The Botanical Identification of Archaeological Cotton

Author(s): S. G. Stephens
Source: American Antiquity, Vol. 35, No. 3, (Jul., 1970), pp. 367-373
Published by: Society for American Archaeology
Stable URL: http://www.jstor.org/stable/278347
Accessed: 01/08/2008 10:56

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=sam.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

http://www.jstor.org
 REPORTS 367

point in his particularoffering. He stopped the if they would apply the questions of artifact
young man, who was to bearall the offeringsto relocation of this variety to their informants.
the cave, and made a quick search of his cere- These questionscould include queriesabout the
monial gear (into which the point had been types of artifactsfrom other culturaltraditions
immediately included upon its acquisition). that have been or are being relocated,the types
After finding the point, he stuck it on the of localities in which the artifacts were origi-
bottom of a jicara with beeswax at the side of nally found, the types of localities into which
the center piece, a one peso coin. The offering the artifacts were relocated, and the reasons
bearerwas then sent away and deposited all the why such artifacts are valued and relocated. In
offerings in the shrine just to the side of the fact, I believe that the time has come for eth-
spring. He returned to the rancheriawith sev- nographersto consider the more generalprob-
eral bottles full of holy water drawn from that lems of reuse and/or relocationof materialsfor
spring. economic, ceremonial, or other reasons. Sys-
The /mara'akama/'sreason for includingthe tematic studies of this sort should eventually
point in his part of the offering was straight- allow archaeologiststo apply such knowledge
forward: anything that old, i.e., pre-Huichol, to purely archaeologicalsituations.
was bound to have power which would in turn
Acknowledgments.The ethnographicresearchwas
make the offering both more powerful and at- sponsored by the National Science Foundation,the
tractive. I made attempts to find out if this Wenner-GrenFoundation for AnthropologicalRe-
ceremonial use of non-Huicholartifactswas in search,and the MesoamericanCo-operativeResearch
any way common. The /mara'akama/, and Programof SouthernIllinois University.I would espe-
other informants,all agreedthat it was uncom- ciallylike to thankRichardPailes,who encouragedme
to write this paper,and to CarrollRiley and BerleClay
mon but they could cite some other specific for theirvery helpfulcomments.
examples (one involvinga paintedceramicbowl HILL,W.W.
originally located in the Bolaios valley, 1938 The agriculturaland huntingmethodsof the
anothera fragmentof a stone figurineoriginally NavahoIndians. YaleUniversityPublicationsin
found nearSantaCatarina). Anthropology,No. 38:1-194.
WEIGAND,PHILC.
Thus, the fluted point made another jour- 1967 Huicholethnohistory.Paperpresentedto the
ney, perhaps 40 kilometers(by trail), from its American Society for Ethnohistory at Lex-
place of discovery.Manyexamplesof projectile ington. Ms. on file at the UniversityMuseum,
SouthernIllinoisUniversity.
point reuse for huntingby aboriginalgroupsare 1969 Modem Huichol Ceramics.Mesoamerican
known in the literature. A well documented
Studies, 1969 Series, No. 3. SouthernIllinois
case of the hunting type of reuse, and the UniversityMuseum.
ceremonial reasons for it, is given by Hill
(1938). But this particularexample falls outside
that category entirely, at least in the phase of THE BOTANICAL IDENTIFICATION
its travels recounted in this paper. However, OF ARCHAEOLOGICAL COTTON
after learningof this point's redepositionin the S. G. STEPHENS
holy cave, I wonder if perhapsmy originalfind ABSTRACT
might not also be a redeposition. Two possi- Cotton plant parts recoveredfrom archaeological
bilities come to mind: 1) the point had been a sites usuallyconsist of cordageand textile fragments,
raw cotton, boll segments,and seeds. All of these can
ceremonialoffering once before and was left in be readilyidentified as cotton, but muchless easily as
what today is a milk-cheese processing rock particularspecies of cotton. Some possibilities for
shelter at a date when the shelter had a year- obtaining a more precise botanical identificationare
around spring requisite for ceremonial sites. presented.Peduncles(i.e., the stalksto which the bolls
are attachedin living material)might be particularly
Many such Huichol area springs recently have usefulfor identificationpurposes.
dried up in areas where mesa tops have been
DEPARTMENT OF GENETICS
deforested. And, 2) the point was found, per- NORTHCAROLINASTATEUNIVERSITY
haps in Post-Palaeo-Indiantimes at a distance June, 1969
from the milk-cheese processing shelter, re-
touched and, when broken, finally discardedin The genus Gossypiumincludes about thirty
the talus where I saw it. species of which only four are known to have
In conclusion, I believe that ethnographers been domesticated. These are Gossypium ar-
could be of an immense help to archaeologists boreum L., G. herbaceum L., G. hirsutumL.,
368 AMERICAN ANTIQUITY [Vol. 35, No. 3, 1970

and G. barbadenseL. G. arboreumand G. her- or cultural diffusions. Our present knowledge

baceum are known collectively as the "Asiatic" of the cultivated cotton species and their his-
cottons though their joint ranges extend be- tory and patterns of distribution is consistent
yond the Near East and into Africa.Thereis no with the hypothesis that there was no bar-
conclusive evidence that either species has ever badense in CentralAmerica,no hirsutumsouth
existed in the New World except in recent of the Caribbean coastal regions of South
botanical collections. It is possible that they America, and no hirsutum or barbadensein the
may have been introduced into the New World Old Worldpriorto the 15th Century.But this is
colonies of the 16th-18th Centuries;if so, they only a hypothesis which would be invalidated
have not survived. Early references to "ar- immediately if, for example, pre-Columbian
boreum" and "herbaceum"outside the limits archaeological remains of identifiable New
of the Old World are unreliable, and one can World species should be found in the Old
still find old herbariumspecimensof undoubt- World,or of barbadensein CentralAmerica.
edly New World species that bear these names The taxonomic identification of living cot-
(Stephens 1963). Of the New Worldcottons, G. ton species is based primarilyon the compara-
hirsutum and G. barbadense, the former is tive morphology of perishable parts (flower
found predominantly from Central America parts and the associatedbracteoles, the surface
northwardinto the U.S. Southwest and around texture and shape of living bolls). Even in well
the CaribbeanSea, the latter in tropical South preserved herbarium specimens much useful
America. It is probable that the extension of information is reduced, lost unavoidably in
barbadense into the Antilles and Central processing, or not included in the preserved
America is relatively recent (Stephens 1967a). sample. The material obtained from archaeo-
It may have been associated with the Carib logical sites is still more restrictive from the
movements into the Antilles from the South point of view of identification, since it usually
Americanmainland(circa A.D. 1000 according consists of fiber samples, dried boll segments,
to Rouse 1964), and subsequentlyinto British and, less frequently, seeds. All these are readily
Honduras in 1797 (Firschein 1961). In post- identified as cotton but not as a particular
Columbian times the New World cottons, like species of cotton. A better understandingof
other New World cultigens, were disseminated what can, and what cannot, be determined
widely in the Old Worldtropics and subtropics, from such material would be helpful both to
and today there are few dry, frost-freeareasof the botanist and to the archaeologist.
the world in which they are not found.
The oldest archaeologicalrecordsof cotton FIBERPROPERTIES
are widely separated in space, though not in A very small fragmentof cordage or textile
time. The locations and approximate datings materialcan be identified as cotton because of
are: MohenjoDaro, WestPakistan-ca. 3000 BC the characteristic flattened convoluted struc-
(Gulati & Turner 1928); Tehuacan Valley, ture of the component fibers. Small samplesof
Mexico-3400 - 2300 BC (Smith & MacNeish raw (unprocessed) cotton can often be sub-
1964); Coastal Peru-3600 - 2500 BC (Bird & jected to a range of standardizedtests by the
Mahler 1951; Lanning 1967; Moseley, cited in fiber technologist and, because of the remark-
Pickersgill1969). able development in instrumentationthat has
Based largely on analogy with the present occurredover the past twenty years, such tests
distribution of the cultivated species, and on can be performed quite rapidly. They would
general similarities in fiber and other charac- include measurementsof fiber length, uniform-
teristics, the archaeologicalmaterialfrom West ity, fineness, wall thickness, and orientation of
Pakistan, Mexico, and Peru is usually assumed cellulose in the fiber wall. But from the point
to be arboreum, hirsutum, and barbadense, of view of botanical identification, these
respectively. However, it is possible that a measurementsare not very informative unless
careful reexamination of plant remains from they can be comparedwith standardsobtained
these sites might lead to a much more reliable from present-day material. Ideally these stan-
identification. If cotton plant remains from dards should be obtained from presentday cul-
dated archaeological sites could be reliably tivated species as representedby primitivecul-
identified as to species, they could provide tigens and their wild relatives. Unfortunately,
convincingevidence for pre-Columbiancontacts most of the data availablehave been collected
 REPORTS 369

from modern commercial cottons which in no times the fuzz cover is incomplete, giving the
sense represent the species from which they seed coat a piebald appearance.Fuzzy, tufted,
were developed. and naked seed forms occur as simple alterna-
The fiber propertiesof a modern Sea Island tive mutants in the cultivated species and,
or American "Egyptian-type"cotton are not because it is much easier to remove the lint by
typical of G. barbadenseas a whole, nor can the hand from naked and tufted seed types, it
fiber propertiesof a modern Upland cotton be seems likely that these would be favored
assumed to represent those of G. hirsutum. whenever they appeared in primitive cultiva-
Thus comparisons between archaeological tion. One might speculate, therefore, that the
samples and commercialsamplesare not mean- earliest cultivated seeds would have a ragged
ingful and may be misleading.Of more signi- fuzzy cover similar to the wild forms today,
ficance would be comparisonsbetween archae- and that tufted or naked formswould appearat
ological material and primitive cultigens and later stages of domestication. In primitivecul-
wild forms existing today in the same general tigens today fuzzy seeds are more common in
area, but the necessary data have not been hirsutum,arboreumand herbaceum,and tufted
collected systematically nor are they readily seeds in barbadense,but there are too many
available. exceptions for this character to be useful in
FIBERCOLOR discriminatingamongspecies.
The seed fibers of wild forms of the culti- LINT COVERAND SEEDAGGREGATION
vated species usually have a mixture of green In order to establish themselves as wild
and brown colors. The green color fades with
forms, cotton plants must have an adequate
aging to a dirty grey so that the general ap- means of seed dispersal.Wild forms of the cul-
pearance of the aged or weathered fibers is a tivated species have very sparse lint, and the
dingy or rusty brown color. In some forms seeds do not aggregateinto compact "locks"
there is very little brown coloration and the when the bolls open. They are blown around
aged fibers appeardirty white or grey. In con- readily by the wind and seeds can usually be
trast, the seed fibers of primitivecultigens are collected from the ground at any season of the
usually clear white or uniformshadesof brown year. Seeds of cultigens, in contrast, are aggre-
(buff, khaki, light brown). These color variants gated in compact locks which tend to remainin
are found in all the cultivated species. A deep the bolls. Eventuallythey fall to the groundas
chocolate color, associated with a very soft locks, not as individualseeds, and they areless
textured fiber, is apparently found only in widely scattered than in the wild forms. "Cul-
primitivecultigens of G. barbadense.This kind tigens" can often be found as escapes where
of fiber was recovered from the Huaca Prieta there is little competing vegetation (e.g., along
site in Peru and it is still found in coastal Peru
hedgerows, ditches, cleared areas, and aban-
and Ecuadoramong primitivecultigens of that doned fields), but they are rarely found re-
region.Thus a deep chocolate colored fiber that establishedin the wild as a component of undis-
leaves a dusty brown deposit on the fingers turbed vegetation. Again one might speculate
when handled indicates strongly that the spe- that the earliest cultivated types would have
cies is G. barbadense.
sparse lint, with seeds not aggregated into
SEED FUZZ compact locks, though apparently such types
The seed fibers of the cultivated species have not yet been found among archaeological
occur in two layers: an outer layer of convo- material.In one form of primitivecultigen, the
luted fibers (lint) which can be pulled from the possibilitiesof seed dispersalhave been reduced
seed, and an undercoat of short unconvoluted still further by the actual fusion of the seeds
fibers (fuzz) which are difficult to remove. In into kidney-shapedmasses. These are the kid-
wild forms of the cultivatedspecies,both layers ney cottons, and in nature they are found only
are usually present and when the lint is re- in one of the cultivated species, G. barbadense
moved the fuzz layer is raggedand irregular.In var brasiliense.Types which mimic the kidney
cultivated forms the fuzz layer may be missing seeded condition can be synthesized in culture
entirely (naked type), restrictedto the apex of from species other than G. barbadense.How-
the seed (tufted type), or form an evenly felted ever, the synthetic types always have naked
cover over the seed coat (fuzzy type). Some- seeds and greatly reduced lint cover-charac-
370 AMERICAN ANTIQUITY [Vol. 35, No. 3, 1970

teristics not associated with natural kidney BOLL SEGMENTS AND PEDUNCLES
types. Differences in seed aggregation are illus- When ripe, the cotton boll splits vertically
trated in Fig. 1. into segments which for a time remain attached
to the boll stalk or peduncle. The number of
segments into which the boll is divided, and the
shape of the boll before it splits into segments
provide useful, though not critical, information
on the species to which it belongs. In a crude
way, boll shape may be expressed as a boll
index (maximum diameter of the boll as a per-
centage of its length). As the bolls dry out, the
individual segments become recurved and even-
tually they become separated from each other
and from the peduncle to which they were
originally attached (Fig. 2). Thus samples re-
covered from archaeological sites usually con-
sist of individual boll segments. The peduncle,
i* : which at this stage bears a superficial resem-
blance to a dried clove, is lost (or perhaps
overlooked). This is unfortunate because a
peduncle, even when very old and dry, can
often furnish a critical means of identifying a
particular cultivated species.
Fig. 1. Upper row: Individual locks of cotton ex-
tracted from newly opened bolls: left, cultivated kid- Appropriate measurements of a single dry
ney cotton (barbadense); middle, cultivated free- boll segment can provide crude estimates of the
seeded cotton (hirsutum); right, wild-form of hir- shape of the boll from which it was obtained
sutum. Lower row: The same with lint removed. and the number of segments into which it was
divided (Stephens 1967b). However, Smith and
MacNeish (1964) have shown that dry seg-
ments, when swollen in hot water, return to
something approaching their original shape.
Measurements of the swollen material can be
made more accurately but, even so, much more
reliable information can be obtained from the
average dimensions of several segments col-
lected at the same level and location.
In living material (Fig. 3), three leafy brac-
teoles surround the boll and are attached to the
upper part of the peduncle. Within the brac-
teoles and closely investing the base of the boll
is a papery cup (calyx). At the base of each
bracteole is a circular depression, which is the
site of an external extra-floral (E) nectary.
When a nectary is present at the site, the base
of the depression is lined with secretory hairs
which under low power magnification (circa
10X) have the surface appearance of a honey-
comb. At the outer base of the calyx and alter-
nating in position with the E nectaries are the
sites of the internal extra-floral (I) nectaries.
When I nectaries are present, they may be in
Fig. 2. Dried and opened bolls ofhirsutum: above,
bracteoles removed, otherwise intact; below, boll the form of inconspicuous pores, horizontal
segments and boll stalk (peduncle) disassembled. slits, or (in some species) large triangular de-
 REPORTS 371

i-:iiiii ._.

i::i?i?-i
-- -

::::::::'
:::
::: i-i
i:i.-
??-::i:
:_:::-:;
--::::
-..--.
- -.

---::--
g IBtPakE_ii: :?
::::::

e L
:::::--:::i-:::;

ES::-
.::: -:::-:
':ii
i iii- --?"-:

;r:

Fig. 3. (a) Young boll of G. barbadense, showing peduncle (P) and one of the three bracteoles (B)
surrounding it. (b) The same, with bracteoles removed to show the calyx cup (C) surrounding the base of the
boll. (c) The same, enlarged, in apical view, with boll removed to show interior of calyx cup. Sites of nectaries
shown as follows: E-external extra-floral nectary; I-internal extra-floral nectary; F-floral nectary seen as a ring
of darker tissue lining the base of the calyx cup.
372 AMERICAN ANTIQUITY [Vol. 35, No. 3, 1970

pressions.As in the case of the E nectariesthey floral nectary and to determinewhether fringe
are lined with secretory hairs. In driedmaterial hairs are present (Smith and Stephens
the leafy bracteoles disintegrate,and the edges n.d.).
of the calyx cup become eroded, but usually If archaeological material should provide
both E and I nectaries can be distinguished informationon boll shape,numberof segments,
around the apex of the peduncle,particularlyif dispositionof E and I nectaries,and presenceor
the latter is swollen in hot water and treated absence of fringe hairs on the floral nectary,
with appropriate clearing agents. Peduncles meaningful comparisons with similar data
processed in this manner are illustratedin Fig. available from living forms of primitive culti-
4. gens could be made. Sometimes this could lead
Lining the inner wall of the calyx cup to positive species identification. The data
around its base is a circularfloral nectarywhich listed in Table 1 have been obtained from a
is present in all species of cotton. Under low representativecollection of about 250 primitive
magnification the circular band of secretory cultigens belonging to the New Worldspecies,
hairs has a honeycomb appearancesimilar to G. hirsutum and G. barbadense, which have
that found in the E and I nectaries. In some been maintainedin tropical cultivation for the
species the secretory band is fringedwith sim- past nine years. The data presented for the
ple hairs;in others the fringe is absent. Even in Asiatic species, G. arboreumand G. herbaceum,
material recovered from archaeologicalsites, it are much less reliable since they are based on a
may sometimes be possible to distinguish the few miscellaneousstrains maintained in green-

Fig. 4. "Heads" of dried peduncles showing positions and forms of nectaries. (The "shanks" of the peduncles
have been removed). Magnification ca. 10 X. (a) Basal view showing external extra-floral (E) nectary of
hirsutum. (b) Side view showing slit-like internal extra-floral (I) nectary of barbadense. (c) Apical view showing
hirsutum floral nectary surrounded by fringe hairs (F). (d) Apical view showing barbadense floral nectary with a
few scattered fringe hairs. In most forms of barbadense fringe hairs are missing entirely.
 REPORTS 373

house culture for other purposes as well as on lecting information of potential value to the
earlier published information (Silow 1941; archaeologist and, to the latter, the kinds of
Hutchinson et al 1947). Clearly, the data are far material which would be useful for identifi-
from complete; they are presented here only to cation purposes.
indicate to the botanist the possibility of col-
Table 1. Boll and peduncle characteristics of living primitive cotton cultigens.

Boll characters Peduncle characters

Boll Number of E I Fringe

index segments nectaries nectaries hairs

hirsutum
Central America 65-95 3-5 usually usually present
present present*
Caribbean 50-80 3-4 usually usually present
present present*

barbadense
General 40-70 3 usually usually absent
present present*
Coastal Peru 50-80 3 usually usually usually
and Ecuador ppresent present absent

arboreum t No data 3 absent usually present

present**

herbaceum - No data 3-5 absent usually ?

present**

*If present, often irregularly developed

**If present, well developed triangular depressions
t Representative measurements not available. In general, arboreium has pyramidal bolls: herbaceum
round or oval bolls.
Acknowledgments. Paper number 2896 of the domestication in Peru. American Antiquity
Journal Series of the North Carolina State University 34:54-61.
Agricultural Experimental Station, Raleigh, North ROUSE, I.
Carolina. Work supported by National Science Foun- 1964 Prehistory of the West Indies. Science
dation Grant number GB 7769. Genetics Department, 144:499-513.
North Carolina State University, Raleigh.
SILOW, R. A.
BIRD, J. and J. MAHLER 1941 The comparative genetics of Gossypium
1951 America's oldest cotton fabrics. American anomalum and the cultivated Asiatic Cottons.
Fabrics 20:73-78. Journal of Genetics 42:259-358.
FIRSCHEIN, I. L. SMITH, C. E., and R. S. MacNEISH
1961 Population dynamics of the sickle-cell trait 1964 Antiquity of American polyploid cottons,
in the Black Caribs of British Honduras, Central Science 143:675-676.
America. American Journal of Human Genetics
13:233-254. SMITH, C. E., and S. G. STEPHENS
GULATI, A. M., and A. J. TURNER n.d. Critical identification of Mexican archaeo-
1928 A note on the early history of cotton. Indian logical cotton remains. Economic Botany (In
Central Cotton Committee Technical Labo- Press).
ratory Bulletin 17. STEPHENS, S. G.
HUTCHINSON, J. B., R. A. SILOW, 1963 Polynesian cottons. Annals of the Missouri
and S. G. STEPHENS Botanical Gardens 50:1-22.
1947 The evolution of Gossypium. Oxford Uni- 1967a Evolution under domestication of the New
versity Press. World Cottons (Gossypium spp.). Ciencia e
LANNING, E. P. cultura 19:118-134.
1967 Peru before the Incas. Prentice Hall. 1967b A cotton boll segment from Coxcatlan
PICKERSGILL, B. cave. In The pre-history of the Tehuacan Valley
1969 The archaeological record of chili peppers Vol. 1:256-260. Edited by Douglas S. Byers,
(Capsicum spp.) and the sequence of plant University of Texas Press.