good approximation of launch speeds
30 i
nch
es
20 degrees
impact with turf
Figure 7B Use of a stimpmeter angled at 20 degrees to the surface of the green.

for particular putts. This brings us to the central difficulty in putting, namely, hit-
ting the ball at the right speed to make the required distance. Most putting coaches
will advise that the principal reason for 3-putting is poor judgment of distance on
the first putt. Error distances to the side, produced by misjudging the target line,
are generally of much lower magnitude. We show a comparison later of target line
errors versus impact speed errors when putting across slopes. However, there is a
fundamental difficulty in controlling distance in putting that stems directly from
Newton’s laws. This results from the difference between how the ball is accelerated
from the ball strike and how it is slowed to a stop when rolling across the green.
We know that the ball is accelerated during contact with the putter face and sep-
arates from the putter with a speed that is directly proportional to the putter impact
speed. From Cochran’s (2002) investigations in the Callaway Laboratories, the
CofR for a metal-face putter is likely to be approximately 0.9. The typical putter head
weight is 0.77 pounds. So following the discussion in Chapter 3, we have a CofR
speed factor of 1.9 and a mass speed factor for the 0.1 pound ball of 0.77/(0.77 +
0.1) =0.885. Thus the “smash factor” for the putt is typically 1.9 × 0.885 = 1.68, and
each mile per hour increase in putter speed will add 1.68 mph to the launch speed.
In contrast with launch speed generation, the slowing of the ball across the green
is due to the steady resistance produced by the deflection of the grass blades as the
ball rolls over the turf. For a given stimp rating, this resistance can be assumed to
be reasonably constant throughout the roll. The “kinetic” energy of the ball as it
leaves the putter face is proportional to its velocity squared. As it transitions from
slipping to pure rolling, the energy is shared between linear and rotational kinetic
energy, but it is still proportional to the forward velocity squared. It loses kinetic
energy because of the energy expended in deforming grass blades. This energy loss
can be represented precisely by the resisting force multiplied by the roll distance;
that is, in energy terms the ball is doing work against the resistance of the grass. The
end result of this energy balance is that the putt distance is proportional to veloc-
ity squared. This has the profound consequence that errors in putter impact speed
have a magnified effect on putting distance. For example, a 15 percent overspeed
error in the impact speed required for a 20 foot putt would increase the putt length
to 1.152 × 20 = 26.5 feet; a 33 percent distance error that would leave a 6.5 foot sec-
ond putt that would be missed 50 percent of the time even by professional players
(Pelz 2000a). With an underspeed error of the same magnitude, the resulting roll
would be 0.852 × 20 = 14.5 feet, with a still-difficult second putt. As we will see later,
this relationship also has quite a profound effect on the results of different putting
swings.
The relationship can be represented by a set of parabolas for the different stimp
ratings as shown in Figure 7C. It seems that physicists have failed to discuss this
relationship in the golf literature, possibly because to them it is so obvious. However
there is another important consequence of this relationship, which can be illus-
trated by the 6.5 foot second putt in the preceding example. If the stimp rating was
say 10, and the green was level, then we know that a ball rolled at 4 mph would roll
10 feet. Since the putt distance is proportional to speed squared, the speed required
to just make the 20 foot putt would be 5.66 mph; that is, (5.66/4)2 equals the required
factor 2 in distance. After striking the ball in error at speed 1.15 × 5.66 = 6.5 mph,
the player is now left with a 6.5 foot putt. However, the required speed to just make
207 Putting
20
18
16
stimp value*
14
12*
Length of putt, ft 12
10*
10
8*
8
6*
6
0
0 1 2 3 4 5 6
Launch speed, mph
Figure 7C Putt length versus ball launch speed on level greens.
this second putt is now 3.22 mph, since (3.22/4)2 = 6.5/10 as required. Thus almost
60 percent of the speed, required for the initial 20 foot putt, is now needed for the
6.5 foot second putt. Combine the need to still strike the ball with conviction and
the fear of an embarrassing third putt from another overrun and you may get the
start-then-hold-back stroke known in all ball sports as a “choke,” and by some
thought to be the cause of the “yips,” which can affect even the highest performing
players.
Before leaving this topic, it should be noted that the speed-squared versus dis-
tance relationship applies just the same to putting up or down a steady incline. In
this case, the resisting force is still constant since it is reduced or increased by the
constant gravitational effect acting down the slope.
Mention should be made of the MoI of the ball and its possible effect on the
difficulties of putting. If the MoI is reduced, with a heavier inner core and lighter
outer layers, then the rate at which the ball is slowed when rolling across a green
will be increased. In principle this could make it easier to control putting dis-
tance. However, from relationships developed later in the chapter, we show that
the changes cannot be significant. If it was possible to decrease ball MoI by say
25 percent, the ball would in fact only slow down 8 percent more quickly. For the
same launch speed off the putter face, a green with a stimp rating of 12 would play as
if it had a stimp 11 rating. However, the smaller MoI ball, just like a smaller wheeled
bicycle, would have less rolling stability. In consequence, it would be deflected more
easily by tiny imperfections in the turf and by gravity on sloping greens. From the
work by Pelz (2000a), it seems that the former would likely outweigh the advantage
of the reduction in the “speed” of the green. Note that the difference in apparent
speed of the green would not be detected by the stimpmeter because the lower MoI
ball would accelerate faster down the ramp, and the increased launch speed would
make the ball roll almost exactly the same distance. In contrast, the change in MoI
of the ball has no significant effect on the launch speed off the putter face.
If, alternatively, the MoI of the ball was increased by 25 percent, then a stimp
12 green would play like a stimp 13 one, but with some improvement in rolling sta-
bility. However, a larger MoI would certainly make distance control more difficult.
This would seem to end any ideas for making a better-putting golf ball.
THE M ECH A NICS OF LAU NCHING THE BA LL

To illustrate the most logical way to control distance in putting, it is useful to con-
sider the stimpmeter. It is actually the perfect instrument for “putting.” It is shown
in the later science modeling that including the effect of the ball impacting the
ground at a 20 degree angle, the launch speed squared is proportional to the dis-
tance up the stimpmeter from which the ball is released. This is to be expected since
the roll down the stimpmeter occurs under a constant gravitational accelerating
force; that is, the relationship is just the opposite of the constant decelerating force
of the turf. Gravitational force multiplied by distance down the stimpmeter equals
energy gained. Grass resistance multiplied by roll distance equals the same energy
lost. This is illustrated in Figure 7D, where twice the distance up the stimpmeter
produces twice the rolling distance along the green.
So we could imagine a stimpmeter with ball-roll distance scales marked along it
for different stimp ratings. It would require a steeper angle or very long stimpmeter
to cover the range of putting distances usually encountered. Adjustment would still
need to be made for upslope or downslope “putts,” which would just be a small ramp
angle change according to the amount of downslope or upslope along the target
line. Of course, this is not a serious proposal. It is just a preliminary discussion for
considering the putting swing.
In Search for the Perfect Swing the authors studied the putting styles of profes-
sional golfers, taking high-speed video of 16 “first-class” players, in an attempt to
identify common features. They found only the one common feature, namely that
most of them placed the ball opposite the left foot and had their eyes almost directly
above the ball. As for the swing, they found “long backswings and short backswings,
long follow-throughs and short follow-throughs,” “all combinations of length of
backswing and length of follow-throughs,” and “both speeding up of the club head
through the ball and slowing down.” They described the range of swings from the
“smooth flowing stroke” to the “short sharp tap.” All of these styles can still be seen
on televised Tour events. We will see how the physics of the launch of the ball and
//
A //
B B A
/// ///
Figure 7D Roll distance is proportional to start position on the stimpmeter.

209 Putting
the resulting roll of the ball might lead us to simple extremes that could accept these
descriptions.
If we think of the performance of the stimpmeter as an analog for putting desired
distances, then we might be led to the following concept. A putting swing should
be one for which different swing arc lengths produce proportionate changes in putt
distances. This brings us to the simplest putting model as shown in Figure 7E. In
contrast to the full golf swing, which has the characteristics of a double pendulum
to generate maximum speed, the putting swing should have the one characteristic
of being mechanically simple, and this is properly represented by a single pendulum.
By its simplicity, this allows better control over club head speed. We see later that
even very young children adopt a stiff-arm pendulum swing when asked to do accu-
rate underarm throws. The model shown in Figure 7E can represent either shoul-
ders, arms, and putter moving as a single piece, or a long putter anchored somewhere
on the chest with the left hand. If a constant average torque is applied to this model,
irrespective of the swing distance, then we are back to the same relationship as for
the stimpmeter. Now, instead of the relationship being between the gravitational
force and distance along the stimpmeter, this is the relationship between torque and
the swing angle. Torque multiplied by angle, just like force multiplied by distance,
is the input of energy. If the angle is doubled, and the same average torque is main-
tained throughout the swing, then the energy input has doubled and the kinetic
energy will have doubled. Thus the speed-squared of the putter at impact will have
doubled. The launch speed squared will also have doubled since the smash factor
can be considered to be the same for both ball strikes. This will result in a doubling
of the distance. We will refer to this as the constant average torque model.
The practice required to keep a constant average torque through the putting
swing would be to putt alongside a scale with marks at say 4 inch intervals. The
scale would then be used to position a block as a backstop for required lengths
constant
torque
B A B
A
Figure 7E Application of constant average torque, over different swing lengths gives
proportional roll distances: level, upslope, or downslope.
of backswing. The practice would involve putting to equi-spaced distances from

equi-spaced positions of the backstop. Practicing in this way would indirectly teach
the brain to apply approximately constant average torque over different swing
lengths. Figure 7E is illustrated on an upslope to emphasize that the equal back-
swing to putt length relationship remains intact for constant up- or downslopes.
The main reason for presenting the model is that it represents the simplest possi-
ble relationship between the physics of the ball strike and the physics of the roll. It
also seems that it may represent one extreme of the range of putting styles that are
seen on the golf course; that of the long “smooth flowing swing.” If, for example, a
15 degree backswing (approximately 15 inches backward along the ground) was
used for a 12 foot putt then a long 30 degree backswing (approximately 30 inches
backward along the ground) would be required for a 30 foot putt.
The other extreme of putting styles would be a constant relatively short back-
swing, with varying average torque applied for different length putts. For the longer
putts this would fit the category of the “short sharp tap.” In this case, the varying
average torque, over the same swing length, gives the same variation of energy
input, and so increases the speed-squared of the putter head proportionately. So we
are back to a one-to-one relationship, in this case between average applied torque
and putt distance, as illustrated in Figure 7F. The practice in this case would involve
putting sets of different distances with a fixed backstop; the backstop set at a dis-
tance that feels best for the player for both short and long putts. We will refer to
this as the constant swing length model. The main issue with this model is that of
accelerating from a reasonably short backswing, appropriate for short putts, but suf-
ficiently powerful for the long putts. The key to do this is to focus on the length of
the through stroke, although there may certainly be a point beyond which the back-
swing has to be lengthened. However, a significant advantage of this style is that
from a short backswing it is easier to keep the clubface square at impact, which is of
course of primary importance.
It should be noted that the effectiveness of any variant of the putting swing is
extremely difficult to measure. Because of the variability of the individual player’s
swing, and the wide variation across any group of players, this can only be carried
torque
torque B = double torque A

A B
approx.
constant
Figure 7F Application of proportionately more torque over a constant swing length gives
proportional roll distances: level, upslope, or downslope.
211 Putting
out by statistical analysis of large numbers of player tests under very careful control.
So the intention here is simply to offer alternatives for discussion, based only on the
physics of the strike and roll of the golf ball.
The method suggested most in the literature is in some way the middle ground
between these two extremes. The relationship between the swing and the roll dis-
tance is more complex than for the other two, since in this case swing length is
increased less for longer putts than for the constant torque model, but the average
torque must also be increased as putt length increases. So it involves manipulating
two variables instead of one for the previous two models. This is the “constant swing
time model.”
This method is based on constant total swing time, combined with constant
tempo (backward to forward swing time ratio). The swing time that feels “natu-
ral” to the particular player is usually recommended. This is somewhat qualitative
but probably depends on the build of the player, and particularly on the length and
bulk of the arms. The basis for this method is that the putting swing is not merely
modeled well by a powered simple pendulum, but it should actually behave like a
clock-spring-driven pendulum, which just like a freely swinging gravity pendu-
lum, has a constant swing time irrespective of the swing angle. We will discuss the
method first then consider its positive and negative attributes. Many proponents
of this swing style can be found in any on-line search. Two of the proponents with
great credibility are Pelz (2000a) and Thomas and Melvin (2012).
Thomas and Melvin (2102) suggest that golfers should precede their tryout of
the constant time model by swinging the putter with one hand, and preferably with
the right hand for a right-handed player. The swing should start from the address
position and swing naturally back and then forward through the ball. They advise
that this gets the player into “the ballpark of the correct ‘rhythm’ and close to what
should be used for all putts.” Of course in the actual putting swing a larger aver-
age torque would be required to obtain the same rhythm as the one-arm “natural”
rhythm. As for tempo, they suggest that the backswing should usually be about twice
the forward swing time up to impact. In contrast Pelz (2000a) suggests a tempo of
one to one, just like the clock pendulum. An illustration in his book shows a golfer
practicing “tick” then “tock” swinging back and forth between two fixed stops.
For this method, practice with a metronome is sometimes suggested. But it
probably should be preceded by first finding the rhythm that feels best to the indi-
vidual player as recommended by Thomas and Melvin. The practice procedure is to
take away the putter on the “tick,” execute the backswing during “tock” then “tick,”
and strike the ball on the following “tock”; or maybe better expressed as: from
the takeaway, count “tick,” “tick” for the backswing, then “tock” at the ball strike.
Alternatively, following the Pelz (2000a) advice it would be simply back on the
“tick” and forward to strike the ball on the “tock.”
A clock or gravity pendulum is subjected to the maximum torque at the
beginning of the swing when the pendulum weight is most offset from the pivot.
The torque then decreases to zero at the bottom of the swing where the speed of
the pendulum weight is greatest.
It is easy to make a freely swinging pendulum putting device, as used in The
Search for the Perfect Swing investigations, or by Pelz (2000a) in the investigations
for his Putting Bible. Such a device will certainly launch the ball straight off the face
but it contains a serious weakness when it comes to putting accurate distances. This
stems from the fundamental physics of the pendulum.
This maximum speed of any freely swinging pendulum, at the bottom of its
swing, is governed by the root of (1 – the cosine of the swing angle). By the prop-
erties of the cosine function this can be very closely approximated by just using
the swing angle for angles up to 30 degrees. That is, the pendulum speed increases
directly with the swing angle for the range of swing angles used in putting. If the
speed for a swing angle of 5 degrees is set to 1.0, then for swing angles of 10, 15, 20,
25, and 30 degrees, the corresponding speeds calculated precisely are 2.0, 3.0, 4.0,
5.0, and 5.9 to the nearest single decimal; almost exactly in proportion to the swing
angle. Assume that the 5-degree swing of a pendulum putter produces a 3 foot putt.
So in this case using the putt length to velocity squared relationship, the putt lengths
for the six equal-spaced swing angles of 5, 10, 15, 20, 25, and 30 degrees would be
3, 12, 27, 48, 75, and 108 feet respectively. This is not a very user-friendly scale with
which to do accurate putting to different distances.
More generally for any player, constant time for the swing back and forth with
a constant tempo, making the reasonable assumption of the same torque variation
throughout the swing, will always give an approximate distance-squared relation-
ship to different swing lengths. This property is illustrated in Figure 7G.
An example of putt distance to swing length for the constant time method is
given by Pelz (2000a). He advises that a 6-inch backswing is “about perfect for a
2-foot putt on most greens,” a “12-inch backswing is” about what you would make
for a 12-foot putt, and “for a 25 footer 18 inches behind the ball” is a “good refer-
ence stroke.” With a pendulum length of about 60 inches, each degree of swing is
approximately one inch of backward movement of the putter head. With our pre-
dicted “scale,” starting at 3 feet (including a 1-foot overrun for the 2-foot putt) with
a doubled 12-inch backswing, we would predict 22 × 3 = 12 feet. For an 18-inch
backswing, 3 times the initial one, we would expect a putt distance of 32 × 3 = 27
feet. These are in reasonable agreement with Pelz’s reference putt lengths, which
with his suggested overrun distances would be 13.5 feet and 26.5 feet.
The fundamental problem with this non-linear relationship of putt distance to
swing length is that it magnifies errors. Since putt length is proportional to swing
angle squared, a +10 percent error of only +1.5 degrees in setting the required
B B
A A
putt B = 4 × putt A
equal times
Figure 7G Application of proportionately more torque over increasing swing lengths gives equal
time swings and roll distances increasing with the square of the swing length.
213 Putting
15 degree swing angle for a 27 foot putt as just discussed, will produce a (1.1)2 = 1.21,
or 21 percent, increase in distance for an overrun error of 5 feet 8 inches. With the
constant torque or constant swing length model the overrun error for a +10 percent
error in torque or swing length would be 10 percent. This doubling of errors from
swing length to putt length for a fixed time swing is completely general. We can
write for the constant time swing:
putt length = a constant value multiplied by swing angle squared
where the particular constant value depends on the speed of the green.

The swing angle (or equivalently swing length) is taken to be subjected to
random errors. In statistics the standard measure of the error band or the level of
variation of a parameter is the “standard deviation.” If the standard deviation of the
swing angle for a particular player is SD-angle, then because of the squared rela-
tionship we get the general result that SD-putt-length = 2 × SD-angle. This stems
from a branch of statistics call “error propagation,” which is nicely described by
Navidi (2011).
It is very interesting that the advice of a 2:1 tempo by Thomas and Melvin
(2012) corresponds with an insightful mathematical analysis of the putting swing
by Grober (2011). Grober found that with reference to the pendulum model, errors
in impact speed resulting from variations in the driving forces are minimized
if: (a) the average magnitude of the applied torque produces twice the natural pen-
dulum frequency, or equivalently half of the swing time; (b) the tempo of back-
swing to forward swing times is 2 to 1. The constant swing time was built into the
modeling, and experimental measurements of one professional golfer “whose put-
ting stroke is representative of those with the properties” described in the analysis
showed almost complete agreement with the predicted velocity profile during the
swing. So the results support the advice of Thomas and Melvin (2012) provided a
constant swing time has been adopted. It does not address the more general issue of
putting distance errors, which was the reason for setting up the other models. It may
be noticed that setting up a rhythm with the natural swing of one arm and the club
will mean that when the golf swing is used with the same rhythm it will be driven
at a higher speed than for its natural rhythm, in qualitative agreement with finding
(a) by Grober.
Thomas and Melvin (2012) compare the swing to the feeling of throwing a ball
underarm. This is a good analogy to the swing. However, the physics of underarm
throwing presents a more complex relationship between the swing and distance for
the thrower because the launch angle also changes with the swing length. The point
in this off-topic excursion is that despite the more complex relationship of swing
to distance, we seem to have an innate ability to throw balls underarm with preci-
sion. This probably arises from the survival skill of early humans to throw stones
and later spears with precision; a task that involves the same judgment concerning
(overarm) swing speed, launch angle, and distance. This innate skill was shown in a
very interesting study on the “underarm throw accuracy of children” (Van Rossum
and Bootsma 1989) that demonstrates quite remarkable ability in children as young
as 4 to 6 years. In the study, groups of 4- and 6-year-olds, without any prior tuition,
were asked to toss balls underarm at a target. The authors categorized 4 types of
swings: (1) arm kept stretched; (2) arm bent, but bend is kept fixed; (3) arm in con-
tinuous bending throughout the swing; and (4) arm is stretched first and then bent
during the throw.
For the 4-year-olds, two-thirds of the group were in categories 3 or 4. At 6 years
old, also with no tuition in precision underarm throwing, 70 percent were in cat-
egories 1 and 2. That is, the 6-year-olds had simplified their swings, reducing the
“degrees of freedom” to that of a simple pendulum for increased accuracy. A good
question might be: what strategy is most naturally adopted by young children asked
to throw, or better still roll underarm at predefined targets. Since humans have been
perfecting throwing skills from about 2 million years ago, the hardwired strategy
might provide some very useful information. If they adopt near equal time swings,
that would be much stronger justification than the clock pendulum. More detailed
modeling of this topic is given in Appendix B.
CA PTU R ING THE BA LL IN THE HOLE

Putting coaches advise players to always putt to reach a point some distance past
the hole. The typical suggested overrun distance is somewhere between 1 foot and
18 inches. This strategy reduces the number of short putts, which are never given a
chance of being made. If missed, it also leaves a second putt that is short enough to
be made with confidence. However, if the chances of making the putt are very slight,
then it may be better to try for the exact distance to the hole, since this would reduce
the likely length of a second putt.
A second consideration arises when we follow the strategy to aim for a target
past the hole. That is the conditions under which a ball will be captured if it strikes
the front edge of the hole with the speed required for the overrun. This quite amaz-
ingly complex problem has been put to bed in the most complete manner in a single
paper by a single author (B.W. Holmes 1991). Holmes considered all possible condi-
tions for capture and escape. His statement of what happens to the ball after passing
over the rim explains perfectly the difficulty of the problem:
“When the ball first encounters the rim of the hole, it will either lose contact with
the rim and enter ballistic motion, or it will continue to roll on the rim. If the ball
is rolling on the rim, it will continue to do so until it escapes rolling or is captured
rolling, or until it loses contact with the rim. If the ball is in ballistic motion, it will
continue until it is captured by the hole or until it strikes the rim again. When it
strikes the rim, a no-bounce, no-skid approximation is invoked. Having struck the
rim, the ball will either continue rolling on the rim or lose contact with the rim. If
it loses contact a second time, it enters ballistic motion, which will be terminated
either by capture, escape flying, or striking the rim again.”
Whew! Figure 7H illustrates one of these alternatives, namely, a ball rolling around
the rim. The ball reaches the edge of the hole at a speed that will be referred to as the
“crossing speed.”
The diameter of the standard hole is 4.25 inches, approximately two and a
half times the 1.68 inch ball diameter. The only parameters of importance in this
215 Putting
escape
center of
ball arc
crossing
speed
offset
Figure 7H Ball reaching edge with too much speed to be captured.
problem, in addition to mass and MoI of the ball, are the velocity with which the
ball crosses the edge of the hole and the offset position from the center. In the par-
ticular scenario illustrated, the horizontal component of force, from edge contact,
forces the ball to enter a curved rolling path around the edge of the hole. For the
ball to be captured, the radius of curvature of the path must reduce to that of the
hole radius. With that radius, or less, it will roll around the rim until slowed enough
to drop inside. The higher the ball speed, the greater will be the initial radius of
curvature of its path around the rim, and thus the higher the likelihood that the
ball will escape.
Holmes’ results of analyzing the variety of escape mechanisms, from becoming
airborne off the back edge, to rolling around and then out from the lip, can be sum-
marized in a single diagram as shown in Figure 7I.
The position of the ball on the left in Figure 7I is at the maximum offset from
the center at which it will be captured with the given crossing speed. In this case,
hole diameter
effective
diameter
3.65 mph
crossing
speed
speed profile
for capture ball would escape ball will escape unless it
outside of effective diameter stops and rolls over sideways
Figure 7I Effective decrease of hole diameter for a given edge crossing speed.

any offset position, between the one shown and the same position on the opposite
side, will result in capture. This span across the center of the hole is the “effective
diameter” for the particular crossing speed. For the effective diameter to equal the
actual hole diameter, the ball would need to be captured when it rolls exactly in line
with either far edge of the hole. The only way this will happen is if the ball just slows
to a standstill at that point and topples over sideways. Because of its rolling stability,
any small forward velocity would take it on past the edge. This scenario is shown on
the right of Figure 7I.
In contrast, the maximum crossing speed for a ball exactly on the centerline is
3.65 mph. At this speed the ball would of course fail to be captured for any slightest
offset from the center; that is, the effective diameter has decreased to zero. Moreover,
any ball approaching the hole in excess of 3.65 mph will never be captured.
This provides an understanding of the problem of the diminishing hole size, but
its definition in terms of speed is not particularly helpful to the player. Much more
useful is knowledge of the effective hole size for putts intended to go, say, 12 inches
past the hole on different stimp-rated greens. To answer this question, we first need
to discuss the speed profile across the front edge of the hole that would ensure cap-
ture in each offset position. It turns out that the profile, which is actually composed
of intersecting segments for the different escape mechanisms, is almost precisely
a parabola as shown by Penner (2002a). This parabolic distribution of crossing
speeds is superimposed in Figure 7I.
We first consider two examples to illustrate a very surprising result. On a very
high “speed” stimp 12 green, a 4 mph putt will take the ball 12 ft. So the cross-
ing speed required to travel one foot further is such that the ratio of the crossing
speeds squared must equal 1/12 and so equals (1 /12 )(4) = 1.15 mph. The effec-
tive diameter from the parabola relationship for this crossing speed is 3.45 inches,
which is just over 80 percent of the full 4.25 inch diameter. However, if we are play-
ing on a municipal course with slow stimp 6 greens, the crossing speed to travel
another 12 inches is (1 / 6 )(4) = 1.63 mph. The corresponding effective hole
diameter in this case is 3.05 inches, just over 70 percent of the hole size. Figure 7J
illustrates the general relationship, for different stimp ratings, between distance
past the hole for a miss and relative hole size. The preceding example is illustrated
by the arrows.
These results present a compelling argument for putting larger diameter holes
in slower greens intended for recreational play. The changes to allow equiva-
lent hole capture for 18 inch overrun putts are in fact quite modest as shown in
Table 7A.
These changes would probably make very little difference to the scoring rate of
long putts. In Search for the Perfect Swing (Cochran and Stobbs [1968] 1999), experi-
ments were reported for an 8.5 inch hole, double the normal size. For skilled play-
ers, they found that the average number of putts from 6 feet went for 1.56 to 1.12,
that is, approximately 90 percent chance of making 6 foot putts. However, from
15 feet, the improvement was quite modest, going from an average of 1.88 putts to
1.60 putts. For a high-handicap player, it is likely that the change would make insig-
nificant differences from 15 feet. However, it would of course dramatically decrease
the number of 3-putts.
217 Putting
4.5
4.0
Effective hole size, inches stimp value*
3.5 10* 12*

8*
6*
3.0
2.5
0 3 6 9 12 15 18
Distance past hole for miss, inches
Figure 7J Effective hole size for varying amounts of overrun for different stimp ratings; arrows
indicate different effective hole sizes for 12″ overruns on stimp 6 and stimp 12 greens.
PUTTING U P OR DOW N THE SLOPE

Next we need to address the effect of slope on putts. The easiest situation to
deal with is when the putt is directly up- or downslope. This is illustrated in
Figure 7K, where two forces are acting on the ball, which is in a condition of pure
rolling. The vertical gravitational force on the ball has a component acting down
the slope, which is labeled “gravity.” Opposing this is the resistance to rolling of
the grass, which acts uphill. We can assume that the grass resistance is always
more than the component of gravity acting down the slope, otherwise the ball
Table 7A Equivalent hole sizes for equal capture potential of 18 inch overrun putts
Stimp rating 5.00 7.00 9.00 11.00 13.00

Equivalent hole size (in.) 5.38 4.82 4.54 4.36 4.25
Hole size increase (%) 26.5 13.4 6.8 2.5 0.0
gravity
velocity
grass resistance
slope
Figure 7K Golf putt directly downslope.

Table 7B Maximum slopes for different stimp ratings
Stimp rating 5 7 9 11 13
Max slope (%) 15.7 11.2 8.7 7.1 6.0
Max slope (deg) 8.9 6.4 5.0 4.1 3.5
Table 7C Equivalent flat distances for up- and downslope putts: stimp 7; slope 4%
Upslope Downslope
Slope distance (ft.) 21.0 16.0 11.0 11.0 16.0 21.0

Initial velocity (mph) 8.3 7.2 6.0 4.1 5.0 5.7
Flat distance (ft.) 28.5 21.7 14.9 7.1 10.3 13.5
would continue to roll at increasing rates or stop and roll back for an uphill putt.
The maximum slope values for balls to slow and have a chance of stopping, for
different stimp ratings, are given in Table 7B. From available data concerning
putting on sloped greens, it seems that a 4 percent slope is the largest likely to be
encountered, except on intentionally steep approaches or side slopes.
For upslope or downslope putts, the most helpful data for the player is the “equiv-
alent flat” distance to the hole. Table 7C shows these equivalent flat distances for
three different putt distances on both a steep (4 percent) upslope and a downslope
with a stimp rating of 7. The holes are at distances 10, 15, and 20 feet, with the putts
intended to go 1 foot past the hole if missed. It can be seen that even with a relatively
low stimp rating, the effect is pronounced.
The first column shows that the 21 foot upslope putt should be putted as if it was
a 28.5 foot level putt. The speeds in the center row are part of the calculation and
are included for comparison with the initial speed for flat putts of 11, 16, and 21 feet,
which would be 5.1, 6.2, and 7.1 mph, respectively. These values and others shown
later were obtained from equations developed in the later modeling section of the
chapter.
To give an indication of how difficult it is to putt on championship greens,
Table 7D gives the same data, again from the equations developed later, but for a
stimp rating of 13. Note particularly that a 21 foot downslope putt must be exe-
cuted as if it was a 7 foot putt on level ground.
Table 7D Equivalent flat distances for up- and downslope putts: stimp 13; slope 4%
Upslope Downslope

Initial velocity (mph) 6.7 5.9 4.9 2.2 2.6 3.0
Flat distance (ft.) 34.9 26.6 18.3 3.7 5.4 7.1
219 Putting
PUTTING ACROSS SLOPING GR EENS

If we consider making the same distance putts as in Tables 7C and 7D but angled
across the slope, we have to decide on both the equivalent flat putting distance and
the initial putt direction. We refer to the initial direction as the target line, which
simplifies the problem to putting to a particular point on the green.
Figure 7L illustrates the ball trajectory for putts at 30 degrees to the upslope
direction, at 30 degrees to the downslope direction, and directly across the slope.
The dashed lines show the equivalent 21 foot putts on level ground. The end points
of these lines are the target points for the three long putts; each involves estimat-
ing both an equivalent flat distance to the hole and the required offset to the target
point. The speed, distance, and offset values for these nine putts are given later in
the modeling section. Note that none of these putts are ones a player would expect
to make. The average number, of even the shorter 10 foot putts, that are holed by the
average amateur player is less 10 percent. As mentioned previously, the objective is
to give the ball a chance to be captured by aiming beyond the hole, but to try to leave
a safe follow-on putt.
One surprising aspect of making such putts is the existence of a target line indi-
cator point which always lies directly upslope from the hole. This is shown for the
three 21 foot putts in Figure 7L. If the putting start point for any of these three
putts was moved sideways left or right, then the correct line still passes very closely
through the indicator point. This is discussed by Thomas and Melvin (2012) and
supported analytically by Grober (2011b). So it can be advantageous to take a few
steps to each side of the ball in order to identify the indicator point with more than
one observation.
To complete our discussion, we will apply the modeling equations to test two
often heard comments about putting:
1. Speed is more important than the target line.

2. Straight putts downslope are easier to make than straight upslope putts, since
errant downslope putts turn toward the hole, while upslope ones turn away.
Slope 4 percent (2.3 degrees)
20
Stimp rating = 7
15
10
0
–20 –15 –10 –5 0 5 10 15 20 25
Distances in feet
Figure 7L Trajectories predicted to roll 12 inches beyond the holes for 10, 15, and 20 foot putts
on a 4 percent slope with a stimp rating of 7 feet.
Slope 4 percent (2.3 degrees)
Stimp rating = 7 Stimp rating = 7

25
50% direction 10%
errors
speed
20
errors
15
10
0
–5 0 5 –5 0 5
Distances in feet
Figure 7M The effects of ± 50 percent errors in the initial trajectory (left) compared to the
effects of only ± 10 percent errors in the putt speed (right).
The first is easy to test by introducing speed and initial direction errors sepa-
rately into selected putts.
Figure 7M shows one set of error comparisons. The plot on the left shows the
effect of errors in selecting the correct offset value. The required break for this putt is
predicted to be 37 inches to the right of the hole as indicated by the dot in the figures.
The two missed putts were started with breaks of 18 inches and 56 inches, respec-
tively. The longest second putt for these major miss-reads, of ± 50 percent in the
estimated break, is less than 3 feet.
In contrast, the plots on the right of Figure 7M are the result of only ± 10 percent
errors in the desired initial speed of 7.2 mph. Even the shorter putt leaves a second
putt of 3 feet, while the longer one has a 6 foot return putt down the slope. So there
is no doubt that advice No. 1 has great validity. Note that the putts are ± 4.5 feet
from the intended target at 21.5 feet. This illustrates the point made earlier that if
the putt is very unlikely to be made then aiming for the exact distance to the hole is
possibly a better strategy.
Figure 7N shows the trajectories of upslope and downslope putts on a 4 percent
slope, fast championship green, with a stimp rating of 13 feet. In these cases the target
is the hole. Putts from left to right are upslope. All four putts have a 5.0 degree error in
initial launch direction. Putts started to the left of the target line have a +10 percent ini-
tial speed error; those started to the right have a −10 percent speed error. As expected,
the effect of gravity on the downslope putts is seen to move the ball toward the hole,
while the effect is opposite for the uphill putts. Even highly skilled professional players
make direction errors of this magnitude by misreading the slope direction. The issue
221 Putting
Slope 4 percent Stimp 13

5
–5
–15 –10 –5 0 5 10 15
Distances in feet
Figure 7N The effects of ± 5 degree errors in the initial trajectory; putts to the left of the hole
have +10 percent speed error while those to the right have -10 percent speed error.
here, of course, is speed control. In this case, the launch speed to reach the hole for the
upslope putt is 5.7 mph, so a 10 percent speed error is 0.57 mph. For the downslope
putt, the perfect speed is 2.55 mph, thus the error limits in this case are only 0.26 mph.
The same error of +0.57 mph for the downslope putt as for the upslope one would in
fact leave the ball 7.5 feet below the hole, with a high chance of a third putt, even for a
professional player. This is clearly the reason why players favor the upslope first putt.
This suggests strongly that advice No. 2 is not good, while reinforcing the value of
advice No. 1.
EFFECT OF I M BA LA NCE OF THE GOLF

BA LL (W ITH DAV ID M A R SH, 2013)
This issue was brought to my attention by Dave Marsh (2013), who has developed a
proprietary method for assessing the degree of imbalance of golf balls. His scale of
balance ranges from 1 to 10, with 10 representing the most perfectly balanced balls
he has found and 1 representing the most out-of-balance ones. His contribution to
this topic was acknowledged at the beginning of the book.
It has long been suggested in the popular golf literature that golf balls diverge
from spherical symmetry by amounts sufficient to affect the accuracy of putts.
This conjecture has never gained widespread acceptance due to the absence of any
published investigations and the disinclination of golf ball manufacturers to make
claims of purer rolling golf balls. However, there is a serious fringe interest in this
topic. Most commonly this involves the floating of golf balls in a salt solution and
the marking of the center of the spherical crown above the liquid. If the marked
point then returns to the top, after the ball has been turned to a different position,
the ball is deemed to be “unbalanced.” That is, the CoM of the ball is displaced
from the spherical center of the ball. In this case the CoM, of course, settles at the
lowest point below the surface of the liquid as shown in Figure 7P.
The recommended procedure in this case is to scribe any circle on the ball that
passes through the marked point. The CoM of the ball will necessarily lie in the
plane of this circle. Thus, if the ball is rolled along this circle, then the CoM will
simply rise and fall as the ball rolls along a true path. Most importantly the resultant
gravitational force on the ball will continuously pass through the track line of the
ball along the ground. Thus it will produce no turning moment, which could deflect
liquid surface
center of mass
offset
.
center of mass
Figure 7P Unbalanced ball floating in salt solution.
the ball from a straight trajectory, or on a slope change the expected curved trajec-
tory. This situation is illustrated in the left pane of Figure 7Q.
In contrast, if the ball had not been “balanced” and scribed, then it would most
likely be rolled with the CoM offset to one side. The right pane of Figure 7Q illus-
trates the worst condition where the CoM is offset horizontally to the side. Although
the likelihood of placing a ball in exactly this position is diminishingly small, there
is a 50 percent chance of placing the ball with the CoM offset from the vertical cen-
terline more than 70 percent of the extreme distance. This is a simple probability
calculation. It can easily be inferred by splitting the ball into four equal quarters as
shown on the right side of Figure 7Q. The ball, placed randomly, has a 50 percent
chance of the offset CoM being in either of zones A, since they constitute half of the
ball. The least effective position in these zones would be on the chain-dotted lines
separating A from B. At this position, the offset distance sideways is reduced by the
cosine of 45 degrees, which is 0.707. So in randomly placing an unscribed ball, the
player has a 50 percent chance of getting 70 percent or more of the worst effect from
an unbalanced ball.
With the CoM offset to the side, the resulting moment of the gravitational force
about the center of the ball will, through a property called gyroscopic precession,
cause the ball to follow a curved path; that is, in the right pane of Figure 7Q the
ball will be induced to turn to the right when rolling into the page. Most of us have
experienced this effect on bicycles. While freewheeling down a slight incline (pref-
erably at a bend), just lean your weight over to one side. This tilts the wheels side-
ways and so offsets your CoM. The important point is that when the offset CoM is
made, the bike will just keep turning without your position shifting sideways any
further.
This situation can be readily modeled in mechanics and has been published in
two separate papers from the University of Sydney, Australia. Guest (1964) showed
A . A
.
Figure 7Q Golf ball rolling into the page: (left pane) along a balance circle, and (right pane)
with the center of mass offset to the side.
223 Putting
Table 7E Sideways movement of 7 ft. putt on pool table with stimp rating 33.4
Degree Average displacement left and right (in.)

of balance
#1 #2 #3 #4 #5 #6
Ball A 10 0.88 −0.5 −1 −0.13 −0.13 −0.25

Ball B 1 8.38 9.63 10.25 10.13 8.5 9.5
Tack ball 1 11.5 10.6 9.5 10.3 10.3 9.3
that balls with a flat on one side, thus displacing the CoM toward the other side,
would follow a curved path on a flat surface represented by an equi-angular spiral.
Cross (1998), whom we came to know in Chapter 4, extended the model for applica-
tion to the game of lawn bowling. Lawn bowling balls have a bias weight on one side
so that they can be made to curve around balls that are in front of the target. In the
models used by both of these investigators, the balls were assumed to be perfectly
spherical, and the imbalance resulted from the CoM of the ball not being at the
spherical center. In the experiments, mass imbalance was deliberately introduced
into otherwise perfect balls. In molding golf balls, variations in the manufacturing
process will certainly include density distributions that are not symmetric about
the center. These can stem from slight density variations in the molded parts them-
selves or from slightly non-uniform weld lines between the two halves of the cover.
Before going further, it should be noted that the level of precision achieved by the
ball manufacturers is amazingly high. This is particularly the case when we consider
that the balls are molded at high speed in large, multicavity molds, and very precise
control of both pressure and temperature must be maintained throughout all of the
spherical cavities. Nevertheless, we will see that just the tiniest amounts of mass
imbalance can cause significant deviations in the roll.
To exaggerate the effect of the golf ball imbalance, Marsh and I carried out tests
on a pool table. A small grooved ramp, like the stimpmeter, was constructed to roll
the ball from the rail cushion down the length of the table. It was first determined
that the pool table had a stimp rating of 33.4. Balls rated by Marsh to be between
1 and 10 in degree of balance were rolled a distance of 7 feet, and the side movement
was measured. Every roll was carried out twice: once with the imbalance on one side
then again with the imbalance on the opposite side. Between each roll, the table was
brushed to remove any effect of the previous ball track. The average offset from the
target line of pairs of opposite rolls was taken in each case.
Table 7E shows the results of 6 pairs of rolls of three different balls. Balls A and
B were rated by Marsh to be the most balanced and most unbalanced categories,
respectively. The third ball, labeled “Tack ball,” was one of the most balanced balls
with a very small mass added to one side to give it approximately the same degree
of imbalance as ball B. In this case, the weight was a small brass tack pin available
from craft stores. The tack has a round flat head that fits flush to the ball surface. The
tack was weighed on a precision balance, and its weight was determined to be 0.23 g.
Since the ball was well balanced before the tack was added, we can determine the
amount of the CoM offset. This is determined to be only 0.11 mm (0.004 inches) as
illustrated in Figure 7R.
CoM offset = 0.11 mm
ball weight
45 grams
small
craft pin radius
21.34 mm
0.23 grams
Figure 7R Degree of imbalance produced by a 0.23 gram craft pin.
It is easy to demonstrate that this offset position is the balance point. The offset is
so small that we need to use the precise value of 0.1085 mm to make the demonstra-
tion give identical balance numbers. The radius of the ball is 21.34 mm, so the tack
is 21.34 – 0.1085 = 21.2315 mm from the predicted balance point. The ball weighs
45 g, and its weight can be taken to act through its center, which is 0.1085 mm away
from the predicted balance point. So if we could suspend the ball from the predicted
balance point, the moment or torque on one side from the weight of the tack would
be 21.2315 × 0.23 = 4.88 in g-mm units. The applied torque from the weight of the
ball would be 0.1085 × 45 = 4.88 g-mm. So we have the remarkable results that a ball
whose CoM was displaced by only 0.11 mm, or four-thousandths of an inch, will roll
an average of 10 inches sideways over a distance of 7 feet on a surface with a stimp
rating of 33.4. Obviously, the next question is how far would this ball likely deviate
on a much slower putting green?
Marsh and I carried out testing at the Meadow Brook course, one of the pre-
mium golf courses in Rhode Island. In the tests, we used a grooved wooden “stimp-
meter,” designed to attach to the platen of a tripod. The angle of the “stimpmeter”
was increased until a roll of 18 feet was established. The balls were released by hand
from a mark on the ramp. The green was mowed and rolled, and then the flattest area
was sectioned off for the testing. The green tested at stimp 8.5. Care was taken not
to walk on the designated 18 foot track. The ball used was Ball B, for which the pool
table results are given in Table 7E. A significant portion of the balls tested for degree
of imbalance, from a batch of 18 balls from premium manufacturers, showed levels
of imbalance near to that of Ball B, although Ball B was the worst one tested in that
batch. From comparison with the “Tack ball,” it can be estimated that this degree of
unbalanced roll could be produced by a perfectly spherical ball with the CoM only
0.1 mm (0.004 inches) offset from the center of the sphere.
The quite startling results of this testing are shown in Figure 7S. The ball final
resting positions are represented by reduced size circles to make the picture more
comprehensible.
The hole was positioned in the diagram at the front and at the center position for
the entire set of rolls. This can be taken to be the end position of the ideal trajectory.
From this datum, the average of the putts with right and left CoM is approximately
3 inches to the right and left, respectively. This is very clear evidence of a significant
effect of the smallest amount of non-uniformity in the ball.
A predefined target zone was established using pre-measurement rolls. Strays
outside of this zone were considered outliers. A number of short and long rolls were
also ignored. Short ones will naturally curve less and the opposite for long ones.
225 Putting
CoM left CoM right
Stimp = 8.5
18.6
18.4
Roll distance, feet
18.2
18.0
17.8
17.6
17.4
8 6 4 2 0 2 4 6 8
Lateral ball position, inches
Figure 7S 18 foot rolls from a grooved ramp using an out of balance, commercial, 3-layer ball;
each roll was repeated twice with the offset center of mass (CoM) on opposite sides.
It seemed that the outliers to the side were caused by less than smooth releases
from the “stimpmeter,” from which the ball had not stabilized to pure rolling before
reaching the ground.
Besides showing a distinct separation between the left- and right-biased balls,
the results obviously display quite a wide scatter from other sources. These are no
doubt the invisible imperfections in the turf, referred to by Pelz (2000a) as a major
source of missed 6 foot putts.
We can remove the bias of the CoM offsets from Figure 7S, and isolate these
other effects, by removing the average left and right offsets. This means moving all
of the left-offset balls approximately 3 inches toward the center, and likewise for
the right-offset ones, although the actual average offset values were used. This is
perfectly acceptable because the left and right bias of these two sets can have no
other source than left and right CoM offsets. It leaves the scatter of each of the two
subsets of rolls unchanged; that is, the scatter due to other sources remains. The
results of this are shown in Figure 7T. So this is the result we would expect with a
perfectly balanced ball, but in this case the exact launch conditions for each ball are
unchanged. From counting on Figure 7T the number of circles mainly inside the
two chain-dotted lines, it can be determined that 49 of the 80 “balanced” ball rolls
would have crossed inside the edge of the hole, compared to 31 of the 80 unbalanced
balls counted on Figure 7S. A very slight cross slope from right to left was probably
responsible for the wider sideways variation of the left-biased rolls.
The Cross model predicts an offset distance of 6.3 inches to the left and right for
this amount of imbalance on the stimp 8.5 green. I believe that this divergence of
the experimental rolls from the model is most likely due to the height of the grass.
This is sufficient for the ball to be modeled correctly as rolling through rather than
CoM left CoM right
Stimp = 8.5
18.6
18.4
Roll distance, feet
18.2
18.0
17.8
17.6
17.4
8 6 4 2 0 2 4 6 8
Lateral ball position, inches
Figure 7T Results as in Figure 7S, with the left and right bias of the left and right center of mass
position removed.
over the grass. We will deal with some of the real complexities of this at the very
end of the modeling section. However, it is clear that the grass displaced around the
bottom of the ball, at any given instant in the roll, will necessarily provide a resisting
torque to the turning of the ball. In this particular case, because of the very small
driving torque resulting from the tiny CoM offset, any such resistance is likely to be
significant. In the work of Guest and Cross, dealing with much heavier metal balls
and bowling balls, the resistance to twist, on other than a very soft surface, is likely
to be negligible compared to the large CoM offset.
The rate of turn of a ball rolling with an offset CoM, increases directly as the
offset increases and as the speed decreases. We can see the latter behavior in putts on
side slopes, where the curve of the ball sharply increases as it slows; see, for example,
the right pane of Figure 7M. For a given CoM offset, and a given stimp rating, the
side movement increases in direct proportion to the distance of the putt. So, for
example, we can predict that a 9 foot putt, on the stimp 8.5 green with Ball B, would
produce a biased turn of approximately 1.5 inches to the side.
Finally, the side movement increases directly with the stimp rating. Therefore,
having one set of values allows predictions of any others. We can work from my pre-
diction with the Cross model of 6.3 inches for an 18 foot putt on a stimp 8.5 green
with a 0.1 mm CoM offset. On a stimp 13 championship course, the predicted side
movement of Ball B, for the same 18 foot putt, is (13/8.5) × 6.3 = 9.6 inches. If we
take the experimental results for the 18 foot stimp 8.5 rolls, we might expect the
actual value to be as low as half of this, or 4.8 inches. However, as the grass is short-
ened, the area of contact at the bottom of the ball decreases, and the resisting torque
will decrease. We would expect the Cross model to move toward the actual behav-
ior. I tested the limiting case of a polished granite surface for which the stimp rating
227 Putting
was tested to be 150. The rolling resistance in this case is mainly the slight rise and
fall of the ball over the dimples plus perhaps some small amount of deformation of
the cover. The predicted trajectory of the tack ball was in almost complete agree-
ment with the test rolls, which gave a sideways movement of 20 inches in 3.5 feet.
We can predict this from the 6.3 inches over 18 feet on a stimp 8.5 surface. With the
distance reduction, we get (3.5/18) × 6.3 = 1.225 inches of side movement. Applying
the stimp increase gives (150/8.5) × 1.225 = 21.6 inches of side movement.
We can conclude with some degree of certainty that for putting on faster greens,
it is worthwhile adding balance lines to the golf ball. For competition at the highest
level it would seem to be essential. A worst-case ball selection, with a worst-case
orientation, on a fast championship green, could give somewhere between 4.8 and
9.6 inches of deviation on an 18 foot putt, and between 1.6 and 3.2 inches on a 6 foot
putt. Clearly this topic is worthy of more research.

7.1 The science of ball striking with the putter
Figure 7.1 illustrates a ball strike with a putter and the resulting launch of the ball.
The loft angle δ c of putters is typically 2 or 3 degrees and is exaggerated for clar-
ity in the figure. The putter is shown in two shades of gray. The lighter gray repre-
sents the putter blade and the supporting structure for the hosel and any additional
items. In this case, the only added item is the remainder of the mass, illustrated
in dark grey, and positioned low and at the rear of the structure. This is intended
to move the center of mass backward and below the line of action of the normal
impulse, p y , between the ball and the putter face. The normal impulse, p y, with
moment arm, s z , will thus kick the putter head counterclockwise about its center of
mass (indicated by the black dot toward the back of the putter), resulting in angular
velocity, Ω x , of the putter head at the instant of separation. These effects will cause,
through the gear effect, a topspin component to be added to the ball. At the same
time, the wedge effect caused by the small loft angle will impart a small backspin
component to the ball rotation. For good putter design, we need to consider the
combinations of loft and CoM position that will maximize the top spin rate, ω x.
The simple case to deal with first is when s z = 0; that is, when the CoM is directly
behind the ball strike position. For this sweet spot hit, Eqs. 3.5 and 3.10 in Chapter 3
can be applied with the approximation that for δ ≤ 3 degrees, cos(δ) ≈1, to give
v1  1 + e n 
= (7.1)
V0 1 + m / M 
and
ωrb / v 1 = − sin(δ c − α vb )/ λ . (7.2)

In addition, Eq. 4.4 in Chapter 4, with e t = 0, gives
V  m 
(δ c − α vb ) =  0 sin(δ c )/  1 + λ(1 + )  , (7.3)
v
 1  M 

ωx
Ωx
αvb
δc py
V0
v1
y
pz
sz
sy
Figure 7.1 Impact velocity and resultant ball velocity with a generic putter.
229 Putting
where the MoI of the ball is defined as before by λmrb2 ; and unless stated otherwise,
λ = 2 / 5 for a uniform ball is used in the calculations. The dimensionless quantity
ωrb / v 1 indicates the amount of slipping of the ball, and is referred to as the “slip”
parameter. If we define topspin as positive, then ωrb / v 1 = 1 is the condition of pure
rolling; that is, the peripheral velocity of the ball is the same as its forward velocity.
When ωrb / v 1 = 0 , the ball is in a condition of “pure” slipping, without any spin.
When ωrb / v 1 is negative, the ball starts with backspin, slipping is more severe,
and the ball will travel further before achieving the desired condition of pure roll-
ing. Note that for a sweet spot hit, the putter loft will produce backspin, hence the
expression on the right of Eq. 7.2 is negative as indicated.
Table 7.1 shows the calculations using Eqs. 7.1 to 7.3, for 350 g putters with lofts of
2 and 3 degrees. The calculated values for a 3.5 m/s putter strike would give a 5.4 m/s
ball speed, and in turn an approximate 8 m putt on an average speed green. Since the
slip parameter is close to zero, the putt is likely to slip for just beyond the first meter.
In these calculations we will assume SCof R = 0, as expected with the very small
loft angle. The putter head mass is taken to be 0.35 kg and a Cof R of 0.75 is assumed.
We have ignored the gear effect resulting from the tangential impulse, p z , between the
ball and the putter face; see Figure 7.1. This produces some backspin, Ω x, to the putter
head as shown in Figure 7.1; and referring to Eq. 3.15 in Chapter 3, we write
Ω x = − mv 1 sin(δ − α vb )s y /I x , (7.4)

where I x is the moment of inertia of the putter head about the sideways x axis. Also
from Eq. 3.17 in Chapter 3, we can determine the added topspin this gives to the
ball from
ω gear = −(s y /rb )Ω x . (7.5)

Taking a nominal value of I x = 1500 g-cm 2 for the putter, and s y = rb , Eqs. 7.4
and 7.5 give a topspin component, for the 3.5 m/s strike, of 0.23 rad/s (0.04 rev/s)
for the 2-degree loft and 0.35 rad/s (0.06 rev/s) for the 3-degree loft. These values
are an order of magnitude less than the backspin produced by the wedge effect. We
ignore the gear effect produced by the tangential impulse and consider only that
produced by the normal impulse, the latter effect being for hits above the sweet spot.
With this simplification, we can use a correspondingly simplified relationship
from Eq. 4.19 in Chapter 4; that is
Ω x = mv 1 cos(δ − α vb )s z / I x . (7.6)

Table 7.1 Speed and backspin parameters for a 3.5 m/s putter strike
Putter loft ν1/V1 (δc–α vb) ωrb /ν1 Putter speed Ball speed Backspin
(deg) (deg) V0 (m/s) ν1 (m/s)
(rad/s) (rev/s)
2 0.64 0.38 −0.015 3.5 5.43 −3.81 −0.61

3 0.64 0.57 −0.022 3.5 5.43 −5.72 −0.91
Table 7.2 Effect of height of center of mass of putter on a 3.5 m/s ball strike
2-degree Wedge Gear Resultant spin Slip

putter effect effect rate parameter
sz/rb (rad/s) (rev/s) (rad/s) (rev/s) (rad/s) (rev/s)
0.00 −3.81 −0.61 0.00 0.00 −3.81 −0.61 −0.01

0.25 −3.81 −0.61 2.17 0.35 −1.64 −0.26 −0.01
0.50 −3.81 −0.61 8.69 1.38 4.87 0.78 0.02
0.75 −3.81 −0.61 19.54 3.11 15.73 2.50 0.06
This equation, together with Eq. 7.1, can be used to solve the results of a high face
hit as shown in Figure 7.1. Table 7.2 shows the results of applying the wedge effect to
Eqs. 7.1 to 7.3 and the gear effect to Eqs. 7.5 and 7.6 for a gradually lowering CoM;
that is, the moment arm, s z in Figure 7.1, increasing from zero to 75 percent of the
ball radius. Note that the moment arm equal to the ball radius is impossible since it
would require the CoM to be on the sole of the putter.
From a putter design perspective, the results of these calculations, shown in
Table 7.2, are disappointing. The maximum topspin of 0.06 is insignificantly differ-
ent than pure slipping.
The only other apparent strategy for generating topspin is to strike the ball above
the equator with the bottom edge of the putter. This was discussed earlier. However,
unlike a high-strike pool shot on a hard table, which can create significant topspin
(Daish 1972), this would simply drive the ball downward into the green with some-
what unpredictable launch conditions.
The small amount of loft is applied to putters to give a small upward launch and
so start the ball predictably from its resting nest in the short grass. This has a nega-
tive effect on the desired topspin. However, it can be seen in column 5 of the last row
in Table 7.2 that even a very low CoM (s z = 0.75rb ) with a loftless putter would only
produce 3.1 revolutions per second of topspin.
7.2 The roll of the golf ball on a level green

We must start our modeling with the stimpmeter, for which the calculation of the
launch speed, including the effects of the groove and the inclined impact with the
ground, can be found in an article in The Physics Teacher by B. W. Holmes (1986).
Holmes used conservation of energy to calculate the velocity of the ball at the bot-
tom of the stimpmeter, that is, at the point of impact with the turf. He calculated
that for a uniform ball, with MoI (2 / 5)mrb2 , this terminal speed will be 1.88 m/s.
We use gravity and opposing friction to obtain the same result. This will provide
useful relationships for later calculations of putting on inclined greens. We also
show later that even large differences in the MoI values of golf balls make little
difference in putting performance. For this reason, we will also use (2 / 5)mrb2 as
the MoI since, as in previous chapters, it provides some simplifications of the equa-
tions of motion.
231 Putting
ω
72.5°
rc
rb
v
fg
fr fh
Figure 7.2 Views of ball in a 145° stimpmeter groove rolling down the 20° incline.
Figure 7.2 shows the parameters needed to calculate the ball speed on the
stimpmeter incline. Parameters v and ω are the linear and angular velocities. The
forces of interest are the component of gravitational force acting down the incline,
f g = mg sin(20), and the friction resistance between the ball and the stimpmeter, f r .
If we define the linear acceleration down the incline as v , we can write,
mv = f g − f r . (7.7)

Since the ball is in a condition of pure rolling down the stimpmeter, we have
v = rc ω and so v = rc ω , (7.8)

where ω is the angular acceleration, and rc is the actual rolling radius as shown in
the view down the incline in Figure 7.2. The moment of the resisting force, f r , about
the ball center is responsible for the angular acceleration according to
(2 / 5)mrb2 ω = (2 / 5)mrb2 (v / rc ) = f r rc . (7.9)

Eliminating v between Eq. 7.7 and the right-hand equation in 7.9 gives
 1 
fr = fg   = 0.3054 f g . (7.10)
1 + (5 / 2)sin (72.5) 
2

Finally, from Eqs. 7.7 and 7.10, we get
v = 0.6946 g sin(20) = 2.3304 m / s2 ;
and from v = 2 vs  , where s = 0.762 m (30 inches) is the distance along the stimp-
meter, we get the velocity of the ball impacting the ground as v i = 1.88m/s, as
given by Holmes. Note that if the ball was rolling on an inclined flat (un-grooved)
surface, then the sine term in Eq. 7.10 equals 1, and the resisting force becomes
f r = (2 / 7)f g . Equivalently, if a golf ball is rolling on a sloped green surface, and
the component of gravitational force down the incline is f g , then in the absence
of other frictional forces, the acceleration down the incline will be given by
v = (f g − f r )/ m = (5 / 7)f g /m. We meet this result later when considering putting
on inclined greens.
ω
0
vi
rb
v0
ph
pv
Figure 7.3 Impulses acting on the ball at impact with the turf from the stimpmeter.
The ball reaches the end of the incline and impacts the turf with speed
v i = 1.88m/s, but travelling downward at an angle of 20 degrees. The impact with
the green must convert the inclined roll to a horizontal one. To do this it must
eliminate the downward velocity component v i sin(20) and re-impose the con-
dition of pure rolling. Thus the impact includes a vertical and horizontal impulse
as shown in Figure 7.3.
Note that following ground impact, the horizontal impulse will point in the
direction of motion since it must reduce the angular velocity to the value for pure
rolling on the flat surface.
The vertical impulse simply eliminates vertical momentum and thus satisfies the
relationship p v = mv i sin(20). The horizontal impulse changes the horizontal veloc-
ity component and the angular velocity according to
p h = mv 0 − mv i cos(20) (7.11)

and
p h rb = (2 / 5)mrb2 (ω 0 − ω i ), (7.12)

where ω i is the angular velocity prior to striking the green. Substituting ω 0 = v 0 /rb
and ω i = v i /rc , and eliminating p h between Eqs. 7.11 and 7.12 gives
v 0 = (5 / 7)v i [cos(20) + (2 / 5)/ sin(72.5)] = 1.83 m/s. (7.13)

This is the result given by Holmes (1986), which he also validated experimen-
tally. The value converts to 4.09 mph, applied for convenience as 4 mph earlier.
Note that if the roll length along the stimpmeter is changed, since v = 2 vs  ,
the square of the velocity changes in proportion. This result was used earlier in the
chapter.
Next we need to determine the roll of the ball across the green after it exits the
stimpmeter. Penner (2002a) analyzed this problem using the model illustrated in
Figure 7.4.
With the compression of the grass ahead of the rolling golf ball, the location of
the reaction force from the ground moves ahead of the ball: distance s x , which is
greatly exaggerated in Figure 7.4. The vertical component of the ground force, f g ,
supports the weight of the ball and provides a counterclockwise moment f g s x. The
233 Putting
sx
rb
v
mg
sz fg
fr0
Figure 7.4 Model of golf ball rolling over grass; adapted from Penner (2002a).
horizontal force component, f r 0 , caused by the resistance of the grass, produces

clockwise moment f r 0 s z. The suffix zero refers to the fact that this is the resisting
force of the grass on a flat green, that is, zero slope. We see later that, very surpris-
ingly, for the same grass but on a slope, the grass resisting force changes.
For this case, angular deceleration ω  satisfies the relationship
I bω = (f g s x − f r 0 s z ). (7.14)

The resisting force component must produce the corresponding linear decelera-
tion v ; that is,
mv = − f r 0 . (7.15)

The vertical force component is supporting the ball weight, mg, and so
f g = mg . (7.16)

Finally, since the ball is in a condition of pure rolling, the velocities and decelera-
tions must satisfy
rbω = v and so rbω = v. (7.17)

Eliminating f r 0 , f g , and v between Eqs. 7.14 to 7.17, and making the
simplification s z ≈ rb, gives
v = − mgs x rb /(I b + mrb2 ). (7.18)

Using the MoI for a one-piece uniform density ball (I b = (2 / 5)mrb2 ), Eq. 7.18
gives
v = −(5 / 7)(s x / rb )g ; (7.19)

and so
f r 0 = (5 / 7)(s x / rb )mg . (7.20)

This result does not allow us to establish the resisting force from first principles.
However, it provides an understanding of the mechanism of deceleration and estab-
lishes a basis for analyzing putts across sloped greens.
To get a first estimate of the resisting force, we can use the stimp rating S m,
expressed in meters, and the launch velocity v 0 (= 1.83 m/s) from the stimpmeter.
Newton’s law then gives
f r 0 = mv 20 /(2S m ). (7.21)

Slow greens are generally characterized as those with roll distances
around 4 feet (1.22 m), while fast championship greens are usually in excess of
12 feet (3.66 m) of roll. The average resisting forces in these two cases are thus
(0.045)(0.5)1.832 /1.22 = 0.0618 N and (0.045)(0.5)1.832/ 3.66 = 0.0206 N, respec-
tively. The corresponding values of s x from Eq. 7.20 are s x = 0.196 rb for a slow green
and s x = 0.065rb for a championship green; that is, a range for slow to fast greens of
0.065 < s z / rb < 0.196 as given by Penner (2002a).
One final question should be addressed with regards to the stimpmeter rating.
Is the rating value affected by the MoI of the ball? This was discussed earlier. To
answer this question, it seems reasonable to assume that for a given green, the value
of s x will depend on the weight of the ball and its radius, since it is a function of
the deflection of and friction between the grass blades. As such, it is not likely to
be affected by the MoI of the ball. However, the deceleration of the ball on a given
green is affected by the MoI of the ball according to Eq. 7.18. To check if this is a
significant factor, we can estimate the changes that would occur in the stimp read-
ing if we were to increase or decrease the ball MoI to even unrealistic limits. The
preceding calculations are for an idealized uniform-density ball for which the MoI,
I b , is (2 / 5)mrb2 ; or I b/mrb2 = 0.4. In Table 7.3, the center row represents the uni-
form ball, and the rows above and below that are for a reduction and increase of
50 percent in MoI.
These are extreme cases; the former would require the outer region of the ball to
be very low density, and the latter would only be possible with a shell design filled
with the lightest foam. The previous equations, but with the different MoI values,
were used to get the velocities, in columns 2 and 3 of Table 7.3, before and after
impact with the ground. The deceleration rates in column 4 were obtained from
Eq. 7.19 with s x = 0.065rb for a 12 foot stimp-rated green. The deceleration values
were then used to get the stimp rating distances in columns 5 and 6.
It can be seen that the stimp rating with all three balls is virtually the same. The
additional question, however, is would the larger deceleration of the low-MoI ball
making putting on fast greens significantly easier? Assume the ideal 12 foot (3.66 m)
Table 7.3 Stimp ratings for a fast green with varying ball

moment of inertia
T b/mrb2 νf νS v. stimp rating

m/s m/s (m/s)/s
m ft.
0.20 2.05 1.96 0.53 3.60 11.82

0.40 1.88 1.83 0.46 3.66 12.00
0.60 1.76 1.72 0.40 3.70 12.14
235 Putting
putt would travel 13 feet (3.96 m) if it missed the hole. Using the deceleration val-
ues in column 4, the low-MoI ball would require a launch velocity of 2.05 m/s, and
the medium-MoI one would require 1.88 m/s. With the same speed error of say
+0.25 m/s, the high-MoI ball would run 5.0 m (16.4 feet), while the medium-MoI
ball would run 5.1 m (16.7 feet); an insignificant difference. These comparisons are
for a fast green where any differences due to MoI would be maximized. As discussed
earlier, there is also the effect on stability to be considered, which no doubt greatly
outweighs any advantages of a low-MoI ball.
7.3 Putting straight up or down a sloping green

We are dealing with this as a separate topic from putting in any direction across
curved greens to establish a simple model for application to the more general prob-
lem. For this, we simplify the Penner model slightly, as shown in Figure 7.5, by mov-
ing the center of force to the lowest point of contact. The sloped green lies on the x–y
plane and the components of gravity in the x and z directions are
f gx = mg sin(α) and f gz = mg cos(α). (7.22)

Because we have moved the normal force, f gz , backward by distance s x , we
must introduce moment τ, with magnitude
τ = mg cos(α)s x , (7.23)

where s x is the moment arm in Figure 7.4.
The deceleration, v , must satisfy
mv = − f r ± f gx , (7.24)

where the plus sign is for a downhill putt as shown in Figure 7.5, and the sign is
reversed for uphill ones.
Assuming the condition of pure rolling gives
v = rbω ans so v = rbω ; (7.25)

so angular acceleration, v / rb , must satisfy
(2 / 5)mrb2 (v / rb ) = − τ + f r rb . (7.26)

ω
τ
fgx
v
α fgz
fr
Figure 7.5 Model of golf ball rolling down an incline.

Eliminating v between Eqs. 7.24 and 7.26, and substituting for τ from (7.23),
gives
f r = (5 / 7)(s x / rb )mg cos(α) ± (2 / 7)mg sin(α). (7.27)

We can note that f r = f r 0 cos(α) ± (2 / 7)mg sin(α), where f r 0 is the resisting
force on level ground and can be obtained from the stimp rating in Eq. 7.21.
Finally, from Eq. 7.24, we obtain
v = −(5 / 7)(s x / rb )g cos(α) ± (5 / 7)g sin(α). (7.28)

The first term in Eq. 7.28 is the deceleration caused by the resistance of the grass,
as derived in Section 7.2, but reduced according to the decreased gravitational force
normal to the turf. This term will be labeled v r . We can note that the slope of greens
is never large enough for cos(α) to be significant, so we can estimate v r directly
from the stimp rating. The second term in Eq. 7.28 is the resultant gravitational
deceleration or acceleration acting parallel to the incline.
Note that by setting v = 0 in Eq. 7.28, we can determine maximum slope, α max,
which would allow the ball to stop on a downslope or stop and not roll back on
an upslope. For this condition, Eq. 7.28 reduces to α max = tan −1 (s x /rb ). This gives
maximum slope values ranging from 11 degrees for stimp = 4 to 3.7 degrees for
stimp = 12. Expressed in percentage slope terms, these values are 19.4 percent and
6.5 percent, respectively. A range of these values was presented in Table 7B.
To apply these results in a form meaningful for the golfer, it is important to
include the stimp rating, and therefore establish the model in terms of putt dis-
tances. For example, if the stimp rating is defined by S m in meters, and an up- or
downslope putt has length S p , the advice to start the ball at a certain speed v 0
would be of little value. A much more practical approach would be to suggest that
the putt should be struck so that it would travel distance S p0 on a flat green with the
same stimp rating. This appeals directly to the skill of the golfer who can then pick a
point beyond or before the hole, depending on the slope direction, and putt to those
points as if the slope did not exist. Using our equations, the approach would take the
following steps:
(a) From the stimp rating distance S m , and the stimp launch speed of 1.83 m/s,
determine the deceleration v r caused by the grass resistance; that is, from
Eq. 7.21, v r = f r 0 / m = (1.83)2/(2S m).
(b) Determine the launch velocity, v 0 ,required to make the desired putt up- or
downslope to distance S p from v 0 = 2 vS  p ,where v& = v& r m (5 / 7)g sin(α).
(c) Use v 0 and v r to determine the equivalent flat distance; that is,
S p 0 = v 20 /(2 v r ).
Predictions for up- and downhill putts on a 4 percent (2.29 degree) slope, with
stimp rating 7 feet (S m = 2.134 m) are given in Table 7.4.
We consider enhancements to our model of golf ball rolling in Section 7.5: mainly
to understand the mechanisms of grass deformation better and why a ball rolling
237 Putting
Table 7.4 Predicted initial velocities and equivalent distances for up- and

downslope putts
Stimp 7 ft; slope 4% Upslope Downslope
Slope distance (ft.) 21 16 11 11 16 21

Initial velocity (m/s) 3.69 3.22 2.67 1.84 2.22 2.54
Flat distance (ft.) 28.5 21.7 14.9 7.1 10.3 13.5
through grass may be deflected significantly off course so easily. For now, we extend
the model to predict trajectories across sloped greens.
7.4 Putting across a sloping green

Penner (2002a) applied an extension of his rolling model to this general problem.
To obtain the expected behavior, he introduced an additional degree of freedom
into the model by inclining the resisting force to the velocity vector. In this analysis,
we will adopt simpler approach by assuming that the resisting force acts as a drag
force, always directly opposing the forward velocity. In addition, the component of
the gravitational force, down the slope at right angles to the roll direction is assumed
to act as a centripetal force causing the trajectory to curve about an instantaneous
center. Figure 7.6 shows an overhead view of the ball, travelling on an uphill trajec-
tory, at angle β across the slope, with the instantaneous center of curvature to the
left. The slope has angle α, sloping upward in the positive x direction.
The gravitational force on the ball, in the negative x direction, is mg sin(α).The
component of the gravitational force acting in the direction of the resisting force is
mg sin(α)sin(β); therefore, from Section 7.3, we can write a more general expres-
sion for Eq. 7.27 as
f r = f r1 ± f r 2 , (7.29)

where
f r1 = (5 / 7)(s x /rb )mg cos(α *) (7.30)

Ωz
y
v
β
Rc
x
mg sin (α) ω
fr
Figure 7.6 Overhead view of putt angled up a side slope.

sb0
sfo
Ωz0
v0
y
β0
Rc0
x
Figure 7.7 Initial conditions of putt angled up a side slope.
and
f r 2 = (2 / 7)mg sin(α)sin(β), (7.31)

and α * = sin −1 [sin(α)sin(β)] is the slope along the angled incline. Note that
f r1 = f r 0 cos(α *), where f r 0 is the resisting force on a level green given by Eq. 7.21.
In this case, the deceleration along the inclined path must satisfy the more gen-
eral expression for Eq. 7.15 as
mv& = − f r m mg sin(α)sin(β). (7.32)

The component of the gravitational force, in the plane of the slope and normal
to the instantaneous rolling direction of the ball, is mg sin(α)cos(β). This force pro-
duces a centripetal acceleration with respect to an instantaneous center of curva-
ture. This defines the current radius of curvature, R c, of the curved path of the putt.
We can thus write
v 2/ R c = R c Ω2z = g sin(α)cos(β). (7.33)

Since the rate of change of direction is equal to Ω z, from Figure 7.7 we have
β = Ω z . (7.34)

These equations allow us to set up the following numerical scheme:
(a) From the stimp rating distance Sm , and the stimp launch speed of 1.83 m/s,
determine the deceleration caused by the grass resistance; that is,
v r = f r 0 /m = (1.83)2/(2 Sm ).
(b) Select an initial launch angle β = β0 and launch velocity v = v 0 .
(c) Determine radius of curvature R c and the rotational speed Ω z about the center
of curvature from Eq. 7.33.
(d) Determine the deceleration v from Eq. 7.32.
(e) Apply a time increment to find the changed velocity, v.
239 Putting
Table 7.5 Predicted speeds and breaks for Figure 7J trajectories;

slope = 4%; stimp = 7
Direction to Hole distance Initial speed Equivalent Predicted break

fall line (deg) (ft.) (m/s) distance (in.)
(ft.)
10 2.622 14.4 4.0

+60 15 3.164 20.9 7.9
20 3.625 27.5 11.8
10 2.285 10.9 15.1
0 15 2.762 15.9 25.6
20 3.170 21.0 36.4
10 1.906 7.6 10.1
−60 15 2.302 11.1 17.7
20 2.638 14.5 25.2
* for putt to 1 foot past hole; total putt distances 11, 16, and 21 feet.
** from edge of hole; add 2.125 inches for break from center.
(f) Determine the new position from the time increment, v and β = Ωz.
(g) Loop back to (c) until the velocity reaches a prescribed small value.
(h) Loop back to (b) using an iterative scheme to adjust the launch conditions
unless the end of the trajectory is close enough to the target point.
(i) From the derived launch conditions, determine the offset of the initial velocity
from the hole (called the break) and the equivalent distance of the putt on level
ground (as in Section 7.2).
All of the trajectories on sloped greens, given earlier in the chapter, were deter-
mined by the application of this numerical scheme. The predicted putt launch con-
ditions, for the trajectories in Figure 7L, are shown in Table 7.5. Note that these
solutions are unique. Any alternative trajectories to the holes would travel different
distances with a miss-hit.
7.5 More sophisticated models: rolling with skidding

and slipping; velocity dependence on turf resistance
It should be noted that biological fiber materials are known to behave, under
dynamic loading, in the same way that synthetic long-chain polymers behave; that
is, they show a greater resistance to deformation as the speed of loading increases.
There is a huge literature on the mechanical properties of polymers showing
substantial increases in both elastic modulus and yield stress values as the rate
of straining increases; see, for example, the numerous papers in Proulx (2010).
The structural cellulose fibers in plants have been investigated mainly as renew-
able reinforcing fibers in polymer systems. However, one article addressing the
strain rate sensitivity of leaves themselves (Mukhopadhyay et al. 2008) shows the
modulus of the cellulose fiber approximately doubling as the rate of straining goes
from 0.1/s to 1.0/s. With this result, an increase in grass structural stiffness with
increases of the ball speed does not seem unreasonable.
To model this behavior, we assume that the average resisting force varies linearly
with velocity differences from the stimp launch speed; that is, according to
f r = f r 0 [(1 − µ) + µ(v 0 / v s 0 )], (7.35)

where µ is a speed coefficient, v s0 = 1.83m/s is the stimp launch speed and v 0 is
the launch speed. Note that when v 0 = v s 0 the conditions of the stimp test are still
satisfied.
We can modify the application of Eqs. 7.22 to 7.28 as follows:
(a) From the stimp rating distance Sm , and the stimp launch speed v s0 = 1.83m/s,
determine the basic resisting force f r 0 ; that is, f r 0 = mv 2s 0 /(2Sm ).
(b) Make an initial estimate of the launch velocity, v 0 , required to make the
desired putt up- or downslope to distance s p, from v 0 = 2 vs  p , where
v = f s 0 /m ± (5 / 7)g sin(α).
(c) Change the resisting force to f r = f r 0 [(1 − µ) + µ( v 0 /v s 0 )], for a selected small
value of coefficient µ.
(d) Adjust the launch velocity, v 0 , required to make the desired putt up- or
downslope to distance s p , to v 0 = 2 vs
 p , where v = f r /m ± (5 / 7)g sin(α).
(e) Loop back to (c) until a negligible change in v 0 occurs.
(f) Calculate the equivalent distance over a flat green from s p 0 = v 20 /(f r / m).
(g) Loop back to (c) and adjust the value of µ until agreement is obtained with a
set of accepted data.
The results of this speed correction are shown in Table 7.6 for the up- and
downslope case considered in Section 7.3. Note that the only changes of signifi-
cance are for longer, and particularly for uphill putts, where roll speeds are higher.
The value of µ = 0.15 represents an increase of 17 percent in the resisting force for
the longest upslope putt, with diminishing increases for the other putts.
With negative values of µ the opposite possibility of resisting force decreasing
with speed can be modeled. This extension of the analysis is only included to facili-
tate future investigations of the likely speed effects of grass rolling resistance.
Table 7.6 Predictions with μ = 0.15
Stimp 7 ft.; Slope 4% Upslope Downslope

Flat distance (ft.) 27.4 21.1 14.7 7.1 10.5 14.0
241 Putting
Separate from the grass resistance issue, there are two concerns regarding
assumptions made in the basic mechanics. First there seems to be no tractable
method for analyzing the transition from slipping to rolling in the early phase of a
putt. As a consequence, the putt is analyzed for the launch condition of pure rolling,
with the additional initial slipping resistance distributed over the entire putt. There
is also the matter of the upward launch of the ball, off the 2 or 3 degrees loft of the
putter face, to be considered. This may in some cases give a more rapid transition to
pure rolling through initial small bounces. Moreover, we should also recognize that
the golf ball is rolling through the grass. The analysis up to this point has assumed
that the ball is rolling around its equator. Instead it is more reasonable to define an
intermediate radius as rb*, as illustrated in Figure 7.8. This leads to a further obser-
vation of the Penner model that the center of the resultant force, from the pressure
distribution over the surface of contact, will not lie on the equator of the ball. We
assume, with likely negligible error, that the moment arm s z of the resisting force is
equal to the roll radius rb*.
With this change, the analysis of rolling up or down an incline, in Section 7.3,
Eqs. 7.24 and 7.25 must change to
v = rb* ω and do v = rb* ω ; (7.24)′

and angular acceleration, v / rb , must satisfy
(2 / 5)mrb2 (v/rb ) = − τ + f r rb*. (7.25)′
Solving the equation set 7.21 to 7.26 with these changes gives the deceleration as
mg cos(α)(s x /rb* ) mg sin(α)

v = − ± . (7.27)′
1 + (2 / 5)(rb /rb* )2 1 + (2 /5) (rb/rb* )2
This can easily be seen to reduce to the earlier expression for deceleration
if rb* = rb . The first term in this expression is somewhat academic because, as in
Section 7.3, we are forced to approximate it from the stimp rating as f r 0 /m. In the
second term, we can see that the change serves to decrease the effect of gravity. All
other things being equal, this will result in slightly longer upslope putts and slighter
shorter downslope ones.
The results, of applying Eq. 7.27′ to the up- and downslope cases analyzed in
Section 7.3, are shown in Table 7.7. For these results, the roll radius was assumed to
be 19.7 mm, 1.5 mm less than the ball radius.
sx v
rb
r*b fv ω
fh
Figure 7.8 Rolling “through” the grass.

Table 7.7 Initial velocity and equivalent distances from Eq. 7.27′
Stimp 7 ft.; Slope 4% Upslope Downslope
Slope distance (m) 21 16 11 11 16 21

Flat distance (ft.) 28.1 21.4 14.7 7.3 10.6 13.9
The concept that the ball is rolling through the grass and must settle on an inter-
mediate roll radius has some implications for the stability of the putt. It is easy
to appeal to a minimum energy principle and contend that the roll radius will be
selected to minimize the frictional energy loss rate below the surface of the grass.
However, there is no proof that such a principle applies to this complex situation.
For the simpler situation of a cylinder rolling over a smooth perfectly deformable
surface, such limit theorems do not apply because the extent of the boundary itself
is part of the solution; see Hill (1951). The author has shown that for the even sim-
pler situation of pushing a wedge along a perfectly deformable surface, the precise
solution depends on the starting conditions and that different steady conditions can
exist or can change later with even the slightest disturbances (see Dewhurst 1978).
These observations were later to be subsumed into a general theory of chaos. We
should expect the same kind of uncertainty with the rolling golf ball, where even the
smallest of disturbances may change the rolling characteristics beyond reasonable
expectations. This brings us back to the comments by Pelz (2000a) about unseeable
factors, such as the shallowest invisible footprints that can nullify the most perfectly
stroked putts, and possibly change the basic rolling condition for a significant dis-
tance beyond the disturbance.
7.6 Conditions for capturing the ball

As seen in Section 7.4, sinking the putt is first a matter of selecting the launch direc-
tion across the green and striking the ball at the required speed. However, it is not
sufficient to just roll the ball over the edge of the cup. It is also essential that the
ball is actually captured in the hole. Holmes, who published a charming dialog on
the mechanics of the stimpmeter (1986), also published an exhaustive analytical
and experimental study on the conditions for ball capture (Holmes 1991). Penner
(2002a) used this work to evaluate the initial launch speeds necessary for the ball
capture. We first review this body of work. Then, for consistency with the prior sec-
tion, we consider the relationship between recommended putt lengths past the hole
and the probability of capture.
Holmes analyzed all of the alternative scenarios to construct a complete map
of capture and escape regions in the ( v e , d off ) space, where v e is the velocity with
which the ball crosses the edge of the hole, and d off is the offset position from the
hole center; see Figure 7.9. Holmes determined that the capture conditions range
from ( v e )max = 1.63 m/s for a center hit (d off = 0) to v e = 0 for a glancing miss
(d off = rh ); that is, for the latter, the ball must stop and drop in sideways. The
243 Putting
hole diameter, dh
effective
diameter, deff
ve
doff
Figure 7.9 Effective hole size corresponding to crossing speed v e .
separation of Holmes’ capture regions from escape regions, defining the maximum
edge crossing speed for different offset position, follows a slightly wavy parabola.
This can be represented almost exactly by
v e = (v e )max [1 − (d off / rh )2 ]. (7.36)

This is the approximation used by Penner (2002a) to evaluate the launch veloci-
ties for successful capture. Equation 7.36 can be viewed to represent a changing hole
“size” for different speeds of the ball on reaching the hole. If, for example, the speed
is 1.0 m/s, then solving Eq. 7.36 for offset doff gives d off / rh = 0.62. This means that
a ball, crossing at 1.0 m/s, will be captured if it crosses within a center band of width
2d off = 0.62d h , where d h is the hole diameter. The width 2d off will be defined as the
effective hole size, d eff , for any particular edge crossing speed v e ; see Figure 7.9.
We can transform the speed limits given by Eq. 7.36 to distance past the hole
for different stimp ratings. Assuming a constant resisting force, the distances are
proportional to velocity squared; therefore, we can predict the maximum distance
past the hole S past, which would have resulted in hole capture for different offsets as
(S past )max = (v e / v s )2 [1 − (d off /rb )2 ]2 S m , (7.37)

where S m is the stimp rating in meters, and v s = 1.83 m/s is the stimp launch speed.
Figure 7.10 shows a plot of Eq. 7.37 for stimp ratings between 5 and 13. We can
read approximate values for effective hole sizes from these plots. For example, if a
putt would go 1 m past the hole if it misses, then the effective hole size is approxi-
mately 3 cm with a stimp rating of 5, and 7 cm with a stimp rating of 13. Thus, if the
player strikes the correct contact launch speed, then he or she has more than twice
as much chance of holing this putt on a fast green than on a slow one. Of course,
3
Stimp 13
Stimp 11
Stimp 9
Stimp 7
Distance past hole, m Stimp 5
2
0
5 4 3 2 1 0 1 2 3 4 5
Offset ball position, cm
Figure 7.10 Maximum distance past hole (overrun) for an off-center putt to achieve ball capture.
on the fast green, there is a much higher chance of playing a shot that would travel
further than 1 m past the hole if it missed, with an associated reduction in effective
hole size.
We can invert Eq. 7.37 to obtain these values directly; that is,
1/2
  v 2s S p  
d eff = 2 rh 1 −  2   . (7.38)
  v e S m  
This of course raises the question as to the optimum overrun distances that
should be targeted. The trade-off is quite simple: a faster putt will suffer less from
minor disturbances, so putting accuracy will increase; however, effective hole size
will decrease. Pelz (2000a) reports, from “testing for optimum speeds on differ-
ent putts and different greens” all around the United States, that the optimum dis-
tance past the hole is always near 17 inches. The optimum in this case is the value
for which the highest percentage of balls are captured when they are rolled along
the same line, with the same launch speed, but nevertheless follow slightly variable
paths because of the sensitivity to “unseeable factors.” The tests were carried out
using a calibrated inclined ramp to first find the “line and speed to make the highest
percentage of putts.” The hole was then removed to measure the average overrun
distance.
If we substitute an overrun distance of 17 inches (0.4318 m) into Eq. 7.38, then
we obtain the effective hole sizes shown in Table 7.8.
From Pelz’s work with tour professionals, we can safely assume that the testing
was carried out mainly on faster greens and that the recommended overrun dis-
tance provides between 75 and 80 percent of the hole size for capture.
Table 7.8 Effective hole sizes for the Pelz optimal overrun distance
Stimp rating 5 7 9 11 13
Effective hole size (%) 63 70 74 77 79
A P P E N DI X A
Energy transfer during club-ball collision*

This discussion is in support of the conclusions reached in Chapter 3 concerning the power
of the modern driver. This is rather complex; and like the modeling sections of each chapter,
it can be skipped without any loss of understanding of the science principles of the game.
The approach is to use the simplest possible viscoelastic ball collision model established
by Simon (1967). This model has an elegant closed-form solution, which allows us to quite
easily investigate energy transfer and loss during ball impact with a flexible club face. This
will take us directly to the mechanism for achieving high impact efficiency. We validate the
model with reference to the work by Cochran (2002), who conducted a series of experiments
involving collisions between a golf ball and a rigid surface at speeds from 5 m/s to 50 m/s, in
each case measuring the contact time and CofR. We address this first.
R IGID SU R FACE COLLISIONS

The Simon model is defined by
F = mx = −κ | x |1/2 ( x + c | x | x ), (A1)
where the dots represent single and double derivatives with respect to time, and where
κ = elastic stiffness, N/m3/2 , and

c = damping coefficient, s/m
x = normal relative displacement of the impacting bodies, m
x = normal relative velocity of the impacting bodies, m/s
F = contact force, N
m = mass parameter of the impacting bodies, kg.
Note that in general, the coefficients κ and c represent damping and stiffness of the com-
bined colliding bodies. Also, when c = 0, the equation reduces to the Hertz relationship for
the perfectly elastic deformation between two bodies (Hertz 1882); that is
F = mx = −κ | x |3/2. (A2)
Equation A1 is referred to in the golf literature as a two-parameter model, but this is only
the case for ball collisions against a rigid heavy plate to which it has been applied. This is illus-
trated in Figure A1. In this case κ and c describe only the ball behavior, and κ is given by
* SUPPORTING PHYSICS—skip without any loss of understanding of the science principles

of the game
245
246 Appendix A
rb
Figure A1 Impact of golf ball against a rigid barrier.
κ = 4(E b /(1 − υ b )2 )(rb )1/2/3, (A3)
where Eb and υ b are elastic modulus and Poisson’s ratio, respectively, of the ball.
Three papers can be referenced to help in checking the appropriateness of the elastic
modulus and Poisson’s ratio values used in our application of Simon’s model. Tanaka et al.
(2006) determined that a finite element model of a 3-layer ball gave excellent agreement
with a series of experimental measurements of contact time, contact force, and ball rebound
speed using the values shown in Table A1. Also shown in the table are values obtained
through material testing by Mase (2003) and Tamaogi and Sogabe (2010) for two-part con-
struction balls, the latter using high-speed testing with the split Hopkinson bar method;
Kolski (1949). Note that although the ball cover is a relatively small proportion of the total
ball volume, it must stretch to allow significant ball deformation, so its higher modulus has
a disproportionate effect on the overall stiffness of the ball. This is particularly the case
because of the high Poisson’s ratio of the ball core material and the fact that the bulk, or
volumetric, modulus K is related to the elastic modulus E by K = E /(3(1 − 2υ)). Thus, with
an average measured value of 0.433 for poison’s ratio for the core material, the volumetric
modulus is 2.5 times larger than elastic modulus E. This means that the core will experience
relatively little volume change. Therefore, as the core flattens from a sphere, its surface area
will increase, requiring significant stretching of the thinner cover, giving a large overall
increase in ball stiffness.
By trial and error it was found that the values of Eb = 145 MPa and υ b = 0.433 in the
Simon model gave combined values of CofR and contact time in excellent agreement with
Cochran’s experimental results (2002) of golf ball collisions with a rigid heavy block. The
modeling results are described following. It can be noted that these values of elastic modulus
Table A1 Experimental measurements of ball material properties
Core Mid-layer Cover
Elastic modulus (Mpa) (1)50, (2)60, (3)65.5 (1)25 (1)400, (2)276, (3)310
Poisson’s ratio (1)0.49, (2)0.43, (3)0.38 (1)0.49 (1)0.45, (2)0.41, (3)0.32
(1) Tanaka et al. (2006), (2) Mase (2003), (3) Tamaogi and Sogabe (2010).
247 Appendix A
20
18
16
14
Contact force, kN
12
10
8
6
4
2
0
0 1 2 3 4 5 6 7
Club face and ball displacements, mm
Figure A2 Simon model predictions of the compression and restitution of ball impacts against a
rigid barrier at speeds 4, 22, 35, and 50 m/s.
and Poisson’s ratio are reasonable ones with reference to the measured material properties
of modern two-layer balls in Table A1. In this case, it would appear that the average cover
modulus of 293 MPa has the effect of more than doubling the stiffness of the bulk of the ball,
residing in the core, from an average value of 63 MPa to 145 MPa. It is best to consider the
core as providing an elastic foundation under the cover, and the pronounced effect of this is
well understood in the 2-dimension analog of elastic foundations under beams; see Hetenyi
(1946). The value of 0.433 for Poisson’s ratio is near the mean of the values for two-part con-
struction balls in Table A1.
Figure A2 shows solutions for contact force, F, plotted against ball displacement, x, for
four impact speeds (4, 22, 35, and 50 m/s) against a rigid plate. The speeds are those used
for experimental measurements of contact time and CofR by Cochran (2002). The slowest
speed produces the tiny hysteresis loop in Figure A2, with maximum ball compression less
than 1 mm and maximum force less than 1 kN. For the highest speed, the ball compression
reaches 6.4 mm, with a maximum force of 18.4 kN. In every case, the energy lost in the ball,
equal to the area within the hysteresis loop, is exactly the value corresponding to the respec-
tive CofR through the “kinetic” definition. Note for these cases, the ball comes to a complete
rest against the rigid barrier at the end of the compression stage, so all of the initial kinetic
energy is stored in the ball at this point. The area within these hysteresis loops is exactly
equal to the energy lost in the collision. Note the disproportionate increase in this area with
impact speed. There is some confusion in the literature with simple compression testing of
these materials, which gives completely erroneous results for ball striking. The otherwise
excellent paper by Michal and Novak (2001) unfortunately relies on low-speed compression
testing for some of their conclusions. The force displacement plots used there can be seen to
involve relatively small energy losses.
Table A2 shows the parameters and calculated values for these four cases. Note that for
this simple model, the damping coefficient, c, is a property of the particular collisions and not
a fixed material property of the ball. In each case, it is adjusted incrementally until the ball
hysteresis loop closes. This is the same as making the adjustments until the final ball velocity
from restitution occurs at the same time as the ball displacement returns to zero.
Table A2 Simon model calculations for golf ball strikes against a heavy plate
Ball speed (m/s) κ (106Nm−3/2) c (10−2sm−1) Fmax (kN) X b_max (mm) W k0 (J) Wcb (J) Wrb (J) tc (μ sec) e*
4 32.39 5.11 0.92 0.88 0.36 0.36 0.28 687 0.88

22 32.39 1.25 7.00 3.38 10.88 10.88 7.77 491 0.85
35 32.39 1.00 12.09 4.82 27.56 27.56 18.07 449 0.81
50 32.39 0.99 18.32 6.23 56.25 56.25 31.62 424 0.75
249 Appendix A
20
18
16
14
Contact force, kN 12
10
8
6
4
2
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Time, milliseconds
Figure A3 Simon model predictions of the compression and restitution of ball impacts against a
rigid barrier at speeds 4, 22, 35, and 50 m/s.
Equivalently, for any value of damping coefficient, c, the rebound speed can be incremen-
tally increased until the hysteresis loop closes. The rebound velocity at this point determines
the corresponding value of CofR.
Equation A1 can also be solved for velocity, x , from which the time steps can be calcu-
lated and the force, F, in turn plotted against time. The plot for each of the impact speeds
is given in Figure A3. The highest curve representing the 50 m/s impact is identical to the
force time graph obtained experimentally by Cochran (2002) using a pressure transducer
embedded into the steel barrier plate. Cochran also measured the contact times for the four
different ball speeds. These are compared with the Simon model in Figure A4.
Details of the analytical solution of the Simon equation can be found in Impact
Mechanics by Stronge (2000). I used Matlab (www.mathworks.com) computer code to
generate Figures A2 and A3.
We can now proceed to assess the difference when the collision is between a golf ball and
a club head with a spring face rather than with a heavy rigid plate.
0.70
Simon model
Contact time, milliseconds
0.65
Experimental
0.60
0.55
0.50
0.45
0.40
0 10 20 30 40 50
Ball speed, m/s
Figure A4 Effect of impact speed on contact time. Simon model predictions and experimental
measurements by Cochran (2002).
250 Appendix A
DR I V ER BA LL I MPACTS
To understand the nature of the restitution with the modern driver, it only matters that the
club should store energy elastically during compression. It does not make any difference to
the resulting CofR whether the energy is stored in a thin face through diaphragm deflection
or in a solid club through localized face compression. If we adopt the latter as illustrated in
Figure A5, then a simple expansion of Simon’s model will allow us to represent both the
club and the ball; see Timoshenko and Goodier (1970) for a full description of the theory of
curved contacting bodies to be used next.
We assign elastic modulus Ec, Poisson’s ratio υc , and radius of curvature R c to the club;
and elastic modulus Eb, Poisson’s ratio υ b , and radius rb to the ball. Stiffness κ in the
Simon’s model (Eq. 2A) is then given by
κ = 4 E * (R * )1/2/ 3, (A4)
where
E * = [(1 − υc2 )/ E c + (1 − υ2b )/ E b ]−1 = (1 / Ec* + 1 / E *b )−1 (A5)
and
R * = (R c−1 + rb−1 )−1 . (A6)
E *b and E c* are simply shorthand notation for E b/(1 − υ2b ) and Ec/(1 − υc2 ), respectively.
The single mass parameter in Eq. A1 is given by
m = (1 / M c + 1 / m b )−1 , (A7)
where M c = mass of club head, kg

m b = mass of ball, kg.
We consider two extreme cases: one case in which the club suffers the same proportional
energy loss as the ball and the other case with the club acting as a perfectly elastic body, that
is, with no internal energy loss.
Rc
rb
Figure A5 Impact between a solid elastic driver head and golf ball for application of the Simon
collision model.
251 Appendix A
For the first case we will assume that we have discovered a new material which can be
used to make a traditional wood driver with energetic CofR = 0.83, and that it is used with
a more resilient ball that has energetic CofR = 0.83. Thus the CofR for the combination,
using the normal velocity definition should turn out to be 0.83. The radii of curvature of
the impacting bodies, R c and rb, are assigned values 0.20 and 0.0213 m, respectively, cor-
responding to the average of bulge and roll for a traditional wood driver face and the radius
of a standard ball.
From the hertzian relationships used in the Simon model, the separate displacements of
the ball and the club during impact (x b and x c , respectively) are inversely proportional to
their respective elastic parameters E *b = E b/(1 − υ2b ) and E*c = Ec/(1 − υc2 ); see Timoshenko
and Goodier (1970).
Since x c + x b = x , the total deflection, we can write the separate displacements as
x c � x /(1+ E*c / E*b ) (A8)
and
x b � x /(1+ E *b / E *c ). (A9)
In solving this set of equations for a ball/club collision, the independent ball elastic
parameters E b and υ b will be 145 MPa and 0.433 as for the rigid barrier collisions. The
elastic parameters of the club were selected to satisfy a contact time of 0.45 ms. Interestingly,
if a value of 0.19 for Poisson’s ratio is chosen, representative of a plain persimmon wood
head, the required modulus is Ec = 430 MPa, two-thirds of the value of actual persimmon.
As for the rigid barrier collisions, the damping coefficient, c, is selected to satisfy the condi-
tion that the ball recovers at the instant of separation with the required ball speed for the
selected CofR value; that is, at time 0.45 ms and ball velocity x = ev 0 , where v 0 = 44.73 m/s
and e = 0.83 for a R&A/ USGA conforming driver.
Applying Simon’s model, with Eqs. A4 and A7, and with these initial and final conditions,
and assuming a CofR value for both the ball and the club head, gives the separate plots
of contact force versus club and ball displacement, x c and x b , respectively, as shown in
Figure A6. The principal predictions of the analysis are given in Table A3.
The parameters in Table A3 are
Fmax = maximum contact force, kN;

Wcc = energy used in compression of club head, J;
Wcb = energy used in compression of ball, J;
Wrc = energy recovered from club head compression, J;
Wrb = energy recovered from ball compression, J;
e *x = CofR of club head;
e *b = CofR of ball;
e * = CofR of the entire collision;
t c = contact time, s;
x c = displacement of club face, mm;
x b = displacement of ball, mm;
d cont = maximum diameter of contact area, mm.
252 Appendix A
14
12
10
Contact force, kN
8
4 Ball
Club
2
0
0 1 2 3 4 5
Figure A6 Application of the Simon collision model to 44.7 m/s, 0.83 coefficient of restitution
(CofR) impact in which both the ball and the club have a 0.83 energetic CofR.
It can be seen that the energy stored in the club and ball at the end of the compression
phase of impact is
Wcc + Wcb = 12.46 + 31.16 = 43.62 joules.
The energetic CofR measurement of impact efficiency can be applied separately to the club
and ball. The energy recovered during restitution is 8.58 J from the club and 21.46 from the
ball. Thus the CofR is e c* = 8.58 /12.46 = 0.83 for the club and e *b = 21.46 / 31.16 = 0.83 for
the ball, and so e * = 0.83 for the collision, in exact agreement with the values of initial and
final relative velocity.
It seems unlikely that a modern hollow titanium club head is as inefficient during the
impact as the ball. Also, energy losses in the hollow titanium shell clubs do not result from
internal viscoelastic frictional losses, as would be the case for an “improved” solid driver
head. The energy losses in this case are from deflections of the shell body and resulting vibra-
tions that continue after ball separation. However, the intention here is to bracket the per-
formance of the modern driver to illustrate the likely range of energy transfers. To do this we
now consider the changes for the same 0.83 CofR collision with no energy losses in the club,
that is, for a perfectly elastic club head.
For the case of a perfectly elastic club, we need to first consider the relationship between
the Simon model and a perfectly elastic hertzian collision. Referring back to Simon’s equa-
tion, we can easily obtain the maximum total deflection since it occurs at the end of compres-
sion when x = 0. Substituting this into Eq. A1 gives
x max = (Fmid / κ)2/3 , (A10)
Table A3 Simon model calculations for ball strikes with an elastic driver head—impact
speed 109 mph; coefficient of restitution = 0.83 for both club and ball
Fmax Wcc Wcb Wrc Wrb e c* eb* e* tc xc xb dcont

13.74 12.46 31.16 8.58 21.46 0.83 0.83 0.83 0.45 1.95 4.88 23.2
253 Appendix A
14
12
10
Contact force, kN 8
4 Ball 0.77 CofR

Club1.00 CorR
2
Club and ball both 0.83 CofR
0
0 1 2 3 4 5
Figure A7 Application of the Simon model to a 0.83 coefficient of restitution (CofR), 44.7 m/s
collision using a perfectly elastic club. Dashed lines represent 0.83 CofR for both the ball and
the club.
where Fmid is the value of the contact force at mid-impact, that is, the end of the compression
phase. We can solve the values of the contact force, from the Simon model, for the entire
range of the compression phase from x = 0 to x = x max . If we then assign all of the damp-
ing to the ball, the club must in consequence be perfectly elastic, and it will respond to this
increasing contact force in accordance with the hertzian equations for elastic contact. Thus
we can write
x c � (F/κ)2/3/(1+ E c* / E *b ), (A11)
and then the ball displacement is simply
x b = x − x c . (A12).
Figure A7 shows the result of solving Simon’s equation, using the same material proper-
ties and damping coefficient, but with the division of the total displacement according to
Eqs. A11 and A12. This seems to be completely consistent since both the total deflection and
the maximum force are unchanged from the Simon model. This of course involves a different
ball. It is less efficient because it contains all of the damping.
For comparison, the previous results for equal impact efficiency of the club and ball are
shown with gray lines. Note that with internal damping, the force during compression rises
faster than for a purely elastic body and “leads” the deflection. Maximum deflection thus
occurs after maximum force, as the body continues to slow down to a stop at the maximum
deflection point while the force drops. During restitution, the same thing occurs, and the
displacement reduces more slowly than in an elastic body. With reference to the club, maxi-
mum displacement of the impacting bodies occurs where the solid and dash lines cross, and
the force at this point is Fmid . Thus, at this point, the elastic club head has unloaded back
along the loading line, by the amount outside of the gray hysteresis loop. It then continues
to unload along the original loading path during restitution. The principal results from this
analysis are given in Table A4.
254 Appendix A
Table A4 Simon model calculations for ball strikes with an elastic driver head—impact
speed 109 mph; coefficient of restitution = 1.00 for the club and 0.77 for the ball
Fmax Wcc Wcb Wrc Wrb e c* eb* e* tc xc xb dcont

13.74 10.39 33.23 10.39 19.65 1 0.77 0.83 0.45 1.95 4.88 23.2
It can be seen that the energy stored in the club and ball at the end of the compression
phase of impact is
Wcc + Wcb = 10.39 + 33.23 = 43.62 joules.
In this case, e c* = 10.39 /10.39 = 1.00 for the club and e *b = 19.65 / 33.23 = 0.77. For the
collision, e * = (10.39 + 19.65)/(10.39 + 33.23) = 0.83, again in exact agreement with the
values of initial and final relative velocity.
In summary, the Simon “ball” is used to bracket the performance of a modern conform-
ing ball to illustrate the likely range of energy transfer to the elastic face of the driver. For
simplification of modeling, a solid driver head has been assumed. For such a head to pro-
duce 0.830 CofR, its modulus must be approximately 430 MPa. This value can be found
among the rigid thermoset elastomers. The only material to be used in golf club design
that has a modulus within the same order of magnitude is wood, used with the rings ori-
ented to the face. The persimmon used by the Louisville Golf Company has modulus
of E c* = 644 /(1 − 0.192 ) = 668 MN/m 2 . However, face inserts are used by the Louisville
Golf Company. These are made from high molecular-weight polyethylene (HDPE) with
approximately the same modulus as the persimmon. It is interesting to postulate that the
use of inserts manufactured with a relatively stiff thermoset elastomer (modulus around
400 MN/m 2) might provide better performance from a wood-head driver.
A P P E N DI X B
Distance error propagation in putting*

In Chapter 7 we discussed three putting swing models which may be taken to represent the
range of putting styles seen on the course. The discussion of error there was deliberately sim-
ple to make one main point of potentially large importance; namely that the widely taught
method of maintaining a constant swing speed incorporates a magnifying effect of swing
length errors on putt distances. We consider this here in a more formal manner.
Errors in a particular putting style may be separated into two types. The first type is the
random variability of the parameters which are intended to be fixed. We will assume such
random errors are relatively small. For the three models under consideration these fixed
parameters are swing angle A, average applied torque T, and swing time t.
In contrast, for each swing style there is a single parameter which is intended to be adjusted
to produce different putt distances. These adjustments involve difficult decision making and
as such are subject to much greater variability, when compared to the unknown exact value
to be used. For the three models these are either A or T.
For the constant swing length model, illustrated in Figure 7F, there will be some ran-
dom variation in the swing angle A. We define this error as ε cA , where superscript c denotes
“practiced to be constant.” The error is expressed in the units of the parameter, in this case,
degrees. There will also be errors of judgment in the applied torque for any particular putt,
and we define this as ε T , the absence of superscript c denoting that this may be subjected
to significantly larger errors of judgment for particular putts. The product of the swing angle
and the average torque applied during the swing is equal to the energy change in the swing-
ing system, which can be expressed as (1 / 2)Iω 2 where I is the MoI of the shoulders, arms,
and club about the pivot, and ω is the angular velocity at impact. The effect of gravity is
taken to be negligible. Since the entire system is a single pendulum, the amount of energy
transferred to the club head is a fixed proportion k 0 of this total amount. So we can write
2 TA = Iω 2 = (1/k 0 )MV 2, (B1)
where M = club head mass and V = impact velocity.

Next we apply the CofR and mass factors for the strike to relate the kinetic energy of the
club head to the kinetic energy of the ball as
(1/ 2 ) MV 2 = (1/ 2 ) M  v (1 + m / M ) / (1 + e n ) = k 1 (1/ 2)mv 2, (B2)

2
where m = ball mass, v = ball launch speed, and where the CofR, e n , can be treated as a con-
stant and in consequence so can k 1 .
* SUPPORTING PHYSICS—skip without any loss of understanding of the science principles

of the game
255
256 Appendix B
Substituting Eq. B2 into Eq. B1 gives
TA (k 0 / k 1 ) = (1/ 2)mv 2. (B3)
If the rolling resistance, including the gravitational effect on a downslope or upslope, is f r

and the putt length is L, then all of the ball energy at launch is absorbed in the work against
the rolling resistance, equal to f r L. Finally, from Eq. B3, we have putt length given by
L = (k 0 /(f r k 1 ))TA = k 2 TA , (B4)
where k 2 is a constant for any particular putt. Applying the accepted multivariate propaga-
tion of error formula in statistics (Navidi 2011) we can write the error in putt distance as
ε L / L = (ε cA / A )2 + (ε T / T)2 . (B5)
For example if the normal error band in the fixed angle of swing A is (ε cA / A ) = ±0.05 and the
normal error band in the applied torque is (ε T / T) = ±0.1 then Eq. B5 gives (ε L / L) = ±0.11,
or ±11 percent.
With the constant torque model, Eq. B4 still applies but now the practice is to keep the
torque constant and the more difficult judgment is making the correct swing angle. So the
change in the error propagation formula is simply
ε L / L = (ε cT / T)2 + (ε A / A )2 . (B6)

Appropriate error bands might be (ε cT / T) = ±0.05 and (ε A / A ) = ±0.1. Then Eq. B6 gives
(ε L / L) = ±0.11 as before. If we assume that keeping the torque fixed may be a more difficult
challenge than maintaining a fixed backswing angle, then we might write (ε cT / T) = ±0.1 and
(ε A / A ) = ±0.1, and so the expected putt length error range becomes (ε L / L) = ±0.14. Note
that the separate errors sometimes act in concert and sometimes act in opposition, which is
why their combined effect is reduced through the root-square-addition formulation.
In contrast, maintaining a constant swing time t, both swing angle and torque must
change in the same proportion to keep the swing angle and club head velocity in the same
proportion. Thus we have
T = Ak 3 t , (B7)
where k 3 is a constant for a particular player. Following the steps for the other models, we
can write from Eq. B4,
L = k 2 TA = (k 2 k 3 )tA 2. (B8)
Because the angle effect is raised to power 2, the multivariate propagation of error formula
(Navidi 2011) becomes
ε L / L = (ε ct / t)2 + (2 ε A / A )2 , (B9)
where the effect of ε A has been multiplied by the exponent 2.

257 Appendix B
If we now apply the same error bands as before, that is, a normal swing time error band
(ε ct / t) = ±0.05 and a corresponding angle error band (ε A / A ) = ± 0.1 then Eq. B9 gives
(ε L / L) = ±0.21, or 21 percent as discussed in Chapter 7.
Note that even if a player’s rhythm is perfect, that is, (ε ct / t) = 0 , the putt length error in
this example only reduces from ±0.21 to ± 0.20. Also, although the discussion has been
presented in terms of likely error bands, occasional large swing angle errors will produce a
double distance error with substantial consequence.
As stated earlier, the testing of the relative effectiveness of any different swing styles must
be done on a large scale with careful controls and the application of sound statistics. The
fact that players with the extraordinary ability to visualize the roll and control their muscles
accordingly may use the constant time method to great effect, is not evidence that it is the
most effective means of controlling distance. However, if humans do naturally use constant
time swings to project objects, we may have a hardwired ability both to keep the swing time
constant and to visualize the nonlinear relationship between swing angle and the projection
distance, which could clearly shift the balance of effectiveness. The limited study of 4- and
6-year-olds by Van Rossum and Bootsma (1989) does not provide clear evidence either way.
The study does show that some 6-year-olds naturally adopt the added simplification of fixing
the release angle and changing only the launch speed when throwing for accuracy. However,
it was not reported whether they changed the launch speed by changing backswing length,
torque, or both.
It should be noted that the nonlinear behavior of the constant time swing allows a very
large range of putt distances without unreasonably large swing angle and torque values. In
contrast, the two alternative models may require an increase above the fixed torque or the
fixed backswing beyond a particular level of putting power. If we were to examine in detail
the swings of a large number of players, we would no doubt find a range from shorter swings
with larger increases of average torque as the swing length increases in small steps, to longer
swings with smaller increases in average torque as the swing length increases in larger steps.
At the very middle of these two classes lies the constant time swing, and at the outer limits
lie the other two models.
R EFER ENCES
Arakawa, K., Mada, T., Komatsu, H., Shimizu, T., Satou, M., Takehara, K., and Etoh, G.
“Dynamic Contact Behavior of a Golf Ball During Oblique Impact.” Experimental
Mechanics 46 (2006): 691–697.
Ashby, M. F. Materials Selection in Mechanical Design. 3rd ed. Amsterdam: Elsevier, 2005.
Bearman, P.W. and Harvey, J. K. “Golf Ball Aerodynamics.” Aeronautical Quarterly 27
(1976): 112–122.
Benson, T. Shape Effects on Drag. NASA, Glenn Research Center, August 2010.
Bernoulli, D. Hydrodynamica. 1738.
Boothroyd, G., Dewhurst, P., and Knight, W. A. “Selection of Materials and Processes.”
Design for Manufacture and Assembly. 3rd ed. (Ch. 2). Boca Raton, FL: CRC Press, 2011.
Brevard, D. “The Stimpmeter: Friend or Foe?” USGA (www.usga.org), November, 2010.
Butler, J. H. and Winfield, D. C. “The Dynamic Performance of the Golf Shaft During the
Downswing.” In Science and Golf II: Proceedings of the World Scientific Congress of Golf,
edited by A. J. Cochran and M. R. Farrally, 259–264. London: E & FN Spon, 1994.
Cochran, A. J. “Development and Use of One-dimensional Models of a Golf Ball.” Journal of
Sports Sciences 20, no. 8 (August 2002): 635–641.
Cochran, A. and Stobbs, J. Search for the Perfect Swing. Chicago: Triumph Books, 1999. First
published 1968 by The Golf Society of Great Britain.
Cornish, J., Otto, S., and Strangwood, M. “Modeling of Oblique Impact of Golf Balls.” In The
Engineering of Sport 7, (2009): 669–675, 2009.
Cross, R. “The Trajectory of Lawn Balls.” American Journal of Physics 66, no. 8 (1998).
Cross, R. “The Bounce of a Ball.” American Journal of Physics 67, no. 3 (1999).
Cross, R. “Measurements of the Horizontal Coefficient of Restitution for a Superball and a
Tennis Ball.” American Journal of Physics 70, no. 5 (2002a).
Cross, R. “Grip-slip Behavior of the Bouncing Ball.” American Journal of Physics 70, no. 11
(2002b): 1093–1102.
Cross, R. “Bounce of a Spinning Ball near Normal Incidence.” American Journal of Physics
73, no. 10 (2005).
Cross, R. “Impact of a Ball on a Surface with Tangential Compliance.” American Journal of
Physics 78, no. 7 (2010): 716–720.
Cross, R. Physics of Baseball & Softball. New York: Springer Science & Business Media, 2011.
Cross, R. “Oblique Bounce of a Rubber Ball.” Experimental Mechanics (2014). doi:10.1007/
s11340-014-9938-3.
Cross, R. and Bower, R. “Effects of Swing Weight on Swing Speed and Racket Power.” Journal
of Sports Sciences 24 (2006): 23–30.
Cross, R. and Nathan A. M. “Performance Versus Moment of Inertia of Sporting Implements.”
Sports Technology 2 (2009): 7–15.
259
260 References
Daish, C. B. The Physics of Ball Games. London: The English Universities Press, 1972:
821–828.
Davies, J. M. “The Aerodynamics of Golf Balls.” Journal of Applied Physics 20 (1949).
Dewhurst, P. “On the Non-uniqueness of the Machining Process.” Proceedings of the Royal
Society of London, A 360 (1978): 587.
Dewhurst, P. and Reynolds, C. R. “A Novel Procedure for the Selection of Materials in
Product Design.” Journal of Materials Engineering and Performance 6 (1997): 359–364.
Dieter, G. E. Engineering Design. London: McGraw Hill, 1983.
Farrally, M. R. and Cochran, A. J., eds. Science and Golf III: Proceedings of the World Scientific
Congress of Golf. Champaign, IL: Human Kinetics, 1998.
Garwin, R. L. “Kinematics of an Ultraelastic Rough Ball.” American Journal of Physics 37,
no.1 (1969): 88–92.
Goldsmith, W. Impact. New York: Dover Publications, 2001.
Greenwad, R. M., Penna, L. H. and Crisco, J. J. “Differences in Batted-ball Speed with Wood
and Aluminum Baseball Bats: A Batting Cage Study.” Journal of Applied Biomechanics
17 (2001): 241–252.
Grober, R. D. “Resonance in Putting.” Open access article; arxiv 0903.1762, 2011a.
Grober, R. D. “The Geometry of Putting on a Planar Surface.” Open access article; arxiv
1106.1698, 2011b.
Guest, P. G. “Rolling Precession.” American Journal of Physics 33 (1965): 446–448.
Haake, S. J. “Apparatus and Test Methods for Measuring the Impact of Golf Balls on Turf.”
PhD thesis, University of Aston, 1989.
Haake, S.J. “The Impact of Golf Balls on Natural Turf: A Physical Model of Impact.” Applied
Solid Mechanics 4, edited by A. R. S. Ponter and A. C. F. Cocks, 72–79. London: Elsevier
Applied Science, 1991a.
Haake, S. J. “The Impact of Golf Balls on Natural Turf II.” Journal of the Sports Turf Research
Institute 67 (1991b): 128.
Hertz, H. “On the Contact of Elastic Solids.” Journal für die reine und angewandte Mathematik
92 (1882): 156–171.
Hetenyi, M. Beams on Elastic Foundations. Ann Arbor: University of Michigan Press, 1946.
Hill, R. “On the State of Stress in a Plastic Rigid Body at the Yield Stress.” Philosophical
Magazine 42 (1951): 868.
Holmes, B. W. “Dialogue Concerning the Stimpmeter.” The Physics Teacher 24, no. 7
(1986): 401–404.
Holmes, B. W. “Putting: How a Golf Ball and Hole Interact.” American Journal of Physics 59
(February 1991): 2.
Horak, K. “Determination experimentale du coefficient apparent de restitution tangen-
tialle.” Jednota Ceskosl. Mat. Fys. 66 (1937): 131.
Horwood, G. P. “Golf shafts—A Technical Perspective.” In Science and Golf II: Proceedings
of the World Scientific Congress of Golf, edited by A. J. Cochran and M. R. Farrally,
259–264. London: E & FN Spon, 1994.
Hurrion, P. D. and Hurrion, R. D. “An Investigation Into the Effect of the Roll of the
Golf Ball Using the C-Groove Putter.” In Science and Golf IV: Proceedings of the
World Scientific Congress of Golf, edited by E. Thain, 531–538. Abingdon, England:
Routledge, 2002.
Johnson, S. H. and Ekstrom, E. A. “Experimental Study of Golf Ball Oblique Impact.” In
Science and Golf III: Proceedings of the World Scientific Congress of Golf, edited by
M. R. Farally and A. J. Cochran, 519–525. Champaign, IL: Human Kinetics, 1998.
261 References
Johnson, S. H . and Lieberman, B. B. “An Analytical Model for Ball Barrier Impact—II: Oblique
Impact.” In Science and Golf II: Proceedings of the World Scientific Congress of Golf, edited
by A. J. Cochran and M. R. Farrally, 315–320. London: E & FN Spon, 1994.
Johnson, S. H. and Lieberman, B. B. “Experimental Determination of Apparent Contact
Time in Normal Impact.” In Science and Golf IV: Proceedings of the World Scientific
Congress of Golf, edited by E. Thain, 524–530. Abingdon, England: Routledge, 2002.
Johnson, S. H. and Lieberman, B. B. “Normal Impact Models for Golf Balls.” In The
Engineering of Sport, edited by S. Haake, 250–256. Balkema, Rotterdam: Taylor and
Francis, 1996.
Kolski, H. “An Investigation of Mechanical Properties of Materials at a Very High Rate of
Loading. Proceedings of Physical Society of London B62 (1949): 676–700.
Lieberman, B. B. “The Effect of Impact Conditions on Golf Ball Spin Rate.” In Science and
Golf: Proceedings of the First World Scientific Congress of Golf, edited by A. J. Cochran,
225–230. London: E & FN Spon, 1990a.
Lieberman, B. B. “Estimating Lift and Drag Coefficients from Golf Ball Trajectories.” In
Science and Golf: Proceedings of the First World Scientific Congress of Golf, edited by
A. J. Cochran, 187–192. London: E & FN Spon, 1990b.
Lieberman, B. B. and Johnson, S. H. “An Analytical Model for Ball Barrier Impact.” In Science
and Golf II: Proceedings of the World Scientific Congress of Golf, edited by A. J. Cochran
and M. R. Farrally, 309–314. London: E & FN Spon, 1994.
Lyons, R. G. Understanding Digital Signal Processing. New Jersey: Prentice Hall, 2011.
Marsh, David “A Method for Quantifying the Degree of Imbalance of a Golf Ball.” Personal
communication to Dewhurst, 2013.
Mase, T. “Experimental Benchmarking of Golf Ball Mechanical Properties.” Unpublished
manuscript, East Lansing: Michigan State University, 2003.
Maw, N. “A Theoretical and Experimental Investigation into the Impact and Rebound of
Elastic Bodies.” PhD thesis, Sunderland Polytechnic, England, 1976.
Maw, N., Barber, J. R., and Fawcett, J. N. “The Oblique Impact of Elastic Spheres.” Wear 38,
no. 1 (1976): 101.
McGinnis, R. S. and Nesbit, S. “The Golf Club Deflection and a Function of the Swing Hub
Path.” The Open Sports Sciences Journal 3 (2010): 155.
Mehta, R. D. and Pallis, J. M. “Sports Ball Aerodynamics: Effects of Velocity, Spin and
Roughness.” In Materials and Science in Sports, edited by F. H. Froes, 186–197.
Warrendale, PA: TMS, 2001.
Michal, G. M. and Novak, M. D. “Designing for Ball Impacts.” Journal of Advanced Materials
and Processes 159, no. 9 (2001): 31–35.
Milne, R. D. and Davis J. P. “The Role of the Shaft in the Golf Swing.” Journal of Biomechanics
25, no. 9 (1992): 975–983.
Monk, S. A., Davis, C. L., Otto, S. R., and Strangwood, M. “Material and Surface Effects on
the Spin and Launch in Golf.” Sports Engineering 8, no. 1 (2005): 3–11.
Mukhopadhyay, S., Fangueiro, R., Arpac, Y., and Senturk, U. “Banana Fibers—Variability
and Fracture Behavior.” Journal of Engineered Fibers and Fabrics 3, no. 2 (2008): 39–45.
Navidi, W. Statistics for Engineers and Scientists. New York: McGraw Hill, 2011.
Nesbit, S. M. “A Three Dimensional Kinematic and Kinetic Study of the Golf Swing.” Journal
of Sports Science & Medicine 4 (2005): 499–519.
Nesbit, S. M. and Serrano, M. “Work and Power Analysis of the Golf Swing.” Journal of Sports
Science & Medicine 4 (2005): 520–533.
262 References
Pelz, D. Dave Pelz’s Putting Bible: The Complete Guide to Mastering the Green. New York:
Doubleday, 2000a.
Pelz, D. Dave Pelz’s Short Game Bible: Master the Finesse Swing and Lower Your Score.
New York: Doubleday, 2000b.
Penner, A. R. “The Physics of Golf: The Convex Shape of the Driver.” American Journal of
Physics 69 (October 2001): 1073–1081.
Penner, A. R. “The Physics of Putting.” Canadian Journal of Physics 80, no. 2 (2002a): 83–96.
Penner, A. R. “The Run of a Golf Ball.” Canadian Journal of Physics 80, no. 8 (2002b): 931–940.
Pringle, M. M. “Characteristic Spring Constants for Golf Club Heads.” In Science and Golf
IV: Proceedings of the World Scientific Congress of Golf, edited by E. Thain, 402–409.
Abingdon, England: Routledge, 2002.
Proulx, T., ed. Dynamic Behavior of Materials. Vol. 1. Proceedings of the 2010 Annual
Conference on Experimental and Applied Mechanics. New York: Springer, 2010.
R&A/USGA. Interim Report on the Study of Spin Generation, 2006 (available at http://www.
randa.org/).
R&A/USGA. Second Report on the Study of Spin Generation, 2007(available at http://www.
randa.org/).
Roach, N. T., Venkadesan, M., Rainbow, M. J., and Lieberman, D. E. “Elastic Energy Storage
in the Shoulder and the Evolution of High-Speed Throwing in Homo.” Nature 498,
no.7455 (2013): 483–486.
Sharpe, R. S. “On the Mechanics of the Golf Swing.” Proceedings of the Royal Society of
London, A 465 (2009): 551–570.
Simon, R. “Development of a Mathematical Tool for Evaluating Golf Club Performance.”
Proceedings of ASME Design Engineering Congress, New York, 1967.
Smith, L., Broker, J. and Nathan, A. “A Study of Softball Player Swing Speed.” In Sports
Dynamics, Discover and Application, edited by A. Subic, P. Trivailo, and F. Alam, 12–17.
Melbourne: RMIT University, 2003.
Smits, A. J. and Smith, D. R. “A New Aerodynamic Model of a Golf Ball in Flight.” In Science
and Golf II: Proceedings of the World Scientific Congress of Golf, edited by A. J. Cochran
and M. R. Farrally, 340–347. London: E & FN Spon, 1994.
Stronge, W. J. Impact Mechanics. Cambridge, England: Cambridge University Press, 2000.
Tamaogi, T. and Sogabe, Y. “Determination of Viscoelastic Models for 2-Piece Golf Balls
Using Polymeric Split Hopkinson Pressure Bar.” Journal of the Japanese Society for
Experimental Mechanics 10 (Special Edition 2010) S174–S179.
Tanaka, K., Sato, F., Oodaira, H., Teranishi, Y., Sato, F., and Ujihashi, S. “Construction of the
Finite-Element Models of Golf Balls and Simulations of Their Collisions.” Proceedings
of the Institution of Mechanical Engineers, Part L. Journal of Materials: Design and
Applications 220 (2006): 13–22.
Thain, E., ed. Science and Golf IV: Proceedings of the World Scientific Congress of Golf,
461–473. London: Routledge, 2002.
Thomas, F. Just Hit It. Champions Gate, FL: Frankly Golf Publications (ISBN:
978-0-615-18109-7), 2008.
Thomas, F. From Sticks and Stones. Champions Gate, FL: Frankly Golf Publications
(ISBN: 978-0-615-46171-7), 2011.
Thomas, F., and Melvin, V. The Fundamentals of Putting. Champions Gate, FL: Frankly Golf
Publications (ISBN: 978-0-615-72691-5), 2012.
Timoshenko, S. P. and Goodier, J. N. Theory of Elasticity. New York: McGraw-Hill, 1970.
263 References
Tuxen F. Trackman Newsletters Numbers 1 to 10, Nov. 2007 through Jan. 2014. http://
trackmangolf.com/company/newsletter
Tuxen, F. “Trackman at Wendy’s 3-Tour Challenge.” TrackmanTM News, January 2008.
http://trackmangolf.com/company/newsletter
Tuxen, F. “Focus: Smash Factor.” TrackmanTM News, May 2008. http://trackmangolf.com/
company/newsletter
Tuxen, F. “Focus: The Secret of the Straight Shot II.” TrackmanTM News, July 2009. http://
trackmangolf.com/company/newsletter
Tuxen, F. “Insight: PGA Tour Data.” TrackmanTM News, January, 2010. http://trackman-
golf.com/company/newsletter
Tuxen, F. “Trackman’s Ten Fundamentals.” TrackmanTM News, October 2010. http://track-
mangolf.com/company/newsletter
Tuxen, F. “Trackman Innovation: Partial wedges—pitch shots” TrackmanTM News, January
2014. http://trackmangolf.com/company/newsletter
Tuxen, F. TrackMan A/S, U.S. Patents 2007/0293331, 2008/0139330, 2009/0295624,
2009/0075744.
Van Rossum, J.H.A and Bootsma, R.J. “The Underarm Throw for Accuracy in Children.”
Journal of Sports Sciences 7 (1989): 101–112.
Wishon, T. The Search for the Perfect Driver. Sports Media Group, 2006.
Wishon, T. The Right Sticks. Sports Media Group, 2008.
Wishon, T. A series of three shaft fitting articles for Golf Digest. www.golfwrx.com/author/
tomwishon/, 2013 (Tom Wishon Golf Technology, Durango, CO).
Wood Handbook: Wood as an Engineering Material. Madison, WI: Forest Products Laboratory,
United States Department of Agriculture Forest Service, 1999.
Wu, C.-Y., Thornton, C. and Li, L.-Y. “A Semi-analytical Model for Oblique Impacts of
Elastoplastic Spheres.” Proceedings of the Royal Society of London, A 465 (2009): 937.
A B OU T T H E AU T H O R
Peter Dewhurst’s career in mathematics and engineering research was unantici-

pated. From high school he started as a student apprentice with an engineering
company in the North of England. Through co-operative studies at a local commu-
nity college he obtained a Higher National Diploma in Production Engineering and
became a production equipment designer. Unsatisfied with his work in industry,
with the financial support of his young wife Iris, he gave up a starting career as a
Production Engineer to enroll as a freshman in the Department of Mathematics
at the University of Manchester. Just over three years later, after graduating first in
class in the mathematics program, he was the winner of the Sir Charles Reynolds
Research Fellowship with the freedom to choose his own research program for
a doctoral degree. At the same time he was appointed to a part-time position as
Regional Tutor in mathematics for the Open University.
Dewhurst’s initial research was on the subject of short duration dynamic events,
in particular concerning the impulsive effects of contact explosions and the pos-
sibility of using the energy in large scale forming of aerospace parts. He continued
to work in theoretical and experimental mechanics for a number of years, concen-
trating on problems in material deformation. He moved from Britain to the United
States in 1980, and while maintaining an interest in engineering mechanics, he
switched his attention to the problems of conceptual product design. With former
colleague Geoffrey Boothroyd, he combined his earlier experience of production
processes with mathematical modeling to establish novel methods for evaluating
conceptual designs of products for their ease of manufacture. He is co-founder
of Boothroyd Dewhurst Inc., a concept design consulting and software company
(www.dfma.com) that continues to have a major impact on product design.
Dewhurst spent 1999 consulting for NASA on manufacturing issues with
the design of the Fastrac rocket engine, an initiative to lower the cost of satellite
launches. This sparked his interest in optimal aerospace design; and on returning to
University research in 2000, he changed the direction of his research interests once
more to develop methods for the design of absolute minimum-weight structures.
In more recent years, he has been the leader of three research programs in optimal
structural design: two funded by the National Science Foundation and one by the
satellite group of Sandia National Laboratories. Most recently he became interested
in the impact mechanics of sports and the design of sports equipment.
Since the start of his career in research, Peter Dewhurst has won numerous
national and international honors and awards, including the United States National
Medal of Technology, presented to him by President George H. W. Bush.
265
266 About the Author
In addition to teaching and research, Dewhurst has consulted for many corpora-
tions including: Abbot Laboratories, Apple, Ford, General Electric, General Motors,
Hewlett Packard, IBM, Isuzu Motors, McDonnell Douglas, Motorola, Samsung,
Seiko-Epson, Sikorski Aircraft, United Technologies, and Vitro Corporation.
Dewhurst is Professor Emeritus in Theoretical and Applied Mechanics, and
Industrial Engineering, at the University of Rhode Island, where he was two-time
winner of the Carlotti Award for research excellence.
I N DE X
action and reaction, 19, 39 ball restitution, 4, 63, 118, 125, 137, 192,
address position, 39, 49, 50, 149, 250, 252
187, 211 ball speed factors, 42, 67, 79, 83, 206
aerodynamic force, 16–30, 31–38 ballistic screen, 9
aerodynamic performance, 10–11 baseball, 9, 32, 69, 88, 135
air density, 24–25 blade, irons, 5, 6
air mass, 30, 32 bounce and roll, 3, 4, 10, 43, 75, 96–98, 108,
air speed, 10, 17, 38, 85 154, 155
critical, 18 braking power, 118
supercritical, 18 braking torque, 6, 7, 21, 89, 99, 126
aircraft, 16, 18–19, 27–28, 109
athletic shoes, 4, 100 Callaway, 6–9, 11, 72, 123, 172,
atmospheric conditions, 24–25, 32, 35 178–182, 206
atmospheric pressure, 24 carry, 13
attack angle, 8, 26, 40, 58, 62, 74, 86, 113, cavity back, 5, 173, 191
147, 162, 193, 198 center of mass (CoM), 9, 58–59, 60, 62
automobile, 24 centrifugal force, 51, 52, 55, 58–60
dampers, 72 components, 61
tires, 100 chip shot, 149
club, effective impact mass, 70–71, 73,
backswing, 39, 47–48, 50–51, 55, 60, 78, 86–89, 183
208, 210–213, 256–257 club face, 4, 6, 39, 40
balata ball, 96–97, 107–108 angled, 26, 27, 37, 57
ball compression, 77, 89, 247 bulge, 122–124, 173, 180, 192
ball diameter, 18, 214 closed, 147
ball flight, 1, 6, 8, 10–11, 13, 15, 17–18, friction, 95
20–23, 25, 28, 31–32, 35–39, 44, frictionless, 63, 101
174, 180 grip or stick, 4, 6, 95, 96, 100, 110, 118
curved, 26–28, 48, 147, 150, 166–172 open, 147, 177
distance, 44, 74, 96, 177, 178 performance, 8
side movement, 124, 152–159, 166–170, roll, 122–124, 130, 173, 180
175, 180, 223, 226–227 rotation, 60, 154
time, 26, 30, 43, 122, 152–154, 156, 158, slip/skid, 113
159, 192, 193 speed variations, 121
trajectory, 109 spring effect, 71, 76
ball mass-MoI factor, 111, 130 sweet area, 44
ball radius, 34, 35 sweet spot, 81
267
268 Index
club head drag, 10, 16–18

cannon testing, 103 induced, 10
rotation, 5, 120, 131, 174, 176, 183, 187, driver design, 6, 173, 177, 179
189, 198 CofR, 187
club length, 52, 53 MoI, 187
club path, 26, 27, 29, 41, 48, 57, 62, 147, 158, driver head
159, 173, 175, 177, 193, 204, 205 bulge, 122–124, 173, 180, 192
horizontal, 58, 148, 153 roll, 122–124, 130, 173, 180
vertical, 148 driver strikes, 179
coefficient of restitution, CofR, 4, 9, 41, 43, drives
89, 173, 179, 185, 206 launch conditions, 8, 10, 74, 97
effect of impact force, 89 longest, 192
effect of impact speed, 43, 72, 93 driving power, 77
effect of loft, 73, 89 driving torque, 48, 49
estimating the value, 67, 70, 71 draw, 147
factor, 44 DuPont, 96
kinetic, 76, 77
measuring the value, 68, 69 eagle, 4
normal CofR, 63, 76, 83, 91, 93 effective impact mass, 68–71
spin CofR, 95, 98, 100–107, 109–112, 193 energy storage, 77, 78, 96, 110, 119
tangential CofR, 85
CofR-MoI index, 185, 187, 198 face angle, 27
CofR limit for competition, 68 horizontal, 147–150
compression test, 9 vertical, 62, 147–150
contact force, 40–41 face friction, 63, 95
average, 41 fade, 147
maximum, 41 fairway, 27, 40, 43, 57, 58, 96, 97, 107–109,
normal, 99 122–123, 154, 192
profile, 41 width, 30
tangential, 84 Ferrari, 730
contact time, 3, 37 flight time, 22, 25, 26, 30, 38, 43, 122,
consequences of short time, 39–40 152–154, 156, 159, 192, 193
co-ordinate system, 35 flight trajectory, 8, 11, 15, 21, 23, 26–30
global, 194 force transducer, 93
local, 194 friction force, 71, 89, 137, 191
cricket, 32 friction reversal, 99, 137
curved ball flight, 26, 147
curved shots, 146, 158 gear effect, 120, 121–124
golf ball
de-lofting, 57, 58, 74, 79–81 British ball, 34
Denver, 24 cannon testing, 98
deviation, lateral, 29, 151 construction, 107
diaphragm spring, 7, 8, 44, 68, 77 core, 109, 117, 118, 135, 139
dimples, 4, 10 dimples, 4, 10, 15, 16, 18, 19, 32, 227
effect on drag, 16–18 distance, ball, 105
effect on lifting (lateral) force, 18–20 floating, 24, 25, 85, 221
game without dimples, 15 gutta percha, 18, 95
269 Index
high-spin ball, 104, 105 ionomer, 96, 104, 105

internal friction, 8, 72, 118 investment casting, 5, 6
moment of inertia (MoI), 103, 104,
111, 207 kick point, 42, 59, 60
orientation, 27 kinetic energy, 16, 52, 75–77, 126,
premium, 5, 11, 23, 24, 96, 109, 118, 206, 209
135, 224
smooth, 4 lateral deviation, 29, 151
Surlyn, 96, 104 lateral force, 32, 33, 34
symmetry, 27 launch angle, 8, 10, 11, 24–26, 30, 62
trajectory, 109 horizontal, 147
unbalanced, 221–225 vertical, 147
urethane, 4, 96, 105, 107, 108, 118 launch
weight, 24, 28, 34 conditions, 8, 10, 74, 97
grass juice, 107, 111, 118 optimum, 8, 10, 15, 96, 97
Great Big Bertha, 11, 123, 173, 178, 179, launch to loft ratio, 62, 63
182, 187 lie angle, 49, 149, 153
Greens, 15, 22, 26, 43, 96, 98, 109, 118, 201 lifting force, 19, 20, 22, 28, 150
invisible imperfections, 203 links courses, 26
slope, 205, 207, 217, 219 Liquidmetal, 8
speed, 99, 202, 204, 205 loft
stimp rating, 205 de-lofting, 57, 58, 74, 79–81
grooves, 6, 104, 105, 107–108, 118 determining the value, 83
groundspeed, 25, 30 dynamic loft, 7, 57–59, 60, 62
gyroscopic stability, 27 effect on ball speed, 63–66
effect on CofR, 67
hardwood, 9 effect on launch angle, 62, 63
Harrier, 4 horizontal loft, 148
Haskell ball, 95 increase of driver loft, 7, 59
hexagonal dimple, 10, 11, 32 spin loft, 85
Honourable Company of Edinburgh Louisville Golf Company, 9
Golfers, 1 LPGA, 5, 11, 12, 23, 111, 112,
hook, 25, 27, 29, 122, 149, 151, 152, 154, 118, 120
158, 175, 177, 193
hosel, 56, 57, 59, 69, 173 Magnus force, 31
hybrid club, 65, 152 Magnus, Gustav, 31
matrix equation, 198
impact conditions, 118, 174, 176 micro slip, 100
impact performance, 9, 179 modulus, 9, 118, 239, 240
impact period, 39 MoI limit for competition, 187
impact speed, 4, 10, 47 molasses, 3, 4, 117
effect on backspin rate, 73, 103, 111, 112 moment of inertia (MoI), 5–7, 9, 46, 47, 51,
effect on ball speed, 42 103, 111, 173, 229
impulse, 2, 40 principal components, 173, 180
normal, 86 momentum, 2, 40, 42
tangential, 120 conservation, 42, 44, 64
impulse moment, 141, 196 muscle, 51, 62, 77–78, 257
270 Index
Newton, 2, 3, 6, 19, 21, 42, 44, 68, 189, 202, constant swing length, 210
206, 234 constant time swing, 211, 213, 257
normal contact force, 84, 86, 99, 194 gravity pendulum, 211
rhythm, 211
off-center ball strike, 5, 7, 8, 39, 44, single pendulum, 209
120–122, 134, 171, 173, 179–180, 183, tempo, 211–213
185, 189, 191–192, 194, 201 underarm throwing, 209, 213, 214
offset ball strike, 179, 183
optimum launch conditions, 8, 10, 15, 16, radian, 27
35, 96, 97, 192 Ryder Cup, 52
overall distance standard (ODS), 10,
96, 192 scratch golfer, 54
shaft bending, 60–61, 79
patent, 7, 8, 118, 119 shaft kick point, 42, 59–60
perimeter weighting, 5, 7 shaft weight, 46, 77
persimmon, 9, 44, 251 effect on ball speed, 68–72
PGA, 5, 11, 12, 23, 111, 112, 118 effect on swing speed, 46
piezo plate, 137 shafts
pitch shots, 113 angular velocity, 79
polybutadiene rubber (PBR), 109, 118, 119, graphite shaft, 5, 88
135, 139, 144 impulsive wave, 69
power of the drive, 75–78 role during impact, 41
pressure, contact, 100, 110, 111, 126 steel shaft, 7, 79, 88
projectile, 16, 18, 19 shear force, 105, 106, 110, 117, 120, 121, 138
PTFE, 101 side movement, 124, 152–154, 156,
pure rolling 158–159, 166–169, 180, 223, 226–227
on club face, 94, 96, 101, 102, 109, 154 normalized, 152, 170, 175
in putting, 201–206, 217, 225, 229, sidespin, 120, 122–124, 147, 164, 175, 177,
231–233, 235, 241 178, 204
putter, 5, 7 slice, 25, 27–29, 103, 147, 149, 151–152,
putting, 201 154, 156, 158–160, 175, 177–178, 180,
break, 220, 239 182, 193, 204
capture conditions, 214 slip/skid parameter, 113–117, 120, 130,
inability to launch with topspin, 203 163–164, 166, 178, 197
initial slipping stage, 202 spin
launch speed, 205, 206 angular deceleration, 99, 233
roll of the ball, 209, 232 cannon testing, 103, 104, 106–108, 111,
rolling resistance, 201, 202, 205 127, 128, 143
slope of the green, 201–203, 219, 207, gear effect, 120–124, 131–135
217, 219 grassy lies, 98, 106–109, 117
speed of the green, 207, 213 energy release, 109, 111, 138
stimp rating, 205 overspin, 95, 96, 98, 104, 109, 113, 118, 125
unbalanced balls, 221 skidding, 5, 35, 99, 104, 111, 113, 163,
putting errors, 206, 212, 213, 220, 221 197, 239
putting swing, 206, 209 spin axis tilt, 27–28, 31, 35, 150–154,
clock pendulum, 211, 214 156–161, 166–171, 175–177, 188,
constant average torque, 209 192–193
271 Index
spinning and gripping, 98 swing plane, 49–50, 54, 175

types of golf balls, 95, 105, 144 torque, 48–49, 51, 52, 55, 209–213
wedge effect, 66, 121–123, 130, 192, wrist cock, 51
228–230 swing path, 29, 50
spin axis swing weight, 47
horizontal, 146
tilted, 27–28, 31, 35, 150–154, 156–161, tee, 3, 16, 25, 27, 43, 85
166–171, 175–177, 188, 192–193 tennis, 32, 69, 88, 135
spin coefficient of restitution, 95, 98, test plate, 9, 68
100–107, 109–113, 117–120, 125–126, titanium alloy, 6, 7, 8, 9, 179
128, 229 torque, 48–49, 51, 52, 55, 209–213
higher spin rates in ‘grassy’ lies, 106 Trackman, 5, 8, 11, 12, 23, 34, 38, 47, 56,
mechanism of energy release, 109, 58, 63, 67, 71, 74, 85, 86, 89, 91, 92, 97,
111, 138 104, 108, 109, 112–116, 120, 131–133,
spin rate loss, 21 144, 151, 152, 155–157, 174, 178, 180,
spring quality, 6, 7, 8, 109 181, 183, 185, 187, 188, 191, 193
squealing, 2, 4, 100 Tuxen rule, 83, 85, 153, 158
stability, 1, 27, 28, 48, 202–204, 207, 208,
216, 235, 242 U-grooves, 6, 107, 108
of stance 1, 4, 50 underarm throwing, 1, 209, 213–214
stimpmeter, 203–205, 207–208, 223–225, Uniroyal ball, 10, 11, 32, 33
230–234, 242 USGA/ R&A overall distance standard
instructions, 205 (ODS), 10, 96, 192
pure rolling, 205
strain energy (energy storage) V-grooves, 6, 107
normal, 71, 76, 77, 198 vibrations
shoulder, 78 club head, 76, 198, 252
tangential or shear, 85, 111, 117, 126, 138 shaft, 39
superball, 109, 119, 135, 140–142, 144 viscoelastic materials, 117–119, 245, 252
swing viscosity, 35, 72
centrifugal force, 51–53, 55, 59–60
circular arc, 51–52 wind
double pendulum, 48–49, 52–54, 60, 79 crosswind, 29–30
hinge angle, 81 head or tailwind, 25–26, 27, 29
pendulum arm, 49, 79, 81 wind tunnel, 4, 5, 10, 19, 20, 23
pivot, 47–48, 50, 52, 59, 211, 255 World Scientific Congress of Golf,
power, 46, 48, 51, 75–78 34, 72, 98
precision, 1, 27, 47, 50 wrist cock, 50
primary rotation point, 174 delay, 51, 52
single pendulum, 59, 81, 209, 255 release, 50, 53, 59, 60

The Science of The Perfect Swing

Uploaded by

Copyright:

Available Formats

The Science of The Perfect Swing

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Science of The Perfect Swing

Uploaded by

Copyright:

Available Formats

What are some of the important developments in golf equipment over the past 50 years discussed in the introduction?

What are some of the important developments in golf equipment over the past 50 years discussed in the introduction?

How does the loft of a club affect ball launch angle and ball speed?

How does the loft of a club affect ball launch angle and ball speed?

The Science of

Oxford is a registered trademark of Oxford University Press

Published in the United States of America by

© Oxford University Press 2015

All rights reserved. No part of this publication may be reproduced, stored in

You must not circulate this work in any other form

Library of Congress Cataloging-in-Publication Data

Preface—important notes to the reader xi

3. Basic fundamentals of ball striking 39

Effect of impact speed on CofR 72

4. The generation of ball spin 95

5. Analysis and formulation of rules for curved ball flight 147

6. A general model of driver impact and ball flight 173

Optional Reading—The Supporting Physics 194

Appendix A: Energy transfer during club-ball collision 245

150 mph 1,500 lb

Recall that Newton told us that

force × time = mass × acceleration × time = mass × final speed.

Now it’s just a matter of using the language of impact:

force × time = impulse; mass × final speed = momentum.

More generally, impulse is always equal to change of momentum.

150 mph 1,500 lb

I MPORTA NT DEV ELOPM ENTS IN GOLF

A BR IEF DISCUSSION OF MOM ENT

BACK TO THE HISTORY

Driver 112.0 −1.3 165 11.2 2,685 31 39 269

Club Attack Ball Vertical Spin rate Max Landing Carry

Driver 94.0 3.0 139 14.0 2,628 25 36 220

THE GA M E W ITHOUT DI MPLES

20 average PGA drive

THE GOLF BA LL AS A PROJECTILE

Relative Drag Force bullet

50 100 150 200

Figure 2C Separation of air flow behind smooth and dimpled non-spinning balls.

THE GOLF BA LL TA K ES FLIGHT

Figure 2D Rotating golf ball acting like a spherical fan.

Relative Lift Force all

Figure 2E Relative lift of a golf ball and same-diameter smooth ball.

Relative Drag Force

Figure 2F Relative drag of a golf ball and same-diameter smooth ball.

Spin-rate loss/second = 0.0004 × (speed) × (spin rate)

Ball Speed, mph

Lift and Drag forces/ball weight

THE EFFECTS OF ATMOSPHER IC

THE EFFECTS OF HE A D- A N D TA ILW IN DS

Figure 2K Effect of a headwind on ball flight.

0 20 40 60 80 100 120 140 160

Figure 2M Short 5-iron shot with 10 mph head- and tailwind.

Figure 2P Effect on slice of 17 mph tail- and headwinds.

EFFECT OF HE A D- A N D TA ILW IN DS

Figure 2Q Effects of a crosswind on ball flight.

OPTIONA L R E A DING—THE SU PPORTING PH YSICS*

Figure 2.1 Forces acting on a spinning golf ball in flight.

ball cross-sectional area = s

Figure 2.2 Air flow disturbance over a spinning ball.

Preface—important notes to the reader xi

3. Basic fundamentals of ball striking 39

Effect of impact speed on CofR 72

4. The generation of ball spin 95

5. Analysis and formulation of rules for curved ball flight 147

6. A general model of driver impact and ball flight 173

Optional Reading—The Supporting Physics 194

Appendix A: Energy transfer during club-ball collision 245

ρ = density of air = 1.204 kg/m3 for standard atmospheric conditions at sea level of