27 October 1967, Volume 158, Number 3800 SCIENC:E

early nuclearity and retained it through-

out the sequence? Palerm and Wolf
presented a corollary hypothesis which
they felt should be checked by future
Farming Systems and [Link] investigators. They pointed out that all
the areas which were nuclear only in
Growth in Ancient Oaxa[Ca the early part of the sequence had pre-
dominantly slash-and-burn agriculture
of lowland (roza) type. All areas which
Physiographic features and water-control techniclues rose to prominence only late in the
sequence were arid regions demanding
contributed to the rise of Zapotec Indian civilizati i[on*. very efficient irrigation systems. The
five perennially nuclear regions are
lkby ~~ ones in which, today at least, virtually
Kent V. Flannery, Anne V. T. Kirkby, Michael J. Kir k'b
every farming technique known in
Aubrey W. Williams ; Jr. Mesoamerica is applicable. They hy-
pothesized that farming had begun in
Mesoamerica as slash-and-burn, and
that through time a series of new tech-
During the last 15 years an increas- american civilization vvas that of Pal- niques had been worked out: irrigation,
ing number of anthropologists and ge- erm and Wolf (4, pp. I1-37), who over flood-water farming, chinampas, and so
ographers have turned their attention a 10-year period in thLe 1950's sought on. As such technological innovations
to the pre-Hispanic civilizations of to find correlations between social sys- appeared, they were "applicable to an
Mexico and Guatemala. The evolution tems, agricultural systtems, and their ever-decreasing number of areas" (4,
of these ancient complex societies is of environmental settings. A major proc- p. 36). The areas in which the great-
general theoretical interest because it ess they observed at work in ancient est variety of techniques could be as-
seems to have taken place independent- Mesoamerica was the formation and similated remained nuclear; those in
ly of the early Old World civilizations. dissolution of "key areas" or "regional which only the older techniques could
Given the limitations of the archeologi- nuclei." These they deefined as "areas be applied gradually lost their influence
cal data, there has been considerable of massed power in both economic and assumed a marginal role.
latitude for varied and competing theo- and demographic terrms" (4, p. 9), None of Palerm and Wolf's "peren-
ries about the origins of the early which at various point,s in Mesoameri- nially nuclear" areas has ever been in-
New World states. can prehistory had atcted as "nodal vestigated with their hypothesis in mind,
Some authors have theorized that points of growth" or as nuclei for though a number of related theories
Mesoamerican civilization arose in the "symbiotic areas";these3 nuclei were in- have now been tested in the Valley of
arid highlands, because of the need for strumental in stimulati: ng cultural evo- Mexico (6). However, excavations in a
a strong centralized governmentto con- lution over wide geo'graphic regions few of Mesoamerica's"fringe"or "mar-
trol large-scale irrigation projects (1). (5). Each time Mesoanlerica moved up ginal" areas-ones which never became
Others have argued that civilization be- to a higher level of soc:ial and political nuclear-have been carried back to the
gan first in the humid tropical low- complexity, this move seems to have very beginnings of agriculture by Mac-
lands, where irrigation is not necessary been accompanied by a shift of power Neish (7, 8). One of these, in the Val-
(2). Still others have sought a middle and influence from o:ne area to an- ley of Tehuacan, Mexico, has now
ground between these positions, main- other. Some regions wcere nuclear only yielded the longest single stratified se-
taining that Mesoamerican civilization in the early (Formative) and middle quence in all of Mesoamerica. Publica-
began through the "intertwining of (Classic) periods; otherr areas were nu- tion of the Tehuacan sequence (9)
many regional strands," both highland clear only in the late (Post-Classic) places Mesoamerica in a better position
and lowland (3). periods. Only five reigions in Meso- than ever before for the testing of
One of the most intriguing hypoth- america were listed 1by Palerm and theories about the processes involved
eses of the evolution of early Meso- Wolf (4, p. 30) as having "maintained in the establishment of village life, and
their key importance from Archaic about the evolution of chiefdoms and
Dr. Flannery is an archeologist in the Office of times right up to the ti:me of the Span- early states (10).
Anthropology, Smithsonian Institution, Washing- ish Conquest."These arreas are the Val- In 1966 we selected the Valley of
ton, D.C.; Mrs. Anne Kirkby is a geographer in
the Centre of Latin American Studies, University ley of Mexico, the reg;ion of Cholula- Oaxaca (Fig. 2) as a natural laboratory
of Cambridge, England; Dr. Michael Kirkby is
a geomorphologist in the Department of Geog- Puebla, the Mixteca 1klta, the Valley in which to investigate a number of
raphy, University of Cambridge; Dr. Williams is of Oaxaca, and the re-gion of Guate- these hypotheses (11). Oaxaca was cho-
an ethnologist in the Department of Sociology and
Anthropology, University of Maryland, College
mala City (see Fig. 1). sen partly because the outlines of its
Park. Why had these five regions attained later prehistoric sequence (see Fig. 3)
27 OCTOBER 1967 445
 had been worked out by Caso and Ber-
nal (12), and partly because it lay close
enough to Tehuacan to be related to
it during all or part of its prehistory.
This eliminated many of the preliminary
steps that would have been necessary
were the area totally unknown, and
permitted us to concentrate from the
very beginning on problems of cultural
and ecological process. It also allowed
us to compare and contrast the agricul-
tural potential of a nuclear valley
(Oaxaca) and a marginal valley (Tehua-
can).

Location of the Valley of Oaxaca

The Valley of Oaxaca lies in the

southern highlands of Mexico, between
16?40'-17?20'N and 96015'-96055'W.
It is drained by two rivers: the upper
Rio Atoyac, which flows from north
Fig. 1. Outline map of Mexico and Guatemala, showing regions mentioned in the to south, and its tributary, the Rio
text. "Nuclear" areas listed by Palerm and Wolf (4) are indicated by squares or large Salado or Tlacolula, which flows west-
black circles; Tehaucan, a "marginal" area, is indicated by the small black dot. For ward to join the Atoyac near the pres-
the sake of brevity we have adhered to Palerm and Wolf's original classification, ent city of Oaxaca. The valley is shaped
although recent archeological data indicate that it needs to be revised and updated.
like a Y or three-pointed star, whose
center is Oaxaca City and whose south-
ern limit is defined by the Ayoquesco
gorge, where the Atoyac River leaves
the valley on its way to the Pacific
Ocean. The climate is semiarid, with
500 to 700 millimeters of annual rain-
fall, confined largely to the summer
months. The valley-floor elevation aver-
ages 1550 meters.
Situated in the mountainous central
part of the state of Oaxaca, the region
is surrounded by valleys with steep sides,
narrow floors, and perennially flowing
streams. In contrast, the Valley of Oax-
aca is a wide, open plain with abundant
flat land and streams which are dry
most of the year. Yet it was this valley,
where moisture is scarce and man must
devise means to control it, which be-
came the most powerful nuclear area
in the southern highlands. It is general-
ly believed that this development was
the work of the Zapotec Indians, who
now inhabit the valley and whose his-
tory can be traced back many thou-
sands of years in that region (13, p.
788; 14).

Physiography and Vegetation

A typical cross section of the valley

(Fig. 4) shows four distinct physio-
graphic zones: (i) the "low alluvium,"
or present river flood plain; (ii) a zone
Fig. 2. Map of the upper Atoyac River drainage basin, indicating major towns and of "high alluvium," which is mainly
archeological landmarks in the Valley of Oaxaca. The rectangle formed by dashed lines de- an abandoned flood plain of Pleisto-
limits the survey area shown in Fig. 5. Line A-B locates the cross section shown in Fig. 4.
446 SCIENCE, VOL. 158
cene-to-Recent age, formed by the Ato- piedmont is still one of the most com- tural techniques worked out by the
yac River and its tributarieswhen they plex vegetation zones, with varying Zapotec Indian inhabitants of the val-
flowed at a higher elevation; (iii) a communities of tree legumes, prickly ley gave them an early advantage over
piedmont zone flanking the high allu- pear (Opuntia), organ cactus (Lemai- their neighbors. Finally, there are vari-
vium; and (iv) the surrounding moun- reocereus), maguey (Agave), Dodo- ous additional factors, only indirectly
tains. naea, and-at elevations of 1800 me- related to agriculture,which contributed
The river channel is incised no more ters and above-scattered oaks (Quer- to the rise of stratified societies in
than 1 to 2 meters into its present cus spp.). The high mountains have Oaxaca.
floodplain, which is only locally pres- forests of oak, pine, and manzanita In the sections which follow we out-
ent and nowhere more than 600 me- (Arctostaphylos). It is to be hoped line some of the environmentalfeatures
ters wide. The main part of the flat that future work in Oaxaca will greatly of the Valley of Oaxaca which may
valley floor, which varies from 1 to enrich this tentative and oversimplified be considered advantageous. We then
15 kilometers in width, is formed by reconstruction. attempt to show their relevance to the
the high alluvium; this zone is sepa- periods of food-collecting and incipient
rated from the low alluvium by a 1- cultivation (8000 to 1500 B.C.), early
to 3-meter rise in elevation. Between Agricultural Potential village farming (1500 to 600 B.C.),
the high alluvium and the mountains and the rise of towns and ceremonial
lies the piedmont zone, where the We feel there are several environ- centers (600 to 200 B.C.).
land has a slope of 1 to 2 degrees mental aspects of the Valley of Oaxaca The Valley of Oaxaca has 700
and has been dissected by tributary which make it a better place for agri- square kilometers of relatively flat land,
streams to form low rounded spurs culture-and especially for primitive, the largest such expanse in the Mexi-
and isolated hills with up to 30 meters pre-Hispanic types of agriculture-than can highlands south of Cholula-Pueb-
of relief. The piedmont was originally many adjacent parts of highland Mex- la. Until recently, it was generally be-
formed as a series of coalescing alluvial ico, including some of the other areas lieved that the valley was the bed of
fans, and remnants of this deposited described by Palerm and Wolf as nu- a former lake, which had dried up
material remain as fan gravels of clear. Second, some of the agricul- prior to 600 B.C. This lake was men-
probably Pleistocene age. Later stream
dissection has exposed underlying rocks
which are pre-Jurassicto Miocene. The
DATES TEHUACAN VALLEY OF CENTRAL OLMEC
piedmont zone grades eventually into VALLEY OAXACA CHIAPAS REGION
the true mountain zone, where valleys
have up to 1000 meters of relief and
A.D. 1500 Monte
slopes are steep. The mountains are 1400 Alban V
formed mainly of pre-Jurassic meta- 1300 Venta
1200 Salada
morphic rocks, Cretaceous limestones, 1100 Monte
and Miocene ignimbrite tuffs (15). The 1000 Alban IV
tuffs are most extensive in the extreme 900
800
eastern end of the valley, between Mit- 700
la and Tlacolula, where they abound 600 Monte _
in small caves and rock-shelters which 500 Palo Alban III Laguna
400 Blanco UpperTres
were occupied during the earliest pe- 300 Jiquipilas Zapotes
riods of Oaxaca prehistory. In the ex- 200 ____
100 Monte Istmo
treme western part of the valley, where 0 Alban II Horcones MiddleTres
rocks of the basal metamorphic com- B.C. 100 Zapotes
200 __
plex are most widely exposed, there are 300 Monte Guanacaste LowerTres
deposits of magnetite and mica which 400 Santa Alban I Francesa Zapotes
were used and traded by the later occu- 500 Maria
600 _ Escalera
pants of the valley as exotic raw mate- 700 Guadalupe Dili La Venta
rials. 800
900
Originally, each of these physiogra- 1000 San Jose Cotorra San Lorenzo
phic provinces would have had its own 1100 Ajalpan
distinct vegetational cover. Today, af- 1200
1300
ter thousands of years of intensive cul- 1400
tivation, so little remains of the original 1500
2000 Purr6n
valley-floor vegetation that it can only 3000 Abejas
be hypothetically reconstructed from
4000
pollen grains and carbonized seeds in 5000 Coxcatlin Coxcatlin
archeological sites of that zone (16).
The present Atoyac floodplainmay have
6000
had phreatophytic species like bald cy- 7000 El Riego Guilt Santa
press (Taxodium), willow (Salix), and 8000 Naquitz Marta
wild fig (Ficus), while the high alluvium
was probably characterized by a more
Fig. 3. Pre-Hispanic cultural periods in the Valley of Oaxaca, compared with those of
open cover of grasses and woody le- adjacent regions, as determined by radiocarbon dating and interregional similarities
gumes like mesquite (Prosopis). The in artifacts.
27 OCTOBER1967 447
tioned in Zapotec legends, and casual dislocations have occurred in the val- vegetation cover, clearing of land does
inspection of the valley revealed sup- ley. Thus three factors-aridity, down- not appreciably increase erosion. The
posed "shorelines" or old "lake ter- faulting, and a low rate of stream deg- gentle slopes of the valley floor further
races," as well as seasonally inundated radation-all may have contributed to discourage erosion, and the presence of
areas which were reputed to be rem- the alluvial expanses which make the an extensive flat area lessens the incen-
nants of the lake (13, 14). In 1960, Valley of Oaxaca unique among its tive to clear land higher in the moun-
Lorenzo (17) presented geological evi- neighbors. tains. All these factors combine to keep
dence to the contrary. Similarly, our soil erosion to a minimum in the Val-
study reveals no evidence of a lake. ley of Oaxaca. It is perhaps worth not-
The supposed lake "shoreline"is not Less Severe Soil Erosion ing that the most badly eroded valleys
horizontal but varies in elevation by of the Mixteca, like Yanhuitlan, occur
over 200 meters within a distance of In the mid-1940's Cook studied a at elevations of 2000 meters in the
some 40 kilometers. What it really con- number of the higher valleys just to pine-oak zone, where the ratio of ac-
sists of is the break in slope formed the north of the Valley of Oaxaca- celerationon cleared land is highest.
where the steeper fan gravels meet the Tamazulapan,Yanhuitlan, and Nochist-
valley alluvium. The fan gravels are lan, in the Mixteca Alta. All these val-
clearly fluvial in origin, and no lacus- leys have suffered extremely destructive Soils and Water Table
trine deposits or fossils have been soil erosion, which Cook traced to agri-
found, so the hypothesis of a perma- cultural activities extending far back The flat valley floor and the thick
n,ent lake in the valley during the last into the pre-Hispanicera (18). alluvial deposits offer clear advantages
10,000 years must be rejected. The By comparison, soil erosion is not as a site for early agriculturalists,but
seasonally flooded localities reputed to a severe problem in the Valley of these factors are partially offset by the
be "lake remnants" are in reality low- Oaxaca, although there is some local relative aridity, which limits both avail-
lying areas where the water table is gullying of hillslopes. One reason is able water and soil fertility. Soil pro-
close to the ground surface. Many of that, relatively speaking, little land has files are poorly developed, and the allu-
these occur in clay areas of low so far been cleared in the higher moun- vial structure is retained almost unal-
permeability, which further tend to tains. In the Valley of Oaxaca, acceler- tered below the A horizon. Prismatic
maintain standing bodies of water for ated erosion due to clearing of the structure and some salt accumulation
long periods. natural vegetation for agriculture is is found in the B horizon. Most valley-
With the rejection of the lake hypoth- most severe not only on the steepest floor soils belong to the Brown Soils
esis, other explanations should be sug- slopes but also where the vegetation is group, but they tend toward Gray Des-
gested for the Valley of Oaxaca's un- densest. This is strikinglyshown tby our ert soils in the most arid areas, such
usually wide and flat floor. It is known erosion measurements on steep slopes as the Mitla end of the valley.
that relatively arid climates favor allu- near Mitla, made in 1966 over a period Humus and nutrient concentrations
vial-fan deposition, and this tendency of 2 months during which 237 milli- in the soils are so low that it is the
toward alluviation in combination with meters of rain fell. Table 1 gives the differences in water availability which
the inability of the upper Rio Atoyac ratios of the rate of erosion on cleared constitute the most important determi-
to downcut (because of the high re- land and the rate of erosion on land nant of the usefulness of the soils for
sistance of rocks in the Ayoquesco with natural vegetation. man. The finest grained soils with the
gorge) could explain the great width It would appear that agricultural best water-holding characteristicsoccur
of the valley. Furthermore, this exten- land clearance in the oak-pine forest on the high alluvium, in a band run-
sive deposition could have been initiated zone results in an accelerated erosion ning parallel to the river, but at dis-
by downfaulting, for local deformation of 45 times the natural rate, whereas tances of 500 to 1500 meters away
of Miocene (but not later) sediments in the cactus-scrub zone, where slopes from it. Except where the high alluvium
shows that some late-Miocene/Pliocene are already in equilibriumwith a sparse is more than 2 kilometers wide, this

2>
2:
j
ALLUVIUM PIEDMONT
M:
MOUNTAINS PIEMONTHIGH PIEDMONTZONE MOUNTAINS
ZONE 3 SPUR <I

A -J I-
ENE
B

Son AgustinEtla /
Fa~ngrave/s (modernvillage)

,X,, \\ \ \ ,,31 Gne/ss

Miocene sediments n . /

I Km' x t
L - (VerticalexaggerationIOx)

Fig. 4. Cross section of the northwesternValley of Oaxacanear Etla, showing major physiographicareas discussed in text.
448 SCIENCE, VOL. 158
band of fine-grainedsoils extends to the Table 1. Comparison of erosion ratios in three may receive almost 1000 millimeters an-
outer edge of the alluvium, where it different vegetation zones near Mitla, in the
Valley of Oaxaca. [Erosion ratio = C/N,
nually. Open-waterevaporation depends
meets the piedmont zone. where C = erosion rate on cleared land and principally on temperature, and de-
Soil-grain size influences water reten- N = erosion rate on land with undisturbed creases with elevation. On the valley
tion and is thus important both for natural vegetation. Both C and N were cal- floor it averages 2000 millimeters an-
culated, for slopes of comparable steepness
(i) dry farming and (ii) commercial during the same rainy season (1966), by nually (three to five times the precipita-
crops with high irrigationrequirements. measuring the percentage of surface material tion), while at 3000-meter elevations it
In between these two extremes, in cases moved downslope from a previously estab-
lished 50-meter line.] is only 340 millimeters (one-third the
where more limited types of irrigation precipitation).Hence the growth of per-
are practiced, the depth to water table Ratio for manent pine forest on the high moun-
rate of
and the yields from wells are more erosion: tains, and the sparser cactus and mes-
importantthan soil texture in determin- Erosin Eleva- Natural cleared quite-grassland cover of the lower
tion vegeta- land/land
ing the value of the soil for farming. rosit
(meters) tion with slopes.
This brings to mind recent comments natural For the growth of annual crops with-
by Stevens (19) on Indian farming in vegetation
cover
out irrigation, the ratio of rainfall to
general: it is not necessarily the best open-water evaporation must remain
soils which are the most intensively No. 1 1750 Cactus- 1.0 close to 1.0 throughout the summer
used, because factors of technology and scrub months, with June to August the most
water table may be the primary ones. No. 2 2000 Oak- 3.8 critical period. On the floor of the Val-
Within the present flood plain of Dodonaea
ley of Oaxaca this ratio ranges from
the Atoyac River, well water is within No. 3 2300 Pine-oak 45.4 0.50 near Tlacolula to 0.93 near Ocot-
3 meters of the surface; within the lan. This range may be contrasted with
zone of high alluvium, it lies between similar figures obtained for the floor
2 and 10 meters down. In both these existent, and the present-day Zapotec of the Valley of Tehuacan, which vary
zones, water yields are usually adequate of this area cultivate sugarcane, which between 0.45 and 0.65 (21). Thus con-
for small-scale irrigation of the special-requires an 18-month frost-free period. ditions in the western part of the Valley
ized types described below. In the fan These favorable conditions deterio- of Oaxaca are somewhat more favor-
gravels of the piedmont zone, water is rate rapidly as one ascends the hills able for dry farming than are those in
generally more than 10 meters below to either side of the Valley of Oaxaca. Tehuacan, a fact which was probably
the surface, and well yields are only Above altitudes of about 2300 meters, important during the early stages of
sufficient for immediate household summer temperatures are low enough agriculture.
needs. to inhibit cultivation of maize, and Even more striking contrasts between
wheat is at present a more reliable Oaxaca and Tehuacan may be seen,
crop. At elevations of 3000 meters, however, when one examines their ir-
A Relatively Frost-free Climate mean daily minima in January are rigation potential. The Tehuacan Val-
about 0?C, and the dominant cultivar ley has an extremely low water table
At elevations of 2000 to 2800 meters, is the potato. (about 20 meters), but the valley lies
in areas like the Valley of Mexico or The two ancient indigenous races of just to the south and east of a block
Cholula-Puebla,winter frosts may have maize known so far for the Early For- of limestone-travertinemountains which
been a real deterrentto year-round cul- mative period in Mesoamerica-Nal- constitute a major aquifer. Very large
tivation of maize until frost-resistant Tel and Chapalote-do poorly in cold quantities of subsurface water emerge
strains were developed, some time after conditions and are sensitive to highland from springs at the base of this range,
the Middle Formative period. For ex- rusts (20). The essentially frost-free na- near the western outskirts of the city
ample, Sanders' figures for the Valley ture of the southern Valley of Oaxaca of Tehuacan (22). Thus, shallow-well
of Mexico (6, pp. 20, 23) indicate that probably gave it considerableadvantage irrigationis impossible;much more fea-
between October and February the area over the higher valleys of the nearby sible is a large-scale canal-irrigation
may have temperatureswhich are detri- Mixteca at this early period (1500 to system to carry water from the pro-
mental to maize. In contrast, tempera- 900 B.C.), when only those primitive lific springs out to the central and
tures on the floor of the Valley of races of maize were known. In later southern parts of the valley. Such a
Oaxaca are well suited to year-round periods, with frost-resistant strains of canal system was indeed developed at
growing of maize, even the primitive maize, this difference was probably less Tehuacarnduring later periods of its
strainsof the Early Formative. crucial, as high population densities in prehistory, when the population of the
At the level of the valley floor (1420 the Mixteca and the Valley of Mexico valley was already high (8, 9, 23).
to 1740 meters, with an average of suggest. Canal irrigation on a large scale is
about 1550), the mean annual tempera- nowhere practical in the Valley of Oax-
ture is 20?C, with an annual range of aca, where springs are small and sur-
6?C and a daily range of 15?C. Ex- Precipitation and Hydrology face flows are insufficient for irrigating
treme minimum temperaturesover a re- more than a small area. However, be-
cent 12-year period are close to 0?C. Mean annual rainfall on the floor cause of the unusually high water table,
In any one year there is only slight of the Valley of Oaxaca varies from shallow-well irrigation is widely prac-
probability of frost, and this largely 490 millimeters at Tlacolula to 740 ticed, and this technique, which requires
in the higher parts of the valley. In millimeters at Ocotlan. There is a gen- relatively little effort and can be per-
the main Atoyac River floodplain south eral rise in precipitation with increas- formed on an individual family basis,
of Oaxaca City, all of which lies below ing altitude, so that the surrounding can be traced back to at least 700 B.C.
1550 meters, frosts are virtually non- mountains at elevations of 3000 meters and probably earlier.
27 OCTOBER 1967 449
Food-Collecting,. Incipient-Cultivation ly effective experiments with the grow- vium, which narrows to 500 meters
Periods: 8000 to 1500 B.C. ing of maize, beans, and squash. near Etla and expands to 2 kilometers
Not only is it impossible to speak in the broad plain just south of Oaxaca
The oldest archeologicalmaterialsre- of "key" or "nuclear" regions at this City, the Zapotec practice a kind of
covered in 1966 came from a series time, it is also virtually impossible to rudimentary water control known as
of caves and rock shelters near Mitla. define individual "culture areas" within riego a brazo or "pot-irrigation"(Fig.
These shelters occur in volcanic-tuff the southern highlands. This suggests 7). This technique was described in
cliff faces at elevations of 1900 meters, that, while cultures were still primarily 1960 by Lorenzo (17), and we have
near the transition from the piedmont food-collecting, the individualpeculiari- since studied it in detail in the muni-
to the higher mountains. ties of the various valleys were not cipio of Zaachila.
For years it had been known that especially significant. It was full-time "Pot-irrigation"involves the digging
this elevated region, 200 meters or agriculturewhich brought about special- of a series of shallow wells right in
more above the valley floor, was richer ized adaptations to local peculiarities the cornfield, tapping the stratum of
in surface finds of the food-collecting of soil, rainfall, and water table and water which lies between 1.5 and 3.0
era than any other; when a recent lake gave each valley its regional character. meters from the surface. An acre of
on the valley floor was still considered At this point, even slight differences in land may have ten of these small wells,
a possibility, the hypothetical lake was agriculturalpotential may have started which are filled in during the plowing
often used to explain the restriction of certain valleys, like Oaxaca, on the path season and then reopened when water
these early cultures to the upper pied- to nuclearity. is needed. Water is drawn up from
mont (24). The real reason why this each well in a 3-gallon pot and poured
zone was so consistently used in early gently around the individual corn
times is that it has the richest and Early Village Farming Period: plants. By means of this system, farm-
most varied assemblage of edible wild 1500 to 600 B.C. ers within the 3-meter water-tablezone
plants of the entire region. For the often achieve three harvests a year. At
most part, shelters immediately over- Several parts of the valley were se- any time of the year, dry season or
looking the valley floor were used infre- lected as "pilot areas" in which to sur- wet, this belt of pot-irrigated alluvium
quently or not at all before 1500 B.C.; vey for early village farming communi- resembles a huge patchwork of small
it was full-time agriculture, with its ties, with subsequent test excavations. but highly productive gardens. Riego a
need for flat land and fine-grained soil, Chief among these was a 10-kilometer brazo requires no large labor force or
which eventually diverted attention strip in the extreme northwest corner centralized control; it is carried out on
from the upper piedmont and allowed of the valley, near Etla. We concen- an individual-household basis. How-
the high alluvium to emerge as the trated on the Early Formative San ever, the zone where this technique can
zone of major utilization. Jose phase (1200 to 900 B.C.) and the be used constitutes a very small per-
Between 7840 and 6910 B.C. (as esti- Middle Formative Guadalupe (900 to centage of the valley-floor area in
mated on the basis of radiocarbon de- 600 B.C.) and Monte Alb'an I (600 to Oaxaca, and, as mentioned above, it
terminations), the Indians who camped 200 B.C.) phases (Fig. 3). cannot be used at all in low-water-table
seasonally in Guila Naquitz Cave col- In the Etla region, the most favor- areas like the Valley of Tehuacan.
lected acorns, pinyon nuts, mesquite able agricultural land is that part of The association of San Jose and
beans, prickly pear and organ-cactus the high alluvium where the water table Guadalupephase villages with this zone
fruits, wild onion bulbs, hackberry, is within 3 meters of the surface. As of pot-irrigation was very suggestive,
maguey (Agave sp.), nanche (Malpighia shown in Fig. 5, most Formative sites but until recently no actual well to dem-
sp.), susi (Jatropha sp.), and a dozen thus far located (including all the Early onstrate the existence of the technique
other species, all of which were pre- Formative sites) are concentrated in or in the earlier part of the Formative
served within the cave by dessication adjacent to this zone. In the narrow had been found. In August of 1966,
(25). Toward the end of this period, parts of the valley, prime localities were Richard Orlandini and James Schoen-
small black beans (Phaseolus sp.) and the tips of piedmont spurs, which raised wetter of the Oaxaca Project discovered
squash seeds (Cucurbita sp.) appear in the villages just high enough above the a Formative well which had been ex-
the refuse; thus Guila Naquitz is added alluvium so they would not flood in posed by adobe-brickmakersin a bank
to the list of sites known to belong to the rainy season. In wider parts of the some 50 meters back from the river
the "incipient cultivation"period in an- valley, where the piedmont spurs are at Mitla (Fig. 7). Associated pottery
cient Mexico (26). too far from the 3-meter water-table dated the well to the Guadalupe phase,
A nearby cave, Cueva Blanca, dated belt, villages were built on the high allu- considerablystrengtheningour evidence
at about 3295 B.C. by the radiocarbon vium in areas where sandy soils pro- for water control in the early village
technique, yielded a later food-collect- vided them with the best-drainedloca- farming period.
ing, incipient-cultivationhorizon which tions available (see Fig. 6). Our pre- Agriculture within this high-water-
is in most respects identical to the Cox- liminary surveys in other parts of the table zone supported villages of large
catlan phase (5000 to 3000 B.C.) de- valley suggest that the pattern observed size and material wealth. The best-
fined by MacNeish at Tehuacan (9). at Etla is probably typical of the ear- known site of the San Jose and Guada-
It would appear that at this period the lier part of the Formative. In areas lupe phases is San Jose Mogote, which
whole of the southern Mexican high- of low water table, such as north and we tested in 1966 (see Fig. 5). Here
lands was occupied by a series of re- east of Tlacolula, evidence of Early Early Formative artifacts can be picked
lated, seminomadic bands who moved Formative occupation is correspond- up over 40 acres of a piedmont spur
seasonally from resource area to re- ingly sparser. surroundedon three sides by alluvium.
source area and engaged in increasing- In this belt of high-water-tableallu- Rows of post molds and burned wall
450 SCIENCE, VOL. 158
fragments suggest that houses were
large and rectangular, with partial stone
foundations and wattle-and-daub walls
which were plastered with mud and
whitewashed. Besides the usual internal
features, like hearths and bell-shaped
sub-floor cooking pits, one San Jose
phase house had a recessed circular
area a meter and a half in diameter,
which had been plastered and painted
red. Around this circle were scattered
fragments of figurines, exotically deco-
rated pottery, fragments of black and
white mica, raw chunks and small pol-
ished mirrors of magnetite, and orna-
ments and discarded fragments of im-
ported marine shell (27). In levels be-
longing to the early Guadalupe phase,
such scatters gave way to an artificial
platform of earthen fill with stone re-
taining walls oriented almost due north-
south, and presumably having had a
ceremonial function. Such orientations
characterize later ceremonial structures
in the valley as well (28).
The evidence of long-distance trade
in the San Jose phase (which is lacking
in earlier periods) reflects two things:
an increasing interest in status differ-
entiation (with artifacts of imported
materials serving as insignia of status) Fig. 5. "Pilot" survey area in the northwestern part of the Oaxaca Valley, showing
and formalized contacts with other In- the distribution of Formative archeological sites with regard
--r to nhvsiogranhic
r areas
dian groups in differing environmental and water resources (see text).
zones of Mesoamerica. Marine pearl
oyster and Spondylus shell were im- irrigation in that part of the high allu- Moreover, the actual area irrigated is
ported from the Pacific, while Neritina vium, came the first sizable spread of relatively small, and it is the communi-
and pearly freshwater mussels came settlement up the more permanent trib- ties upstream that get most of the wa-
from the Gulf Coast. Anomalocardia utaries of the Atoyac into the piedmont ter. For this reason, villages both in
subrugosa, a mollusk eaten by Forma- (Fig. 5). These latter sites are of two the piedmont and in the pot-irrigation
tive villagers in the estuary zone of the types: "habitation" sites on the first zone augment their water-control farm-
distant Chiapas-Guatemala coast (29), terrace of the stream or a low ridge ing by cultivating the nearby hillsides.
was also imported. near it, and "ceremonial" centers on The technique used is simple dry farm-
Most important are the chunks and hilltops nearby. ing with fallowing, called variously
mirrors of magnetite, a raw material We doubt that this pattern of settle- tlacolol or barbecho in different parts of
native to the Valley of Oaxaca, for ment was random. Most sites outside Mexico (4), and it profits from the
nodules of this metal are known to the high alluvium at this period are on low erosion rate of the lower piedmont
have been polished into concave mir- perennial streams, and, like the present zone. Such an agricultural pattern, com-
rors and buried in ceremonial caches villages of the piedmont zone, they are bining an intensively cultivated (often
by the "Olmec" peoples of the south- located not downstream, at the point irrigated) core area with a less inten-
ern Gulf Coast (30). At present, the where most water is available, but up- sively cultivated hinterland, has been
Valley of Oaxaca must be considered stream, where the water can be most called the "infield-outfield" system (31).
a possible source for the Olmec mag- effectively diverted for irrigation. To- The piedmont areas into which these
netite. day, these villages divert the water into later Formative farmers expanded have
canals which follow the natural con- been cultivated for so long that traces
tour of one of the piedmont spurs of early irrigation canals are virtually
Rise of Towns and Ceremonial downstream until they come to the eradicated. They remain only in in-
Centers: 600 to 200 B.C. crest of the spur. Here the village and stances where the water used for irri-
the "master canal" are located, and gation was so rich in dissolved traver-
During the later stages of the Middle water is distributed to fields on both tine that the canals themselves have ac-
Formative period, villages within the sides of the spur. tually been "fossilized" through deposi-
3-meter water-table zone increased in This technique of small-scale canal tion of this calcareous material (Fig. 8).
size and number. Coupled with this irrigation is only feasible along the up- In 1966, James Neely of the Oaxaca
population increase, which we attribute per edges of the piedmont zone, where Project investigated one such area in
to the success of dry farming and pot- streams have good perennial flows. the mountain zone near Mitla.
27 OCTOBER 1967
451
 This site, called Hierve el Agua (Fig. gote grew to more than 100 acres, and gions swung quickly away from pre-
8), is a complex of "fossilized" ancient in the process it differentiated internally vious traditions and featured, during
irrigation canals covering a square kilo- into ceremonial and secular precincts, this and subsequent periods, provincial
meter of hillside below a spring par- cemetery areas, and probably precincts imitations of the Valley of Oaxaca
ticularly rich in travertine. A series of of craft specialization as well. Similar styles (9).
dry-laid stone terraces had been irri- developments took place throughout the While much of the surrounding area
gated by means of small canals which valley, where there were now more than may willingly have entered into a sym-
carried the water down to the fields 30 ceremonial mound groups in opera- biotic relationship with the Valley of
and along the tops of the terrace walls. tion (13, p. 797). The most impressive Oaxaca for the economic advantages
Neely's 40 test pits dug into these ter- of these was the mountaintop elite it offered-such as a ready market for
races reveal an occupation beginning center of Monte Alban, which (although their surplus and their locally special-
before 300 B.C. and expanding through still in its initial building stages) already ized products-there are hints that not
all subsequent periods of Oaxaca pre- had, according to archeological evi- all the marginal valleys joined peace-
history. It is probably no accident that dence, monumental construction, bas- fully. Caso (32) believes that at least
this evidence of small-scale canal irri- relief carving, a stela-altar complex, ca- one set of glyphs carved in stone at
gation begins during the first sizable lendrics, and hieroglyphic writing (13, Monte Alban represent conquered
expansion out of the 3-meter water- p. 788; 14). neighboring towns, and Coe (33) has
table zone and up the perennial tribu- By now, this area of "massed pow- recently suggested that an even earlier
taries. er" had begun to extend its influence series of bas-reliefs, the so-called dan-
With at least four agricultural sys- into the surrounding valleys, bringing zantes of Monte Alban I, depict slain
tems operating-dry farming and pot- them rapidly into its sphere. This in- and mutilated captives. While these war-
irrigation in the high-water-table zone, fluence can be seen over an area of like interpretations remain to be proved,
canal irrigation and hillside fallowing tens of thousands of square kilometers, they are in no way inconsistent with
systems in the piedmont-the Valley of from the Pacific Coast to the Tehuacan what is known ethnographically of
Oaxaca reached another plateau on its Valley. In fact, techniques of pottery groups of a chiefdom stage of organiza-
climb toward civilization. San Jose Mo- design in Tehuacan and adjacent re- tion (10).

452 SCIENCE, VOL. 158

Evaluation of Hypotheses high alluvium where only dry farming which are as varied as the techniques
or flood-water farming is possible; by themselves. In Oaxaca, canal irrigation
With these data in mind, we can A.D. 1300 the area farmed included not of the Hierve el Agua type is applicable
now tentatively evaluate the relevance only the entire valley floor but also the to a very tiny portion of the valley,
of various developmental hypotheses to lower slopes of the mountains, which and it is no more productive (in terms
what happened in the Valley of Oaxaca. were frequently terraced. The terraces of labor input relative to crop yield)
First of all, it no longer seems likely apparently depended on rainfall and on than pot-irrigation. Moreover, many
that slash-and-burn was the sole early- the lower evaporation rate of the north- canal-irrigating villages are so high in
farming technique in Mesoamerica; in and east-facing slopes. These data sug- the piedmont that only summer crops
fact, given the locations of the cave gest that the assimilation of new farm- can be grown. What canal irrigation
areas where farming began, such as the ing techniques was also a way of did do was to open up an additional
moist barrancas of the arid Tehuacan bringing into cultivation more and more niche within the valley which had not
Valley, it is even possible that some of the previously unproductive physi- previously been agriculturally produc-
kinds of water control (like the terrac- ographic units of the valley: the greater tive.
ing of wet arroyos) are as old as agri- 'the number of systems used, the greater Perhaps most significantly, in canal-
culture itself (34). However, our work the acreage producing at top capacity. irrigation communities there tends to
in Oaxaca strongly supports Palerm Our evidence for early irrigation in be less equitable distribution of land
and Wolf's view that the ability to as- the Valley of Oaxaca will undoubtedly and property rights than in pot-irrigat-
similate new agricultural techniques please advocates of the "hydraulic the- ing communities. As past writers have
through time is a key factor in retain- ory" of state formation (1). However, observed, irrigated land is "improved"
ing nuclearity. we see no evidence in Oaxaca for irri- land; it represents an investment which
Furthermore, the four systems we gation systems so large that they would makes it a more scarce and competed-
postulate for the Monte Albain I pe- necessitate a strong centralized au- for resource than it previously was,
riod did not represent the final stage thority. In fact, we fear that the hy- and leads to problems of inheritance
of Zapotec agriculture. In later periods draulic theory-at least, in its purest of property and differential access to
(A.D. 100 to 900), large mound groups form-may at times obscure the real good land. If a developing society has
and habitation sites cover areas of the effects of irrigation in ancient society, tendencies toward status differentiation

Fig. 6 (opposite page, left). Archeological sites in the Valley

of Oaxaca, showing contrasts in land-use patterns. (Top) Cueva
Blanca, a cave near the transition from the mountains to the
upper piedmont. This zone, which is particularly rich in edible
wild plants, was extensively used in the food-gathering era.
(Bottom) San Jose Mogote, a site covering the tip of a low
piedmont spur next to the 3-meter water-table zone of the
Atoyac River alluvium. The latter was the zone of most in-
tensive utilization during the early village farming era. (The
artificial mounds seen just above and to the right of the present-
day village indicate the center of the site, which fills the upper
two-thirds of the photograph.)

Fig. 7 (opposite page, right). Primitive water control in the

Valley of Oaxaca, new and old. (Top) Zapotec farmer engaged
in "pot-irrigation"in the high-water-table zone near Zimatlan.
Behind him is the shallow well from which the water has been
drawn. (Bottom) Prehistoric well belonging to the Guadalupe
phase (900 to 600 B.C.), exposed in a high bank near the
river at Mitla. [Courtesy, R. Orlandini and J. Schoenwetter]

Fig. 8 (right). Prehistoric irrigation canals "fossilized" by traver-

tine deposition, showing contrasts between regions at different
periods. (Top) Small-scale canals used to irrigate terraced hill-
side at Hierve el Agua, near Mitla, Oaxaca [Photo by Chris L.
Moser.] Irrigation systems of this type date back to before
300 B.C., clearly antedating the origins of the state. (Bottom)
Massive irrigation canal which runs for several kilometers
across the Valley of Tehuacan, Puebla. Such canals were ap-
parently constructed only during later periods of Mesoamerican
prehistory, after the formation of the state had already taken
place. [Courtesy, respectively, J. A. Neely and R. B. Woodbury]

27 OCTOBER 1967 453

 already, which the Formative peoples The Oaxaca Project has only begun, 13. I. Bernal, in Handbook of Middle American
of Oaxaca did have, these tendencies and our reconstruction must remain a Indians, vol. 3, Archaeology of Southern
Mesoamerica, G. R. Willey, Ed. (Univ. of
can be aggravated by control and in- tentative one. We are especially aware Texas Press, Austin, 1965).
14. J. Paddock, in Ancient Oaxaca, J. Paddock,
heritance of irrigation systems. It may that we cannot as yet integrate into Ed. (Stanford Univ. Press, Stanford, Calif.,
be that early irrigation in Oaxaca can this scheme the botanical and paly- 1966).
15. H. Williams and R. F. Heizer, Contrib. Univ.
be most profitably viewed in these nological materials from Oaxaca which Calif. Archaeol. Res. Facility (Berkeley,
terms. are currently undergoing study. They Calif., 1965), pp. 41-54.
16. The pollen is being analyzed by J. Schoen-
will make the story still more complex, wetter; the modern flora, by W. Ernst; and
the ancient plants preserved in dry caves, by
but civilization is a complex process; C. E. Smith. We acknowledge their generosity
Summary and Conclusions single-cause theories, no matter how at- with the botanical data, but they should not
be held responsible for any errors in our
tractive, are inadequateto explain it. reconstruction.
The Valley of Oaxaca's large flat 17. J. L. Lorenzo, Rev. Mex. Estud. Antropol.
16, 49 (1960).
floor, high water table, low erosion rate, 18. S. F. Cook, Ibero-Americana 34, 2 (1949).
and frost-free floodplain give it a higher References and Notes 19. R. L. Stevens, in Handbook of Middle Ameri-
can Indians, vol. 1, Natural Environment and
agriculturalpotential than that of most 1. K. A. Wittfogel, Oriental Despotism (Yale Early Cultures, R. C. West, Ed. (Univ. of
Univ. Press, New Haven, Conn., 1957). Texas Press, Austin, 1965), pp. 289-290.
surroundingareas. The development of 2. M. D. Coe, Smithsonian Inst. Misc. Collec- 20. E. J; Wellhausen, L. M. Roberts, X. E.
the pot-irrigation system early in the tions 146, 33 (1963). Hernandez, in collaboration with P. C. Man-
3. G. R. Willey and P. Phillips, Method and gelsdorf, Races of Maize in Mexico (Bussey
Formative period gave it a head start Theory in American Archaeology (Univ. of Institute, Harvard Univ., Cambridge, Mass.,
over other valleys, where the low water Chicago Press, Chicago, 1958), p. 151. 1952), pp. 54-62.
4. A. Palerm and E. R. Wolf, "Ecological Po- 21. M. J. Kirkby and A. V. Kirkby, unpublished
table did not permit such farming; tential and Cultural Development in Meso- studies, 1966; meteorological data courtesy
Oaxaca maintained its advantage by america," Pan American Union Soc. Sci. of the State of Oaxaca and the Comision
Monograph No. 3 (1957). del Papaloapan.
assimilating canal irrigation, barbecho, 5. R. Wolf, Sons of the Shaking Earth (Univ.
E. 22. C. E. Smith, Fieldiana: Botany 31, 68 (1965).
of Chicago Press, Chicago, 1959), p. 18. For
infield-outfield systems, flood-water the original statement of the "symbiotic area"
23. R. B. Woodbury, personal communication.
24. D. P. Jewell, Amer. Antiquity 31, 874 (1966).
farming, and hillside terracing as these concept, see W. T. Sanders, Viking Fund
25. Plant identifications by C. Earle Smith, U.S.
Pub. Anthropol. 23, 115 (1956).
methods arose. With the expansion of 6. W. T. Sanders, "The Cultural Ecology of the Department of Agriculture, personal commu-
nication.
population in the high-water-tablezone Teotihuacan Valley" (Pennsylvania
Univ., University Park, 1965).
State
26. R. S. MacNeish, in Handbook of Middle
of the high alluvium, competition for 7. R. S. MacNeish, Trans. Amer. Phil. Soc. 48, American Indians, vol. 1, Natural Environ-
ment and Early Cultures, R. C. West, Ed.
highly productive land and manipula- pt. 6 (1958).
(Univ. of Texas Press, Austin, 1965), pp.
tion of surplusesmay have led to initial 8. Archaeological-Botanical
Second Annual Report of the Tehuacdn
Project (Peabody 413-426.
27. The magnetite,
disparities in wealth and status; com- Foundation, Andover, Mass., 1962).
known Etla or Zimatlfn
probably from one of the
sources, was identi-
9. R. S. MacNeish, Science 143, 531 (1964).
petition probably increasedwhen canal- 10. For the definition of the terms band, tribe,
fled by Dr. P. Desautels, Division of
Mineralogy, Smithsonian Institution. Identifi-
irrigationsystems were added duringthe chiefdom and state as used in this article, as cation of marine and freshwater mollusks
well as the characteristics of each type of
Middle Formative, improving some lo- social grouping, see E. R. Service, Primitive
was made by Dr. J. Morrison, Division of
Mollusks, Smithsonian Institution. The
calities to the point where one residental Social Organization (Random House, New
"exotically decorated pottery" referred to
York, 1962).
group owned land more valuable than 11. The first field season of the project (1966), en- would be technically
ican
described by Mesoamer-
as zoned
archeologists punctate and
that of its neighbors. titled "The' Prehistoric Cultural Ecology of zoned rocker-stamped neckless jars (teco-
the Valley of Oaxaca," was sponsored by the
Trade in exotic raw materials, which Smithsonian Institution. The second field sea-
mates),
and cylinders
white-rim
with
black
excised
ware, and bowls
de-
(raspada)
appear to have served as the insignia son (1967) was sponsored by National Science signs in the "Olmec" tradition.
Foundation grant No. GS-1616 to the Uni- 28. J. R. Acosta, in Handbook of Middle Ameri-
for status over much of Formative versity of Maryland. The staff included can Indians, vol. 3, Archaeology of Southern
K. V. Flannery (archeologist); M. J. Kirkby
Mesoamerica, increased the wealth of and A. V. Kirkby (geomorphologists); A. W. Mesoamerica, G. R. Willey, Ed. (Univ. of
Texas Press, 1965), pp. 814-836.
the Oaxaca communities, and their Williams, Jr. (ethnologist); J. Schoenwetter,
Austin,
29. M. D. Coe, Papers Peabody Museum
Museum of New Mexico (palynologist); C.
elite made contact with the elite of Earle Smith, U.S. Department of Agriculture, Archaeol. Ethnol. Harvard Univ. 53, 115
(1961); - and K. V. Smith-
other cultures, such as the Olmec. Such and W. Ernst, Smithsonian Institution (bota-
sonian Contrib. Anthropol.,
Flannery,
in press.
nists); J. A. Neely, University of Arizona,
contact probably stimulated exchanges and F. Hole, Rice University (archeologists); 30. P. Drucker, R. F. Heizer, R. J. Squier,
of the "lore" known only to the elite M. E. King, Howard University (prehistoric Bur. Amer. Ethnol. Bull. No. 170 (1959),
textiles); and C. Moser, R. Orlandini, S. plates 43 and 45.
-calendrics, hieroglyphic systems, Maranca, M. Winter, S. Lees, K. Vaughn, J. 31. E. R. Wolf, Peasants (Prentice-Hall, Engle-
wood Cliffs, N.J., 1966), p. 21.
and symbolic art-thus widening the James, S. Kitchen, and E. Martinez (field as-
32. A. Caso, in Handbook of Middle American
sistants). We are indebted to J. L. Lorenzo,
gap between farmer and chief. Through Departmento de Monumentos Prehispanicos, Indians, vol. 3, Archaeology of Southern
Instituto Nacional de Antropologia e Historia Mesoamerica, G. R. Willey, Ed. (Univ. of
cooperation or coercion, the Oaxaca (Mexico City), and L. Gamio, representative Texas Press, Austin, 1965), p. 937.
Valley chiefdom drew together a sym- of the I.N.A.H. in the Oaxaca archeological 33. M. D. Coe, Mexico (Thames and Hudson,
area, for official permission and good advice. London, 1962), pp. 95-96.
biotic area of 50,000 square kilometers, Special thanks go to I. Bernal, J. Paddock, 34. There are some indications that moist bar-
commemorating its accessions with D. Quero, C. R. Welte, R. B. Woodbury, E. rancas in the arid highlands of Mexico were
Salzberger, and W. T. Sanders for their help. the original habitat of wild maize; it is caves
stone monuments carved in bas-relief. Nancy H. Flannery drafted Figs. 1, 2, 4, and in such settings which have produced the
5.
By the start of the Christian Era it was 12. On the basis of their stratigraphic work at
earliest evidence of cultivated maize. C. E.
Smith [Fieldiana: Botany 31, 80 (1965)] sus-
the dominant political entity in the the site of Monte Alb,n during the 1930's and pects that maize cultivation began in such
southern highlands of Mexico, and had 1940's, Alfonso Caso and Ignacio Bernal es- barrancas and only later spread to the
tablished a five-period sequence for the time alluvial fans at the mouth of the barranca
become a true state. span between 600 B.C. and A.D. 1500. and from there out onto the valley floor.

454 SCIENCE, VOL. 158