The Origins of Agriculture

Author(s): Kent V. Flannery

Source: Annual Review of Anthropology, Vol. 2 (1973), pp. 271-310
Published by: Annual Reviews
Stable URL: http://www.jstor.org/stable/2949273
Accessed: 06-02-2017 03:11 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Annual Reviews is collaborating with JSTOR to digitize, preserve and extend access to Annual Review of
Anthropology

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
Copynight 1973. All rights reserved

THE ORIGINS OF * 9531

AGRICULTURE

Kent V. Flanneryl
University of Michigan Museum of Anthropology, Ann Arbor, Michigan

It is now becoming increasingly clear that the domestication of weeds and cultivated
plants is usually a process rather than an event.
Edgar Anderson (1, p. 766)

Perhaps no aspect of prehistory has received as much attention over the last 15
years as the origins of agriculture. Archeological expeditions in the Near East, in
Thailand, in Mexico, in Peru, and elsewhere have unearthed specimens of man's
earliest crops. Botanists have argued over what the specimens mean. Geographers
have analyzed the ranges of the wild ancestors of today's crops, and told us where
they should originally have been domesticated. Anthropologists young and old have
presented models for the way agriculture might have begun-some reasonable, some
preposterous. Surely at this stage we could declare the origins of agriculture a
bandwagon.
I have already traveled many thousands of miles on that particular wagon, and
I yearn to step down and tackle other problems. I agreed to undertake this final
review, however, because I feel a critical, skeptical view of the whole bandwagon
is badly needed. The first archeologists to work seriously on the origins of agricul-
ture were a cautious and circumspect lot. Unfortunately, they were followed by a
number of botanically naive, sensation-seeking opportunists who were more con-
cerned with finding "the oldest domestic plant" than with clarifying the processes
by which agriculture began. Their ingenuousness spread even to the botanists who
worked with them, and soon we had claims for domestication based on a single
burned seed, a single trampled rind, or a single crumpled pod. In cases where the
range of variation of the wild ancestor was not known (in fact, even in cases where
the actual species of wild ancestor was not known), we had prestigious botanists
assigning a single crushed specimen to a modern cultivated race-a race which, in

'A number of archeological and botanical colleagues contributed unpublished data to this
paper. I would like to thank in particular G. Beadle, K. C. Chang, R. Drennan, R. I. Ford,
W. Galinat, L. Kaplan, T. Lynch, R. S. MacNeish, B. Pickersgill, and C. E. Smith Jr. Any
errors in the text, however, are my responsibility.

271

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
272 FLANNERY

some cases, may have taken thousands of years to stabilize. These were botanists
who, under normal conditions, would have argued that nothing less than 100
specimens-with a mean and standard deviation-was an adequate sample; but
perhaps the search for agricultural origins is not a normal condition. And what the
botanists claimed was usually nothing compared to what the archeologists claimed.
Thanks to Robert Braidwood and Richard MacNeish, I have been lucky enough
to work in Iran, Mexico, and Peru with some of the best "paleoethnobotanists"
working on the origins of agriculture: Hans Helbaek, Jack Harlan, Willem van Zeist,
Paul Mangelsdorf, Walton Galinat, George Beadle, Earle Smith, Barbara Pickers-
gill, and Larry Kaplan, among others. Consequently, I have received "personal
communications" and overheard comments which have never appeared in print. I
have also been made aware of changes in opinion that have appeared in journals
which archeologists ordinarily never read. I think that if all archeologists could have
overheard the off-the-cuff comments I received from these botanists, they would
have a more profound and healthy skepticism of what we know about early agricul-
ture. When they see a chart showing "domestic beans" at 5000 B.C., and only one
single cotyledon is indicated-separated by 2000 years from the next oldest bean-
they might take it with a grain of salt.
In return, I wish I could explain to every botanist that while an archeologist looks
the other way for one minute, a pack rat can bury an intrusive bean 50 centimeters
deeper in his favorite dry cave. Of course, botanists who work in the field with
archeologists do learn (from bitter experience) not to accept anything on faith; I
would like to single out Barbara Pickersgill in particular for her outstanding skepti-
cism.
In this review, I will not try to cover the whole world; that would be impossible.
Nor will I try to cover every cultivated species. What I have done is to select four
areas-two in the Old World, two in the New World-and discuss the current
evidence for the major staples domesticated there. Where possible I have also
reviewed the current hypothetical models for the way agriculture began in a given
area. Please note that the model for each area is different; I do not believe that
agriculture began the same way, or for the same reasons, in all four areas, nor do
I believe one model can explain them all.
My coverage of all four areas (Southwest and Southeast Asia, Mesoamerica, and
the Andes) will not be balanced, because we know far less about some areas than
others; I have spent the most time on Mesoamerica because the new "maize vs
teosinte" debate is one of the most exciting theoretical issues of 1973. But first let
me begin with a discussion of two major types of early agriculture.

ECOSYSTEMS AND PRIMITIVE AGRICULTURE

Geographer David Harris has made a useful distinction between two types of
"paleotechnic" or primitive agriculture. In his typology, Harris regards farming
systems as "man-modified ecosystems" and analyzes them in terms of species diver-
sity, productivity, and stability or homeostasis (Harris 26).

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 273

1. Seed-crop cultivation-including early wheat farming in the Near East or

early maize agriculture in Mesoamerica and Peru-is a relatively simple ecosystem
because it consists of very few species (often only one) growing in nearly pure stands
composed of large numbers of individuals. It is highly productive (as the later
discussion of wheat and maize will show) but also very unstable because of its low
species diversity; it requires constant human attention.
2. Vegeculture-such as the cultivation of manioc in Amazonia or yams and taro
in southeast Asia-is often a very complex ecosystem, because many different
cultivated species may be grown in a single field. Although sometimes less produc-
tive than intensive seed-crop cultivation, it is far more ecologically stable because
of its species diversity; it more closely approximates the complexity of natural
vegetation.
Because most early centers for seed-crop farming were in arid regions with good
archeological preservation, a great deal is known about early farming there. More-
over, morphological changes in the seeds following domestication can usually be
detected. On the other hand, most early centers for vegeculture were humid tropical
regions with poor archeological preservation; and since many of these crops are
planted by cuttings, roots, or some other vegetative part which shows little or no
morphological change after domestication, we know much less about early vegecul-
ture. Archeologists like Donald Lathrap (40) have long bemoaned the fact that we
know so little about the origins of root crops, and their importance is so frequently
ignored.
That is, it is ignored by archeologists working in arid regions; but archeologists
working in the tropics, in spite of the absence of data, frequently present theories
of root crop cultivation which are at present untestable. For example, Lowe (45) has
argued for early manioc cultivation on the Pacific Coast of Chiapas and Guatemala,
citing as evidence small obsidian chips which are said to resemble those used in
manioc graters in South America. Yet a sample of more than 50 preserved plants
from 1000 B.C. on the Guatemalan coast, which Coe and I recovered in 1962 (Coe
& Flannery 9), contains no manioc; it is, like all arid Mexican highland samples of
the same time period, mostly maize. Complicating the picture still further is the fact
that botanist Earle Smith (MacNeish 46) identified a seed of wild Manihot from one
of MacNeish's Tamaulipas caves. The species was one which has never been domes-
ticated (Smith 67), but overeager archeologists continue to cite it as an example of
cultivated manioc. The result is that many archeologists in Mesoamerica ignore root
crops, while others argue strenuously for them with no supporting data. I will try
not to ignore them, but I have little concrete data to offer.

THE ORIGINS OF AGRICULTURE IN SOUTHWEST ASIA

Southwest Asia, one of the first regions of the world to have agricultur
most intensively investigated area from the standpoint of early cultivation. Thanks
to the pioneering efforts of archeologists like Robert Braidwood (Braidwood &
Howe 6) and paleoethnobotanists like Hans Helbaek (28, 29), we probably know

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 274 FLANNERY

more about early farming in the Near East than anywhere else. One or two valleys
in Mesoamerica have produced superb evidence for early agriculture; but in South-
west Asia we have carbonized plant remains from Israel, Jordan, Syria, Turkey,
Iraq, and Iran, all antedating 5000 B.C. The group of sites shown in Table 1 is only
a tiny sample of what could have been listed.
So much has been published on early agriculture in the Near East that I will give
here only a short resume of the archeological evidence, and briefly restate one
current hypothesis for the way farming may have begun. A more extensive synthesis
of the archeological data can be found in a paper by Jane Renfrew (61), and a more
detailed statement of the hypothesis can be found in an earlier paper of my own
(Flannery 15).
The Near East is one of those parts of the world where sedentary life-in "hut
compounds" or actual villages-seems to have begun before agriculture (Flannery
16). The earliest permanent settlements there are either without evidence of domes-
tic plants, or have only plants which are "wild" in the phenotypic sense.2 Currently
it appears that these early sedentary communities relied on a mixed strategy of
harvesting wild cereal grasses (wheat, barley, and ryegrass), nuts (pistachios, wal-
nuts, almonds, and acorns), and legumes (Astragalus, Trigonella); hunting herd
ungulates (sheep, goat, and deer); and collecting a wide variety of smaller species
like fish, water turtles, land snails, cockles, mussels, and crabs. This adaptation
seems to have crystallized during the second half of the Wurm glacial advance of
the Pleistocene period (20,000-9,000 B.C.)-a period during which the Near East is
now believed, on palynological grounds, to have been colder and drier than today
(van Zeist & Wright 74).
There can be no doubt that this prolonged cold, dry period affected man's envi-
ronment in southwest Asia. Areas of the Zagros Mountains in Iran and Iraq, which
today would have oak and pistachio woodland (were it not for overgrazing and
deforestation) became virtually treeless during the late Wiirm. This would not only
have reduced wild nut crops like acorn, pistachio, walnut, and almond, but also wild
cereal grasses which are part of the woodland floral community. Only in the Levant
(Israel, Lebanon, Jordan) and the Ghab basin of Syria did relict woodlands of oak
survive (Niklewski & van Zeist 54). It was probably from these relict populations
that the rest of the Near Eastern uplands became reforested after the post-Pleisto-
cene climatic amelioration. These woodlands were the ancestral home of wild wheat
(Triticum), barley (Hordeum), lentils (Lens), vetch ( Vicia), vetchling (Lathyrus),
peas (Pisum, Cicer), linseed (Linum), and many of the other ancestors of Southwest
Asia's staple crops.
Wild wheat, barley, and their relatives have a long history in Southwest Asia.
Some of the hardier species surely weathered the cooling and drying of late Wiirm
times; in the sediments of Lake Zeribar in the Iranian Mountains, some "cereal
type" pollen was present before 12,000 B.C. (van Zeist & Wright 74). But several of

2That is, the plants do not differ morphologically from wild races. This does not prove they
were not cultivated-it merely shows they had not yet undergone any genetic change in the
direction of the domestic races.

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 275

n;~~~~~~~ X X X X
o o . .S

a))

s S . w ~~X X X XE

0 - ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~
c E s3 X XX X X X X z

e X
-Z sw 0 0
0

C)~~~~~~~~~~~~~~~~~~~~~~~~~~~a

a) ~ ~ ~ ~~~~~~~~~~~
Xa) t<~ a). O O O O *

0
w ? N c X~~~ X X X u
a) 0~~~~~~~~~~~~~~

> ~ ~ K K K K K ooooo o oo o_

~~d, 0 *a)~0

0 --. >K K. KK
a) ~ ~ ~ ~ ~ ~~ - :3 )c

_ o mo ooo m oo oo 00 o _ cK
0~~~~~~~~~~~~~~~~~~.
_ sa S i')00000 0 00 C

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
276 FLANNERY

the species most exploited by prehistoric man prefer warmer habitats, and would
not have done well until temperatures began to rise toward the end of the Wurm.
By late Paleolithic times, when scattered oaks and pistachios were reinvading the
Zagros, conditions for the cereals had significantly improved. After 12,000 B.C., over
much of the Near East, the possibilities for extensive use of wild wheat and barley
were not only good, but steadily improving.
There is no evidence, however, that this climatic amelioration caused agriculture
to begin. Surely there had been periods just as favorable during earlier interglacials,
without any resulting agriculture. Two factors, however, were different during late
Wiirm: (1) the population of Southwest Asia was higher than it had ever been before
and (2) for the first time human groups were using on a large scale a whole series
of foods which had previously been ignored or insignificant in their diet: the snails,
crabs, fish, mussels, and other "harvestable" invertebrates already mentioned. In
such a context, even wild cereal grasses might have looked appetizing, in spite of
the labor involved in their harvest and preparation.
It must be admitted that our evidence for the first use of wild wheat and barley
is circumstantial. Not a single carbonized grain has ever been recovered from the
periods we would most like to know about-not because the sites lack ash deposits,
but because previous archeologists usually lacked interest in saving any. In addition,
it appears that some late Wurm ash deposits are so mineralized that the carbonized
seeds cannot even be floated out with water, which is one common recovery method.
These technical obstacles are gradually being overcome by techniques like Cam-
bridge University's froth flotation (Jarman, Legge & Charles 32), which has now
yielded some of the world's oldest cereal remains from Nahal Oren, Israel. But until
such techniques are widespread, we must draw on other lines of evidence. These
have been summarized as follows (Flannery 15):
1. To convert most wild cereals into a satisfying meal, you need processing
implements and storage and cooking facilities which the previous Paleolithic peoples
of the Near East do not seem to have had. These implements and facilities appear
with explosive rapidity after 10,000 B.C.
2. The wild cereals come ripe at a wholly different season from the wild tree crops
(acorn, almond, pistachio, walnut): nuts in the autumn, cereals in late spring.
Moreover, the harvest season for wild cereals is much shorter and more restricted
and cannot be approached with the casualness permitted by the long nut crop
season. An appropriate shift in settlement pattern and social organization was
required in order to focus on wild cereals, and seems to have appeared by 10,000
B.C.

An Introduction to the Wild Cereals

Now let us examine the characteristics of five cereal species of three genera-
Triticum, Aegilops, and Hordeum-whose habitats and ranges have recently been
outlined by botanists Jack R. Harlan and Daniel Zohary (25).
1. Wild barley (Hordeum spontaneum) is the most widespread of these cereals,
tolerating the greatest range of environmental conditions. Its primary wild range is
the uplands of the Lebanon-Judean, Taurus, and Zagros Mountains, where it forms
an important component of the summer-dry deciduous oak woodland. Inhibited by

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 277

extreme cold, it only rarely occurs in the wild above 1500 meters (and during the
late Wurm period may have stayed even lower); hence it is unlikely that man would
have found useful stands in the higher mountains or on the plateaus of Turkey and
Iran. On the other hand, wild barley tolerates heat and aridity better than wheat,
and will follow seasonal stream beds down from the mountains and out into the hot
steppes and desert regions. In fact, a slender, small-seeded race occurs as an annual
grass of wet stream beds in the barren Negev and Syrian-Transjordan deserts, while
a much more robust race inhabits the moist uplands at the head of the Jordan Valley.
In addition, wild barley does so well as a "weed" in disturbed habitats or wheat fields
that its range has been greatly increased by agriculture; today, semiwild "weed"
barley "grows in the mountain forest, on the coastal plain, in the shade of rock
outcrops in semidesert areas, and as a weed in the fields of every conceivable
cultivated crop" from Morocco to Turkestan (Helbaek 28, p. 112). Wild barley
therefore flourishes rather than retreating under conditions of human exploi-
tation.
2. Wild einkorn wheat (Triticum boeoticum), as its name implies, usually has
"one grain" on each internode of its head; it is what geneticists call a diploid species,
which in the case of wheat means it has 2 X 7, or 14 chromosomes. Actually, the
name einkorn is not wholly apt, since one race has evolved a "two grain" internode
to survive in arid regions: one grain of the pair germinates the first year, while the
second is delayed until later in case the first fails.
Einkorn, in contrast to wild barley, tolerates cold conditions well. It grows above
1500 meters in Turkey and up to 2000 meters in the Zagros Mountains of Iran and
Iraq. Its primary range seems to be the Taurus-Zagros Mountain arc and immedi-
ately adjacent steppe; it grows in the Levant only as a weed in disturbed habitats.
It prefers basalt, marl, and limestone as substrata, and will form massive stands "as
thick as a cultivated field" where it finds these conditions in the heart of its primary
range.
3. Aegilops speltoides, known as "goat-face grass," is a wild cereal closely related
to wheat; its natural range overlaps with wild einkorn. In the Tigris headwater
region of Turkey's southern Taurus region, wild einkorn and Ae. speltoides form
dense stands on basalt ridges and uplands where they can be harvested together.
Botanists feel that at some time in the very remote past, diploid (2 X 7) einkorn and
diploid (2 X 7) Ae. speltoides hybridized to produce a tetraploid (4 X 7, or 28
chromosome) wheat known as wild emmer (see below).
4. Wild emmer wheat (Triticum dicoccoides) is the most sensitive and demand-
ing of all the wild cereals to be discussed here; it will not tolerate cold as well as
einkorn, nor heat and aridity as well as wild barley. It is therefore the most localized
in its "primary range," the area where it will form dense stands. To complicate
matters, its range is split into two distinct segments, separated by a gap in the Syrian
Mountains where no wild emmer will grow.
A robust, large-seeded race of emmer occurs in Palestine, where it forms massive
stands on basalt and limestone slopes in Galilee, Mt. Hermon, and the Golan
plateau. Here Zohary estimates that harvests of up to 500-800 kilograms of grain
per hectare (450-700 pounds per acre) could be made from untended, native wild
stands in a rainy year. A smaller-seeded race occurs sporadically in the lower oak

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
278 FLANNERY

woodland of the Zagros and Taurus ranges, where rainfall is between 300 and 500
millimeters. This smaller race, which seemingly never forms dense stands, could
eventually prove to be a "weed" emmer which spread to disturbed habitats in the
Zagros-Taurus arc following cultivation. This cannot as yet be shown, nor is it clear
precisely how this eastern race is related to Triticum araraticum, another wild
tetraploid wheat occupying the same region.
Perhaps most significant is the fact that the large-seeded Palestinian wild race
forms consistently fertile hybrids with cultivated emmer (Triticum dicoccum),
while the small-seeded Zagros race usually will not. This strongly suggests that the
Palestinian race is the wild ancestor of today's domestic emmer, and it focuses
attention on the Israel-Jordan-Lebanon area as an early center for emmer cultiva-
tion.
5. Aegilops squarrosa, another species of "goat-face grass," has its principal
range along the Caspian sea coast and the Kopet Dagh mountains on the Iranian-
Turkmenistan border; there are also sporadic occurrences in the Zagros, but these
may be "weed" strains in disturbed habitats. Aegilops was eaten by prehistoric man
in the Zagros, but seems never to have been as important as wheat. We mention
Ae. squarrosa primarily because genetic evidence suggests that, long cpnturies afte
cultivation had begun, it was diploid (2 X 7) Ae. squarrosa which hybridized with
tetraploid (4 X 7) emmer to produce a hexaploid (6 X 7, or 42 chromosome) wheat,
Triticum aestivum-the world's first "bread wheat." In general, the wheats men-
tioned above are unsuitable for breadmaking, and tend to be eaten as porridge.

The Potential (and Problems) of Wild Grain Harvests

The tremendous productivity of wild wheat and barley within the heart of their
"primary" range has recently been stressed by Harlan & Zohary (25). They com-
ment that "over many thousands of hectares" on suitable basalt or limestone sub-
strata "it would be possible to harvest wild wheat today from natural stands almost
as dense as a cultivated wheat field." Zohary's estimates of 500-800 kilograms per
hectare for mixed stands of wild emmer and barley in rainy years on the east Galilee
uplands have already been mentioned. Under such conditions, a single hectare of
land could provide more than two million calories-or about the number required
by a family of three during the course of a year.
On a basalt mountain range near Diyarbakir, in the upper Tigris drainage of
Turkey, Jack Harlan set out to find out just how much a prehistoric man might be
able to harvest (Harlan 24). Diyarbakir lies almost directly in the track of the rain
winds passing through gaps in the Syrian mountains, and in 1966 the basalt uplands
were a sea of wild einkorn wheat mixed with goat-face grass. After five separate
periods of hand-stripping the wild wheat heads from their stalks, Harlan found he
was collecting an average of just over 2 kilograms of grain per hour. Such hand-
stripping is still practiced by the Bedouin of Southern Jordan, but it takes tougher
hands than the average university professor has; Harlan's were soon lacerated by
the coarse and raspy exterior of the cereal heads. He then switched to harvesting
by means of a "prehistoric" sickle, made from flint blades set in a wooden handle;
this enabled him to harvest nearly 2.5 kilograms per hour with less wear and tear

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 279

on his hands. In the end, Harlan concluded that a family of four, harvesting for the
whole of the 3-week period during which the Diyarbakir einkorn came ripe, could
easily harvest a metric ton-one year's supply. Let us hasten to add, however, that
such a family would face many obstacles, not the least of which are the physical
characters of the plants themselves.
Look now at Figure 1, Hans Helbaek's diagrammatic drawing of a cereal "spike"
or head. The black, composite vertical column is the axis or rachis of the spike,
composed of many individual internodes separated by white interstices. Each inter-
node bears at its upper end a "spikelet," consisting of one or more mature cereal
grains (shown in white), encased between tough "glumes" or husks (shown as
stippled). When a cereal plant reaches maturity, the axis begins to disarticulate,
starting at the top. One by one, individual internodes detach themselves at the white
interstices and fall to the ground; the whole process takes 1 or 2 weeks on the
average. Finally, only the stalk is left; each seed has buried itself in the ground where
it waits, encased in its protective glumes, for the next rainy winter. The "brittle
rachis" of the wild cereals thus acts to disperse the seeds one at a time, preventing
the unfavorable condition which would result if the whole head of grain fell to the
ground in one place. In the latter case, half the seeds would never be buried and
the rest would sprout in a dense mass of competing seedlings.
The brittle rachis is of no help to the collector of wild wheat or barley, however.
In the first place, he cannot harvest before the grain is fully ripe, or a dried, wizened,
unpalatable head results. It is thus not surprising that carbonized archeological
specimens show that mature heads were harvested (Zohary 80). On the other hand,

Figure I Diagram of a spike of wheat or barley. The

black, composite column is the axis or rachis. Its
component parts, the internodes, are separated by white
interstices where, in the wild, the axis disarticulates. In
domestic cereals the interstices toughen and do not
disarticulate. Three internodes near the bottom of the

r-\ rl/ drawing are shown as having mature florets (white ov

enclosed between glumes (stippled areas); in domestic
cereals these glumes become easier to remove from the
floret, in which the fruit or grain develops (see text).

After Helbaek (29, Figure 138).

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 280 FLANNERY

the collector cannot wait too long after the cereals ripen, or else he will be greeted
by a field full of empty stalks. Even during the latter part of the harvest season, the
heads may be brittle enough to shatter at the slightest touch.
Here timing becomes critical, for final ripening is brought on by the onset of dry
hot weather at the end of the growing season, and this is quite variable from year
to year. Under ideal conditions, a family of cereal collectors might have 3 weeks in
which to harvest their metric ton; but Zohary points out that if an extremely dry
warm spell (known in Israel as the hamsin) happens to occur at maturation time
"shedding of fruit can be completed in a matter of 2 or 3 days" (80, p. 57). Being
in the right area at the right time is therefore far more important than for collection
of wild nut crops, and the harvest may be a fast and furious affair.
Some of the problems facing prehistoric collectors might thus be listed:
1. The brittle rachis, which, combined with unpredictable hot dry weather,
makes timing essential: a year's supply of grain could hinge on a crucial period of
3 days or less.
2. The tough glumes which hold each kernel in a tight grip long after the rachis
had disintegrated. In the wild cereals, these inedible glumes adhere so strongly to
the grain that conventional "threshing" will not remove them.
3. The annual fluctuations in rainfall characteristic of the Near East. Zohary's
estimate of 500-800 kilos per hectare was based on a "rainy" year. In a "dry" year,
the yield could be 100 kilos or even less-a reduction which could bring some
families to the brink of starvation.
4. The extreme localization of dense cereal stands. Even in a rainy year, it is only
certain limestone massifs or basalt plateaus which have had stands as dense as those
described by Harlan & Zohary. In between there are alluvial plains, wadis, and
uplands of less favorable substratum where the cereals, if present at all, grew very
sparsely.
5. Harvesting tools. Hand-stripping is an efficient way of selecting only the ripe
heads and keeping the brittle internodes from scattering, but it can be slow and
painful. More efficient implements for cutting and facilities for transport were
needed.
6. Storage. After all, where can you go with a metric ton of cleaned wheat seed?
It requires storage facilities, and it requires that they be sufficiently waterproof so
the grain does not sprout during the moist winter season.
Not all of these problems could be solved by the prehistoric cereal collectors of
the Near East prior to domestication. But we know what a few of their solutions
were, and we can make some educated guesses about the rest on the basis of
archeological data.
1. The development ofharvesting tools. Sickles made from flint blades set in slots
in a wooden or bone handle and cemented with natural asphalt or gum became
widespread in the near East after 10,000 B.C. Sickle blades can be identified by a
characteristic "sheen" or luster which develops on the cutting edge through its use
on grass stems.
2. Collecting and transporting facilities. On the basis of limited archeological
evidence, we suggest that grain was collected and carried from the field in baskets.

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 281

One of our earliest clues, dated by radiocarbon to before 8500 B.C., consists of
fragments of possible basketry from Level B1 of Shanidar Cave in Iraq (Solecki 69).
Somewhat later sites (6500-7500 B.C.) have produced clear fragments of twined
baskets, sometimes waterproofed with bitumen (Hole, Flannery & Neely 30).
3. The development of processing facilities. Sometime around the end of the
Pleistocene, man discovered that by roasting the grain he had collected he could
render the glumes so dry and brittle that they could be removed by abrasion. At
several sites this was accomplished by roasting the cereals over heated pebbles in
a pit or subterranean earth oven (cf. van Loon 73).
Removal of the glumes by abrasion (either directly or after roasting) was appar-
ently done first with mortar and pestle. Harlan, following the experimental harvest
already mentioned, found he could effectively dehusk his einkorn wheat with the
type of wooden mortar and pestle used by the Osage Indians of the central USA.
Wooden mortars have not been preserved at sites in the Near East (they could have
existed well back in the Paleolithic), but limestone or basalt mortars and pestles
appeared at the end of the Pleistocene. They were joined, after no great interval of
time, by shallow basin-shaped or saddle-shaped grinding slabs on whose upper
surface grain could be abraded by means of a flat or loaf-shaped second stone held
in the hands (like the metate and mano of New World archeology). Further grind-
ing by mortar or slab reduced the cereals to "groats," or coarse grits of grain which
could be cooked up into a mush or gruel. Such groats are present in some of the
earliest carbonized grain samples (antedating 7000 B.C.), which even seem to have
included the woody base of the cereal spikelet; evidently this nearly inedible part
of the plant was ground up and eaten right along with the grain (Helbaek 29).
4. The expansion of storage facilities. Below-ground storage pits were already
present in some Upper Paleolithic sites, but the Natufian people of Palestine (9000-
7000 B.C.) made them widespread in the Near East. Conical or bell-shaped (wider
at the base than at the mouth), these pits are extremely numerous at Natufian sites,
where they are sometimes coated with lime plaster, presumably for waterproofing.
The process of roasting or parching the cereals, already mentioned, besides weaken-
ing the glumes also kills the germ so the cereal will not sprout during the damp
winter. A single Natufian site, Ain Mallaha, contains more storage facilities than
are known from all previous sites in the Near East put together (Perrot 57).
5. Harvesting along the altitude cline represents one method of increasing the
harvest in spite of the brief, rapid ripening season of the brittle-rachis wild cereals.
It depends on the fact that differences in maturation time do exist between different
cereals and between different altitude zones. Zohary points out that in Israel, for
example,

in mixed stands, H. spontaneum matures from 1 to 3 weeks earlier than T. dicoccoides.

Plants growing in sites with deep heavy soil mature somewhat later than those occupying
shallow soils. Considerable differences in the ripening time occur, however, at different
altitudes. Maturation of wild emmer at the sea level belt near the Sea of Galilee occurs
around the end of April, while higher up in the Safad area (alt. 700-800 m) stands
mature around 1 5th-20th May. In the adjacent east-facing slopes of Mt. Hermon (alt.
1400-1600 m), ripening occurs still later-in early June. Similar altitude clines occur

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 282 FLANNERY

in Turkey, Iraq, and Iran. Therefore, in regions with varied topography, altitudinal
amplitude compensates for the abrupt shedding. Collectors can start their harvest in lower
elevations and proceed gradually to climb the higher slope, and effective harvesting time
is extended to last 4 to 6 weeks (Zohary 80, p. 57).

We do not know for sure if such a pattern was followed during the 10,000-7,000
B.C. period, but if so, it might be detectable in the settlement pattern: harvesting
camps at higher altitudes should date to later in the year.
6. Problems solved by domestication: finally, we come to a series of problems
which were not (indeed, could not be) countered purely by the stategy and technol-
ogy of harvesting and processing. These include (a) the brittle rachis, and (b) the
annual variability in rainfall which kept densities of cereal stands unpredictable.
Only the beginnings of agriculture-which brought the cereals down from their
rainfed upland habitats, placed them on permanently humid floodplain soils, and
reduced selection pressures for the wild phenotype-directly attacked these prob-
lems. Although the exact nature of Natufian subsistence is poorly understood, at
least one village of this period, Tell Mureybit, although yielding a large carbonized
seed sample, seems to have only grain of the "wild" brittle-rachis type (van Loon
73).
An important genetic change following domestication-one controlled by per-
haps as few as two genes-was the shift to a tough rachis, one that held together
at maturity. Tough-rachis mutants must always have existed, but in the wild they
would have been selected against because of their poor seed dispersal mechanism.
Early farmers evidently selected for them, for shortly after 7000 B.C. tough-rachis
strains of emmer wheat or two-row barley appear in archeological deposits from
Jordan to Iran. In some cases, three internodes from a single spike have stayed
together through threshing, carbonization, 9000 years of burial underground, and
the rigors of archeological flotation!
Another genetic change-this one apparently controlled by a single gene (Stubbe
72)-was the shift to a "naked" kernel, one whose glumes did not adhere as strongly
as do wild glumes. "Naked" barleys appeared sporadically as early as 7000 B.C., but
did not become common until after 6000 B.C. This must have reduced the amount
of labor in threshing by a considerable factor. Still a third change was the evolution
of two-row barley into six-row barley of both "hulled" and "naked" strains. This
change, which occurred sometime before 6500 B.C. and became widespread by 5500
B.C., converted the normally sterile lateral kernels in each barley spikelet to large,
fertile kernels like the central spikelet. Such lateral kernels develop facets where they
meet the central kernel, as well as a slight twist; these characters allow six-row
mutants to be identified in the archeological plant remains.
Finally, domestication allowed man to plant cereals selectively on permanently
humid soils where moisture in the ground would help them survive the fluctuations
in annual rainfall characteristic of the Near East. At Ali Kosh in southwest Iran,
early farmers planted their wheat and barley so close to swamp margins that seeds
of the club-rush Scirpus were mixed with the carbonized cereals (Flannery 15).
Since wild wheat and barley do not grow in such marshy habitats, the native
vegetation must have been removed to make way for them-the first steps in man's
deliberate modification of the Near Eastern landscape.

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 283

A Model for the Origins of Near Eastern Farming

If, as Lee & deVore (43) tell us, hunting-and-gathering was the most stable adapta-
tion man ever had, why did he ever give it up for a life of toil in the wheat fields?
We may never know why agriculture began in the Near East. All we know is that
it was initiated toward the end of a long, cool, dry climatic phase by people with
a recently evolved tradition of "broad spectrum" resource utilization; that it proba-
bly began among people who were at least partly sedentary, and whose population
density (although far from the highest in the world at that time) was probably higher
than it had ever been previously; and that it began in a region of relatively great
environmental diversity over a relatively small area.
Binford, Birdsell, Wynne-Edwards, and others (Binford 5) have argued that most
hunting-gathering populations, once reasonably well adapted to a particular envi-
ronment, tend to remain stable at densities below the point of resource exhaustion.
Lee (42) has suggested that emigration was one of the main mechanisms for keeping
human populations low during the two million years prior to 10,000 B.C. By 10,000
B.C., however, man had occupied virtually every major land mass in the world, from
Australia to the southern tip of South America. In other words, after that point
emigration probably decreased in importance as a population-limiting device, while
other mechanisms-long lactation, infanticide, senilicide, and so on-increased in
importance. Some groups may also have attempted to increase food supply relative
to population density by cultural means: turning to previously ignored sources of
food like small aquatic or invertebrate species, or attempting to increase local stands
of useful plants. The latter could be one path to agriculture.
In 1968, in a paper on post-Pleistocene adaptations, Binford (5) proposed a
"density equilibrium" model for cultural change which I later applied to the Near
East (Flannery 15). Binford argued that adaptation would change only in the face
of some disturbance in the equilibrium between population and environment, such
as (a) a change in the physical environment which would bring about a reduction
in the density of chosen plant or animal foods, or (b) a change in population density
or distribution which would raise human populations too close to the carrying
capacity of the immediate area. A change of the first kind does seem to be reflected
in the palynological record in Southwest Asia-the cold, dry late Wiirm, which
restricted woodland resources to relict areas. And there are changes in population
reflected in the archeological record, which we understand only poorly, but which
might be examples of the second kind of change. For example, judged purely on the
basis of numbers of sites per period (a slippery statistic at best), some areas of the
Lebanon coast would seem to have decreased in population in the very late Pleisto-
cene, while certain areas of Israel and Jordan enjoyed increases (Copeland & Wes-
comb 10, Marks 53). However, such site distributions may actually reflect shifts in
population from one region to another, without any real overall population
increase.
The Near East is a mosaic of "favorable" habitats (e.g. oak-pistachio woodland)
and "marginal" habitats (e.g. gravel desert); the wild cereals have definite "op-
timum" zones in which they grow densely and marginal zones in which they do
poorly. If one follows Binford's model to its logical conclusion, the "optimum"
habitats should have been the centers for population growth, with the marginal areas

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 284 FLANNERY

receiving the emigrant overflow. It is in these marginal zones that man-land disequi-
librium and stress would have been felt first. Thus in 1969 I suggested that farming
might have begun first, not in the optimum area of wild cereal growth (where, as
Harlan and Zohary point out, wild wheat already does as well as it would in a
cultivated field), but around the margins where it was necessary to raise the available
food per capita. In the marginal zone, wheat and barley would only produce 500
kg per hectare if artificially planted in selected humid soils such as stream beds and
swamp margins. Rather than repeat the whole argument in detail, I refer the reader
to the original presentation (Flannery 15).
Let me now raise some objections to this model. First, although it has won an
almost frightening acceptance among some of my colleagues, it is still unproven and
highly speculative. Second, as my comments above indicate, our archeological data
(such as they are) do not show strong population increases in "optimum" areas like
the Lebanese woodland, but the very opposite-some of the most striking increases
are in "marginal" habitats like the Negev! Clearly, there were a number of very
interesting processes at work in the late Pleistocene Levant which could only be
handled by a more multivariate model. Finally, the model comes too close to making
population growth and climatic change into prime movers. It may be that the
"demographic change" which made cultivation seem like a good idea in Southwest
Asia was an increase in sedentary communities-and the latter may have begun in
response to changes in socio-political organization which had nothing to do with
either climate or population density.

THE ORIGINS OF AGRICULTURE IN SOUTHEAST ASIA

As far as the origins of agriculture go, China and the Far East are an enigma. It
will surely be decades before we have more than a sketchy outline of what happened
there, and I will touch only briefly on some of the latest discoveries.
East Asia has examples of both of Harris' artificial ecosystems, temperate seed-
cropping and tropical vegeculture. In the north, the Hwang-ho River valley of China
has been singled out as an early center for the cultivation of foxtail millet (Setaria
italica) and perhaps broomcorn millet (Panicum miliaceum). Although present
archeological discoveries trace these cultivated grasses back to only the fifth mil-
lenium B.C. (Chang 8), earlier stages in their history will almost certainly be filled
in some day. In addition, wheat and barley were introduced into China from western
Asia to augment this temperate seed farming. It is by no means clear to what extent
early Chinese agriculture was independent of neighboring regions, and I will offer
no model for its origins.
Southeast Asia is an area of root crops-yam and taro-none of which have yet
been preserved in archeological deposits of any antiquity. Nevertheless, it is tropical
Thailand which has produced the oldest archeological plant remains from the Far
East (see below). Southeast Asia could also prove to be the homeland of rice (Oryza
sativa), and it seems to me, as an outsider, that the "64 dollar question" in the rise
of Far Eastern agriculture is the origin of domestic rice. Asia has countless other
domestic plants, but none comes close to rice in providing an economic base for the

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 285

populations of the East. Sadly, we probably know less about the origins of rice than
any other major cereal, although a few new data are now at hand.
Of the more than 20 species of Oryza distributed through the Old World tropics,
one wild perennial form, Oryza perennis, is the likely ancestor of Oryza sativa. The
latter, although not the only species of cultivated rice, is the one which feeds more
Asians than any other (Peebles 56). Domestication has changed it to an annual
grass, given it a tough rachis and larger grain size (Heiser 27). Specimens of Oryza
sativa are known from archeological deposits of ca. 3000 B.C. at Harappa and
Mohenjodaro in the Indus River Valley and of ca. 2500 B.C. on Formosa (Peebles
56); the history of rice on the Chinese mainland is still longer, going back to the
fourth millenium B.C. Archeologist K. C. Chang (8) hypothesizes that rice was first
domesticated by early yam and taro growers of Southeast Asia, who might originally
have encountered it as a weed in their gardens. While the archeological data to test
this suggestion are not yet available, Chang does focus attention on Southeast Asia,
a region which has recently been the scene of several new breakthroughs in paleoeth-
nobotany.
The oldest radiocarbon-dated plant remains from Southeast Asia come from
Spirit Cave in northwest Thailand; they were recovered in 1966 by archeologist
Chester Gorman and identified by botanists D. Yen and P. van Royen. Gorman's
excavation through a complex series of living floors was masterful, and Yen and van
Royen's analysis careful and circumspect; I stress this for reasons which will be
apparent below. First let us examine the raw data from Spirit Cave as they originally
appear in print (Gorman 23).
The plant remains come from layers 4, 3, and 2 in ascending order of stratigraphy.
Associated radiocarbon dates are:
Hearth in layer 2: 8550 B.P. ? 200 (6600 B.C.)
Charcoal in layer 2a: 8750 B.P. ? 140 (6800 B.C.)
Charcoal in layer 4: 9180 B.P. ? 360 (7230 B.C.)
The plants themselves are identified as:
Almond (Prunus)
Butternut (Madhuca)
Candlenut (Aleurites)
Local wild nuts (Terminalia and Canarium)
Betel nut (Areca)
A bean ( Vicia or Phaseolus)3
Bottle gourd (Lagenaria)
Water chestnut (Trapa)
Black pepper (Piper)
Cucumber (Cucumis)
In addition, some radiocarbon samples are said to be "charred bamboo."
Gorman's presentation of the Spirit Cave data was cautious and reasonable. He
was well aware that none of the plants recovered had been shown to differ genetically

3Many authorities now feel that east Asian Phaseolus beans should be assigned to the genus
Vigna (Heiser 27).

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
286 FLANNERY

from their wild phenotypes; indeed, one "bean" might be a palm, and another could
not be identified to genus. Conspicuously absent from the collection of nuts and
condiments was rice-or any other real staple of the Far East. Such absence,
however, in no way prevented this from being one of the most important collections
of plant remains ever recovered from an archeological site. It was our first look at
the way man had used the floral environment of Southeast Asia in that remote,
pre-pottery, cave-dwelling period. To those of us interested in such things, it seemed
that the Spirit Cave data needed no "window dressing": they were exciting enough
at face value.
We were therefore surprised a few years later, by an article in a popular journal
by W. G. Solheim, overall director of the Thailand project of which Spirit Cave was
one part (Solheim 70). While conceding in one sentence that the material recovered
might all be "merely wild species gathered from the surrounding countryside,"
Solheim spent various sections of the article claiming that "an advanced knowledge
of horticulture" characterized the occupants of Spirit Cave "about 10,000 B.C."; and
his chronology chart (70, p. 38) pushes "incipient horticulture" in Southeast Asia
back to 20,000 B.C. Consequently, the subtitle of the article tells us that the agricul-
tural revolution "began some 5000 years earlier" in Southeast Asia than in the
Middle East. This seems to be a bit of an overstatement until we know, among other
things, whether we are dealing with a pea or a palm! It is not, however, the only
confusing aspect of the chronology chart; although Spirit Cave was still aceramic
at 6800 B.C., the chart tells us that pottery was invented in Southeast Asia at 13,000
B.C. But then, as Solheim says, his reconstruction "is largely hypothetical."
The problem with this overstatement is that it has created a widespread "credibil-
ity gap" between an important and reliable excavation by Gorman (which everyone
accepts) and the inflated claims in the popular press (to which professional archeolo-
gists react with skepticism). There is a real danger that the significance of Spirit
Cave-our first solid evidence for the plants early man was eating at 7200 B.C.
in the forested uplands of Thailand-will be lost somewhere in a smokescreen of
exaggerated claims.
Let us turn now to the problem of rice. Here Solheim's Thailand project has
contributed a major breakthrough. Imprints of cereal grains and husks of Oryza
sativa, the widespread Asian cultivated rice, have been identified in potsherds from
the site of Nok Nok Tha in central Thailand; as is frequently the case at prehistoric
villages, rice chaff was probably mixed into the pottery clay to give it body. The
potsherds involved are dated by radiocarbon to "before 3000 B.C.," to which Solheim
adds "possibly before 4000 B.C." (emphasis mine).4 Stated most conservatively, rice
must have been domesticated some time after the preceramic occupation of Spirit
Cave (6800 B.C.) and before the founding of Nok Nok Tha (4000 B.C.?). Its origins,
of course, could go back still farther elsewhere.
Let me close this section on the Old World on an optimistic note. My criticism
of Solheim's writings in the popular press should not be construed as a criticism of
4He shows "rice" on his chronology chart at 6000 B.C. (70, p. 38)

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 287

the archeological work itself, or as an underestimation of its importance. On the

contrary, his project has transformed Southeast Asia into one of the most exciting
frontiers for research on early domestication in the Old World, and I hope that
political conditions permit him to continue his work there for many years to come.

THE ORIGINS OF AGRICULTURE IN MESOAMERICA

Sometime between the close of the Pleistocene and the start of the fifth millenium
B.C., the Indians of Mexico first began the cultivation of a series of native plants
which would later become the staple foods of ancient Mesoamerican civilization.
For centuries these prehistoric inhabitants of the semiarid basins and valleys of
Mexico, Puebla, Oaxaca, Morelos, Guerrero, and Hidalgo had lived off the land,
learning the secrets of the wild vegetation-how to roast Agave to make it edible,
how to make wooden tongs for picking the spiny fruit of the organ cactus, how to
extract syrup from the pod of the mesquite, how to leach tannic acid from the acorn,
how to find wild bean and wild onion flowers in the dense underbrush, and how to
predict when they would be ready to harvest (Flannery 14). They survived on the
basis of a collecting strategy with many alternate moves and alternate food sources,
depending on whether the rains came too soon or too late, the spring was too cool
or too hot, the deer were in the valleys or up in the forest, the pinyon nut crop was
heavy or meager. Finally, by 5000 B.C., one of their ultimate strategies became the
artificial increase of certain edible plants by selection and planting. Beans, squashes,
pumpkins, amaranths, chiles, tomatoes, avocados (and perhaps even prickly pear,
maguey, and a whole series of semitropical fruits for which we have only Indian
names) came under cultivation not long after this date. But the most important of
these was maize or Indian corn, which they so modified that in the words of George
Beadle (3) these prehistoric Indians "can be credited with having produced the
greatest morphological change of any cultivated plant (assuming I am right about
its wild ancestor) and with having adapted corn to the widest geographical change
of any major crop plant."
Not so very long ago, the whole period of earliest agriculture in Mesoamerica was
hypothetical; it existed only on paper. Then Richard S. MacNeish, after years of
work on the northern periphery of Mesoamerica, began to explore the dry caves of
Tehuacan, Puebla, in the central Mexican highlands. In the years 1960-1964, these
caves yielded the preserved remains of many of the stages in the development of
Mesoamerica's most important crops, from wild plants to highly advanced domesti-
cates. Since then, many of us who were lucky enough to work for MacNeish have
radiated into adjacent regions of Mesoamerica to add some footnotes to the available
data. Techniques of flotation, adapted to archeology by Stuart Struever in the
midwestern United States, are now being used to recover carbonized seeds in regions
where dry caves simply do not exist. At the moment we know a good deal about
the order in which various plants were domesticated in several regions (Table 2).
We still do not know why they were domesticated, and it will certainly be a long
time before we do.

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
288 FLANNERY

0~~~~
-~~~~~~~~~~~~~ ~ ~ ~ ~ >0

to >)) ~ 44
~~~~ 0 ~~~~~~~~~~~~~.4)~~~~~~~~- -)C'

0. ~~~~~~~~~~~~~~~~~~~~~~

kr 0 (5 C'

0 O 0 0 >~~~~~~ u 0 )
0 O 0~ 4)O C' 0 u
0 4O0 )0 O Y 'sC -,'
..o k ) C'S en oo >

0 ~~~~~ 4)~~~~~1 = c t

0 ~~~~~~~~~~C' 0 kr A

O 04) >< u ~~~~~~~~~~~o~~~~0 Q~~~~9 kr t * :

~~ N ~ ~ ~ ~ 4)Q )

H 0~~~~~~~~~~~~~~~~~~'

4) c~~~~~~Cs r C ) >
z ~~~~~~~~~~~

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 289

Co

>
U
u 0 r-
'Co

co en co C>
W)
co en o co
7; wo
Ei
> cn 4L) .- . 0 u
2 > bo U
0 > u C's
ta. X C's
- l-- C's
S.- tf)
Q C> 'D -
> 0 ID
cl E tn
03> cl
> > U M
C's U 0o "V
U -
m >
u 5 6 uQ CAE ,
> u
0 ta. C,
0 c:
C)E c:
0 0 0 0
'CO m
CA
m tn 00
o r.
E tn M . -
la U
C:) E
I (=)
0 0
C) >Q t4
JD
8 cn 0
x 0
'T:0
-

A X
0;7,- r- CO > 0 0 co
"
u u

0 CA

m M
0 (U 0 Em
u m CA

> > CA
0 0
Cl
0 -
la >, CA
-ti -c: CA 0 CA
co E m E co CA 0
0 o m E o CA z
CA CA CA
m m
0 2?
0 =1
co- I...o
'Abo Cd co
E0.
- 5 E - cl
0 m 0 0 0 Q cd cn
> co CA 0 1 .0
cl 0= > > o - = I.-.
.0 c Co E 0
o -0
cn .0 co o u co M -o cn

>
u
>

u co
C's 'cl
en
rqC4 >
> 03 >
la.0 0
C's 00
bo " > m X E
o>1 U
cd 0 4) u UX
p > 7D &4
m M S= 0 > > -- 0U 'm C's co 0
'O C' E U r- 0
Cd 14E cl
>cnEU
0
- 0 0
UE >
E
bo
co o' 0 tti
C6 co
> 0. >
C)
o E00
w 0C)
C> > > 0 > .0 %D
0
0 rq
a-0 rj E
m m M C13 M m CO
rq u

qq
X
Z;
E
m t3 0

q :43m
14i 14i ;l, 0
" E m z
0
lu
0 9 E
C'k u Q "t M I I I.-, -

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
290 FLANNERY

The Botanical Problems

One of the major problems in telling the story of agricultural beginnings in
Mesoamerica is the fact that not even the botanists involved can agree on the origin
of maize, Mesoamerica's most important economic plant. There are two conflicting
views. The most recent theory-and that perhaps best known to archeologists-is
that of Paul Mangelsdorf (50). Mangelsdorf believes that cultivated maize de-
scended from a now extinct wild pod-popcorn in which individual kernels were
enclosed in and protected by chaff rather than by a cupulate fruit case. An older
theory, which has been vigorously revived in recent years by botanists such as
George Beadle, Jack R. Harlan, and Walton Galinat, holds that maize (Zea mays)
may be descended from the widespread Mexican grass teosinte (Zea mexicana), its
nearest relative. Most archeologists are now familiar with Mangelsdorf's views,
since he has been involved in the truly heroic task of analyzing virtually every
important collection of early archeological maize from Mexico or Guatemala (cf.
Mangelsdorf, MacNeish & Galinat 52). Beadle's recent work, however, is known
only from a 9-page paper in a popular museum bulletin (Beadle 4) and from "word
of mouth" among the archeologists who are in correspondence with him. In view
of this, I will try to sketch in the outlines of the argument from an anthropologist's
point of view. In the course of this, I will refer to three recent articles which should
be read by all anthropologists interested in the origins of maize.
The first of these articles is Garrison Wilkes' review of the wild relatives of maize
(Wilkes 79), which has greatly helped me in drawing the following "thumbnail
sketches" of teosinte and Tripsacum.
1. Teosinte (Zea mexicana) is the nearest relative to cultivated maize, having an
identical chromosome number. Indeed, as Wilkes puts it, "to the casual observer,
maize and teosinte are so similar in appearance, with nearly identical staminate
flowers borne in tassels and pistillate flowers enclosed in a system of husks in a lateral
position on the stem, that the most reliable characteristic separating the two is the
pistillate fruit-a distichous spike in teosinte and a polystichous structure (the
familiar ear) in maize." Teosinte is a native annual grass of the semiarid, subtropical
zones of Mexico and Guatemala, from southern Chihuahua to near the Guatema-
lan-Honduran border. It is a "short-day" plant which likes no more than 12 hours
of sunlight a day, combined with warm temperatures. The teosinte fruit has 7 to 12
seeds, enclosed in very hard cupulate fruit cases and set in sequence on a brittle
rachis. Like the wild wheat in the Near East, it shatters naturally and is hence very
difficult to harvest efficiently. Nevertheless, it is used by some Mexican Indians as
a "starvation food." Parenthetically, I might add that its brittle rachis and short
period of peak maturation make it most efficient to harvest by large work gangs or
"macrobands" (MacNeish 47); small "microbands" or individual families would
take too long to get the whole crop before it shattered.
It is this plant which Beadle suggests was the wild ancestor of cultivated maize.
His evidence will be discussed in a later section. But before proceeding to Wilkes'
description of Tripsacum, let me add the following observations which Richard
Ford and I (17) were able to make in 1971 when we accompanied Beadle and Wilkes
on their "wild teosinte harvest" in Guerrero, Mexico.

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 291

Teosinte is a weedy, pioneer plant which colonizes natural scars in the landscape;
in Guerrero the landscape was a semiarid thorn forest with Lucaena, Acacia, and
Setaria. When corn fields are abandoned today, they are rapidly invaded by teosinte;
at one point we drove for 20 kilometers without losing sight of massive teosinte
stands, up to 2 meters high. If a group of hunter-gatherers cleared a campsite in such
a thorn forest, the following year they would return to find their former campsite
a teosinte field. Moreover, to our surprise, wild runner beans (Phaseolus sp.) and
wild squash (Cucurbita sp.) occur naturally in such fields, with the beans twining
around the teosinte (Figure 2). The Zea-bean-squash triumvirate is thus not an
invention of the Indians; nature provided the model.
Due to the long "dormant" period needed by teosinte seeds, dense stands would
occur only about every 2 years. This would continue until regrowth of the thorn

Ift

Figure 2 Wild runner beans twine around a stalk of wild teosinte in second-growth
vegetation near Chilpancingo, Guerrero, 1971.

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
292 FLANNERY

forest eventually shaded out the teosinte. When this happens, relict populations
survive on scree or landslide areas, and on patches cleared by accidental
fire.
2. Tnipsacum-a genus composed of nine diverse perennial grass species scat-
tered from Texas to South America-is a more distant relative of maize, the two
having different chromosome numbers. Although crosses between maize and Tnip-
sacum can be accomplished in the laboratory, according to Wilkes "all evidence
indicates that the recent evolution within the genus Tripsacum has been indepen-
dent and distinct from that of maize." And although Tnipsacum may share the same
semiarid mountain habitat with teosinte, the two do not hybridize in the wild and
have not yet been successfully hybridized in the laboratory. Nor is there any evi-
dence that the Indians of Mesoamerica ate Tnipsacum or, in fact, used it in any way
(except possibly as grass for thatching). This evidence suggests that Tripsacum has
not played as important a role in ancient Mesoamerica as was once thought, and
that the frequently used term "tripsacoid maize" (Mangelsdorf, MacNeish & Gali-
nat 52) is probably a misnomer.
Perhaps equally important, it now appears that Tnipsacum has played no role in
the origins of teosinte either. It was once believed that teosinte was a derivative of
a maize- Tripsacum hybrid, rather than a true wild grass (Mangelsdorf 50). In a 1972
conference at Harvard, Paul Mangelsdorf withdrew his support from this theory of
some 30-years' standing (64). It now seems clear that teosinte is a true wild grass,
independent from Tnipsacum and far more closely related to maize.
Teosinte and Tnipsacum are wild species which can still be studied by botanists
today. We must now, however, consider the possibility of a third "wild species"
which has never been seen growing or collected by a botanist.
3. Wild maize-a presumably annual grass once native to Mexico-was the
hypothetical ancestor of domestic maize originally proposed by Mangelsdorf (50).
This wild maize was supposed to have had small (female) cobs from the top of which
a small (male) tassel grew. It was supposed to have been a "pod" corn (in which
individual kernels were enclosed in and protected by chaff rather than by a cupulate
fruit case) and a "pop" corn-one which would explode when heated. After giving
rise to domestic corn, it was thought to have become extinct, either by overgrazing
from introduced European animals or by "swamping" from domestic corn pollen.
Mangelsdorf's theory was seemingly confirmed by the discovery of tiny corn cobs
in a Tehuacain cave, dated to 5050 B.C., in which the tassel did grow out of the top
of the cob (Mangelsdorf, MacNeish & Galinat 52). These and later cobs had a
number of teosinte-like characters, but these Mangelsdorf explained as the result of
(a) hybridization with Tripsacum to produce teosinte, and (b) subsequent back-
crossing with the latter to produce "teosinte introgression."
The Mangelsdorf hypothesis was also seemingly confirmed by the discovery of
60,000-year-old Zea pollen in a lake core from the Valley of Mexico (Barghoorn,
Wolf & Clisby 2). This pollen was assigned to maize because of its large size relative
to teosinte pollen.
Although Mangelsdorf's hypothetical wild ancestor was widely accepted by ar-
cheologists, many botanists never believed in it. The skeptics included George

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 293

Beadle, Jack R. Harlan, and many others. In particular, Beadle was skeptical
because his own cytological and genetic work suggested that corn and teosinte were
in fact the same species. Over the last 3 years, Beadle's renewed laboratory research
and field work on teosinte have led him to the following conclusions, which appear
in the second of my three recommended articles (Beadle 4):
1. The number of significant independently inherited gene differences between
maize and teosinte is "not great-perhaps 6 to 10." This is based on the rate of
recovery of pure "parental" types among 30,000 second generation hybrids of corn
and teosinte growing at the International Center for Maize and Wheat Improvement
near Mexico City.
2. Teosinte-which can be eaten after grinding in a mortar or "popping" like
popcorn-can be made a more easily usable food plant by just two mutations: "one
to a non-shattering rachis, so the fruits will not be scattered and lost as food; the
other to a soft fruitcase, so the kernels can be threshed free of them." It is not known
how many genes control the toughening of the rachis, but a single gene-the
so-called "tunicate allele"-can produce soft glumes and a lessened tendency to
shatter. (Note the striking similarity between this model and the one Helbaek has
presented for wheat in the Near East.)
3. The tiny, inch-long corn cobs from the Tehuacain caves, which have been
interpreted as "wild maize", can as easily be interpreted as representing "stages in
the transition of teosinte to corn through human selection." Cobs closely matching
them are readily recovered in second (and later) generations of corn-teosinte hy-
brids. They are said to have "fragile" cobs-possibly a persistence of the brittle
rachis of teosinte. Some of them are two-ranked (distichous), i.e. having two rows
of kernels-another teosinte trait. Finally, the early cobs have long, soft glumes
suggestive of a tunicate allele. For Beadle it is not necessary to propose "teosinte
introgression" for these early corn specimens-merely "teosinte ancestry."
4. Mangelsdorf's postulated wild maize could not have survived in the wild
because it had no seed dispersal mechanism; the cob is a "man-made monstrosity"
which no other grass has, and this is one of the reasons domestic corn cannot survive
without human intervention. Yet we are asked to believe that it survived until the
Spanish Conquest and then became extinct within 400 years.
5. It should be pointed out that Mangelsdorf's theory has not remained static in
the face of criticism. Whereas he formerly regarded teosinte as too different from
maize to have been its ancesto-, he now proposes that teosinte arose through
mutation from wild corn (Mangelsdorf 51). To this Beadle replies "if corn could
have given rise to teosinte, the reverse must also be possible-and I would say much
more probable, for teosinte is a highly successful wild plant and corn is not."5
The third article which should be read by all archeologists interested in the origins
of maize is Walton Galinat's recent synthesis in the AnnualReview of Genetics (20).
Galinat, a long-time associate of Mangelsdorf's and perhaps the leading authority
on maize anatomy, makes the following points, which I have placed in quotes to
distinguish them from my own comments:

5Readers interested in this controversy should also read Mangelsdorf's reply to Beadle (51).

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
294 FLANNERY

1. "Teosinte and maize hybridize almost freely, and their hybrids are fully fertile
on both selfing and backcrossing in either direction." Teosinte and maize have the
same chromosome number (n = 10); as shown by R. A. Emerson and George Beadle
as early as 1932, their chromosomes are of equal length and similar in arm ratios;
their chromosomes pair closely at pachytene; they have essentially the same frequen-
cies of crossing over; where known, the position of various genes on the chromo-
somes is similar.
2. Anatomical studies (by Galinat himself, 19) show a "clear-cut connecting link
between the maize cob and the cupulate fruit case of teosinte." In other words, the
cupule which holds the kernel in both archeological and modern maize cobs has
almost certainly evolved from the cupulate fruit case of teosinte. This would strongly
support the Beadle view, since Mangelsdorfs postulated wild maize did not have
a fruit case.
3. It is simply not true that all maize has pollen which is larger than (and hence
distinguishable from) teosinte pollen. This is true for only four out of ten teosinte
varieties; other studies "not only show that the size of pollen in most races of teosinte
overlaps that of the Chapalote maize but that one teosinte, Jutiapa, has significantly
larger pollen than this race of primitive maize." Galinat refers to Chapalote (an
"ancient indigenous race" of Mexican maize) because of its great genetic similarity
to the earliest maize from Tehuacain, and he further refers to a demonstrated
"correlation between pollen diameters and ear or style length in the races of maize
in Mexico." Thus there is an inconsistency between the large size of the 60,000-year
old "fossil wild maize pollen" found in the Valley of Mexico (Barghoorn, Wolfe &
Clisby 2) and the smaller size of primitive maize pollen. Galinat points out that the
various preparation techniques used for pollen analysis-glycerin, lactic acid, ace-
tolyis, and so on-may have a great effect on grain size, making the results of two
palynological studies noncomparable. More to the point, when a scanning electron
microscope is turned on the surface of maize and teosinte pollen, no significant
difference can be detected-so far, at least (Galinat 21). This strongly suggests they
are the same species.
4. The idea that "wild maize" became extinct because of competition from do-
mestic maize or overgrazing by animals introduced by the Spanish is shaky; it
assumes that "the original wild maize was unable to adapt to forces that teosinte
has so successfully resisted." Not only does teosinte survive in spite of heavy maize
competion, but "livestock, far from completely destroying teosinte, tend to spread
it through their fecal droppings" (cf. Wilkes 78).
5. "The results of recent studies of electrophoretic patterns of storage proteins
are consistent with the hypothesis that maize is domesticated teosinte." This com-
ment, based on work by J. G. Waines (Galinat 20, p. 475), is strengthened by recent
work by Robson & Konlande (62) and Cowan (11), which show that maize and
teosinte have similar nutritional makeup and amino acid balances.
6. "The oldest known archeological maize cobs from Tehuac'an, Mexico, about
7000 years in age, which have been assumed to be those of wild maize, can also be
interpreted as in the early stages of transformation of teosinte to maize. All of these

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 295

primitive cobs have cupules that could have evolved directly from the fruit case of
teosinte and some of them are two-ranked and have cupulate interspaces as in the
pistillate spike of teosinte."
This final comment, from a botanical anatomist who has handled every one of the
most ancient maize specimens from Mexico, is perhaps the best statement of where
we stand at the moment. If it ultimately turns out that maize is a domesticated,
highly evolved descendant of teosinte, the history of maize becomes a good deal
simpler (involving less hybridization to account for its teosinte-like characteristics),
and it will no longer be necessary to postulate the complete extinction of its ancestor.
But our troubles will still be far from over, since it remains to be shown what precise
genetic changes led to recognizable prehistoric maize.
Whether, as now seems reasonable, teosinte is the ancestor of maize (or even if
"wild maize" actually existed), it now appears that Zea was originally domesticated
prior to 5000 B.C. By the fifth millenium B.C., human selection of rare or mutant
genes had progressed so far that the occupants of the Tehuacain Valley had plants
with distichous or polystichous cobs with soft glumes and a partially toughened
rachis. Further selection and hybridization over the next 3000 years produced corn
cobs which share many genetic characters with two surviving "primitive races" of
maize, Chapalote and Nal-Tel. However, rather than assigning certain second-
millenium corn cobs to either of those races, I would suggest that we simply say that
modern Chapalote and Nal-Tel retain many characters of Formative corn-a corn
that must have been very unstandardized and genetically diverse, with a whole
spectrum of phenotypes from teosinte-like to Chapalote-like.
Where was Zea originally domesticated? Teosinte's former wild range includes
much of the short-day, semiarid, semitropical uplands between Chihuahua and
Guatemala. If we suggest that it may first have been domesticated on the margins
of its best range-as we have for wheat in the Near East-the area is further
increased. But were its "margins" the very arid highland regions (like Tehuacain)
where teosinte does not grow today, or the edges of the humid lowlands where it
gives way to tropical vegetation?
Some botanists feel that the large-seeded Chalco race of teosinte, a native of the
Valley of Mexico, is the most closely related to maize, and that the Guerrero and
southern Guatemalan races are least maize-like (Wilkes 79), while others attribute
the Chalco race's resemblance to maize to hybridization with the latter. Seeds of
Chalco-type teosinte have been found at the archeological site of Tlapacoya in the
Valley of Mexico in levels dating to 5000 B.C. (Lorenzo & Gonzales 44); but by this
time period there are already specimens of true maize from Tehuacain to the south.
Small pollen grains believed to be those of teosinte occur in Guila Naquitz Cave,
Oaxaca, in levels dated to nearly 7000 B.C. (Schoenwetter 63); but there is no
evidence of domestication at this time in the form of actual mutant plant material.
Indeed, actual teosinte material from dry caves in Mesoamerica is hard to come by.
Equally serious is the possibility that surviving races of teosinte may not give a
clue to the origins of maize because (like the Near Eastern cereals described by
Harlan) they have become "weed" races. According to Wilkes (79), at 10,000 B.C.

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
296 FLANNERY

wild teosinte may have been a more suitable ancestor than the sophisticated, surviv-
ing "weed" teosintes, which have had to specialize to survive, and which have
become "maize-mimics" as they invaded the cornfields; the Chalco race is particu-
larly maize-mimetic. And Galinat (20) points out that domestic maize and today's
teosinte, having exchanged genes over thousands of years, have evolved together and
are perhaps more similar than they would otherwise be. Contributing to this is the
fact that, on the basis of archeological plant samples (Smith 66), prehistoric farmers
apparently encouraged teosinte to hybridize with maize in their fields. In addition
to a shot of hybrid vigor, this gave corn drought and frost resistance (Jones 33).
In sum, we have little data with which to pinpoint the origins of Zea cultivation,
except the very early date of the corncobs from Tehuacain-an area where teosinte
has never been collected in the wild. We know even less about why domestication
began, as the tentative model below will show.

A Model for Early Maize Cultivation: Stage I

Suppose that teosinte (or even an equally unappetizing hypothetical wild maize) was
the ancestral Zea. Why would such a plant have been domesticated in the first place?
I am reluctant to propose for Mesoamerica a "density equilibrium" model for
early corn domestication, such as the Binford hypothesis I applied to the Near East.
The fact is that, prior to 5000 B.C., human population densities in those parts of
Mesoamerica which we have surveyed are very low. There is no area in which we
can document a population expanding so fast that it might have affected the density
equilibrium of adjacent regions; however, MacNeish (48) feels that some areas of
northern Mexico (Nuevo Le6n-Tamaulipas) did have preceramic populations
greater than those of Puebla or Oaxaca. There is one other possibility which is raise
by Richard Ford's work in the arid Southwest (Ford 18). Like the Southwest,
highland Mesoamerica has great contrasts in wild productivity between wet and dry
years. Cultivation might have arisen as an attempt to "even out" the difference
between these extremes by increasing the range of weedy, pioneer annuals, but we
just don't know for sure. Whatever the cause, the origins of Zea cultivation amount
to a deliberate increase in the availability of an "emergency ration"-one of the few
that could be increased on a yearly basis.
One of the important small biotopes of the semiarid valleys of the central and
southern Mexican highlands is the tributary barranca. In these slightly more humid
canyons, cut by streams which vary from seasonal to permanent, grow varieties of
herbs and grasses which may be absent or less common out on the main valley floor.
Two of the grasses which grow in such habitats are foxtail grass (Setaria sp.) and
teosinte; both were harvested and eaten by prehistoric Indians. Setaria, which
reaches maturity in the early autumn, was ground and threshed in mortars. Eric
Callen (7), who analyzed the human coprolites from MacNeish's dry caves in
Tehuacain and Tamaulipas, believed he could detect a selection for larger grain size
in some regions, and perhaps even the first attempts at Setaria cultivation. Teosinte,
which matures later in the fall, can be ground up to produce coarse but rather
pleasant-tasting unleavened cakes when cooked on a hot, flat rock.

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 297

There is a good deal of roughage in teosinte-up to 53 percent-and it may be

harder to prepare than Setaria. But Beadle has experimented on himself and his
colleagues sufficiently to show that daily consumption of 150 grams of teosinte flour
has no ill effect (Beadle 4).
Empirical archeological data suggest that such things as Setaria and teosinte were
really a very minor part of a diet that emphasized prickly pear, roasted Agave,
mesquite, acorns and pinyon nuts, hackberry, wild avocado, deer, cottontail, mud
turtle, and dove (Flannery 14). In a wet year, however, food collectors could count
on a good Setaria harvest in the tributary barrancas, and the coprolite data indicates
that they did. In a dry year, on the other hand, the Setaria harvest could only be
raised to its usual level by augmentation with teosinte, which matured slightly later
in the same habitat. But no matter how much you select and plant Setaria, it stays
the same unappetizing weed. Whereas, if Beadle is correct, teosinte responded to
cultivation and selection with a series of favorable genetic changes which moved it
in the direction of maize. And this may have tipped the balance in favor of increased
attention to the genus Zea on the part of man.

A Model for Early Maize Cultivation: Stage II

In attempting to figure out why man would increase Zea cultivation, there are a
number of variables which need to be quantified. These include (a) the productivity
of wild teosinte; (b) the productivity of cultivated teosinte; (c) the productivity of
early maize; (d) the productivity of the competing vegetation which would have to
be removed in cultivation; and (e) the relative man-hours of work involved in
clearing, cultivating, and so on. Various members of the University of Michigan's
Human Ecology project in Oaxaca, Mexico, have tried (in collaboration with
George Beadle) to quantify these variables. Our preliminary results, which should
not be taken as gospel, are as follows.
1. In 1971, Richard Ford and I measured the productivity of wild teosinte sample
plots during Beadle's "teosinte harvest" in Guerrero and Valle de Bravo, Mexico
(Flannery & Ford 17). Our least productive plot-on a scree slope probably approx-
imating natural wild conditions-yielded an estimated 305 kg per hectare. Since
50% of the seed is inedible roughage, this corrects to 152.5 kg per hectare or less
than a wild mesquite grove produces (see below). Our most productive plot-on an
abandoned (fallow) corn field probably approximating the best cultivated conditions
-yielded an estimated 1254 kg per hectare, or a corrected yield (minus roughage)
of 627 kg per hectare. This is comparable to the yields of wild Near Eastern cereals,
but could probably only be achieved under exceptional circumstances.
2. Robert D. Drennan (13) supervised the cultivation of a field of Chalco teosinte
in the Valley of Oaxaca in 1972 (Figure 3). The yield was 161 kg per hectare;
corrected for roughage, the yield would be only 80 kg per hectare-roughly that
estimated by Kirkby for the maize of 5000 B.C. (see below). Two factors, however,
lead Drennan to believe that this is a substandard yield. First, 1972 was a drought
year in which many maize crops in the area failed altogether. Second, in lieu of any
model for prehistoric teosinte cultivation, plants were spaced as far apart as is
typical for maize; in retrospect, the experimental field could have been planted twice

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
298 FLANNERY

Figure 3 R. D. Drennan and Guadalupe Luna H. from Fabrica San Jose, San Agustin
Etla. Oaxaca, standing in experimental cultivated teosinte field, 1972.

as densely. To this we mnight add the fact that Chalco teosinte, adapted to 2300 m
elevation, might not do as well at 1500 m. Nevertheless, the experiment gives us
somne idea of the perils of teosinte cultivation in a dry year.
3. The native vegetation of river fioodplains in the Mexican highlands-mnost
frequently groves of mnesquite (Prosopis juliRlora)-would have to be cleared to
allow cultivation. Near Mitla, Oaxaca, we measured the productivity of edible
mesquite pods from such groves; the highest yield was 184 kg per hectare, with
yields of 160-180 kg per hectare common. Thus, disregarding for the moment any
nutritional differences between Prosopis and Zea, it would hardly seem worth
clearing the native mnesquite fromn the fioodplains unless a yield higher than
per hectare could be expected.
4. Recent field studies in Oaxaca by Anne Kirkby (38) show that Zapotec Indian
farmers do not consider cultivation and land clearance to be worthwhile unless a
yield of at least 200 to 250 kg (shelled maize) per hectare can be expected.
5. Kirkby also discovered a linear regression relationship between mean corn cob
length and yield in kg per hectare for Indian fields in the Valley of Qaxaca (Kirkby
38). Based on this relationship, and using figures on the mean length of corn cobs
recovered by MacNeish's Tehuacain excavations (and ours in Qaxaca), she then
calculated an "estimated yield" for various periods of prehistory.
These estimates are based on the assumption of a cob-to-plant ratio like today's,
whereas Beadle suspects early maize may have had more cobs per plant (like

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 299

teosinte); nevertheless, we offer them in lieu of any other estimates. The earliest cobs
from Tehuacain suggest a yield of only 60-80 kg per hectare; later preceramic cobs
(ca. 3000 B.C.) suggest yields of 90-120 kg per hectare. According to Kirkby's figures,
maize did not cross the critical threshhold of 200-250 kg until sometime between
2000 and 1500 B.C. This is remarkably close to the actual period at which perma-
nent villages on (or overlooking) good alluvial agricultural land became the domi-
nant type of settlement in Mesoamerica. Yields of 500-800 kg per hectare (as in
wild Near Eastern cereals) may not have become common until centuries after
village life had been established in Mexico!
6. We do not yet have reliable data on the man-hours required to harvest a
hectare of wild teosinte, cultivate a hectare of domestic teosinte, or clear a hectare
of mesquite groves. We have only Drennan's figures from his Oaxaca teosinte
farming (13), which indicate that the harvesting and threshing amount to about 50
percent of the labor input. Cultivation may therefore require twice the labor of
collecting the plant in the wild. On the basis of his Guerrero harvest (see above),
Beadle suggests that a man could harvest perhaps one liter of teosinte per hour by
beating out the seeds on a blanket.
Our tentative quantification of some of these variables allows us to complete the
second stage of our model. Let us assume that teosinte had been domesticated
somewhere in the humid barranca-piedmont habitat it shares with Setaria. Only in
the best of conditions, as a second-growth pioneer on well-watered alluvial fans,
would it approach the productivity of the wild Near Eastern cereals, and teosinte
is even harder to harvest and process, being half roughage. In many areas it would
not have been worthwhile to remove the mesquite cover of the main river floodplains
to cultivate such a plant. A more reasonable strategy would have been to leave the
mesquite on the main valley floor to produce its 180 kg per hectare each year, while
growing teosinte in the piedmont barrancas where it was at home. (Our figures
suggest that teosinte probably could have yielded an average of 150-200 kg per
hectare, reaching 600 kg under exceptional conditions but falling to below 100 in
drought years.) That such a strategy was used is suggested by the fact that for
thousands of years after Zea had been domesticated, no valley-floor villages ap-
peared in Mesoamerica, although MacNeish (47) suspects pit-house settlements in
the tributary barrancas. The Indians divided their time among these various bi-
otopes and probably did not regard any one zone as more "crucial" than the others.
But gradual genetic change leading to larger cob sizes eventually raised the
minimum productivity of Zea to 200-250 kg per hectare or more. At some point
prior to 1500 B.c.-the exact date may vary from one valley to another-it crossed
the "threshhold" which the Indians felt made it worthwhile to clear the mesquite
groves for cultivation. By 1300 B.C., permanent villages of wattle-and-daub houses
on (or overlooking) the main river floodplains were widespread from Puebla to
Guatemala. And when the Indians brought maize out of the barrancas, they also
brought its two pioneer companions-beans and cucurbits-with whom it had
coexisted for thousands of years. If you add to this the cultivated avocado (Persea
americana), also originally native to the piedmont barrancas, you have the four most
common cultivated genera found throughout Mesoamerica at 1300 B.C. With maize

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 300 FLANNERY

providing the carbohydrate, beans and squash seeds the plant protein, and avocados
the fats and oils, the early Mesoamerican villagers were on their way.

Other Mesoamerican Cultivars

There is no way to discuss all of Mesoamerica's early cultivars in an article of this

length, but since beans and squash have already been mentioned, let us pursue them
a bit farther. Some light has been shed on the early use of beans and cucurbits by
analyses of plant remains from 8700-6700 B.C. levels at Guila Naquitz, Oaxaca.
During a study to be published later, the nonrandom association of various plant
species with each other on cave living floors was demonstrated by means of an
ordered matrix of Pearson's r values (Whallon 75). The results suggest that wild
runner beans were collected with plants from the oak woodland zone, such as acorns
and pinyon nuts, while wild cucurbits were collected in a somewhat lower thorn
forest zone, along with cactus fruits and Agave. This is a helpful clue to some of
the floral zones preferred by these wild plants within the piedmont-barranca area.
Three species of Phaseolus-common beans (P. vulgaris), runner beans (P. coc-
cineus), and tepary beans (P. acutifolius)-have wild ancestors in Mexico. The
oldest beans archeologically documented are wild runner beans from caves in
Ocampo, Tamaulipas (7000-5500 B.C.) and the already mentioned specimens from
Oaxaca (8700-6700 B.C.); those from Oaxaca belong to a species which was never
domesticated, while the Ocampo runner beans are wild P. coccineus. Domestic
tepary beans and common beans both make archeological appearances between
4000 and 3000 B.C. According to Kaplan (34), three of the critical changes accompa-
nying the domestication of the common bean were (a) an increase in seed permeabil-
ity, so the beans did not need to be soaked in water as long; (b) a change from
corkscrew-twisted pods (which shatter when ripe) to limp, straight, nonshattering
pods; and (c) in some cases, a shift from perennial to annual growth patterns.
Because beans are intimately associated with maize, both in the wild (see above) and
in the diet of ancient Mesoamerica, it is also worth noting that they are rich in the
amino acid lysine. Since maize is deficient in lysine, the combination of corn and
beans makes for a better plant protein (Kaplan 34).
Domestic beans are among the plants which are represented by very few actual
specimens prior to 3000 B.C.; for example, six pod valves are scattered through levels
in Tamaulipas spanning the period 4000-2300 B.C., and the oldest Tehuac'an speci-
men (4000 B.C. ?) is a single pod. The situation is roughly similar with the genus
Cucurbita (squashes and pumpkins), and is complicated still further by the fact that
in most cases the wild cucurbit ancestors are not known for certain.
Botanists H. C. Cutler & T. W. Whitaker (12) regard certain changes in the
peduncle or stem following domestication as crucial for identification, but many of
the earliest cucurbit specimens are seeds, and isolated seeds at that. Little is known
(or at least, published) about the range of variation in seed size and shape in any
of the potential wild ancestors; based on specimens in the Ethnobotanical Labora-
tory at the University of Michigan, I am struck by the considerable size-and-shape
range in seeds from a single wild cucurbit. To me this would suggest that one might
really need a whole "population" of seeds (say, 100 specimens) before he could prove

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 301

that an ancient sample was outside the wild range. To describe an isolated seed as
"probably wild," "probably domestic," or "pumpkin-like, but too small to be a
domestic" may simply be another way of saying that wild cucurbits (and probably
early cultivated cucurbits as well) had a great deal of seed variability and did not
yet belong to any of the standardized domestic races we know today.
Wild cucurbits, for the most part, have flesh which is either so bitter or so thin
and dry (like a gourd) that it cannot be eaten; it was the seeds that were originally
important, while the edible flesh is a product of domestication. Cucurbit seeds occur
as far back as 8000-7000 B.C. in caves in Oaxaca and Tamaulipas (Table 2); these
earliest specimens are probably all wild forms, or "weedy camp followers" (see
below). Morphologically, some resemble seeds of the pumpkin (C pepo), but Cutler
& Whitaker (12) specifically state that the specimens from Tamaulipas, as well as
a later seed from Tehuacain (5200 B.C.), are "wild." Since no one has ever seen a
"wild pumpkin"-its ancestor is, in fact, not yet decided upon (Whitaker 76)-it
is hard to know just what to call these seeds. I am not a botanist, but I would suggest:
"seeds of an unidentified cucurbit, presumably wild, resembling C pepo but consti-
tuting an inadequate sample for species identification."
When peduncles are present, one could presumably assign a species name with
more confidence; but peduncles do not appear until much later in the archeological
sequence. For example, three seeds (two of them marked as dubious) of Cucurbita
mixta squash appear as early as 5000 B.C. at Coxcatlan Cave, but mixta peduncles
do not occur until 3000 B.C. For which of these dates should the first domestication
be claimed?
As Cutler & Whitaker point out (12, 77), cucurbits tend to be "weedy camp
followers" which do well on disturbed soils, like the talus slope of an occupied cave.
Their wild forms resemble the bottle gourd, one of the plants with the longest
documented history of human use. It may be that they were originally domesticated
by foragers who already knew and cultivated the bottle gourd and who therefore
instantly recognized the cucurbits as potentially useful. At any rate, they are one
of the oldest Mesoamerican plants whose use can be documented, from Oaxaca to
Tamaulipas.

THE ORIGINS OF AGRICULTURE IN THE ANDES

Three very different major biomes contributed to the richness and complexity of
Andean agriculture. The first of these was the Amazon jungle and the tropical
eastern slopes of the Andes. This zone contains many of the wild ancestors of the
Peruvian domesticates-manioc, peanuts, guavas, coca, and lima beans-but little
is known of their history because of the poor archeological preservation in this
humid area. The second zone is the mountains themselves, which contributed the
wild ancestors of potato, quinoa (Chenopodium quinoa), oca (Oxalis tuberosa),
olluco ( Ullucus tuberosus), and other crops. Little is known of the history of these
cultivars either, since dry caves are scarce in the Andes and preceramic archeology
there is still in its early stages. Finally, there is the coastal desert to the west of the
Peruvian Andes. The remarkable archeological preservation in this zone, caused by

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 302 FLANNERY

the dessicating effects of the cold Humboldt current, makes it the richest in ancient
plant remains of any Andean region. Unfortunately, very few wild ancestors of any
consequence are native to this desert, and most of the early cultivars which show
up there were probably introduced from the mountains, the Amazonian forest, or
Mesoamerica.
Once again, Richard S. MacNeish has been a major contributor to our knowledge
of early domestication-this time through his 1969-1972 project at Ayacucho in the
south-central highlands (MacNeish, Nelken-Turner & Garcia 49). The caves of
Ayacucho, while not as dry as those of Tehuacain, appear to have settled at least
one of the Andes' thorniest problems: whether maize was locally domesticated or
introduced from Mesoamerica. Arguing for the latter alternative was the fact that
neither teosinte (or any form of "wild maize") had ever been found in South
America. Galinat & MacNeish (22) now report that Zea cobs from 3000-2500 B.C.
levels at Rosamachay Cave, currently Peru's oldest maize specimens, belong to a
teosinte-influenced race, most closely related to Mexico's Nal-Tel and presumably
introduced from that region. It would thus seem to have taken maize about 2500
years to reach the Peruvian highlands after its initial appearance at Tehuacain; it
seemingly did not reach the desert coast until sometime after 2000 B.C. (Table 3).
Another problem is the origin of the common bean Phaseolus vulgaris. According
to Kaplan (36), it is not yet settled whether P. vulgaris has one center of origin or
multiple centers, though he leans toward the latter. There is a wild vulgaris in
Mexico, and the Andean wild form P. aborigineus is so similar to vulgaris that "you
could conceivably lump them all as races of one species." Indeed, some varieties of
aborigineus are less impermeable than wild vulgaris, and may possibly be escapes
derived from cultivated vulgaris.
The available radiocarbon dates do little to resolve this problem. The oldest
domestic vulgaris beans identified by Kaplan are from Thomas Lynch's excavations
at Guitarrero Cave in the north highlands of Peru, dated to 5600 B.C. (Kaplan,
Lynch & Smith 37); vulgaris first appears in the Ayacucho sequence about 2800
B.C. Its oldest occurrences in Mexico are about 4000 B.C. in Tamaulipas and Tehua-
can (see Table 2). Thus it is presently impossible to argue for its introduction into
Peru from Mesoamerica, and its diffusion in the opposite direction could be sup-
ported only by the most tenuous arguments.
Similar problems occur with cotton, chile peppers, and certain cucurbits which
have radiocarbon dates that put them in Mexico some 1500 to 2000 years earlier
than in Peru. Such dates may simply result from the vagaries of preservation,
sampling, and intrusive rodent burrows, and they should not be relied on too
heavily. Almost certainly, the archeological record pertains to different species: in
the case of cotton, hirsutum in Mexico and barbadense in Peru; in the case of chiles,
annuum in Mexico, baccatum and frutescens in Peru (59). Moreover, there are the
problems of population size and range of variation which I mentioned earlier; if
early domesticates were unstandardized and genetically diverse, isolated specimens
which resemble one or another of today's "races" may contribute more to confusion
than clarification.
My lack of first-hand familiarity with the Andes prevents me from proposing a
processual model for the origins of agriculture there; I don't know why it began.

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 303

But let me present a tentative historical model which does nothing more than
account for the fragmentary plants and radiocarbon dates we have.
The oldest reported archeological plant remains from the Andes are rind frag-
ments of bottle gourd (Lagenaria siceraria) from 11,000 B.C. levels at MacNeish's
Pikimachay Cave in Ayacucho (MacNeish et al 49). Botanist Barbara Pickersgill
(59), whom I have already praised for her skepticism, feels that the wild bottle gourd
was distributed pantropically before man came on the scene and that these few rind
fragments do not necessarily prove either domestication or diffusion from another
continent. Moreover, the fact that overlying levels of the same cave (falling in the
11,000-6,000 B.C. range) have no preservation of plant remains makes these early
rind fragments puzzling. Is it remotely possible they came down a rodent burrow
from the 6,000 B.C. levels, which do have good preservation? If not, we are faced
with a surprising situation: a cave with preservation at 11,000 B.C., followed by 5000
years with no preservation.
Lynch's Guitarrero Cave shows us that by 5600 B.C. the occupants of the Peruvian
highlands were cultivating common beans and lima beans. The latter had to have
come from the eastern slopes of the Andes, where their wild ancestors live; and
MacNeish's Jaywamachay Cave yielded a seed of achiote (Bixa orellana) which, if
it proves not to have been intrusive, suggests that the Ayacucho highlanders were
importing this red pigment from the eastern slopes by this time (MacNeish, Nelken-
Turner & Garcia 49). I draw two conclusions from this: (a) agriculture began in
Peru at too early a date to have been stimulated initially by diffusion from
Mesoamerica, and (b) from the very beginning the Amazonian slopes must have
played a significant role, in spite of our lack of direct data from that region.
By 3000-2500 B.C., domesticates from Mesoamerica were reaching Peru. Maize
of Mexican ancestry reached the highlands, then spread to the desert coast; mos-
chata squash may have been an introduction of the same period. During the mil-
lenium 2500-1500 B.C., numerous Amazonian or eastern slope products like guava
(Psidium guajava), manioc (Manihot esculentum), peanuts (Arachis hypogaea),
and lima beans became firmly established on the coastal desert (Lanning 39, Patter-
son 55). In turn, maize must have traveled to the Amazon basin, but we know little
of the time period involved. And we know next to nothing of the early history of
Peru's important high altitude crops, potato and quinoa, although they grow wild
along the routes through which these exchanges between the Coast and the Amazon
must have taken place.
What this all means in terms of process or developmental models is far from clear.
One of the most fascinating aspects of the Peruvian coast has been the possibility
that here, as in parts of the Near East, sedentary or semisedentary life may have
been possible before agriculture. The enormous fishing and shell-fishing resources
of the Humboldt current make it one of the most productive environments available
to ancient man. Could this be another situation where village life began with a wild
food base, populations rose, and agriculture began in response to disturbances of
density equilibrium around the margins of an expanding coastal population?
Such a model cannot be supported by the archeological data. According to
Patterson (55), the oldest site to be occupied year-round on the Peruvian coast dates
to about 5000 B.C.; and this site was virtually unique because its "contemporaries

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 r.

0 > C'*
:3 U d) 0
w - d)
41) .-
0 u
C) 0. u 0
0 CA Cld
CIS
04
C43
U >-.

Cl* Cf$ C43 Cl*

d) 0 U u
cisus-4
u +1
C'*Co >
ri C13
0 Co
c- -cl Co C43
I.-, 0 44 0
0 O -t:l C) U U
C)
en u
C43 = >, z
1-1 C's cis

O 0 0 O 64 CD C13 > E C', >

o C) C) O 0 0 cr$ CO 0
C)C13 --
0 6. C-- 00=1
W) 0
d) C)

C'* Cl*
Cf$ co
co > Z
C
CO >+- XC)
CO 0 O
CO

C13 d) 4.1
4-1 4-1 > C,3
CO 0. =
r. I
E I-'al
,d)-
+., d)
t:l -
= Ml CO
4-1
64 > - 0 '-
0 0 C13
U 0
MlC,3 CO >
0 u C)
CO > 64
>, 4:1
- C13
.- C,3 A4
d) C13
o O > X
CO C,3 C'3
co
z -0 z o
CO
04
0

0 0 0
C., d " 1-4
1+4 - d) d)
0 CO P " $-4
.u
O CO CO O
CO c) c) O
0 CO
0 0;
d) rq >-. >,
W)v) u
C.) 0 > C,* co cq C14
0 C,3 6. 6 u 00
U
a4A
coCo
co co
O u
C-4co
0
>.. co 0 0 I.-, +,0 C,.
to Co 0 U 0 u U
Ij co co m co
co u cl E m co
co o > m u > >
co C)M >
0 'n >
m co
O 0 0
co m CD 0.)
0 U u en
0

co
+1 r3

r3

t3

t3
O& co t3 Co
r3 tzg
P,
Co

0
z t3 co
cA,

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 305

0
Cd Cd +1
.2
bo m 0
0 0. Cd
0
0
cd
Cd +--
0 an u
cd Cdtncn
Cd =
0
u u cq
0

U 0

C> cd
W) 0

Cd
0 C-14 u
,: 0 0 0
N t-4 N
C's Cd Cd Cid 0
Ml

4) 0 0 0 u r. Ml z
u 4
- z d)
4)
+-, Cf$ 4). CIS
0 0
0 0cd0 0 0
u
0 0
> > Cf$ 4)
C'S Cd C's Cd
0 0 +, a; -.4
A4 O Z Z O z O O.,O Cdla.

Ml t:l
> r.

03
0
cd
C)
cri
Cd
4)
C)
Cid
0 0 > >
u cd
> +-- I..
CO CO cn
CO
CO
0-4U
0
CIS
A4
0 0 06
> c)
bo 0 >
0 0 Cid CIS 0 0 CO
> CII U bo
4) A,
4) I
CdCd 1-4 6 Cd
0
= C'S c) U - d)
CO 0 0
A4
U
-U
1-14 = r.
4) 0 cl cr$
u +- t:l C
cq3 d) C-:
0 0 d) u C13 o C) cd CD cl CD
v) = cd 4) w . >-. W) cd 00 0 00 0
rq '06-4 064 -4 U - U CIS
cd

Cd

ta 0
0
w

b4D
.1ti 0
ZL
4-4 0

C'3 0* 0 d) 03
4-,

0 .:S C', 0
CIO Cf$ C's
ZL 'u 0 CO CN
C14

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
306 FLANNERY

who lived 50 miles to the north ... were unable to maintain year-round settlements
because of the spacing of seasonal food resource areas in their territory." Even at
3500 B.C., most coastal populations are believed to have moved from summer camps
near the sea to winter camps further inland.
The first striking increase in overall coastal population is said to have begun about
3000 B.C. (Patterson 55). Such a population increase is far too late to provide the
initial stimulus for agriculture. Indeed, long before the founding of the oldest known
year-round coastal community, the occupants of the highlands were growing beans
from the eastern Andes. Even the introduction of cotton, squash, beans, and chile
peppers to the coast did not initially break down the annual round from summer
to winter encampments. In fact, Lanning (39, p. 53) cites one community-the
Pampa site, dating to the fourth millenium B.C.-whose occupants "began as both
fishermen and [squash] farmers, then gradually gave up their farming and adopted
a diet consisting almost exclusively of sea food." Regardless of the priority of
sedentary or semisedentary life on the Peruvian coast, I would have to conclude at
this point that the forces leading to domestication must be sought in the highlands
and the Amazonian slope of the Andes.

EARLY EXCHANGES OF CROPS BETWEEN THE VARIOUS

CENTERS MENTIONED ABOVE

An important ethnobotanical question is, "How independent was early agriculture

in Asia, Africa, and the Americas? Could one area have influenced the others by
diffusion of early crops?"
All archeologists should read an article on this subject by Barbara Pickersgill &
A. H. Bunting (60). After dismissing earlier arguments for pre-Columbian maize in
Assam, peanuts in China, grain amaranths in the Himalayas, and so on, the authors
critically examine the botanical evidence for the four plants-cotton, bottle gourd,
coconut, and sweet potato-for which the strongest diffusionist case can be made.
They conclude:
1. The bottle gourd in the New World is far too ancient to have been carried there
from Africa or Asia by agricultural man. Bottle gourds are probably pantropical;
they float and are resistant to sea water and probably reached South America long
before man did.
2. The cultivated sweet potato (Ipomoea batatas) is a hexaploid which may have
arisen by hybridization between a diploid and a tetraploid ancestor. Two species
with both diploid and tetraploid forms, I. tiliacea and I gracilis, occur both in
Polynesia and the Americas. The hybridization leading to I batatas may therefore
have taken place independently in both areas.
3. Coconuts (Cocos nucifera) will survive 110 days floating in salt water, and
could also have reached South America from Polynesia without human agents.
4. Cotton genetics is still a complicated business which has not been fully re-
solved. But on the basis of all available new genetic data, Hutchinson (31) feels that
the "AA diploid" strains of cotton (which go into the makeup of "AADD tetra-
ploid" Gossypium or New World cotton) spread from Africa to America well before

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 307

the start of agriculture and "perhaps before the continents drifted apart." Pickersgill
and Bunting also cite Stephens' (71) correction of an earlier report which, like the
"manioc" from Tamaulipas, has been largely overlooked by archeologists: the "ear-
liest" cotton from Tehuacain (5800 B.C.) is from an area of disturbed stratigraphy
and presumably intrusive.
My overall conclusion is that Old World-New World diffusion of cultivated plants
in Pre-Columbian times was very limited, if it took place at all; and certainly it did
not "cause" agriculture to begin in either area.

SUMMARY AND CONCLUSIONS

I said at the outset of this review that I did not believe one model could explain the
origins of agriculture in all four regions discussed. Cultivation may have begun after
village life was already established in some parts of the Near East and Peru. In
Mesoamerica and Southeast Asia, cultivation probably began during a phase of
nomadic hunting and gathering; and in Mesoamerica, at least, nomadism continued
for thousands of years after farming began. What, if any, are the generalities which
could be offered for the origins of agriculture all over the world? I will offer a few
for one type of "paleotechnic" agriculture, Harris's "seed-crop cultivation."
Although hunters and gatherers know and classify hundreds of wild species in
their environment, they do not necessarily use them all. For example, the Kalahari
Bushmen (Lee 41, p. 35) can name some 223 species of animals, but they classify
only 54 species as edible and hunt only 17 of these on a regular basis. Of the 85 plant
species they recognize, 23 species provide them with 90 percent of their vegetable
foods. In other words, they divide food resources into "first choice," "second
choice," and "third choice" foods, turning to the latter only when they run out of
the former. On the basis of archeological data, virtually all the important seed-crop
cultivars were derived from species which were originally "third-choice" foods; they
were, among other things, more work to harvest and prepare, often requiring
grinding which Pleistocene hunters were ill equipped to do. Their wild ancestors
were used little, if at all, by hunters at 20,000 B.C. in the Near East, at 8000 B.C.
in Mexico or Peru, and so on.
But the ancestors of these seed crops had a number of characteristics which most
of the "first-choice" foods of the Pleistocene hunter did not have. They are mostly
annuals; they yield a high return (200-800 kg per hectare); they tolerate a wide range
of disturbed habitats; they store easily; and they are genetically plastic. Thus they
could be used to replace the "native" vegetation with a plant that in less than a year
after planting would cover a disturbed patch with a dense growth of storable food.
And as time went on, they responded with favorable genetic changes which made
them either more productive, easier to harvest, easier to prepare, or all three. If what
you want is a plant which will increase the carrying capacity of each hectare, and
which can be stored to last out the year, they are your obvious choices. The
disadvantages are that (a) farming may be more work than hunting, judging by the
available ethnographic data, and (b) an unstable man-modified ecosystem with a low
diversity index results. Since early farming represents a decision to work harder and

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
308 FLANNERY

to eat more "third-choice" food, I suspect that people did it because they felt they
had to, not because they wanted to. Why they felt they had to we may never know,
in spite of the fact that their decision reshaped all the rest of human history.
Let me close with Edgar Anderson's admonition: the agricultural revolution was
a process, not an event. To search for "the first domestic plant" is to search for an
event; it is poor strategy, it encourages bitter rivalry rather than cooperation, and
it is probably fruitless. We should search instead for the processes by which agricul-
ture began. To do that we need settlement pattern data; well-excavated living floors
with the plants left in situ; and samples of 100 specimens with a mean, standard
deviation, and range of variation. We need to maintain our enthusiasm, but to
temper it with skepticism, not only for our own efforts but for those of the scientists
in other disciplines with whom we collaborate.

Literature Cited

1. Anderson, E. 1956. Man as a maker of 13. Drennan, R. D. 1973. Experimental teo-

new plants and new plant communities. sinte farming in Oaxaca, 1972. Unpub-
Man's Role in Changing the Face of the lished manuscript. Univ. Michigan, Ann
Earth, ed. W. L. Thomas, 763-77. Univ. Arbor
Chicago Press 14. Flannery, K. V. 1968. Archeological sys-
2. Barghoorn, E. S., Wolf, M. K., Clisby, tems theory and early Mesoamerica.
K. H. 1954. Fossil maize from the Valley Anthropological Archeology in the
of Mexico. Bot. Mus. Leafl. Harvard Americas, ed. B. J. Meggers, 67-87.
Univ. 16:229-40 Washington, D.C.: Anthropol. Soc.
3. Beadle, G. W. 1971. Letter to Flan- Washington
nery 15. Flannery, K. V. 1969. Origins and eco-
4. Beadle, G. W. 1972. The mystery of logical effects of early domestication in
maize. Field Mus. Nat. Hist. Bull. Iran and the Near East. The Domestica-
43(10):2-11 tion and Exploitation of Plants and Ani-
5. Binford, L. R. 1968. Post-Pleistocene ad- mals, ed. P. J. Ucko, G. W. Dimbleby,
aptations. NewPerspectives in Archaeol- 73-100. London: Duckworth
ogy, ed. L. R. Binford, S. R. Binford, 16. Flannery, K. V. 1972. The origins of
313-41. Chicago: Aldine the village as a settlement type in
6. Braidwood, R. J., Howe, B. 1960. Prehis- Mesoamerica and the Near East: a
toric investigations in Iraqi Kurdistan. comparative study. Man, Settlement,
Studies in Ancient Oriental Civilization and Urbanism, ed. P. J. Ucko, R. Tring-
31. Oriental Inst., Univ. Chicago ham, G. W. Dimbleby, 23-53. London:
7. Callen, E. 0. 1967. The first New World Duckworth
cereal. Am. Antiq. 32(4):535-38 17. Flannery, K. V., Ford, R. I. 1972. A pro-
8. Chang, K. C. 1970. The beginnings of ductivity study of teosinte (Zea mexi-
agriculture in the Far East. Antiquity 44: cana), Nov. 22-25, 1971. Univ.
175-85 Michigan Mus. Anthropol. (Mimeo)
9. Coe, M. D., Flannery, K. V. 1967. Early 18. Ford, R. I. 1968. An ecological analysis
cultures and human ecology in south involving the population of San Juan
coastal Guatemala. Smithson. Contrib. Pueblo, New Mexico. PhD thesis. Univ.
Anthropol. 3 Michigan, Ann Arbor
10. Copeland, L., Wescomb, P. 1965, 1966. 19. Galinat, W. C. 1970. The cupule and its
Inventory of stone age sites in Lebanon. role in the origin and evolution of maize.
MWlanges Univ. St. Joseph, Beirut, Mass. Agr. Exp. Sta. Bull. 585:1-18
41-42 20. Galinat, W. C. 1971. The origin of maize.
11. Cowan, J. C. 1972. Personal communica- Ann. Rev. Genet. 5:447-78
tion to R. I. Ford 21. Galinat, W. C. 1972. Personal communi-
12. Cutler, H. C., Whitaker, T. W. 1967. cation
Cucurbits from the Tehuacain Caves. 22. Galinat, W. C., MacNeish, R. S. The an-
Prehistory of the Tehuacdn Valley, ed. cient maize of Ayacucho, Peru. Bot.
D. S. Byers, 212-19. Austin: Univ. Texas Mus. Leafl. Harvard Univ. In prepara-
Press tion

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
 THE ORIGINS OF AGRICULTURE 309

23. Gorman, C. F. 1969. Hoabinhian: a peb- 43. Lee, R. B., DeVore, I. 1968. Problems in
ble-tool complex with early plant associ- the study of hunters and gatherers. Man
ations in southeast Asia. Science the Hunter, ed. R. B. Lee, I. DeVore,
163:671-73 Chap. 1. Chicago: Aldine
24. Harlan, J. R. 1967. A wild wheat harvest 44. Lorenzo, J. L., Gonzailes Quintero, L.
in Turkey. Archaeology 20(3):197-201 1970. El mas antiguo teosinte. Boletin
25. Harlan, J. R., Zohary, D. 1966. Distribu- 42:41-43. Inst. Nac. Ant. Hist., Mexico,
tion of wild wheats and barley. Science D.F.
153:1075-80 45. Lowe, G. W. 1967. Discussion. In Al-
26. Harris, D. R. 1972. The origins of agri- tamira and Padre Piedra, Early Pre-clas-
culture in the tropics. Am. Sci. 60(2): sic Sites in Chiapas, Mexico, ed. D. F.
180-93 Green, G. W. Lowe. Pap. New World
27. Heiser, C. B. Jr. 1973. Seed to Civiliza- Archaeol. Found. 20:53-79
tion: The Story of Man's Food. San 46. MacNeish, R. S. 1958. Preliminary ar-
Francisco: Freeman chaeological investigations in the Sierra
28. Helbaek, H. 1960. The paleoethnobotany de Tamaulipas, Mexico. Trans. Am.
of the Near East and Europe. See Ref. 6, Phil. Soc. n.s. 48(6). Philadelphia
99-118 47. MacNeish, R. S. 1964. Ancient Meso-
29. Helbaek, H. 1969. Plant-collecting, dry- american civilization. Science 143:
farming, and irrigation agriculture in 531-37
prehistoric Deh Luran. See Ref. 30, Ap- 48. MacNeish, R. S. 1964. Personal commu-
pendix 1 nication
30. Hole, F., Flannery, K. V., Neely, J. A. 49. MacNeish, R. S., Nelken-Turner, A.,
1969. Prehistory and human ecology of Garcia Cook, A. 1970. Second Annual
the Deh Luran Plain. Memoirs 1. Univ. Report of the Ayacucho Archaeological-
Michigan Mus. Anthropol. Botanical Project. Andover: Peabody
31. Hutchinson, J. B. 1962. The history and Found.
relationships of the world's cottons. En- 50. Mangelsdorf, P. C. 1947. The origin and
deavour 21:5-15 evolution of maize. Advan. Genet. 1:
32. Jarman, H. N., Legge, A. J., Charles, J. 161-207
A. 1972. Retrieval of plant remains from 51. Mangelsdorf, P. C. 1973. Letter to the
archaeological sites by froth flotation. editor. Field Mus. Nat. Hist. Bull. 44:16
Papers in Economic Prehistory, ed. E. S. 52. Mangelsdorf, P. C., MacNeish, R. S.,
Higgs, 39-48. Cambridge Univ. Press Galinat, W. C. 1967. Prehistoric wild
33. Jones, V. E. 1970. Personal communica- and cultivated maize. See Ref. 12, 178-
tion 200
34. Kaplan, L. 1965. Archaeology and do- 53. Marks, A. 1971. Settlement patterns and
mestication in American Phaseolus intrasite variablity in the central Negev,
(beans). Econ. Bot. 19(4):358-68 Israel. Am. Anthropol. 73:1237-44
35. Kaplan, L. 1967. Archeological Phaseo- 54. Niklewski, J., van Zeist, W. 1970. A late
lus from Tehuacan. See Ref. 12, 201-11 quaternary pollen diagram from north-
36. Kaplan, L. 1972. Personal communica- western Syria. Acta Bot. Neer. 19(5):
tion 737-54
37. Kaplan, L., Lynch, T. F., Smith, C. E. Jr. 55. Patterson, T. C. 1971. Central Peru: its
1973. Early cultivated beans (Phaseolus population and economy. Archaeology
vulgaris) from an intermontane Peruvian 24(4):316-21
valley. Science 179:76-77 56. Peebles, J. 1972. Rice. Unpublished
38. Kirkby, A. V. T. 1973. The use of land manuscript. Univ. Michigan, Ann Arbor
and water resources in the past and 57. Perrot, J. 1966. Le gisement Natufien de
present Valley of Oaxaca, Mexico. Pre- Mallaha (Eynan), Israel. L 'Anthropolo-
history and Human Ecology of the Val- gie 70(5-6):437-84
ley of Oaxaca, Vol. 1. Memoirs 5. Univ. 58. Pickersgill, B. 1969. The archaeological
Michigan. Mus. Anthropol. record of chile peppers (Capsicum spp.)
39. Lanning, E. P. 1967. Peru Before the In- and the sequence of plant domestication
cas. Englewood Cliffs: Prentice-Hall in Peru. Am. Antiq. 34:54-61
40. Lathrap, D. 1970. The Upper Amazon. 59. Pickersgill, B. 1972. Personal communi-
New York: Praeger cation
41. Lee, R. B. 1968. What hunters do for a 60. Pickersgill, B., Bunting, A. H. 1969. Cul-
living, or, how to make out on scarce tivated plants and the Kon-Tiki theory.
resources. See Ref. 43, 30-48 Nature 222:225-27
42. Lee, R. B. 1971. Personal communica- 61. Renfrew, J. M. 1969. The archaeological
tion evidence for the domestication of plants:

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms
310 FLANNERY

methods and problems. See Ref. 15, 72. Stubbe, H. 1959. Considerations on the
149-72 genetical and evolutionary aspects of
62. Robson, J. R., Konlande, J. 1972. Un- some mutants of Hordeum, Glycine,
published studies. Dep. Nutr. Public Lycopersicon, and Antirrhinum. Cold
Health, Univ. Michigan, Ann Arbor Spring Harbor Symp. Quant. Biol. 24:
63. Schoenwetter, J. 1972. The pollen 31-40
records from Guila Naquitz cave, Oax- 73. van Loon, M. 1968. The Oriental Insti-
aca, Mexico. To appear in future volume tute excavations at Mureybit, Syria:
of Prehistory and Human Ecology of the preliminary report on the 1965 cam-
Valley of Oaxaca (see Ref. 38) paign. J. Near East. Stud. 27(4):265-
64. Sci. Am. (no author) 1973. 228(1):44-45. 90
See also Mangelsdorf (51) 74. van Zeist, W., Wright, H. E. Jr. 1963.
65. Smith, C. E. 1967. Plant remains. See Preliminary pollen studies at Lake Zeri-
Ref. 12, 220-55 bar, Zagros Mountains, southwestern
66. Smith, C. E. 1969. Carbonized plants Iran. Science 140:65-69
from San Jose Mogote, Oaxaca. To ap- 75. Whallon, R. W. Spatial analysis of occu-
pear in future volume of Prehistory and pation floors I: application of dimen-
Human Ecology of the Valley of Oaxaca sional analysis of variance. Am. Antiq.
(see Ref. 38) In press
67. Smith, C. E. 1973. Personal communica- 76. Whitaker, T. W. 1970. Personal commu-
tion nication
68. Smith, C. E., Kerr, T. 1968. Pre-con- 77. Whitaker, T. W., Cutler, H. C. 1971.
quest plant fibers from the Tehuacan Val- Pre-historic cucurbits from the Valley of
ley, Mexico. Econ. Bot. 22(4):354-58 Oaxaca. Econ. Bot. 25(2):123-27
69. Solecki, R. S. 1955. Shanidar cave, a 78. Wilkes, H. G. 1967. Teosinte: the closest
paleolithic site in northern Iraq. Ann. relative of maize. Cambridge: Bussey
Rep. Smithson. Inst. 1954 (Publ. 4190), Inst., Harvard Univ.
389-425 79. Wilkes, H. G. 1972. Maize and its wild
70. Solheim, W. G. 1972. An earlier agricul- relatives. Science 177:1071-77
tural revolution. Sci. Am. 226:34-41 80. Zohary, D. 1969. The progenitors of
71. Stephens, S. G. 1967. A cotton boll seg- wheat and barley in relation to domesti-
ment from Coxcatlan cave. See Ref. 12, cation and agricultural dispersal in the
256-60 Old World. See Ref. 15, 47-66

This content downloaded from 66.7.210.88 on Mon, 06 Feb 2017 03:11:22 UTC
All use subject to http://about.jstor.org/terms