Seeds of urbanism: palaeoethnobotany and the Indus Civilization

STEVENWEBER*

Palaeoethnobotanical evidence reveals that there was increasing emphasis on greater varieties of species and cropping practices in the changing subsistence of the Indus civilization: agricultural intensification is discussed in relation to social and environmental changes.

Key-words: Indus, , Rojdi, botanical remains, food agriculture, cereals

Within the Harappan or Indus Valley Civiliza- persed throughout northwestern and Pa- tion located in northwest South Asia, a dra- kistan, ranging from village farming commu- matic shift towards more localized cultural units nities and small towns to several fully developed and away from urban complexes is thought to city complexes housing large populations, with have begun at the end of the 3rd millennium tens of thousands of people (FIGURE1). These Be. The reasons for decentralization and locali- larger communities had houses with uniform- zation are still debated, but explanations often sized bricks, granaries, massive city walls, gate- assume that a changing agricultural system was ways, and extensive areas of craft production a causal factor in these processes (Kenoyer 1991; (Kenoyer 1991; 1998; Possehl 1990; Jacobson Jarrige 1985; Possehll986). This paper will focus 1986). Many of the craft products were stand- on recently collected palaeoethnobotanical data ardized and distributed throughout the Indus from two sites, Harappa and Rojdi, which allow region. The subsistence system consisted of food us to examine changes in agriculture during this production with domesticated and ani- critical time, and to understand how these changes mals, some hunting, fishing, and wild are related to shifts in the socio-economic struc- gathering (Meadow 1991; 1996).Supplying the ture. Analysis suggests that while the types of demand for raw materials, food and finished plants at the two sites differed, the agricultural products was a major mechanism in integrat- strategies in both places were changing in simi- ing the widely dispersed settlements (Kenoyer lar ways. The argument to be proposed is that 1991). Similarities in pottery, weights and seals the same socio-political and environmental is- are strong evidence for a shared ideology and sues were affecting inhabitants at both sites, and suggest the existence of an administrative sys- that even though Harappa and Rojdi represent tem to oversee the manufacturing and distri- very different types of occupation in different kinds bution of goods (Kenoyer 1991; 1998;Glover & of environment, their agricultural strategies were Ray 1994). nonetheless modified in a manner not unlike one About 2000 BC, cultural integration begins another. In turn, the way the agricultural strate- to break down. We see a rise in regional sys- gies change suggests that they were a result of tems that were no longer held together by a cultural change, and not its cause. single ideological or socio-economic system, associated with an increase in settlements and The Indus Civilization the abandonment of many larger urban sites At its height, around 2600 BC, the Indus Civili- [Shaffer 1993; Possehl 1990; Kenoyer 1991). zation included nearly a thousand sites dis- These fragmented, regional cultures, that are

* Department of Anthropology, Washington State University, 14204 NE Salmon Creek Avenue, Vancouver WA 98686- 9600, USA. weberQvancouver.wsu.edu Received 15 September 1998, revised 15 January 1998, accepted 18 February 1999, revised 4 March 1999. ANTIQUITY73 (1999): 813-26 814 STEVEN WEBER

FIGURE1. The location of Harappa, Rojdi and other Harappan sites in northwestern South Asia [Sources:Possehl8 Raval 1989; Weber 1991). SEEDS OF URBANISM: PALAEOETHNOBOTANY AND THE INDUS CIVILIZATION 815

El400 El600 El800 E2000 E2200 E2400 El600 E280O

N2000 NZWO

N1800 N18W

N16W N1600

N14W N1400

YI200 N1200

VIw(I NIOOO

YBOO El400 El600 El800 EZOOO E2200 E2400 E26W €2800

FIGLJRE2. Map of Harappa. {Map from HARP/dapted from Meadow et al. 1996.) referred to as Late Harappan, occur at a time in moisture patterns (Kenoyer 1991; Allchin when the inhabitants of the large urban cen- 1995; Chakrabarti 1997).These in turn may have tres lost control of the trade networks that had caused changes in river patterns, resulting in helped integrate this vast region (Kenoyer 1991; flooding and sedimentation. Crop failure would 1998; Glover & Ray 1994).Disruptions and shifts have been followed by settlement abandonment. in agricultural production play a prominent role Population dislocations, disrupted trade net- in explanatory models of the Late Harappan works and new agricultural strategies would (Possehl 1993). It is argued that there was a have then produced new, localized political decline in traditional crops which fed the large units (Kenoyer 1998). population centres, at the same time as the emergence of new agricultural techniques and The sites of Harappa and Rojdi crop plants that spurred the development of Harappa and Rojdi are sites that span the local, independent communities (Jarrige 1985; Harappan-Late Harappan transition and where Kenoyer 1998). Although explanations for these excavations were systematic, extensive and disruptions in the agricultural base tend to be interdisciplinary. Palaeoethnobotanical recon- regional in nature, they point to widespread struction from these sites allows us to compare causes such as tectonic movement or changes agricultural strategies from an urban and a ru- 816 STEVEN WEBER

APPROX SCALE

CULTWATION

FIGURE 3. Map of Rojdi. (Sources: Possehl b Raval1989; Weber 1991 .) ral community where the focus on particular space, through time and by context to counter- plants differed, yet where the same natural and act the inherent biases in the data. Although cultural processes that were affecting all of these two sites are our best source of consist- Harappan society were at work. Most existing ent and unbiased data, obviously common pat- reconstructions of Indus Civilization subsist- terns identified here will need to be tested, using ence draw on data from a number of sites that similar methods and techniques, at other sites. can be overlapped to produce an agricultural The site of Harappa (FIGURE2) was the first sequence (see: Costantini 1985; 1987; Kajale known city of the Indus Civilization, and cov- 1984; 1991; Meadow 1989; 1996; Jarrige 1985; ering 150 hectares, it remains one of the larg- Possehll986; Saraswat 1992; Vishnu-Mittre & est. It is one of only five large urban centres Savithri 1982; Lone et al. 1993). Unfortunately, identified within this civilization (Kenoyer this means incorporating data from different 1998).The site was linked to both near-by towns types of sites, excavated at different times us- and villages and the more distant cities by trade ing different methods, or where varied collec- and ideology (Kenoyer 1998). The occupation tion and analysis strategies were employed. at this site began at least as early as 3300 BC Harappa and Rojdi not only span the time pe- and continues at present (Kenoyer 1998; riod in question, but in both places a palaeo- Meadow, Kenoyer & Wright 1996). The most ethnobotanical research design was used that recent excavations at Harappa began in 1986 included intensive and systematic analysis of and continue today. Though these excavations the soil. In addition, quantitative techniques have exposed only a small portion of the set- have been used to track plant frequencies across tlement, all of the major mounded areas of the SEEDS OF URBANISM: PALAEOETHNOBOTANY AND THE INDUS CIVILIZATION 817 site have been trenched. Using modern tech- Rojdi lies in the southernmost region of this niques and methods, a significant amount of civilization. Its location along the Bhadar River, new information has been uncovered over these in the geographical centre of Saurashtra (, past 13 years that have provided new data on India) places it in a peripheral region of the its formation, character, evolution and decline Indus Civilization, a region based on summer (see: Meadow et al. 1994; 1995; 1996). crops such as (Weber 1991). Centrally located between two major tribu- Based on its size of 7 ha, and the kinds of taries of the , and based on a win- architectural features it exhibits, Rojdi was a ter cultivation strategy using wheat and barley, farming community, trading and interacting with Harappa served as a production hub and de- other sites in its region (Dhavalikar 1995). While pot for both local and regional trade items, in- some sites in Saurashtra, like (Rao 1979) cluding agricultural products. It lies in the and Rangpur (Rao 1963),contain artefacts and highlands, an environment in the northernmost demonstrate city planning that more clearly area of this civilization. With good rainfall and places them within the socio-political system fertile soils, this area can have abundant har- of this civilization, other sites are like Rojdi vests. Over time, and with growth, the inhab- with a greater mix of local and regional styles itants built new, walled neighbourhoods until, (see Possehl & Raval 1989). Based on the co- at its height, Harappa became a fully devel- existence of local and Harappan styles of ma- oped city complex, housing at points a popu- terial culture at these sites and others, there is lation in excess of 35,000 people (Kenoyer clear evidence that the Harappans established 1998). Based on city layout, the styles of control over parts of Gujarat (Dhavalikar 1995; painted ceramics, inscribed seals and weights, Allchin & Allchin 1982). Artefacts recovered the inhabitants of this site shared the same from Rojdi clearly demonstrate the inhabitants culture seen in other Indus civilization sites. were part of this larger regional culture, inter- Harappa is not only the ‘type’site of this civi- acting with other local and regional settlements. lization but it is one of the most important; Fortified cities like (Bisht 1989; 1991), any trends identified here have significance situated on a small island in the Rann of Kutch, for the culture as a whole. probably controlled the flow of goods between Due to the collection and dating of carbon peripheral regions like Saurashtra (including samples over the last 13years, a good chronol- sites like Rojdi) and the core areas of the Indus ogy now exists for Harappa. The first identi- plain (Kenoyer 1998). fied occupation at Harappa belongs to the Early All distinct areas of this walled town were Harappan Period from 3300 to 2600 BC. The trenched, producing a comprehensive database Harappan Period extends from 2600 to 1900 representing the full spatial and temporal range BC and can be subdivided into building phases of activities occurring at Rojdi. Three ceramic 3A (2600-2400 BC), 3B (2450-2200 BC), and 3C phases (A, B and C), were distinguished accord- (2200-1900 BC). This is followed by the Late ing to fabric, design and shape (Herman 1989). Harappan Period which lasts till about 1700 Over time, a decrease in the quality of the ce- BC. Changes in the artefacts, ecofacts and ar- ramic fabric (the appearance of thicker, cruder chitecture have been documented that support ware) is accompanied by an increase in the the perception that the Late Harappan period quality and refinement of the decorative ele- featured a more localized and less integrated ments, indicating greater local influence political-economic system (see: Kenoyer 1998; (Herman 1989).At the same time there is a se- Meadow et al. 1996). ries of occupations or architectural levels, each In contrast to Harappa, Rojdi was a town, with a range of structures, two for Rojdi A, two whose planned architecture and Harappan-like for Rojdi B and one for Rojdi C. Though there artefacts are found alongside localized charac- are some variations in the building style for teristics in its material remains (FIGURE3). The these levels, the overall pattern suggests an artefacts found at Rojdi are stylistically diver- expansion in the settlement toward the south gent from those representing other Harappan at the time of Rojdi C. regional populations, yet still part of the larger The shifts in architecture and ceramics cor- Harappan cultural whole (Possehl & Ravall989). respond nicely to the radiocarbon chronology Located in an arid, shrub savanna environment, developed for this site. Multiple carbon sam- a18 STEVEN WEBER

cropping tlement (see Chakrabarti 1997),it was occupied plant taxon season Harappa Rojdi during the time periods in question (including the transition to the Late period),the occupants cereals were interacting with inhabitants of other sites wheat (Triticum) W T- I - in the region, and both the artefactual and barley (Hordeum) W T- I T-I11 (Oryza) S T-111 - ecofactual records demonstrate significant shifts, just as can be identified at the same time at millets Har app a. Eleusine S T-TI T-I S T-I1 T-I Agricultural, change and the Setaria S - T-I archaeobotanical record Agricultural products are made by people who pulses and vegetables select land, crops and production strategies they peas perceive as appropriate (Hastorf 1993). These Pisum W T- [I T-I1 strategies can be reconstructed through proper Cicer W T-I1 - palaeoethnobotanical analysis (Pearsall 1989; Lathyrus W T-I1 T-I1 lentils (Lens) W T-I1 T-I1 Hastorf 1993; Gremillion 1997). gram Two agricultural strategies occur in South Dolichos S _. T-I1 Asia, and are believed to have been present in Phaseolus S -. T-I1 the Harappan civilization (Meadow 1989; 1991). Vigna S T-I1 T-I1 Each is represented, respectively, at Harappa Medicago S T-.[II T-I1 and Rojdi (TABLE1). One strategy involves crops sown in the autumn, harvested in the spring oilseed and fibre and fed with winter rains and is found in pre- linseed (Linum) W T-I1 T-I1 history in northwest South Asia (Baluchistan, mustard (Brassica) W T-I1 T-I11 Bannu Basin, Sind, Punjab, Swat, Kashmir). cotton (Gossypium) S T-I1 - Crops include barley, wheat, oats, peas, len- fruits tils, chickpea, jujube, mustard and grass pea. melon (Cucumis) S T-I11 T-I11 The second strategy centres on plants sown in date (Phoenix) S T.11 - summer and harvested in the fall, making use jujube (Zizyphus) W T-I1 T-I1 of summer monsoon rains. Prehistorically, the grape (Vitis) S T-I1 - summer sown plants (millets, sorghum, rice, cotton, dates and gram) were used most often W = winter/spring-harvested; S = summc!r/fall-harvested; in Gujarat and western India (Saurashtra, Kutch, T-I =tier I plants; T-I1 = tier I1 plants; T-111 = tier 111 plants. Rajasthan, Maharashtra). Cities in the Indus Valley, like Harappa, are TABLE1. Main cultivated plants recovered from Harappa and Rojdi. thought to have been based on the winter strategy (Meadow 1991; Kenoyer 1989). This strategy ples from secure contexts were collected from is the older of the two, based on local or West all areas of the site (Possehl & Ravall989). Rojdi Asian species and dependent on an irrigation was occupied from about 2500 to 1800 BC.Both system (Costantini 1990; Weber 1991; 1996). Rojdi A (2500-2300 BC) and Rojdi B (2300-2000 The summer strategy is, and was, based on a BC) fall into the Harappan Period, while Rojdi variety of millets, plants regarded as very hardy, C (2000-1700 BC) represents the Late Harappan drought-tolerant and capable of growing in poor Period. During the Late Harappan period, shifts soils (Weber 1998: 267). This was the promi- in architecture and ceramics point to an in- nent strategy in Gujarat and at sites like Rojdi. creased influence of local, non-Harappan ele- The summer strategy was developed later than ments at a time of settlement expansion and the winter strategy, and has been linked with population growth (Possehl & Raval 1989). pastoralism (Weber 1998: 270; Mehra 1997). As Most Indus civilization sites are small set- time passed, more sites in all areas incorporated tlements like Rojdi, rather than large ones like both strategies, although the emphasis was often Harappa. Although artefacts recovered from on the strategy better suited to the local environ- Rojdi suggest it is not a typical Harappan set- ment. SEEDS OF URBANISM: PALAEOETHNOBOTANY AND THE INDUS CIVILIZATION 819

At both Harappa and Rojdi, thousands of litres Harappa Rojdi of soil were systematically collected and floated from a variety of locations and features, yield- no. of samples analysed 66 284 ing tens of thousands of carbonized seeds, rep- no. of seeds identified 23,231 6256 resenting dozens of species (TABLE2). The no. of edible taxa identified 33 29 strategy was to sample as wide a variety of con- average seed density per litre of soil 40 5 texts as possible and to sample these contexts TABLE2. The archaeobotanical data base at multiple times. In order to maximize the value Harappa and Rojdi. of the archaeobotanical material while mini- mizing the inherent biases in this kind of data- intense the activity involving that type of plant base, samples were systematically collected and or the more the activity involves fire, the higher analysed from each phase and within each con- the density of seeds in the archaeobotanical text. If activities relating to the manipulation sample (Pearsall 1988; 1989; Sullivan 1987; of plant products can be assumed to have been Miller 1988). Harappa was the more substan- distributed systematically with respect to con- tial occupation and therefore more likely to text type, then the samples need to represent have had abundant, intensive and continu- as many different structures and features as ous trash deposition. Seeds at Harappa would possible (Hillman 1981). have experienced a more rapid burial, lead- Like artefacts, seeds can be identified, their ing to an enhancement of organic preserva- spatial and temporal distribution determined tion and concentrations of specific species. and their uses inferred. Combining existing This process may also account for why knowledge regarding archaeobotanical data with Harappa yields a greater variety of species, the context of the material to be analysed per- including most of those found at Rojdi as well mits the palaeoethnobotanist to determine if, as those specific to Harappa. how and when a plant was being used (Dennell At both Harappa and Rojdi the most com- 1974; 1976; Thomas 1983; Miller 1997; Hastorf mon crops were cereals, followed by pulses and & Popper 1988). With additional information vegetables, and finally oil-seed, fibre and fruit about plant morphology, place of origin, geog- plants. A three-tiered, hierarchical model of raphy and growing requirements, the palaeo- plant-use can be constructed for each site (TA- ethnobotanist can reconstruct what the BLE 1). The model can be imagined as an in- surrounding habitat may have looked like and verted pyramid (FIGURE4), with the topmost what plants might have been available to the ‘layer’representing the most abundant species, site’s inhabitants during each phase of occu- as well as those most critical to subsistence (re- pation (Doggett 1989; Harlan 1976, 1992; ferred to as Tier I plants). At Harappa, wheat Hillman & Davies 1990; Vavilov 1992).The basic and barley were the mainstay of the agricul- assumption is that when a large, systematically tural system and top the hierarchy. Grains of analysed sample that represents all occupations these winterlspring harvested cereals are found equally is used (as has been done at these two in nearly every sample, implying extensive use sites),then trends and regularities that are iden- of these species (TABLE3). On average they make tified for each period tend to reflect patterned up 80% of all recovered material from a given behaviour for that period. sample. Their relatively high abundance within The preservation of carbonized seeds and most assemblages suggests greater importance seed fragments has been very good at both sites. than other taxa. The second tier of the hierar- Seeds were recovered from most of the flota- chy at Harappa includes cultivated crops of tion samples represented in each database. lesser importance. These include lentils, linum, Fewer than 5% of the samples from either site grape, pea and grass pea, dates, zizyphus, cot- failed to contain seeds. The average density of ton and some millets. The third level or Tier seeds at the two sites is significantly different 111 plants consists mostly of wild species with (TABLE2). At Harappa, there was an average of a few minor crops like gram, rice and melon. nearly 40 seeds per litre of soil, eight times that While rice phytoliths and a few carbonized rice of Rojdi, with an average of 5 seeds per litre of grains have been recovered from each occupa- soil. Density often reflects the intensity of ac- tion at Harappa, rice appears to have played a cidents leading to seed preservation. The more minor role in their agricultural strategy. Rice 820 STEVEN WEBER

TIER-I PLANTS: Most abundant species recovered and those most crucial to the subsistence system.

TIER-I1 PLANTS: Cultivated crops of lesser importance.

FIGURE4. Three-tiered TIER-I11 PLANTS: model demonstrating A mix of cultivated the importance of food plants in prehistoric South subsistence. Asia. See TABLE1 for reference to species that represent these categories for V Harappa and Rojdi. cropping was probably not well suited to the tinuity in the cropping strategy for their par- environment surrounding Harappa, just as it ticular region over the last 4000 or so years. is not the prominent plant cropped today. This is not to say that the dietary system has At Rojdi, millets are the dominant cereals not undergone some change. Both sites show a and are at the top of the hierarchy. These sum- continuous process of incorporating new spe- mer/fall harvested plants were recovered in over cies into the cropping system, but new species half of the samples and make up the largest neither suddenly appear in large numbers, nor category of crop plants [TABLE1). Barley was do they significantly alter the existing subsist- less important and takes its place in the third ence strategy (Weber 1998). Whether it is the tier. No wheat or rice was recovered from Rojdi. summer-cropped plants being incorporated into As a whole, the second and third tiers at Rojdi the system at Harappa [e.g. gram, linseed and are similar to those at Harappa, and involve millets), or the winter-cropped plants at Rojdi most of the same taxa. A multi-cropping strat- [e.g. peas and lentils), all initially played a minor egy is discernible at each site from the earliest role in the dietary system. Over time, most of occupation, although the emphasis was on the the new species increase in number, implying season that better suited their particular envi- a greater role in the diet, yet the process seems ronment, and multi-cropping intensifies over to have been gradual. Further, existing species the course of each occupation. are not greatly reduced with the introduction This inverted pyramid model also works for of these new crops. There seems to be an over- the agricultural practices found in each region all pattern throughout the Indus Civilization historically. In the Punjab, where Harappa is toward gradually broadening and intensifying the located, and in Gujarat, where Rojdi is located, agricultural system by cropping plants more in- the types and proportions of species have re- tensively throughout the year and by using a greater mained fairly stable over time with little vari- variety of plants during any given season. ation from what has been identified during While neither site shows any sudden or rapid Harappan times. Both Rojdi and Harappa dem- change in its use of plants, some significant onstrate that there has been a great deal of con- changes were taking place during the transi- SEEDS OF URBANISM: PALAEOETHNOBOTANY AND THE INDUS CIVILIZATION 821 Harappan Period Late Harappan Period Harappa Rojdi Harappa Rojdi

ubiquity wheat (Triticum) 85 - 90 - barley (Hordeum) 85 - 88 - millets Eleusine Panicum Setaria density wheat (Triticum) barley (Hordeum) millets Eleusine - 6 - 0.1 Panicum - 0.8 - 0.1 TABLE3. Crop Setaria - 0.1 - 0.9 production of the main cereals from the relative percentage Tier-f plant category wheat (Triticum) 44 - 34 - at Harappa and barley (Hordeum) 35 - 41 - Rojdi. The table millets presents the shifts in Eleusine plant occurrence Panicum within this category Setaria during the transition from the Harappan Ubiquity is the percentage of samples from which a specific taxon was recovered. period to the Late Density is the number of seeds per litre of soil. Percentage is the relative abundance of a Harappan period. specific taxon in a given assemblage. tion to the Late Harappan Period. A similar age of food crops relative to other kinds of plants percentage decline in seed density occurs at declines nearly 5% during the Late Period (TA- both sites, with a 62% decline at Harappa and BLE 3). At both Rojdi and Harappa, a signifi- a 70% decline at Rojdi. This is closely related cant shift from one existing taxon to another to the changing richness of edible plants found within the cereals category can be identified at each site. Richness is the number of differ- as occurring during the transition to the Late ent edible plant taxa found in a flotation sam- period. It is worth noting that at neither site ple and hence measures the diversity of food does rice, a crop long associated with the Late plants (Rocek 1995). At Harappa and at Rojdi, Harappan period, become a prominent plant the overall average was almost three times as (see Kenoyer 1998). Rice can be found quite many edible taxa per total number of edible early at a number of sites, including Harappa. plant fragments recovered per flotation sam- But it is never found in large quantities, even ples during the Late Harappan Period than the at Lothal (Rao 1963) and Rangpur (Rao 1979), earlier periods. A wider range of crops was being suggesting it played only a minor role in the cultivated at both sites during the Late Period. subsistence system of all regions (Weber 1991). If we focus on only the main cereal grains The shift to rice as a primary food grain, mainly found at each site, those taxa that account for found in environments associated with sum- the majority of the recovered seeds in most mer monsoon rains, occurs after the beginning samples and top the hierarchy in each model, of the onset of the Late period (see: Jarrige 1985; an interesting pattern emerges (TABLE3). At Weber 1992; Kenoyer 1998). Harappa, the wheat and barleys account for over At Harappa the shift from one taxon to an- 70% of the recovered food plants. At Rojdi, the other is seen in the wheat-barley record. There millets account for over half of the recovered is a gradual increase in the number of wheat crop plants. At both sites these cereal crops and barley species cultivated at Harappa, but remain relatively stable, though the percent- the majority of the wheats are consistently shot 822 STEVEN WEBER wheat (Triticum sphaerococcum) and bread or have drastically altered the way people lived, club wheat (T. aestivum/compactum), while driven them to another area or dramatically most of the barleys are either 6-row naked ‘shot’ affected access to productive resources. Climate barley (Hordeum sphaerococcum) or 6-row and environment of the subcontinent during hulled barley (H. vulgare). At the earliest oc- the Harappan times have drawn interest and cupation, barley is the dominant grain. During provoked controversy since the earliest exca- the Harappan Period, wheat increases in ubiq- vations in the Indus Valley (Misra 1984; Misra uity, density and percentage until it becomes the & Rajguru 1989; Possehl 1997). Environmen- most common species at Harappa (TABLE3). Yet tal change has played an important part in many during the Late Period, wheat declines and bar- theories about Harappan society (e.g. Raikes & ley once again becomes the dominant cereal. Dyson 1961; Possehl 1986; 1997). Based on a At Rojdi, like Harappa, there is an increas- combination of palynological lake-bed studies ing variety of cereal grains, though in this case (Singh 1971; Singh et al. 1972; Singh eta1.1990) it is millets (Table 3). Many of these species - and ancient vegetation reconstruction based on like Sorghum, which is introduced near the end charcoal analysis (Thiebault 1988; Seth 1978), of the Harappan occupation - never account it appears the vegetation was not that different for a significant portion of the archaeobotanical from today. The beginning of the 2nd millen- record (Weber 1998). In fact, three millets, fin- nium may have ushered in an overall drying ger (), little millet trend in South Asia (Possehl 1997), although (Panicum sumatrense) and foxtail millet (Se- there are indications that some regions may have taria italica), make up the majority of the food experienced an increase in the amount of rain grains representing both the Harappan and Late in the summer months (Kenoyer 1991; Weber Harappan Periods. Yet such measurements as 1992; 1993). Although there is no firm agree- ubiquity, density and percentage of seeds per ment on what the climate was like, or how it sample all show a shifting pattern of occurrence might have changed during the course of the (see Weber 1991 for a detailed account of Rojdi Indus civilization, there does seem to be a con- archaeobotanical record). Finger millet and little sensus that the decentralization and localiza- millet, which figure prominently during the tion of Harappan society displayed in the Late Harappan Period, decline in the Late Period, Harappan Period were closely related to changes while foxtail millet is the dominant plant in going on in the environment - the results of this Late Period but appears in very small num- both natural events and human activity (Shaffer bers in the earlier periods. There is clear evidence 1993; Weber 1996;Possehll993; 1997; Kenoyer that, as in the case of Harappa, a significant shift 1998).A combination of shifts in river systems in plant usage or crop-processing activiteis oc- and in climate led to many Late Harappan settle- curred around 2000 BC amongst existing species. ments being established along newly stabilized Continuity and change in the agricultural river systems (Kenoyer 1991). Yet these new set- strategy are patterned similarly at each site. Is tlements depended on crop plants already incor- this similarity related to the breakdown in the porated into their agricultural strategies. socio-political system that led to the decentrali- While environmental conditions may have zation seen in the archaeological record at had a direct impact on Harappan settlement around 2000 BC? And if it is related, how is it systems, their effects on agricultural produc- related? tion at Harappa and Rojdi are less evident. All the shifts in plant use at these sites are toward Explanations for continuity and change species that grow in similar environments. No Similar shifts toward agricultural intensifica- significant climatic shift is necessary to explain tion at both Harappa and Rojdi, at about the them. Although the emphasis on certain plants same time, suggest that region-wide events were changes over time, at neither site do plants dis- at work. A common cause of agricultural shifts, appear from the archaeobotanical record, im- one that can affect all aspects of society, is change plying that whatever was acquired or used was in the climate and environment. Climate, for kept in the dietary repertoire. This tendency example, influences site location and tends to may reflect a style of decision-making particu- delimit the parameters of subsistence and set- larly suited to matters of agriculture, in that tlement patterns. Any change in climate could once a plant is added to the cropping system, SEEDS OF URBANISM: PALAEOETHNOBOTANY AND THE INDUS CIVILIZATION 823 especially in the marginal environments that being brought in to Harappa were more likely abound this civilization, it is unlikely to be for human consumption than for fodder. These removed. Changes in agricultural strategies and grains were probably stored centrally and then patterns of plant use, while subject to climatic redistributed to the inhabitants of the city un- and hydrological constraints, are generally too der the direction of 6lites. complex to be reduced to shifts in the envi- In contrast, the agricultural strategy during ronment. An increase in weedy species and a the Late Period at Harappa involved a greater decline in crop-seed density may be more closely variety of plants at a time when fodder crops related to an increase in population and a denser like barley were increasingly important. There- network of villages and towns, producing some fore, the transition to the Late Period may have form of environmental degradation, like over- been associated with the increasing importance grazing, overwatering of fields and deforestation. of cattle, or the practice of herding closer to The cereals found at both sites were being the city, or even, possibly, increasing use of dung grown for both grain (for people) and straw (for for fuel, a practice that often leads to the char- animals), and are the key to understanding ring of seeds in archaeological contexts (Miller changing agricultural strategies. The environ- & Smart 1984). ment constrains which crops are grown and Similarly, the choice in millets at Rojdi may when they are planted, but it is the cultural be related to increasing importance of fodder factors leading to crop choice that are more likely plants. Finger millets, which are so common to provide an explanation of changes in plant at the early levels of Rojdi, are more often grown occurrence at each site. for human use, while foxtail millet, the domi- Barley is a more drought-resistant plant need- nant grain of the Late Harappan Period, is com- ing less water than wheat, it has a shorter grow- monly grown as a fodder. Samples from the early ing season and can grow in a more saline soil levels contain mostly cleaned charred seeds, (Miller 1986; 1997). Further, barley has softer which are often taken to represent a single food straw, and requires milling to remove the husks, plant, with a few wild, weedy species mixed meaning that it is better suited for animal fod- in. This seems to indicate that their deposition der than for human consumption (Miller 1997). was related to food processing activities. In When given a choice, people in South Asia tra- contrast, samples from the Late Harappan Pe- ditionally grow wheat or rice for human con- riod show an increase of wild grasses mixed sumption and barley for fodder. Wheat was most in with the millets and a greater variety of taxa, common at Harappa between 2600 and 2000 all suggesting that these grains were being grown BC, during the period of integration. This was as a fodder and entered the site as part of a when the city was most densely populated and fuel. At Rojdi there appears to be an increase when agricultural and herding activities may in both cattle and in agricultural activity, evi- have been occurring farther from the site. In dent in the fact that more nearby land was dis- fact, there may have been greater reliance on turbed. During the shift to the Late Harappan food products coming in from outlying areas Period, we see far less arboreal pollen (imply- just as there was on long-distance trade goods. ing fewer trees), less charred wood in hearths, Storage areas like the central ‘granary’ found more weedy species, more crop variety and an at Harappa would have been in use at this time increased dependence on millet-cropping (We- (see Vats 1940). In an urban setting, with more her 1991). people living in close proximity to one another, There is good evidence that over-exploita- agricultural fields may have been located far- tion of resources was affecting the landscape ther from the site. Settlements in areas like and environment of Rojdi, and possibly even Cholistan along the ancient Hakra River (FIG- Harappa. Yet there is no evidence for any dra- URE 1) may have contributed grains of wheat matic or sudden shift in plant occurrence or and barley to the large urban sites like Mohenjo- use. Rather than being a direct cause of social Daro, Ganweriwala and Harappa (Kenoyer 1991; change as proposed by some scholars (see Mughal 1997). Because greater distances were Wheeler 1953; Fairservice 1971),it seems more involved, fewer varieties of plants were being likely that human-induced degradation to the exploited, or at least fewer varieties were be- environment simply increased the rate at which ing brought into the city. The grains that were the local habitat was being exploited. With 824 STEVEN WEBER nearly 6600 sq. km of available farm land around agricultural activities would more intensively Harappa (Fentress 1982) and a decrease in trade have affected the local environment. Smaller or introduced food goods during the Late pe- communities, less dependent upon imported riod, more nearby land would have been ex- goods, would probably have been more suc- ploited. Over time, though after the transition cessful. to the Late period, this would dramatically af- fect the landscape and any cultures residing Conclusion: a model for change in the area. Just as we see the culturally integrated Harappan Of course, the patterns I have identified at civilization fragmenting into decentralized re- each of these sites may only coincidentally seem gional cultures, the sites of Harappa and Rojdi similar, and by themselves only represent lo- are demonstrating change of their own. Not only cal processes. But when the data-sets are viewed can shifts in the material record and settlement together, a case for region-wide patterns is data be identified at these sites, but agricul- strongly suggested. Similar patterns of change tural strategies can also be seen to be evolving. at two far-flung Harappan sites, involving dif- While the palaeoethnobotanical data suggests ferent plant species, in contrasting types of that there was a long history for multi-crop- environments and representing such different ping involving the same species found in each kinds of communities, is a circumstance that respective area today (Weber 1992; Kajale 1991), should not be ignored, but tested and debated. there was a continuous, though gradual, effort The data from these sites suggest that changes to broaden and intensify the cropping strategy in the agricultural system that occurred around through an increased dependence on more spe- 2000 BC were similar throughout this civiliza- cies grown more regularly throughout the year. tion, regardless of the region in which settle- While there was a greater variety of species being ments were located, or whether they were urban grown, the intensity of use of most crops was or rural. At both sites, shifts in cereal use were declining. The most significant change in crop associated with continued efforts at broaden- use occurred within the cereals category, con- ing and intensifying agricultural strategies. sidered Tier I plants. Rather than being associ- While the identification of a pattern involving ated with a major environmental or climatic agricultural intensification has been developed event where survival was the underlying goal, for the Kachi Plain, where a sequence from the changing proportions of existing species in the sites of , Nausharo and Pirak shows archaeobotanical record reflect the kinds of a shift toward multi-cropping in the Late Pe- choices people make as a result of an altered riod (Jarrige 1985; Kenoyer 1998; Chakrabarti socio-economic situation. The data from these 1997), it is only from the sites of Harappa and two sites suggest that the agricultural strategy Rojdi that shifting dependence of existing crops in the Late Harappan period was one better has been identified. This may be due to the suited for local consumption needs and less kinds of analysis being employed at different based on regional systems of the Harappan sites. Crop choice at Harappa and Rojdi inclined period. While changes in agricultural produc- in all periods toward species better suited for tion and cropping strategies are linked to cor- their environment. In the Late Harappan pe- responding shifts seen in the material record riod, as the socio-economic system became less at both sites, these changes should be seen as centralized, a greater range of economic, food- an indicator of, rather than as an instigator for, oriented activities occurred in each commu- cultural change. While the trends identified here nity. Though there is at present little are based on only two sites, they demonstrate, archaeological evidence for patterns of grain first, the complexity of the agricultural strate- storage and redistribution, it is predicted that gies employed by the Harappans and, second, further excavation will reveal they were also that significant though subtle changes were changing at this time. A shift from large cen- occurring during the transition to the Late pe- tral storage facilities to the more traditional mud- riod. It is only through continued, careful work plastered bins associated with individual houses at these sites and others like them, that this would likely be associated with an increasing proposal of shifting crop choice and its emphasis on local farming in the Late period. relatonship to cultural and environmental In the end, the increase in local herding and change can be understood and better explained. SEEDS OF URBANISM: PALAEOETHNOBOTANY AND THE INDUS CIVILIZATION 825

References ALLCHIN.F.R. (ed.) 1995. The archaeo1obyofEarlyHistoric South JARRIGE, J.F. 1985. Continuity and change in the North Kachi Asia. Cambridge: Cambridge University Press. Plain at the beginning of the second millennium BC, in J. ALLCHIN,B. & R. ALLCHIN. 1982. The rise of civilization In In- Schotsrnans & M. Taddei (ed.), Sauth Asian Archaeology dia and Pakistan. Cambridge: Cambridge University Press. 2983: 35-68. Naples: IsMEO. BISHT, R.S. 1989. The Harappan colonization of the Kutch: An KAJALE,M.D. 1984. Ancient grains from India, Bulletin of the ergonomic study with reference to Dholavira and Deccan College Research Institute 34: 55-74. Surkotada, in K. Deva & L. Gopal (ed.), History and arts: 1991. Current status ofIndian palaeoethnobotany: introduced 267-72. Delhi: Ramanand Vidya Bhavan. and indigenous food plants with a discussion of the his- 1991. Dholavira: New horizons of the Indus Civilization, torical and evolutionary development of Indian agricul- Puratattva 20: 71-82. ture and agricultural systems in general, in J.M. Renfrew CHAKRABARTI,D.K. 1988. Theoretical issues in Indian archae- (ed.], New light on early farming: recent developments ology. Munshiram: New Delhi. in pafaeoethnobofany 155-89. Edinburgh: Edinburgh Uni- 1997. The archaeology of ancient Indian Cities. Delhi: Ox- versity Press. ford University Press. KENOYER,J.M. (ed.). 1989. Oldproblems and newperspectives COSTANTINI,L. 1985. Agriculture in Baluchistan between the in the archaeology of South Asia. Madison: Prehistoric 7th and 3rd millennium BC, Newsletter of Baluchiston Press. Wisconsin Archaeological Reports. Studies 2:16-30. 1991. The Indus Valley tradition of Pakistan and Western 1987. Vegetal remains, in G. Stacul (ed.), Prehistoric and India, Journal of World Prehistory 5: 1-64. Protohistoric Swat, Pakistan (c.3000-1400 BC), Appen- 1998. Ancient cities ofthe Indus Civilization. Karachi: Ox- dix B: 155-65. Rome: IsMEO. ford University Press. 1990. Harappan agriculture in Pakistan: the evidence of LONE,F.A., M. KHAN& G.M. BUTH.1993. Palaeoethnobotany: Nausharo. in M. Taddei led.), Sauth Asian Archaeology plants and ancient man in Kashmir. New Delhi: Oxford 1987: 321-32. Rome: IsMEO. IBH. DENNELL,R.W. 1974. Prehistoric crop processing activities, MEADOW,R.H. 1989. Continuity and change in the agriculture Journal of Archaeological Science 1: 275-84. of the Greater Indus Valley: the palaeoethnohotanical and 1976. The economic importance of plant resources repre- zooarchaeological evidence, in Kenoyer (ed.): 61-74. sented on archaeological sites, Journal of Archaeologi- 1991. The domestication and exploitation of plants and cal Science 3: 229-47. animals in the Greater Indus Valley, 7th-2nd millennium DHAVALIKAR,M.K. 1995. Cultural imperialism: Indus Civiliza- BC, in M.M. Jansen & G. Urban (ed.1, Forgotten cities on tion in Western India. New Delhi: Books & Books. the Indus: 51-8. Mainz: Philipp Von Zabern. DOGGETT,H. 1989. A suggested history of the crops common 1996. The origins and spread of agriculture and pastoralism in to Ethiopia and India, in L.Krzyaniak & M. Kobusiewicz northwestern South Asia, in D.R. Harris (ed.), Agriculture (ed.),Late prehistory of the Nile Basin and the Sahara, and pastoralism in Eurasia: 390-412. London: Institute of Poznan Studies in African Archaeology 2: 27-47. Archaeology, University College London. FAIRSERVIS,W.A. 1971. The roots of Indian civilization. Lon- MEADOW,R.H., J. M. KENOYER& R.P. WRIGHT.1994. Harappa don: Allen & Unwin. Archaeological Research Project: Harappa excavations FENTRESS,M. 1982. From Jhelum to Yamuna: city and settle- 1994. Report Submitted to the Director-General of Ar- ment in the second and third millennium BC, in G. Possehl chaeology and Museums, Government of Pakistan, Karachi. (ed.),Harappan civilization: 245-60. Delhi: Oxford IBH. 1995. Harappa Archaeological Research Project: Harappa GLOVER,I.C. & H.P. RAY.1994. Civilizations in Southern Asia, Excavations 1995. Report Submitted to the Director-General in G. Burenhult (ed.), Old World civilizations: The rise of Archaeology and Museums, Government of Pakistan, of cities and states: 61-76. San Francisco: HarperCollins. Karachi. GREMILLION,K.J. (ed.). 1997. People, plants, and landscapes, 1996. Harappa Archaeological Research Project: Harappa studies in palaeoethnobotany. Tuscaloosa (AL): Univer- Excavations 1996. Report Submitted to the Director-General sity of Alabama Press. of Archaeology and Museums, Government of Pakistan, HARLAN,J.R. 1976. Plant and animal distribution in relation Karachi. to domestication, in early history of agriculture, Philo- MEHR4, K.L. 1997. Biodiversity and subsistence changes in India: sophical Transactions of the Royal Society, London - The Neolithic and Chalc,olithic Age, Asian Agri-History Botany 275: 13-26. l(2): 105-26. 1992. Indigenous African agriculture, in C.W. Cowan & P.J. MILLER,N.F. 1986. Ethnohotanical perspectives on the rural- Watson (ed.), The origins of agricuffure: an international urban connection. Paper presented at SAA 1986, New perspective: 59-70. Washington (DC): Smithsonian In- Orleans. stitution Press 1988. Ratios in palaeoethnohotanical analysis, in Hastorf & HASTORF, C. A. 1993. Agriculture and the onset of political Popper (ed.): 72-85. inequality before the Inka. Cambridge: Cambridge Uni- 1997. Farming and herding along the Euphrates: environmen- versity Press. tal constraint and cultural choice, in R. Zettler (ed.),MASCA HASTORF, C.A. 8r V.S. POPPER [ed.). 1988. Current Research Papers in Science and Archaeology 14: 70-78. palaeoethnobotany. Chicago (IL): University of Chicago MILLER,N.F. & T.L. SMART.1984. Intentional burning of dung Press. as fuel: a mechanism for the incorporation of charred HERMAN,C. 1989. Rojdi material culture: The Sorath Harappan seeds into the archaeological record, Journal of ceramics. in Possehl & Raval (ed.): 53-157. Ethnobiology 4(1): 15-28. HILLMAN,G.C. 1981. Reconstructing crop husbandry practices MISRA,V.N. 1984. Climate, a factor in the rise and fall of the from charred remains of plants, in R. Mercer (ed.), Farm- Indus Civilization: evidence from Rajasthan and beyond, ing practices in British prehistory: 123-62. Edinburgh: in B.B. La1 & S.P Gupta (ed.), Frontiers of the Indus civi- Edinburgh University Press. lization: 461-90. New Delhi: Books & Books. HILLMAN, G.C. & M.S. DAVIES.1990. Measured domestication MISRA,V.N. & S.N. RAJGURU.1989. Palaeoenvironment and rates in wild wheats and under primitive cultivation, and prehistory of theTharDesert, Rajasthan, India, inK. Frifelt their archaeological implications, Journal of World Pre- & P. Sorensen (ed.),South Asian archaeology 198.5: 296- history4: 157-222. 320. London: Curzon Press. JACOBSON, J. (ed.). 1986. Studies in the Archaeology of India MUCHAL,M.R. 1997. Ancient Cholistan: archaeology and ar- and Pakistan. New Delhi: Oxford IBH. chitecture. Lahore: Ferozsons. 826 STEVEN WEBER

PEARSALL,D.M. 1988. Interpreting the meaning of microremain west India, Archaeology and Physical Anthropology in abundance: the impact of source and context, in Current Oceania 6(2): 177-89. palaeoethbotany: analytical methods and cultural interpre- SINGH,G., R.D. JOSH1 & A.B. SINGH.1972. Stratigraphic and tations of archaeological plant remains, in Hastorf & Pop- radiocarbon evidence for the age and development of Three per (ed.):97-118. Chicago (IL): University of Chicago Press. Salt Lake deposits in Rajasthan, India, Quaternary Re- 1989. Palaeoethnobotany: A handbook of procedures. New search 2(24): 496-505. York (NY): Academic Press. SINGH,G., J.R. WASSON& D.P. AGRAWAL.1990. Vegetational POSSEHL,G.L. 1986. African millets in South Asian prehistory, and seasonal climatic changes since the Last Glacial in in Jacobson (ed.):237-56. New Delhi: Oxford IBH. the Thar Desert, NW India, Review of Palaeobotany and 1990. Revolution in the urban revolution: the emergence of Palynology 64: 351-8. Indus urbanization, Annual Review of Anthropology 19: SULLIVAN,A.P. 1987. Seeds of discontent,Journal ofEthnobiology 261-81. 7(2): 137-53. (Ed.). 1993. Harappan civilization. 2nd edition. New Delhi: THIEBAUL,S. 1988. Palaeoenvironment and ancient vegetation Oxford IBH. of Baluchistan based on charcoal analysis of archaeological 1997. Climate and the eclipse of the ancient cities of the sites, Proceeding of the Indian National Science Acad- Indus, in H.N. Dalfes, G. Kukla & H. Weiss [ed.), The third emy 54A: 501-9. millennium BC climate change and Old World collapse THOMAS, K.D. 1983. Agricultural and subsistence systems of Series I: Global environmental change vol. 49: 193-242. the third millennium BC in North-West Pakistan: a specu- Berlin: Springer. lative outline, in M. Jones (ed.),Integrating the subsist- POSSEHL,G.L. & M.H. RAVAL.1989. Harappan civilization and ence economy: 279-313. Oxford: British Archaeological Rojdi. New Delhi: Oxford IBH. Reports. International series 181. RAIKES, R. & R.H. DYSON. 1961. The prehistoric climate of VAVILOV,N.I. 1992. The origin andgeographyof cultivatedplants. Baluchistan and the Indus Valley, American Anthropologist Cambridge: Cambridge University Press. 63: 265-81. VATS, M.S. 1940. Excavations at Harappa. Delhi: Government RAO, S.R. 1963. Excavations at Rangpur and other explorations of India Press. in Gujarat, Ancient India 18-19: 5-207. VISHNU-MITTRE& R. SAVITHRE.1982. Food plants of the 1979. Lothal: a Harappan town, 1955-62, Memoirs of the Harappans, in G.L. Possehl [ed.), Harappan Civilization: Archaeological Survey of India 78: vol. I. 205-21. New Delhi: IBH Publishing. ROCEK,T.R. 1995. Sedentarization and agricultural dependence: WEBER,S.A. 1991. Plants and Harappan subsistence: an ex- perspectives from the Pithouse-to-Pueblu transition in the ample of stability and change from Rojdi. Boulder (CO): American Southwest, American Antiquity 60(2): 218-39. Westview Press. SARASWAT,K.S. 1992. Archaeobotanical remains in ancient 1992. South Asian archaeohotanical variability, in C. Jarrige cultural and socio-economical dynamics of the Indian [ed.), South Asian Archaeology 2989: 283-90. Madison subcontinent, Palaeobotanist 40: 514-45. (WI): Prehistory Press. Monographs in World Archaeol- SETH,S.K. 1978. The desiccation of the Thar Desert and its ogy 14. environs during the Protohistoric and historical periods, 1993. Changes in plant use at Rojdi: implications for Early in W.C. Brice (ed.), The snvironmentalhistoryofthe Near South Asian subsistence systems, in Possehl (ed.): 287- East and Middle East during the last Ice Age: 279-305. 93. London: Academic Press. 1996. Distinguishing change in the subsistence and the SHAFFER,J. 1993. Reurbanization: the Eastern Punjab and be- material records: the interplay of environment and cul- yond, in H. Spodek & D. Srinivasaii (ed.), Urban form ture, Asian Perspectives 35(2): 155-64. and meaning in South Asia: 53-67. Washington (DC): 1998. Out of Africa: the initial impact of millets in South National Gallery of Art. Asia, Current Anthropology 39(2): 267-74. SINGH,G. 1971. The Indus Valley Culture seen in the context WHEELER,M. 1953. The Indus Civilization. Cambridge: Cam- of Post-Glacial climatic and ecological studies in North- bridge University Press.