To cite this article: JULIA SHAW & JOHN SUTCLIFFE (2003) Ancient dams, settlement
archaeology and Buddhist propagation in central India: the hydrological background, Hydrological
Sciences Journal, 48:2, 277-291, DOI: 10.1623/hysj.48.2.277.44695
To link to this article: http://dx.doi.org/10.1623/hysj.48.2.277.44695
Hydrological Sciences–Journal–des Sciences Hydrologiques, 48(2) April 2003
277
Ancient dams, settlement archaeology and
Buddhist propagation in central India: the
hydrological background*
JULIA SHAW
Merton College, Oxford OX1 4JD, UK
and
Institute of Archaeology, Oxford OX1 2PG, UK
julia.shaw@merton.oxford.ac.uk
JOHN SUTCLIFFE
Heath Barton, Manor Road, Goring on Thames, near Reading, RG8 9EH, UK
j.v.sutcliffe@talk21.com
Abstract A group of ancient dams (c. second–first century BC) was located during an
archaeological study of the Sanchi area in central India. Comparison of reservoir
volumes with estimated inflows suggests that their design was based on hydrological
understanding.
Key words ancient dam design; water balance; rice irrigation; Buddhism and social change;
urbanization
Barrages antiques, archéologie et propagation du Boudhisme en Inde
centrale: influence de l’état de l’art hydrologique
Résumé Un ensemble de barrages antiques (c. deuxième–premier siècle av. JC) a été
localisé lors d’une étude archéologique de la région de Sanchi en Inde centrale. La
comparaison entre les capacités des réservoirs et les débits entrant estimés suggère
qu’ils ont été dimensionnés à partir d’un savoir hydrologique.
Mots clefs dimensionnement de barrage antique; bilan hydrologique; irrigation rizicole;
Bouddhisme et changement social; urbanisation
INTRODUCTION
Very few ancient dams in India have been studied on the ground; the subject has been
dominated by a textual and epigraphic framework of analysis. Notable exceptions
include the Chola reservoirs in south India (Venkayya, 1906) and the well-known
Junagarh Dam in western India (Mehta, 1968). During an archaeological survey of the
Sanchi area in central India, a number of ancient embankments were located and
studied. Archaeological evidence suggested that these embankments were contemporary with the Buddhist monastic sites near which they are situated, and that more
than half are dated between about the second and first centuries BC, with others
between the first and 10th centuries AD. A joint archaeological and hydrological study
revealed that the dams appear to have been designed not only with a sophisticated
knowledge of dam engineering but also with an understanding of the principles of
basin water balance. This raises important questions about the role of water resources
* Based on a paper presented at: OH2—First International Symposium on the Origins and History of
Hydrology (Dijon, France, 9–11 May 2001).
Open for discussion until 1 October 2003
278
Julia Shaw & John Sutcliffe
management in the spread of institutionalized Buddhism and accelerated urban growth
in ancient India.
BACKGROUND TO FIELD WORK
Although the monumental remains of Sanchi and other Buddhist sites in the area have
been the subject of much attention since their discovery nearly 200 years ago (notably
Cunningham, 1854; Marshall, 1940), relatively little attention has been paid to the
surrounding landscape, the rural settlement pattern and the ancient land use of the area.
An archaeological survey carried out over two seasons in 1998–2000 covered an area
of about 750 km2 within a radius of about 15 km of Sanchi (Shaw, 2001). These field
investigations had the aim of using settlement archaeology to illuminate Buddhist
history, and in particular to provide evidence on social and religious change in central
India from the third century BC. During this study some 120 settlements and 35
previously unrecorded Buddhist monastic sites were noted, together with 16 ancient
dams (Shaw & Sutcliffe, 2001) which form the subject of this paper.
The dam below the major monastic site on Sanchi hill (Fig. 1) had been examined
by Marshall (1918, 1940); the embankment runs for about 350 m between Sanchi hill
and Nagauri hill to the south, and for over 1000 m from Nagauri hill to the west.
Marshall (1940, p.13) noted that “agriculture was not of course an occupation in which
the monks themselves were allowed to take part”. However, the evidence of a further
15 unrecorded embankments within the Sanchi area, a number of which are associated
with monastic settlements, not only provides empirical evidence for understanding the
history of ancient Indian irrigation, but also raises questions about the socio-economic
and religious infrastructure which supported the construction and management of the
Sanchi dams. These topics are discussed more fully elsewhere (Shaw, 2001), but an
analysis of the relative positions of ancient dams, contemporary settlements and
monastic sites provides evidence for reconsidering the separation of Buddhist monks
from the agricultural infrastructure. However, the emphasis of this paper is on the
hydrological understanding implicit in the design and construction of the set of
embankments.
PHYSICAL SETTING
The Sanchi area is situated within the Betwa basin in central India, which was the
subject of a comprehensive hydrological and hydrogeological study by Indian and
British scientists between 1976 and 1980 (Sutcliffe et al., 1981; Sutcliffe & Green,
1986; Hodnett & Bell, 1986). Thus the present physical background to the area is well
known. Underlying the whole Betwa basin are Pre-Cambrian rocks of Bundelkhand
granite, but these are exposed well to the north of Sanchi. Vindhyan Sandstones of
Cambrian age overlie the granite, and outcrop to the southwest of Sanchi, where they
form flat-topped or gently dipping hills with deeply incised scarps. Much of the basin
is occupied by Deccan Trap basalt, whose flows spread over the land surface to a depth
of 300 m; except in the hill areas, the Deccan Trap is covered by a plain of yellow clay
and black cotton soil derived from basalt weathering. There is a shallow aquifer within
the surface weathered zone of the basalt and soils.
Ancient dams, settlement archaeology and Buddhist propagation in central India
279
10
Sanchi
River
Betwa
River Nion
River
Bes
Fig. 1 Distribution of ancient dams (after Shaw & Sutcliffe, 2001).
Sanchi hill is an outcrop of Vindhyan Sandstone at the boundary between the
sandstone hills to the south and west and the basalt-floored valley of yellow and black
cotton soils to the north and east. The area to the south and west of Sanchi, with its
extensive hill ranges separated by short valleys, is more suited to the construction of
low embankments than the predominantly flat areas to the north and east (Fig. 1). The
hills define basins providing runoff into the dam sites, while the soils of the plains are
suitable for irrigated agriculture. Thus a total of 16 ancient dams were concentrated in
an area of about 400 km2 around Sanchi; these embankments would have formed
reservoirs fed by runoff from the hillsides and located close to areas which could be
280
Julia Shaw & John Sutcliffe
irrigated. The flat topography is less suitable for reservoirs to the north and east of
Sanchi; the only dams east of Sanchi are located among the line of hills around the
Buddhist monastic sites at Andher and Morel Khurd.
The soils on the sandstone hills are thin, up to 0.3 m thick, and sandy; the present
vegetation consists of broad-leaved, deciduous forest species such as teak and sal
(Shorea robusta) in the wetter areas tending towards acacia in the drier areas. The
“black cotton soils” of the plains are now used for agriculture. The depth of the black
soils and the underlying yellow clay is generally between 2 and 10 m but may be
deeper in the central part of the basin. The clay content gives rise to high soil moisture
storage capacity; the soils have marked swelling and shrinking cycles, and during the
dry season they crack to depths of 1.5–2 m under cultivation and up to 6 m under
deeper-rooted perennial grass and scrub. The cracks are up to 75 mm wide at the
surface and the soil therefore recharges rapidly at the onset of the monsoon.
PRESENT CLIMATE
The present climate of the Betwa basin, including the Sanchi area, is highly seasonal
with an annual rainfall ranging from over 1300 mm in the hills to the east of Vidisha to
less than 900 mm in the north of the basin near Jhansi, some 250 km north of Vidisha
(Fig. 2); the average over the whole Betwa basin is about 1140 mm, with 90% of the
Fig. 2 Average annual rainfall, mm: Betwa basin, 1926–1975 (after Sutcliffe et al.,
1981).
281
Ancient dams, settlement archaeology and Buddhist propagation in central India
Table 1 Average rainfall over Betwa basin (1926–1975) and evaporation at Bhopal, mm (after Sutcliffe
et al., 1981).
Jan.
15
Rainfall
Potential
transpiration 83
Open water
evaporation 113
Feb.
7
Mar. Apr.
66
2
May Jun.
5 111
Jul.
390
Aug. Sep.
351 196
Oct.
31
Nov. Dec. Year
15
8 1138
102
152
193
241
197
134
115
133
127
90
75
1643
135
193
241
293
237
164
143
167
165
123
103
2077
Fig. 3 Seasonal rainfall and potential transpiration, Betwa basin, mm (after Sutcliffe et
al., 1981).
rainfall occurring between the middle of June and the end of September (Table 1). The
seasonal distribution of average rainfall is illustrated in Fig. 3.
The open water evaporation, and also potential transpiration, were estimated by the
Penman method using climate averages from Bhopal, which can be taken as typical of
the basin. The monthly values and the transpiration averages are included in Table 1
and Fig. 3, respectively. The annual transpiration is 1640 mm, with values low in
winter, rising to a peak in May as the temperature increases and decreasing in July–
September during the monsoon. The seasonal climate cycle is illustrated in Fig. 3 by
the monthly balance between rainfall and potential transpiration. There is a period of
water deficit from January to June, when potential transpiration exceeds rainfall and
deciduous trees lose their leaves, followed by a period from July to September when
rainfall exceeds transpiration during the monsoon. A period of soil moisture recharge
is followed by a period of water surplus when most of the regional runoff and local
groundwater recharge occurs. After October, when potential transpiration again
exceeds rainfall, crop and vegetation transpiration relies largely on the soil moisture
storage, which depends on soil capacity and rooting depth, and can be estimated as
about 200 mm from a programme of soil moisture measurements and water balance
studies (Sutcliffe et al., 1981; Hodnett & Bell, 1986).
282
Julia Shaw & John Sutcliffe
DRAINAGE PATTERN AND RUNOFF
The River Betwa flows through the Sanchi area from the hilly area in the southwest,
and is joined just north of Vidisha by the Bes tributary which drains an area to the
west. To the east of Vidisha the River Nion drains an area of some 1000 km2; this
tributary is important as it was studied as a sample basin during the Betwa investigation; river flows were measured in 1978–1980, in addition to rainfall, soil moisture
storage and groundwater levels over a network of sites (Sutcliffe & Green, 1986). As
well as providing access to water, the high banks of the Bes and the smaller streams
provide settlement sites which are relatively free from monsoon flooding.
The runoff from the Betwa basin was measured over a period of 50 years, from
1926 to 1975, at Dhukwan Reservoir at the lower end of the basin of 20 600 km2; the
average runoff over this period was 351 mm, compared with the average basin rainfall
of 1138 mm. This runoff, like the monsoon rainfall, is concentrated between June and
September, though there is some residual baseflow from groundwater reserves after the
monsoon. The basin rainfall has varied widely from year to year, from 745 mm in
1966 to 1539 mm in 1961; the standard deviation over the 50-year period was 211 mm
or 18.5% of the mean. As the residual of the hydrological cycle, the standard deviation
of the runoff is relatively high at 157 mm or 44.7% of the average runoff. The relation
between basin rainfall and runoff, derived from records from the whole Betwa basin
and measurements from individual tributaries during the Betwa project (Sutcliffe &
Green, 1981), provides a means of estimating runoff from monsoon rainfall.
This can provide an indication of the possible effect on runoff of historical changes
of climate and particularly rainfall. Since the present climate and runoff provide the
basis for studying the ancient dams and reservoirs, the evidence for change over the
past 2000 years should be considered. There is little palaeo-ecological evidence for the
early historic period, and the main focus of enquiry has been a major phase of
aridification attributed to the third/second millennium BC (Singh, 1971). However,
various studies have indicated the beginning of an arid phase in the second millennium
BC, reaching present levels of aridity about 100 AD (Fuller, 2000). Although it is
hoped to obtain evidence during further research in the area, it will be seen later that
the hydrological discussion focuses on the correlation between reservoir capacities and
basin runoff volumes, which should be affected less by historical changes in local
rainfall and thus runoff depths.
The present hydrological behaviour of the smaller streams in the Sanchi area is
illustrated by a study of the Nion at Kuakheri (Sutcliffe & Green, 1986), based on
detailed measurement of the major components during 1978 and 1979. The records for
1978, which include measurements from a network of soil moisture sites and
groundwater observation wells, show that the reaction of the basin to the monsoon may
be divided into four phases: these may be described as periods of initial soil moisture
recharge when evaporation is negligible; combined soil moisture and groundwater
recharge with evaporation and some runoff; runoff and evaporation after soil and
groundwater storage has been replenished; and recession when some runoff is
maintained from groundwater storage and evaporation from soil moisture. The 1978
rainfall was about average and the measured runoff was 555 mm from the basin of
921 km2. The same sequence was observed though the timing was delayed in 1979,
when the monsoon rainfall was exceptionally low at 40% of average; soil moisture
Ancient dams, settlement archaeology and Buddhist propagation in central India
283
recharge was little below the 1978 value, but runoff was much reduced at 80 mm.
After the monsoon the soil moisture storage declined steadily, but the basin
evaporation was maintained by the initial storage of over 200 mm. This hydrological
regime is reflected in the present land use of the area, and throws light on the role of
the ancient reservoirs in the historic land use.
CHARACTERISTICS OF SANCHI DAMS
The dams in the Sanchi area are prominent and easily identifiable archaeological
features of the landscape. However, with the exception of two dams (5 and 10, Fig. 1)
which have been restored, they are all in a ruinous state and show signs of breaching at
sites where natural drainage channels cross the embankments. The dams have not been
excavated, but they consist of an earthen core, generally derived from black cotton
soil, reinforced by facing of dressed sandstone masonry, especially on the upstream
face. In some places the facing slabs are laid horizontally; in other places the slabs are
interlocked, with most blocks laid parallel to the face, with head-on slabs slotted in
perpendicular to the embankment (Figs 4 and 5). Dam heights range from 1 to 6 m,
with flat downstream sections apparently designed to make the dams resistant to
overtopping (Parker, 1909); some dam widths are as much as 60 m. At least two of the
higher dams (7 and 12) have spillway channels cut in the rock adjacent to the artificial
embankment.
The dams are not isolated features, but form part of a series of “early historic
complexes”. Typically a Buddhist stupa or monastic site is found on a nearby hilltop,
with a village settlement on the lower slopes or on an adjacent mound. The Sanchi
Fig. 4 Devrajpur dam: horizontally laid dam facing (after Shaw & Sutcliffe, 2001).
284
Julia Shaw & John Sutcliffe
Fig. 5 Devrajpur dam: interlocked stone facing (after Shaw & Sutcliffe, 2001).
complex is typical of this arrangement, with the monastic monuments on the hill, and
with one part of the embankment linking Sanchi hill with Nagauri hill, where an
ancient stone quarry, painted rock-shelters, scatters of microliths and early historic
pottery attest to a long occupational sequence. A second embankment links Nagauri
with a hillside settlement to the west. A similar relationship links dams, settlements
and monastic sites throughout the area.
CHRONOLOGY
Irrigation structures are difficult to date because of the nature of their construction,
with building material derived from several sources and subsequent restoration. It is
not surprising that Marshall (1940) could not provide a precise date for the construction of the embankment (no. 14) below Sanchi hill, but his rough estimate of second
century BC, in keeping with the second and most prolific phase of building activity at
Sanchi and neighbouring Buddhist sites, is supported by data from the present study. A
number of potsherds dating to c. third–second century BC were found in a modern
cutting through the eastern dam, and the presence of the same ware at nearby
settlements situates the dam within a network of sites including the monastic complex
on Sanchi hill and the hillside settlement at Nagauri. However, the pottery dating
(Shaw, 2001) is open to revision when a proper excavated ceramic sequence is
established for the area; an earlier date cannot be ruled out.
A group of serpent (naga) sculptures, ranging in date from second century BC to
10th century AD, provide additional markers; 21 nagas were noted during the survey,
15 of them previously unrecorded. The nagas were ancient serpent deities, a number of
Ancient dams, settlement archaeology and Buddhist propagation in central India
285
which in the Sanchi area appear to have been incorporated into both the Buddhist and
Hindu “landscape”. However, in most cases their traditional identity as nature spirits is
revealed by their close links to sources of water; of the 16 groups whose original
provenance is known, seven are associated with reservoirs, five with village tanks, and
four with rivers or streams. A number remain in situ beside ancient water bodies, and
provide terminal dates for the latter. Once these links have been used to build a
chronological framework for the Sanchi area, gaps can be filled by other indicators like
dam morphology, shape and size of stone facing, and positioning within the
surrounding archaeological complex.
The earliest surviving serpent sculptures in the Sanchi area are a naga and nagini
(male and female) couple, which originally stood on the ancient Gulgaon embankment
(dam 10). Both sculptures can be assigned to around mid-first century BC (Williams,
1976), providing a terminal date for the embankment itself. Another almost identical
couple, of which only the male is still in its original position on Nagauri hill (Shaw,
2000, Fig. 14), would have stood just metres from the water-body of reservoir 14
before the breaching of the dam. The close similarities of apparel and ornamentation
suggest a similar date to the Gulgaon couple. This suggests a terminal date of c. 50 BC
for the embankment, though an earlier date cannot be ruled out. A previously
unrecorded naga, in the form of Balarama, a prototypical form of the Hindu God
Vishnu, stands on dam 5 at Chandna (Shaw & Sutcliffe, 2002, Fig. 13); the treatment
is similar to the Vidisha Kubera yaksha, datable to around the end of the first century
BC (Chandra, 1966), making it the earliest non-Buddhist sculpture in Vidisha’s rural
hinterland. Follow-up field study has shown that the facing is similar to Nagauri and
other sites, and the dam is probably as old.
Another previously unrecorded naga from a slightly later period is found at Morel
Kala (dam 12; Shaw & Sutcliffe, 2001, Fig. 15); though only the lower half remains,
similarity with early Kushana sculptures from the Mathura region and in particular to
the Sarnath Bodhisattva (Williams, 1982, pl. 6), suggests a date of c. second century
AD. This comparatively late date accords with the advanced design and appearance of
the dam, which at 6 m is the highest in the region. A previously unrecorded threeheaded nagakal or snake deity in animal form (Fig. 6) near dam 7 at Devrajpur is
assignable to around the fifth century AD, though it is possible that the dam belongs to
a somewhat earlier period on the basis of the chronology of the nearby stupa and
settlement. In all, similar evidence was used to establish a tentative chronology for
seven of the dams (5, 7, 9, 10, 11, 12, 14). The remaining dams with no link with the
naga cult were positioned within the same framework on the basis of dam
morphology, stone-facing type and immediate archaeological context. For example,
four of the dams (1, 2, 8, 15) are directly associated with major settlements, clearly
datable to the early centuries BC. Another four (4, 6, 13, 16) are associated with early
monastic sites, which were established in c. second–first century BC during the second
major phase of Buddhist propagation in the area. These dates clearly require
corroboration through excavation, but the provisional chronology presented in Table 2
shows that the majority appear to belong to the c. second century BC time bracket.
The evidence for dating the continued use of the dams is more sparse, and is the
subject of on-going research. It seems that flood damage rather than sedimentation was
the likely physical cause of disuse. Other reasons are provided by evidence for radical
depopulation following a series of recorded famines from the 14th century onward.
286
Julia Shaw & John Sutcliffe
Fig. 6 Devrajpur Dam: Nagakal in animal form (after Shaw & Sutcliffe, 2001).
The earlier decline of Buddhism in the area, possibly around the 10th century AD,
must also have had a devastating effect on the local economy; just as, by extension, the
construction of the dams cannot be divorced from the establishment of Buddhism and
the acceleration of urban development suggested by the archaeological patterns in the
Sanchi area (Shaw, 2001). However, limited archaeological evidence for later use of
dams is provided by fragments of 10/11th century Hindu temples, which could have
been destroyed during Muslim invasions of 13/14th centuries, which were observed in
hydraulic structures at the Morel Kala dam (no. 12).
RESERVOIR AREAS AND VOLUMES
In order to deduce the roles of the reservoirs, their areas and volumes were estimated
by field measurement and map work, though precise estimates would require a number
of detailed contour surveys (e.g. Meigh, 1995). The heights (h) of the dams were
measured, with an allowance of 20 cm added to allow for erosion for well-preserved
sites and higher allowances for others. Dam lengths (l) were also measured in the field;
these were clear as dams were built between hills. The gradients ( ) of the reservoir
floors were estimated from 1:50 000 contour maps, although the contour intervals were
relatively coarse at 20 m. The gross volumes of the reservoirs were estimated from
these three dimensions (h, l, ) using a simple model of the reservoir shape. The
Ancient dams, settlement archaeology and Buddhist propagation in central India
287
288
Julia Shaw & John Sutcliffe
reservoir outline was assumed to be rectangular, with the width corresponding to l and
the length to h/ ; the area is lh/ and the mean depth is h/2 and the volume lh2/2 .
Alternative assumptions include an isosceles triangle with the dam forming the base,
with a mean depth of 2h/3 but a smaller area; a triangle with the dam at the apex would
have a mean depth of h/3 but a larger area. The rectangular assumption provides a
consistent set of estimates which lies between the two extremes. The dimensions of the
embankments are listed in Table 2, and the calculated areas and volumes are included
in the table.
In some cases, the assumptions were adjusted to take account of local conditions.
In dams 5 and 13, the reservoirs had been restored and the areas were estimated
directly from maps. A number of tributaries converge at dam 9, and the estimated area
took account of this. As expected, the relationship is close between reservoir areas and
volumes, expressed by V = 1.81A1.28, with the volume V in m3 106 and the area A in
km2. This is similar to the relationship derived from surveyed small dams in Botswana
(Meigh, 1995), which was V = 2.35A1.25.
RESERVOIR INFLOW
In order to estimate the inflows to the reservoir sites, the areas of the upstream basins
were measured from 1:50 000 maps and are included in Table 2. As the reservoirs are
confined to a small area around Vidisha, where the present 50-year average annual
rainfall is 1334 mm, with 1230 mm in the monsoon months June–September, this may
be taken as typical of the area and corresponds to an average runoff depth of 500 mm
(Sutcliffe & Green, 1981). The annual inflow volumes have been estimated from the
measured basin areas and this depth and are included in Table 2. These volumes are of
course based on modern rainfall and basin conditions, and would have been affected
by climate change. However, it is the correspondence between the two sets of volumes
which is important. The evidence shows that reservoir capacities are strongly related to
average inflows; consistent under- or over-estimation of either would not affect this
result.
Reservoir volumes and estimated runoff volumes are compared in Fig. 7, which
shows that the average runoff is sufficient to fill the reservoirs in nearly all cases. In
fact, the equation linking runoff volumes Q to reservoir volume V, both in m3 106, is
Q = 2.38V + 0.353, with a highly significant correlation coefficient (R2) of 0.792. This
implies that the runoff is sufficient to fill the reservoirs twice in an average year, or
that the average reservoir would be filled in nine out of 10 years if the runoff were
normally distributed with the present basin-wide CV of 45%. It is perhaps more
important to note that the reservoir sizes are reasonably well adapted to the drainage
basins and the available runoff. There is no evidence of serial construction of the
embankments, which suggests that they were constructed to an appropriate size with an
empirical understanding of the principles of water balance.
FLOOD CONTROL AND WATER DISTRIBUTION
Several of the Sanchi dams show evidence of spillways or control structures. Dam 7 at
Devrajpur has a spillway channel cut into the rock (Fig. 8) where the artificial
Ancient dams, settlement archaeology and Buddhist propagation in central India
289
9
8
7
6
5
4
3
2
1
0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Reservoir volume, m3x106
Fig. 7 Sanchi reservoirs: gross volumes and runoff (after Shaw & Sutcliffe, 2001).
Fig. 8 Devrajpur Dam: spillway (after Shaw & Sutcliffe, 2001).
embankment abuts on the original hillside. This is clearly intended as a relief channel
to prevent damage caused by overtopping during exceptional inflows, though it has not
prevented damage to the embankment at some period at the point where the natural
drainage channel has caused a breach. It is interesting to make an approximate
comparison between the spillway capacity and likely flood flows. An adaptation of the
290
Julia Shaw & John Sutcliffe
“rational formula” can make use of a map of India showing 24-h rainfall depths of
50-year return period (India Meteorological Department, 1972). The depth of 240 mm
over a basin of 13.5 km2, with an assumed 70% runoff coefficient and spread over 24 h
to allow for reservoir storage, would correspond to a discharge of 26.5 m3 s-1. With a
spillway of average width of 8.8 m, treated as a broad-crested weir, this discharge
would require a depth of 1.5 m, which is slightly more than the spillway depth of about
1.3 m. Thus the spillway would be likely to pass large but not extreme floods. The size
design of the reservoirs, the detailed masonry construction, and the existence of
spillways at the higher dams, indicate a level of sophistication comparable with contemporary hydraulic expertise on a larger scale in Sri Lanka, whose early Buddhist
history is closely linked to central India.
LAND USE AND IRRIGATION
If it is accepted that the size of the reservoirs implies that they were designed for
irrigation rather than domestic supply, it remains to speculate on the possible land-use
systems with which they were associated. The reservoir volumes suggest that they
were designed for the irrigation of rice, although this needs corroboration by specialist
archaeobotanical research.
The evidence of modern agriculture in the area makes it unlikely that irrigation
would have been required for wheat. This is because the storage in the black cotton
soils provides sufficient moisture for winter wheat to be grown without irrigation. At
the time of the Betwa project, the land use was largely based on autumn cultivation
of wheat, though some pulses and gram were also grown. Field investigation
(Hodnett & Bell, 1986) showed that the soil moisture storage, recharged each year by
the monsoon, was adequate, supplemented by winter rains, to provide water to the
crop, which was planted in October and harvested in March. The seasonal range of
soil moisture under crops was about 200 mm, and it was noted in 1979 that soil
moisture recharge occurred even in a year when monsoon rainfall was less than half
the average.
Because the yield of irrigated rice would be much greater than wheat or other
crops, it is interesting to consider the impact which the introduction of rice would
have had on the local economy. The dams would have played an important role in
supporting increased population levels during the early-historic period, which seem
to be implied by local settlement patterns and indeed the distribution of large
monastic sites. The relative positioning of dams and Buddhist monastic sites
throughout the area raises interesting questions. While Marshall (1940) was reluctant
to challenge the textual prohibition of monks owning or managing agricultural land,
on the basis of a single reservoir, the spatial relationships between the large number
of dams, monasteries and cult-spots throughout the Sanchi area provide an empirical
basis for considering the possible emergence of interactive support systems between
monastic and lay sections of society during the early centuries BC. The dams are
therefore part of a larger body of material for understanding the dynamics of social
and cultural change between c. third century BC and fifth century AD, and for
relating the spread of Buddhism to other key processes such as urbanization and
agrarian expansion.
Ancient dams, settlement archaeology and Buddhist propagation in central India
291
CONCLUSIONS
While these suggestions require further investigation, the study has demonstrated that
the ancient reservoirs in the Sanchi area were designed for irrigation, presumably of
rice. Their size is in harmony with local drainage basins and runoff volumes, while
some of the larger dams reveal an acquaintance with problems of flood control. This
suggests that they were built by engineers with experience of reservoir irrigation. The
deduction that they were built for rice irrigation throws light on the history of rice in
central India. Through an integrated study of irrigation works, land use, settlement
patterns and Buddhist monasteries, interesting questions have been raised regarding
the relations between monastic and lay sections of society.
REFERENCES
Chandra, P. (1966) Yaksha and yakshi images from Vidisa. Ars orientalis 6, 157–163.
Cunningham, A. (1854) Bhilsa Topes. Smith, Elder & Co., London, UK.
Fuller, D. Q. (2000) Fifty years of archaeological studies in India, laying a solid foundation. In: Indian Archaeology in
Retrospect, vol. III, Archaeology and Interactive Disciplines (ed. by S. Settar & R. Korisettar). Indian Council for
Historical Research, New Delhi, India.
Hodnett, M. G. & Bell, J. B. (1986) Soil moisture investigation of groundwater recharge through black cotton soils in
Madhya Pradesh, India. Hydrol. Sci. J. 31(3), 361–381.
India Meteorological Department (1972) Manual of Hydrometeorology. New Delhi, India.
Marshall, J. (1918) A Guide to Sanchi. Superintendent Government Printing, Calcutta, India.
Marshall, J. (1940) The Monuments of Sanchi. Repr., 1982, Swati Publications, London, Delhi.
Mehta, R. N. (1968) Sudarsana lake. J. Oriental Inst. Baroda 18, 20–28.
Meigh, J. R. (1995) The impact of small farm reservoirs on urban water supplies in Botswana. Natural Resources Forum
19(1), 71–93.
Parker, H. (1909) Ancient Ceylon. Repr. 1981, Asian Educational Services, New Delhi, India.
Shaw, J. (2000) The sacred landscape. In: Buddhist Reliquaries from Ancient India (ed. by M. Willis), 27–38. British
Museum Press, London, UK.
Shaw, J. (2001) The sacred geography of Sanchi hill: the archaeological setting of Buddhist monasteries in Central India.
PhD dissertation, University of Cambridge, Cambridge, UK.
Shaw, J. & Sutcliffe, J. V. (2001) Ancient irrigation works in the Sanchi area: an archaeological and hydrological
investigation. South Asian Studies 17, 55–75.
Singh, G. (1971) The Indus valley culture seen in the context of post-glacial climatic and ecological studies in northwest
India. Arch. Phys. Anthropology in Oceania 6(2), 177–189.
Sutcliffe, J. V. & Green, C. S. (1981) The Surface Hydrology of the Betwa Basin, India. Indo-British Betwa Groundwater
Project, Summary Tech. Report, vol. 2. Institute of Geological Sciences, London, UK.
Sutcliffe, J. V. & Green, C. S. (1986) Water balance investigation of recharge in Madhya Pradesh, India. Hydrol. Sci. J.
31(3), 383–394.
Sutcliffe, J. V., Agrawal, R. P. & Tucker, J. M. (1981) The water balance of the Betwa basin, India. Hydrol. Sci. Bull.
26(2), 149–158.
Venkayya, V. (1906) Irrigation in South India. Ancient Times, Arch. Survey of India, Annual Report 1903–1904, Calcutta,
India.
Williams, J. G. (1976) New naga images from the Sanchi area. Oriental Art 22, 174–179.
Williams, J. G. (1982) The Art of Gupta India: Empire and Province. Princeton University Press, Princeton, New Jersey,
USA.
Received 28 March 2002; accepted 11 November 2002