Water, Soil and Seasonality in Early Cereal Cultivation

Author(s): Andrew Sherratt

Source: World Archaeology, Vol. 11, No. 3, Water Management (Feb., 1980), pp. 313-330
Published by: Taylor & Francis, Ltd.
Stable URL: http://www.jstor.org/stable/124253
Accessed: 15/12/2009 18:00

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

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

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

JSTOR is a not-for-profit 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].

Taylor & Francis, Ltd. is collaborating with JSTOR to digitize, preserve and extend access to World
Archaeology.

http://www.jstor.org
Water,soil and seasonalityin early
cerealcultivation
Andrew Sherratt

Horticultureprecededfield culture, as the garden (hortus)precededthe field (ager).

L. H. Morgan, 1877

The purpose of this paper is to review current ideas of agricultural development in the
sub-tropical and temperate parts of the western Old World and to suggest that the small
scale and restricted extent of early cultivation systems gave them a unique character
which has not been generally appreciated. Recent studies of prehistoric settlement in the
western Old World have shown the basic role of soil moisture in early cultivation, and
the association of such sites with riverine and springside locations. The beginning of
such a cropping system marked a significant departure from earlier forms of cereal
exploitation in the natural habitat zone, and may have involved a difference in growth
cycle from that of the wild forms. The kind of cultivation which can be reconstructed as
the earliest stage of agriculture is best described as a form of fixed plot horticulture
dependent on ground- and surface-water, which differentiated in succeeding millennia
into various forms of dry farming and irrigation agriculture. The former path involved
the development of cultivation techniques suitable for interfluvial areas, including plough
cultivation and swiddening; while the latter involved increasing degrees of water manage-
ment to enlarge the area suitable for intensive cultivation.
Because the evidence for early cultivation systems is so varied in character and the
arguments often speculative, three aspects will be considered separately rather than
treated in strict chronological or regional order. After a short discussion of current
schemes of agricultural development, the problem is approached first from the settle-
ment evidence, and then from the viewpoint of crop ecology, before discussing the
development of methods of water-management. The final section sets these observations
in a wider context.

The character of early cultivation systems

Recent discussions of agricultural development, notably those of Boserup (I965), Wolf

(I966), and Sanders and Price (I968) have stressed the transition from extensive to

World Archaeology Volume II No. 3 Water management

? R.K.P. 1980 0043-8243/1103-03I3 $I.50/I
3I4 Andrew Sherratt

intensive forms of land-use. A major theme of this analysis is the move from long-fallow
systems, often involving 'slash-and-burn' or swidden farming, to more intensive cultiva-
tion with shorter fallows made possible by labour-intensive techniques such as manuring
and irrigation.
Such transitions have been studied notably in south-east Asia (Geertz 1963) and in
tropical Africa (Allan 1965). In applying these ideas to archaeological material, swidden-
ing has usually been seen as the 'primeval' system, and assumed to be typical of the
earliest farming groups. This idea has a long history in Europe, where early Neolithic
groups have often been assumed to be 'slash-and-burn' cultivators (e.g. Clark I952: 93).
Yet the very fact that swiddening is still widespread in many areas and was only super-
seded in other areas in historical times should itself argue caution in using this as a model
for the systems which were in existence six thousand years before.
A further argument against seeing swidden cultivation as a feature of great antiquity,
especially in temperate environments, is its restriction in recent times to the most
marginal European habitats, usually under coniferous forest. Its use is known mainly
from the Boreal forest areas, for instance in Finland (Sigaut, I975), or else from upland
areas of coniferous forest in central Europe such as the Carpathians (Lewicka 1972). In
these areas it occurs on soils so marginal that they were not cultivated before the Middle
Ages, as can be seen from the distribution of Late-Medieval clearance-names (Smith
1967: 133). It could plausibly be argued, therefore, that far from being a 'primeval'
agrarian system, swidden agriculture was a characteristic of the most recent phase of the
internal colonization of Europe, when settlement spread to the least fertile soils which
could not withstand any other form of cultivation.
While this phase may be more significant in the tropics, where low-quality soils form a
greater proportion of the landscape, the same arguments apply: it is likely that the
starting point of the Boserup scheme represents an already developed system of cultiva-
tion which is no model for the earliest farming systems.

Inference from prehistoric settlement patterns

The above arguments receive further support from examination of the archaeological
evidence. The most striking feature of early farming settlement is its restricted distribu-
tion. By comparison with the proportion of the landscape under cultivation today, only a
small fraction was used by early agriculturalists. In the Near East this proportion had
grown much larger by the third millennium B.C.; but in Europe it did not attain anything
like its present spread until the first millennium B.C. or even the Christian era, and the
process of infilling continued well into the Middle Ages.
Not only was the distribution of sites in the earlier phases restricted, it was highly
selective; and there are consistent patterns of association with certain environments and
situations. Alluvial, lake-edge and other locations with high groundwater are character-
istic of the Neolithic both in the Near East and in Europe (Allan 1972; Sherratt 1973).
In the earliest stage of farming villages, many important sites like Jericho and Murey-
bit fall outside the zone of possible dry-farming altogether. Indeed, when evidence of
cereal crops was first found at these sites it was commonly assumed that the grain must
 Water, soil and seasonality in early cereal cultivation 315

have been traded from better watered hill-country nearby, or else that the climate must
have undergone radical alteration. What all such sites possess, however, are abundant
supplies of localized surface- or ground-water, such as the seepage area and spring of Ain
es Sultan at Jericho or the seasonal wadis and perennially damp Euphrates floodplain at
Mureybit. Similar observations apply to the more developed phase represented at
Beidha, Tell Abu Hureyra and Bouqras, where local concentrations of surface-water
permitted a limited cultivation on suitable fine-grained alluvial soils, especially at wadi-
mouths.
The late glacial fall in lake-levels provided suitable terrain on the former lake-margins
in many parts of the east Mediterranean. In Anatolia, the earlier Neolithic sites of the
Konya Plain, including fatal Hiiyiik, cluster in the backswamp region where rivers
spread out in the otherwise arid plain; while further west the fluctuating lakes in the
upland basins of Pisdia are ringed with sites like Hacilar and Suberde. Early sites in
Greece and the Balkans are concentrated in areas of high water-table in the tectonic
basins, either by springs at the valley-edge, by stream confluences, or on lower terraces
by rivers and seasonally-enlarged lakes. In central Europe, early agricultural settlements
occur on levees in the Middle Danube and Tisza flood-plains, and by rivers and streams
within the loess area. All of these locations give access to hydromorphic soils, and avoid
the drier interfluves.
Contrary to expectations generated by the model of 'shifting cultivation', such sites
are usually long-lived, lasting for hundreds or even thousands of years in some cases.
This is as true for Bandkeramik sites in north-west and central Europe as it is for the
early tells of the Near East. Moreover they often represent surprisingly large communi-
ties; sites like Jericho and Catal Hiiyiik are only the most spectacular examples of a
general phenomenon. Clusters of agricultural settlement were often widely scattered, but
within them population was locally concentrated around critical resources such as fine-
grained, well-watered soils.
The implication of all this is that early agriculturalists occupied only a narrow zone of
maximum productivity, in an essentially small-scale though locally intensive system of
cultivation. In places this was capable of supporting nucleated communities, though in
others produced a pattern of hamlets following the exploited zone. This zone of high
water-table was of importance not only for cereal-growing but for a variety of resources,
including water itself, fish and freshwater mussels, and the range of plant and animal
species characteristic of the unusual conditions of the alluvium - in the Near East the
game and exploitable perennials of wet conditions, and in Europe the species which
could flourish only in the more open conditions where the Atlantic forests were inter-
rupted by wet ground (Clarke 1976). In a sense, therefore, 'Mesolithic' and 'Neolithic'
adaptations were parallel rather than successive: the main difference was the possibilities
of expansion inherent in systems using cereals (Sherratt I972: 49I).
The characteristics of initial farming systems in Europe are best demonstrated by
comparision with the patterns which emerged in succeeding millennia. The results of
intensive regional surveys indicate the major changes which took place in the later fourth
and third millennia b.c. A classic study of this kind is the work of Janusz Kruk and his
associates in the loess-covered uplands of Little Poland (Kruk I973). Here the restricted
catchments of early Neolithic sites in the river valleys are a clear indication of the focus of
316 Andrew Sherratt

interest in this zone, in striking contrast to the pattern which succeeded it in the third
millennium B.C.Kruk notes the restriction of the earliest (fifth millennium b.c.) agricul-
tural sites to the lower parts of the valley slopes, immediately overlooking the floodplain -
a pattern which continued in the fourth millennium, with some expansion to similar
positions in small side-valleys. The main change occurred in the third millennium with
the TRB culture, when there was a shift of activity to the interfluves and sites were
located on higher ground above the valleys.
Kruk uses these regularities to reconstruct the economy of the initial settlers (255):
'Because of the limited size of the zone where the activities of the Danubian [fifth and
fourth millennia b.c.] people were concentrated, the regular use of burning economy
seems virtually impossible. The characteristic concentration of settlements within
relatively limited areas also seems to speak against this possibility. Any extensive use of
burning as a farming system would involve also the interfluve .... However, the higher
landscape zone was not systematically exploited by the Danubian people.' Instead, 'it is
possible to assume with a fair amount of certainty that amongst the Danubian people the
plant cultivation was closely linked with the lowest-lying parts of the ground. Natural
conditions of this zone, notably the considerable humidity of soils and their productivity,
favoured intensive cultivation of the garden type. In this system, small field plots con-
centrated near larger settlements ... were worked by hand.... Owing to the high and
durable edaphic potential of the soils that covered this zone it was possible to use the
same portion of land continuously even without fertilizing it.' Such a pattern, with its
small amount of forest-clearance and restricted possibilities for grazing, would fit the
lack of stone tools suitable for woodcutting and the suggestions of stalled livestock in
these early contexts.
There is accumulating evidence (Sherratt, i980) that the late fourth and early third
millennia b.c. saw the widespread introduction of the plough, and that the extension of
cultivation and shift in the focus of settlement to the drier interfluves was associated
with plough-cultivation, extensive clearance and increased quantities of livestock. Such
observations can be made in most parts of Europe. Their significance for our present
purpose, however, is to suggest some important characteristics of the preceding regime.
A common feature of early agrarian systems in the western Old World is their spatial
restriction and concentration on limited alluvial zones. Soil moisture was evidently a
critical feature of their cultivation system, and ground-water as significant as rainfall in
the growth of crops. Such environments, while limited in extent, offered unusually high
productivity and allowed considerable continuity of cultivation and, in some areas,
notable concentrations of population. Such a reconstruction prompts some further
speculations about the nature and origins of this type of cultivation.

Cereal seasonality and growing-conditions

Wild cereals belong to the group of grasses which have adapted to strong seasonal stress
by an annual habit and the development of food-storage in the seed. This allows suc-
ceeding generations to 'recolonize' seasonally bare habitats by rapid growth with the
onset of rain. Wild wheats and barleys occur as members of open communities, particu-
 Water, soil and seasonality in early cereal cultivation 317

larly on limestone or basalt slopes and plateaux in the submediterranean park-forest belt.
Their mechanism of dispersal is a simple one, in which the spike shatters when the seeds
are ripe and they fall to the ground near the parent plant, where the sharp spikelet helps
implantation in cracks in the soil. The time of ripening and localized area of dispersal
prevent them from taking advantage of otherwise suitable habitats, for instance on
seasonally open areas of alluvium. They thus grow mainly on dry hillslopes during the
period of winter rainfall and ripen around May, with some variation according to altitude
(Zohary I969).
Under these dry upland conditions their growing season is spread over nine months.
Such a long period is not, however, a physiological necessity: wheat can be grown in only
three months (90-Ioo days) in appropriate conditions, as it was in certain places in
antiquity. Such three-month wheat is often called spring wheat, since it grows between
March and May. The principal sort of wheat grown in Europe and North America at the
present day, however, is winter wheat which is sown in the autumn and harvested in
summer. This has a higher and more stable yield than spring wheats (Klages I949: 345).
In most areas of Europe this type of wheat actually requires exposure to cold weather
(vernalization), though this is not so in the Mediterranean. At the present time, spring-
sown wheat with a three-month growing season is only grown in any quantity on the
northern margin of cultivation where the winter climate is too harsh.
Spring-grown wheat was, however, known in the Ancient World in the Mediterranean.
Semple (I932: 382) describes the hard, heavy spring wheats as being typical of limited
areas of southern Europe, by comparison with the soft, light grains which were then
preferred for food. 'The heavy Boeotian wheat was doubtless spring-sown, as it is
today, on the margin of the Lake Copais basin as the winter flood waters gradually
receded and Lake Copais contracted. The rich lacustrine soil and high water-table
explained the superior quality of the grain.' The rapid growth of this crop, which did not
have to rely on winter rainfall, was possible in the conditions of high ground-water which
existed in restricted areas near to bodies of water, especially those subject to marked
seasonal changes of volume.
It is tempting to ask whether the system documented for the Copais basin is not a
relic of a much more widespread method of cultivation, that was typical of the pre-plough
agriculture of southern and eastern Europe and the Near East. Is this not, in fact, the
groundwater-based cultivation-system inferred from the settlement evidence in the
preceding section? Such a form of cultivation would take advantage of the short period of
optimum water-availability between winter floods and summer desiccation, allowing the
three months' growth in spring which is necessary for a crop. Since less soil preparation
is needed with a system based on ground-water than in one using rainwater, and little
forest-clearance would be required, this pattern would make sense with low population
levels and a simple technology.
This model suggests that the critical innovation in the cultivation of cereals was their
transfer from dry environments with a winter growth pattern to alluvial, lake-edge,
riverine or springside locations in which an accelerated (spring) growth cycle was
possible. The wild cereals themselves were precluded by their simple dispersal mechan-
ism from colonising such habitats without human aid. The cultivation of crops in damp
conditions is indicated by the presence of Scirpus seeds in early cereal samples (Flannery
318 Andrew Sherratt

I969: 8i), and such deliberate sowing in areas away from their natural habitats was a
necessary precondition (along with the use of the sickle for reaping) for the selection of
tough-rachis forms, since the naturally-dispersed forms would always predominate in
self-seeded populations. The early cultivation of cereals in such habitats was thus a
quantum change from other forms of tending and exploitation, in that it operated in effect
as a new means of dispersal. It thus opened up a new range of highly productive habitats
for the species concerned without involving the massive restructuring of the ecosystem
which became necessary in later agriculture; and it involved only a change in the growing-
season. A minimum of cultivation, and no clearance of climax forest would have been
necessary; and moreover this system had the potential to spread very widely beyond the
natural habitat zone of the wild cereals, wherever such alluvial niches were to be found.
The term 'dry-farming', which is often used to describe the phase of cultivation before
the development of formal canal-irrigation, is thus misleading: indeed, in the model
suggested here, 'dry' (rainfall-dependent) farming was a specialized development parallel
to that of irrigation itself. A small-scale horticulture adapted to riverine and lacustrine
conditions explains the pattern of early Neolithic distribution in which even restricted
areas of fine-grained water-retentive soils were sought for settlement. With initially low
population levels, the still relatively sparse settlements could be supported from very
limited patches of suitable soil, which could sustain even quite large villages. Such soils
tend often to be neglected at the present day because of their small extent, their unsuit-
ability for the cultivation of winter cereals, and their equally important role in providing
spring grazing where large numbers of animals are kept and where much of the rest of the
landscape is under cultivation.
The technology which such a horticultural system would require is of the simplest. In
most cases major forest-clearance would not be needed. The seed would be broadcast, and
relatively little weeding would be necessary. Virtually no preparation of the soil would be
required, which in many cases would hardly rank even as hoe- or digging-stick cultiva-
tion. Where soils are subject to winter flooding and summer desiccation, the deep crack-
ing caused by drying-out would provide natural aeration and make them practically
self-cultivating. The labour-costs would thus have been trivial by comparison with later
forms of agriculture, especially with the major forest-clearance required by inter-fluvial
expansion in third millennium Europe which is reflected in the emergence of effective
axes and the widespread trade in stone for these tools (Sherratt I976: 563-7). Such early
horticultural systems would always have been spatially restricted, however, with locally
high population levels but wide intervening uncultivated areas and with little potential
for local growth without radical changes in technique. The main pattern of growth would
be first by rapid budding-off and export of population, followed by expansion to smaller
and smaller patches of the appropriate high-yielding soils within the occupied area. This
is precisely the picture which emerges from studies such as those of Kruk (1973: see also
Sherratt, 1972: 517, 524).
The rapidity of early agricultural dispersal was not due to shifting cultivation, but to
the pattern of restricted (and often linear) habitats suitable for initial farming practices.
The spread of cultivation in similar habitats would have allowed a dispersal through a
variety of environmental zones with minimal adaptation. It has, for instance, been sug-
gested (Butzer I972: 580) that the penetration of inland Europe would have involved a
 Water, soil and seasonality in early cereal cultivation 319

major shift from winter-grown crops in the Mediterranean to spring-grown crops in the
temperate zone. If both were essentially spring-grown, then no such radical change in the
growth period would have been necessary, other than a slight adjustment in the time of
sowing.
Developments beyond this initial stage of cultivation involved adaptation to drier
habitats which required more soil preparation and the use of fallowing or systematic
rotation. Such techniques are likely to have emerged in areas where forest-clearance was
not necessary. One of the first major steps in this direction was the extensive appearance
of sites on the deep brown soils of the moist steppes of northern Assyria in the sixth and
fifth millennia b.c. (Oates n.d.). Such cultivation would require extensive soil preparation
with the hoe, a fallow of one to four years and probably the development of new strains of
cereals as cultivation moved out of the zone of abundant ground- and surface-water.
Rainfall-dependent growth would have necessitated a winter growing season, at least for
the wheats, and it is likely that there was a return to the pattern of seasonality character-
istic of the wild cereals. Such a move would favour the use of barley, as well as bread-
wheat (Triticum aestivum) - the 'soft, light grain' which replaced the older and harder
forms.
The additional labour required for soil preparation under these circumstances would
have given a powerful incentive to the development of new techniques. As cattle were
common domestic animals on these lowland sites, it is possible that the plough came into
use in this area by the fifth millennium, even though it is only definitely attested from the
fourth millennium in southern Mesopotamia (Sherratt 1980). This could be used to
pulverize the soil to prevent loss of moisture through capillary action by the formation
of a dust mulch (Clark 1952: Ioo; Stevens 1966: 93). Only at this stage would the
characteristic pattern of 'dry farming' have emerged.
These techniques seem to have spread only slowly to the forested areas of Europe,
where they would have necessitated extensive clearance. This innovation is demon-
strated by settlement-pattern evidence such as that from Little Poland referred to above
(Kruk 1973) in the mid-third millennium b.c., which coincides with the first examples of
ploughmarks and iconographic evidence of animal traction (Sherratt i980).
While the extension of interfluvial cultivation in the Mediterranean would have
required a winter growing-season, in temperate Europe the higher spring rainfall would
have allowed the continuation of a spring-growth pattern (fig. ia). The major role played
by barley in late Neolithic and Bronze Age agriculture argues in favour of a spring
growing-season, as cultivation proliferated on sandy or other light soils which warm up
rapidly in spring. The frequency of millet (Panicum and Setaria) in these contexts in
southern Europe (Hartanyi and Novaki I975) also supports this suggestion. Spring-
grown crops would leave a greater area for winter grazing, important in Bronze Age
stock economies.
The first definite indication of winter-grown crops in northern Europe is the spread of
spelt in the first millennium B.C. (Applebaum 1954; 104). By this time it is likely that
winter varieties of the other wheat species had also emerged, involving the evolution of
vernalization mechanisms. This was the point at which the primitive original wheats
such as emmer would have been increasingly replaced in Europe by hexaploid species
better able to withstand the cold, and barley became relatively less important. Rye also
 AD/BC MARGINAL NORTHERN CENTRAL MEDI-
NW.EUROPE &W.EUROPE EUROPE TERRANEAN NEAR EAST

a
f

4-
i ! i !
(Withvernatisation)
i I i i
5
Ii
Wintergrowth
6
/
( ;/ / Springgrowth
7
(millet) Newcrops
a

FigureI Rainfalland flood regimes

 Water, soil and seasonality in early cereal cultivation 321

became more common in eastern and central Europe, after a long period of sporadic
occurrence as a weed (Jankuhn 1969: 220). Spring-grown crops cultivated with the ard
would have survived on the northern margins where the winter was too harsh for the
longer growth-cycle. The pressure of grazing caused by winter sowing could have been
alleviated by the provision of hay, which would have been increasingly important for the
keeping of livestock.
At this point also cultivation began to spread to heavier soils, and more robust and
ultimately sod-turning ploughs developed (Sherratt, forthcoming). This was important
in autumn ploughing, to form clods which would prevent the formation of soil-caps and
pans over winter, unlike the spring ploughing where a fine tilth is required. The use of
iron for plough-shares also produced a more effective instrument for land-breaking,
especially on clays. (Iron shares appeared in the Near East in the early first millennium
B.C., in central Europe at the end of the first millennium B.C., and in northern Europe in
the later first millennium A.D.) A probable consequence of these changes, and especially
(as Susan Limbrey has pointed out) the existence of open ploughed fields at the time of
maximum winter rainfall, would have been the massive increase in soil erosion indicated
in the extensive accumulation of valley alluvium in temperate Europe. The cropping-
system which came into use at this time would be essentially the same as that already in
use in the Mediterranean, with a two-course rotation of alternate winter crops and fallow.
This persisted until the later first millennium A.D., when it was replaced in parts of
temperate Europe by the three-course system in which a further field of spring-grown
crops - oats (for horses), barley, and legumes - was added, which suited the developing
needs of animal husbandry (White 1962: 69). The fully-developed pre-industrial
agricultural system of northern Europe was by this stage able to sustain comparable
population levels to the intensive, though necessarily more restricted, irrigation-based
systems of the Near East, and it was associated with the large-scale urbanization of
temperate Europe.

Developed surface-water agriculture: irrigation

During the time that these developments in rain-fed farming were taking place in
Europe, a succession of irrigation-based systems had appeared in various parts of the
Near East. It is commonly assumed that irrigation represented a major departure from
earlier agricultural systems, involving radical changes in technique. In the perspective
adopted here, however, it can be seen as an artificial extension of the conditions which
characterized the earliest forms of cultivation. A much more continuous pattern of
development, by small additional inputs of labour and capital, can thus be postulated.
The earliest forms of irrigation are likely to be small-scale water-spreading involving
minimum amounts of investment in new facilities. Such are indeed possible in pre-
agricultural contexts, as with the Owen's Valley Paiute who divert streams to water
patches of self-propagating wild plants (Steward 1929). Similar practices may have been
employed from the beginning in locations like that of Jericho. Without modifying the
natural drainage by channels, cultivation may also make use of the sheet-flow of flood-
waters. This system provided the basis of agriculture among some groups of Pueblo
322 Andrew Sherratt

Indians of the south-west of the United States, where the rainfall is too slight to grow
crops. As water seasonally overflows from high ground, however, cultivation is possible
either at the sides of valleys below escarpments or on alluvial fans at the mouths of gullies
(arroyos), or else on the valley floor itself where streams overflow their floodplains
(Bryan 1929: 445). Such flood-water farming supported communities of the Hopi and
neighbouring Indian groups, and was a determining factor in the location of such villages,
which might have a population of several hundred. This method of cultivation, however,
requires undissected valley-floors and is very vulnerable to gullying in periods of stream-
incision (Cooke and Reeves 1976). Vita-Finzi has suggested that suitable conditions for
this kind of cultivation obtained in parts of Jordan in the early Holocene during the
deposition of the Hasa formation (I969, 32).
The earliest demonstrated examples of channel-irrigation in the Near East are related
to the braided streams which flow down alluvial fans on the margins of the semi-arid
basins. Cutting transverse trenches to carry the water to a wider area is a simple but
effective way of using the gradient to maintain an adequate supply to the fields, either
from floodwater or perennial flow. Such a system has been documented at Choga Mami
near Mandali in Iraq and dated to the Samarran culture in the sixth millennium b.c.
(Oates and Oates 1976). This form of irrigation is typical of the Early Chalcolithic period
in the Near East, when it was also employed on the fringes of the Zagros and of the
Kopet Dagh in Turkmenia. In the latter area it sustained the settlements of the
Geoksyur culture (Lisitsina i969). It may also have been used in Anatolia at this
time, for instance at sites in comparable positions in the Konya plain (French 1970;
Sherratt 1973).
This system is well adapted to the margins of lowland areas fed by streams from nearby
better-watered mountains. Similar techniques may have been employed in other areas of
low rainfall such as southern Spain by the third millennium b.c. (Schiile i967). They are,
however, vulnerable to erosion producing channel incision and gully formation. Major
extensions of the cultivated area by these methods, moreover, are difficult to obtain and
require more complex engineering and higher investment. The system of underground
channels or qanats in Iran and adjacent areas (e.g. the falaj system of Oman) carrying
ground-water from valley-edge fans, represents a further attempt to extend these types of
water-sources from the first millennium B.C. onwards (Stronach i980).
In more marginal environments it was sometimes possible to make use of ephemeral
run-off and wadi-flow by the use of dams or small channels. In Baluchistan and the
southern Arabia, where brief periods of stream-flow result from melting snow in the
mountains, the water and fertile suspended silts were trapped by boulder-walls (gabar-
bands) across narrow valleys to allow the growth of a spring wheat-crop (Raikes i965;
Leshnik 1973: 67). A somewhat more sophisticated channel-based system dating to the
fourth millennium b.c. has been found in association with the major Late Chalcolithic
settlement at Jawa in the basalt region in the arid zone of Jordan (Roberts 1977). Here,
three systems of channels were used to tap the short-lived winter surplus, both by gravity
flow from the main wadi and by collecting local runoff, and these were used to supply
both fields and cisterns. Comparable systems supported the Bronze Age towns of
Palestine, and highly sophisticated methods for utilizing runoff were developed in the
Levant and western Arabia in the first millennium B.C. (Evenari et al. 1971).
 Water, soil and seasonality in early cereal cultivation 323

The greatest opportunities for the extension of cultivation by irrigation, however, lay
in riverine environments, especially the major alluvial valleys situated in arid plains but
fed by exotic water, like those of the Tigris/Euphrates, Indus and Nile. Despite the
environmental similarities of these three basins and the hydraulic civilizations which they
produced, however, there are significant differences between them. Most important is
the timing of their flood-periods (fig. ib). The Indus and Nile both have an essentially

Figure2 The differentiationof Old World farmingsystems

autumn inundation, allowing crops to be sown as the waters retreat in time for a winter
growing-season, but the Tigris and Euphrates are characterized by inconvenient spring
floods.
In the Indus basin, on the south-eastern edge of the distribution of the west Asiatic
cereals, crops such as wheat and barley with a winter/spring growth cycle (rabi crops)
could be grown in the dry winter season only in low-lying areas where they could make
use of residual soil moisture from the autumn flood and monsoonal rains (Lambrick
1964: 75). Alluvial silts along the rivers thus provided the basis for Harappan agriculture.
Cultivable land could be easily extended by simple methods of inundation irrigation
(sailaba), especially along inlets and creeks or on the inner sides of river-meanders
(Leshnik 1973: 73). 'Thus the whole operation involves an absolute minimum of skill,
labour and aid of implements' (Lambrick 1964: 76), and this limited but productive
land could support the development of Harappan cities even before the introduction of
the complementary summer/autumn (kharif) crops such as sorghum.
324 Andrew Sherratt

The Nile, flowing in a more narrowly-defined trench, allowed cultivation of the flood-
deposited silts in a series of basins formed by old levees, in which the autumn floods
provided natural irrigation for cereals sown as the waters receded by November (Butzer
1976). Simple techniques for retaining the water and increasing the watered area were in
use by the third millennium B.C., involving the breaching of levees and the canalization of
floodwater by digging overflow channels and constructing dikes. The protodynastic
'King Scorpion' macehead shows a ritual scene of this kind (cover). More labour-
intensive methods of actually raising water from its lower, summer levels began by the
mid-second millennium with the use of the shadtf (pole and bucket lever), and larger-
scale operations were only initiated in the later first millennium B.C. In consequence,
summer-grown crops were not cultivated before the second or first millennium.
The growth of irrigation systems in the Mesopotamian lowlands has been extensively
investigated in recent years (Adams I974; Gibson I974; Oates and Oates 1976). The
Tigris and Euphrates differ from the Nile and Indus in that their floods, related to snow-
melt in the mountains, occur towards the end of the winter growing-season, in April and
May. It was thus the perennial flow which was most useful, with floodwater often
presenting a problem of dispersal: what started as irrigation had to continue as drainage
and crop-protection. Because the rivers deposit large quantities of silt and build up high
levees, they actually run above the level of the surrounding plain. It is a relatively simple
matter to spread water through breaches in these levees; though this water had then to be
dispersed through a system of anastomosing channels. Early settlement 'followed the
numerous small, meandering stream distributaries found on any unmodified alluvial
plain. Lengthy, branching canal networks are not in evidence, and cultivation seemingly
was confined to narrow bands along natural levee backslopes and to favourably situated
margins of seasonally filled depressions' (Adams 1974: 2). Mesopotamian irrigation was
an extensive system in the sense that it required an alternate fallow to restore the water-
table and prevent salination, and that there was more potentially irrigable land than
could easily be managed without major investment in flood-control. Indeed, the onset of
local salination forced periodic shifts. The fallow and intervening areas allowed a sym-
biotic pastoralism which was essential to a full exploitation of the region.
These observations apply to lower Mesopotamia - the 'River Plain' of Buringh (I957).
The development of irrigation on the Assyrian Steppe was a later feature, because the
entrenched character of the rivers necessitated the use of lifting-devices rather than a
simple use of gravity-flow. The shaduf (used in lower Mesopotamia for watering
the levees) was one answer to this problem, and it was in use by the early-second
millennium B. C. (Drower I954), though only in a limited area adjacent to the major
rivers.
In each of these major alluvial basins, the Nile, Indus and Tigris/Euphrates, there is
evidence for the use of the plough (technically the ard or scratch-plough) from an early
stage in the development of lowland irrigation: in Mesopotamia at least by the fourth
millennium and in the others by the third millennium (Sherratt i980). It seems that this
instrument was an integral part of these extensive lowland cultivation systems, and
perhaps a precondition for their development.
The construction of major artificial watercourses in Mesopotamia dates to the Sassan-
ian period (Adams I965), when long branching canals argue a significant degree of
 Water, soil and seasonality in early cereal cultivation 325

central control and investment, and are part of the general pattern of intensified inputs of
capital and labour and the more elaborate construction typical of all forms of irrigation
from the first millennium B.C. onwards. This was the period in which summer-grown
crops such as rice and sesame were introduced from further east. Intensive irrigation led
in many areas to substantially higher population densities and an extension of the areas
under state control. Many of the features stressed by Wittfogel (I957) as associated with
irrigation agriculture may thus be more characteristic of these Iron Age developments
than of preceding systems, which were often village-based and local in management.
The growth of irrigation thus followed a natural progression from the earliest surface-
and groundwater-based systems through an increasing management of sources of surface-
water and its distribution to suitable locations. The earlier forms required only a small
input of effort in propitious circumstances, and involved little co-operation beyond
individual communities. As with south-east Asian systems even quite complex arrange-
ments could grow by quite small increments of household investment in each generation
(cf. Leach 1959). Small-scale hydroagriculture (in Wittfogel's term) was a widespread
though sporadic feature of cultivation in the semi-arid and arid zones from the Mediter-
ranean to central Asia in the third and second millennia B.C. Larger-scale projects
involving higher investment and wider co-operation were needed to increase the area
served by these limited systems, and these did not emerge until the growth of imperial
power and investment in rural development by landlords and urban capitalists, especially
from the first millennium B.C. onwards.

Comparison
Starting from similar initial conditions, therefore, cultivation systems in the Near East
and Europe became increasingly differentiated as they exploited more fully the particular
characteristics of the two regions. In the subtropical zone, where lack of water was the
main constraint, large areas of alluvium could be brought into cultivation by irrigation,
to support substantial local populations from an early date. In the temperate zone, the
most significant agricultural resource was the reservoir of nutrients stored as soil-colloids
which could be released by forest-clearance and ploughing. The wide areas of productive
forest soils postponed the need for intensive methods of cultivation; though, once
introduced, these could support higher overall densities of settlement. In northern
Europe the scratch-plough evolved into a heavier and more deeply-penetrating instru-
ment for use on heavy soils whose thick sod required turning to aerate and drain. What
had originally developed as a response to the lack of water had become a method of
dealing with its over-abundance.
By contrast to the ecosystems of the subtropical and temperate areas, those of the
humid tropics are characterized by rapid oxidation and strong leaching: nutrients are
stored in standing vegetation rather than in the soil. While significant areas of alluvial
and volcanic soils can support permanent cropping, the much greater areas of interfluvial
terrain necessitate short periods of cultivation and long fallows. Nutrients can be released
from standing vegetation by combustion, used up by crops in a few years, and the plot
returned to forest. More permanent cultivation can only be imposed on such soils by
326 Andrew Sherratt

large-scale investment of capital and labour. South-east Asian terracing systems with
padi-farming are a specialized development of this kind, using flowing water to carry
nutrients not available in the soil (Geertz 1963).
The kinds of transition to more intensive systems studied by development economists
such as Boserup, therefore, are only partly analogous to the kinds of change inferred in
subtropical and temperate regions, and in any case represent a relatively late phase of
land-use even in the tropics. Recent research has increasingly emphasized that the earlier
stages of tropical cultivation were characterized by systems of fixed-plot horticulture
(Harris I973: 399-402); while ecological arguments and archaeological evidence from
several parts of the world have emphasized the importance of low-lying, alluvial areas in
early cultivation.
In south-east Asia, rice began as a palustrian species, adapted to semi-aquatic habitats,
as did taro. Dry rice, suitable for interfluvial cultivation, was a secondary development.
Gorman (I977) has postulated an initial stage of inundation-cultivation in piedmont
regions, lasting down to c. 2000 B.C., followed by an extension of irrigation-systems in the
open alluvial plains, with iron and plough-cultivation (using the water-buffalo) spreading
in this context in the first millennium B.C. Extensive swiddening could be a relatively
recent development, as drought-resistant strains of rice allowed a spread to the hills in
the first millennium A.D.
In Africa, where agriculture developed at a later date than the other areas considered
here, the situation is less well understood. Nevertheless, Harlan (1975: 197) has pointed
out the importance of fluctuating shallow lakes in the Sahel region of West Africa where
many of the indigenous cultigens emerged, which would have been suitable for decrue
cultivation on moist soil as the floodwaters subsided.
In Mesoamerica (Price 197I) the present importance of swidden contrasts with the
types of cultivation inferred for prehistoric and early historic times. Early crops in the
dry tropical zone of Mexico were cultivated by floodwater along barrancas(wadis). In the
Valley of Oaxaca, one of the most intensively-investigated areas, initial cultivation was
confined to the alluvial zone with a high water-table, and extended over a limited area by
pot-irrigation from wells. In the Late Formative period, settlement spread to the pied-
mont zone where water from springs could be directed by canals to terraced fields, as in
the fossilized system at Hierve el Agua (Kirkby i973). In the Valley of Mexico, canal
irrigation supported the growth of Classic Teotihuacan, while intensive chinampa (wet-
garden) cultivation sustained Postclassic Tenochtitlan. In the Maya area of the humid
tropical lowlands, the supposition that swiddening formed the basis of sites like Tikal has
been replaced by a recognition of the importance of shallow lakes (bajos) exploited by
canals and raised fields as well as a variety of other water-spreading devices (Harris 1978),
and the situation is more complex than was previously imagined. In coastal Peru, flood-
water farming on the coast was replaced by limited canal-irrigation in inland valleys, and
in turn succeeded by larger-scale canal irrigation on the arid coastal plain (Moseley I974).
All these examples show a direct transition from floodwater-cultivation to canal-based
systems, with extensive cultivation as a later, marginal development.
Behind the regional variations, therefore, underlying similarities in the development of
agricultural systems are becoming evident. They emphasize the importance of an often
protracted initial phase of small-scale surface- and groundwater-based horticulture,
 Water, soil and seasonality in early cereal cultivation 327

which ultimately differentiated into hydraulic and rainfed systems of both extensive and
intensive kinds. Further definition of the sequences of local change should allow a
clearer view of the social forces which prompted the development of more intensive
systems of cultivation.

Note on terminology
The term 'spring wheat' is used in a special sense by plant geneticists, to describe any
wheat which lacks a vernalization requirement, including winter-grown varieties in the
Mediterranean and Near East. I am grateful to F. G. H. Lupton of the Plant Breeding
Institute, Cambridge, for this and other observations.

Acknowledgements
I am grateful to Jim Lewthwaite and Neil Roberts for discussing the results of their
original fieldwork and developing some of the basic ideas of this paper: and to Joan Oates
and David Harris for their comments on earlier versions of it.

I4.Viii.1979 University of Oxford

References

Adams, R. M. i965. Land behindBaghdad.Chicagoand London.

Adams, R. M. I974. Historicpatternsof Mesopotamianirrigationagriculture,I-6. In Downing,
T. E., Irrigation'sImpacton Society. Arizona.
Allan, W. I965. The African Husbandman. London.
Allan,W. 1972. Ecology,techniquesand settlementpatterns,211-26. In Ucko, P., Dimbleby, G.
and Tringham, R. (eds), Man, Settlementand Urbanism.London.
Applebaum,S. I954. The agricultureof the early Iron Age as exemplifiedat FigheldeanDown,
Wilts. Proceedings of the Prehistoric Society. 20:103-I4.

Boserup, E. I965. The Conditionsof AgriculturalGrowth.London.

Bryan, K. I929. Floodwater farming. Geographical Review. 19:444-56.

Buringh,P. I957. Living conditionsin the lower MesopotamianPlain in ancienttimes. Sumer.

I3:30-46.
Butzer, K. W. I972. Environmentand Archaeology.London and Chicago.
Butzer, K. W. I976. Early HydraulicCivilizationin Egypt. Chicago.
Clark,J. G. D. 1952. PrehistoricEurope:the economicbasis.London.
Clarke,D. L. 1976. Mesolithic Europe: the economic basis, 449-8i. In Sieveking, G. Long-
worth, I. and Wilson, K. (eds), Problemsin Economicand Social Archaeology.London.
Cooke, R. U. and Reeves, R. W. 1976. Arroyosand EnvironmentalChange.Oxford.
328 Andrew Sherratt

Drower, M. I954. Water-supply, irrigation and agriculture, 520-57. In Singer, C., Holmyard, Jo
and Hall, A. R., A History of Technology, vol. i. Oxford.

Evenari, M., Shanan, L. and Tadmore, N. I971. The Negev: the Challenge of the Desert. Cam-
bridge, Mass.
Flannery, K. V. I969. Origins and ecological effects of early domestication in Iran and the
Near East, 73-I00. In Ucko, P. and Dimbleby, G. (eds), The Domestication and Exploitation of
Plants and Animals. London.
French, D. H. 1970. Notes on site distribution in the (umra area. Anatolian Studies. 20:139-48.
Geertz, C. I963. Agricultural Involution. Berkeley.
Gibson, M. I974. Violation of fallow and engineered disaster in Mesopotamian Civilization,
7-20. In Downing, T. E. and Gibson, M., Irrigation's Impact on Society. Arizona.
Gorman, C. I977. A reconsideration of the beginnings of agriculture in south east Asia, 321-
56. In Reed, C. (ed.), Origins of Agriculture. Paris and The Hague.
Harlan, J. R. 1975. Crops and Man. Madison, Wisconsin.
Harris, D. R. I973. The prehistory of tropical agriculture: an ethnoecological model, 391-417.
In Renfrew, A. C. (ed.), The Explanation of Culture Change. London.

Harris, D. R. 1978. The agricultural foundations of lowland Maya civilization: a critique,

301-23. In Harrison, P. D. and Turner, B. L. (eds), Prehispanic Maya Agriculture. Albuquerque.

Hartanyi, B. P. and Novaki, G. 1975. Samen-und Fruchtenfunde in Ungarn von der Neustein-
zeit bis zum I8. Jahrhundert. Agrartorteneti Szemle. 17:1-65, 1967.
Jankuhn, H. I969. Deutsche Agrargeschichte: Vor-und Friihgeschichte. Stuttgart.
Kirkby, A. V. T. 1973. The use of land and water-resources in the past and present Valley of
Oaxaca, Mexico. Mem. Mus. Anth. Univ. Michigan, 5.
Klages, E. 1949. Ecological Crop Geography. London and New York.
Kruk, J. 1973. Studia Osadnicze nad Neolitem Wyiyn Lessowych. Warsaw.
Lambrick, H. T. 1964. Sind: a General Introduction. Hyderabad.
Leach, E. R. I959. Hydraulic Society in Ceylon. Past and Present. 15:2-25.
Leshnik, L. S. 1973. Land-use and ecological factors in prehistoric north-west India, 62-84.
In Hammond, N. (ed.), South Asian Archaeology. London.

Lewicka, A. 1972. Brandwirtschaft und Brandrodung in den Polnischen Karpathen, 1I9-42.

In Balassa, I. (ed.), Getreidebau in Ost-und Mitteleuropa. Budapest.

Lisitsina, G. N. I969. The earliest irrigation in Turkmenia. Antiquity. 43:279-87,

Morgan, L. H. i877. Ancient Society. New York.
Moseley, M. 1974. Organisational preadaptation to irrigation and the evolution of early water
management systems in coastal Peru, 77-82. In Downing, T. E. and Gibson, M. (eds). Irriga-
tion's Impact on Society. Albuquerque.
Oates, J. (typescript). Land use and population in prehistoric Mesopotamia.
Oates, D. and Oates, J. 1976. Early irrigation agriculture in Mesopotamia, 109-36. In
Sieveking, G., Longworth, I. and Wilson, K. (eds), Problems in Economicand Social Archaeology.
London.
 Water, soil and seasonality in early cereal cultivation 329

Price, B. L. I971. Prehispanic irrigation agriculture in Nuclear America. Latin American

Research Review. 6:3-30.
Raikes, R. I965. The ancient gabarbands of Baluchistan. East and West. 15.
Roberts, N. 1977. Water conservation in ancient Arabia. Seminar for Arabian Studies. 7: 34-46.
Sanders, W. and Price, B. I968. Mesoamerica: the Evolution of a Civilization. New York.
Schiile, W. I967. Feldbewasserung in Alt-Europa Madrider Mitteilungen. 8:79-99.
Semple, E. C. I932. The Geography of the Mediterranean Region: its Relation to Ancient History,
London.
Sherratt, A. G. I972. Socio-economic and demographic models for the Neolithic and Bronze
Ages of Europe, 477-542. In Clarke, D. L. (ed.), Models in Archaeology. London.
Sherratt, A. G. I973. The interpretation of change in European prehistory, 4I9-28. In
Renfrew, A. C. (ed.), The Explanation of Culture Change. London.
Sherratt, A. G. 1976. Resources, technology and trade: an essay in early European metallurgy,
557-82. In Sieveking, G., Longworth, I. and Wilson, K. (eds), Problems in Economic ant
Social Archaeology. London.
Sherratt, A. G. I980. Plough and pastoralism: aspects of the Secondary Products Revolution.
In Hammond et al. (eds), Pattern in the Past: Studies in Memory of David Clarke. Cambridge.
Sherratt, A. G. (forthcoming). The evolution of the plough in temperate Europe.
Sigaut, F. I975. L'Agriculture et le Feu. Paris and The Hague.
Smith, C. T. I967. An Historical Geography of Western Europe before 800o. London.
Stevens, C. E. I966. Agriculture and rural life in the Later Roman Empire, 89-17. In The
Cambridge Economic History, vol. I. Cambridge.
Steward, J. H. 1929. Irrigation without agriculture. Papers of the Michigan Academy of Science,
Arts and Letters. I2:I49-56.
Stronach, D. I980. Achaemenian and Seleucid Iran, 206-II. In Sherratt, A. G. (ed.), Cam-
bridge Encyclopedia of Archaeology. Cambridge.
Vita-Finzi, C. 1969. Geological opportunism, 3I-4. In Ucko, P. and Dimbleby, G. (eds), The
Domestication and Exploitation of Plants and Animals. London.
White, L. I962. Medieval Technology and Social Change. Oxford.
Wolf, E. R. I966. Peasants. Englewood Cliffs.
Wittfogel, K. A. I957. Oriental Despotism: a Comparative Study of Total Power. New Haven.
Zohary, D. I969. The progenitors of wheat and barley, 47-66. In Ucko, P. and Dimbleby, G.
(eds), The Domestication and Exploitation of Plants and Animals. London.

Abstract
Sherratt, A.

Water, soil and seasonality in early cereal cultivation

The development of agricultural systems in the western Old World can be divided into three
main phases. Initial cultivation probably took the form of a localized fixed-plot horticulture
330 Andrew Sherratt

in a narrowrange of alluvial habitats. This was expanded both by diverting surface-waterto

reacha wider area,and by adaptationto dry conditionsby a better use of rainfall.Both required
additional labour either for canal-constructionor soil-preparation.The plough emerged as
part of these developments,and in Europeits use was associatedwith majorclearanceof wood-
land. Truly intensive systems became common in the third phase, from the first millennium
B.C. onwards, which was characterizedin the Near East by the construction of large-scale
irrigationworks, and in Europe by the developmentof the heavy plough.