The Lost River: On The Trail of Saraswati

by Michel Danino

  Amalananda Ghosh was the first to test the waters. Before he was nominated director general of the ASI in 1953 (a post he held for fifteen years), he spent two winters exploring the valleys of the Sarasvatī and the Drishadvatī, as he called the Ghaggar and the Chautang respectively (Figs 6.5 and 6.6). Besides surface explorations, he conducted a few limited excavations. By Ghosh’s reckoning, the ‘sand-banks’ of the lower section of the Ghaggar were 5 to 10 km apart. His initial observation marks a watershed, figuratively as well as literally:

  In view of Stein’s statement which had led us to believe that nothing very ancient would be found in the region, it was a great thrill for us when even on the first and second days of our exploration we found sites with unmistakable affinities with the culture of Harappa and Mohenjo-daro. And a few subsequent days’ work convinced us that the Sarasvatī valley had been really a commingling of many rivers, not only geographically, but culturally.28

  Ghosh’s thrill is understandable: the Indus civilization did extend into the new Indian nation. But how far? In all, he identified no less than 100 sites, of which 25 displayed a true Harappan culture, the easternmost being Kalibangan (Fig. 6.1). He suspected, rightly, that more sites were bound to come to light further east, and concluded in the meantime that ‘the valleys of the Sarasvatī and the Drishadvatī must be regarded as very rich indeed in archaeological remains’.29 The other 75-odd sites belonged to later cultures, mostly the Painted Grey Ware and the Rangmahal. We have met the second, which belongs to the first centuries CE; as regards the first, abbreviated as PGW in archaeological literature, it was a village-based culture extending from Punjab to Uttar Pradesh and dating from the late second millennium BCE, at the start of the Iron Age.

  Ghosh ended his report with an important reflection on the flow of the bygone river: the Harappan sites ‘on the bank of the Sarasvatī’ could not have been established there ‘had the river been dead during the lifetime of the culture’.30 He was the first archaeologist to identify Harappan sites on the Indian side of the ‘Sarasvatī Valley’, as he called it, and also the first to assert that the river must have been in flow while those sites were thriving.

  At this point, further discoveries became inevitable. In 1960, Suraj Bhan began a survey of Haryana—that is, further upstream in the valleys of the Sarasvatī and the Drishadvatī, as he himself called those two dry beds.31 Apart from working out a chronology of the phases involved, Bhan discovered many new sites, including Rakhigarhi, Siswal, Mitathal and Balu (he was joined by Jim Shaffer for a season in 1977). In addition, between the present Yamuna and the Chautang’s bed, he identified three palaeochannels—as many clues to the eastward migration of the Yamunā.32 Further explorations in Punjab and Haryana were conducted by K.N. Dikshit in 1963 and, from 1975 to 1980, by Jagat Pati Joshi, Madhu Bala and Jassu Ram, who added considerably to the list of new sites.33 Other explorers of the region include Katy F. Dalal, R.S. Bisht and V.S. Wakankar.

  Moving downstream, let us return to the Cholistan Desert, where Aurel Stein had spotted the first Harappan settlements along the Sarasvatī’s dry bed. Following in his footsteps, the Pakistani archaeologist Mohammad Rafique Mughal undertook a systematic exploration of this unrelenting expanse of sand dunes and scrub vegetation in 1974.34 Over four gruelling seasons, Mughal covered almost 500 km in Cholistan and came upon numerous spots strewn with potsherds and terracotta cakes, which testified to intense life and activity where, today, even goats and cattle find it hard to survive under the scorching sun. His discovery of 363 pre-urban, urban and post-urban sites of the Harappan tradition, combined with those on the Indian side, transformed the conventional picture of the Indus civilization forever.

  Of those 363 sites, ninety-nine belong to the ‘Hakra Ware’ phase, which is roughly dated 3800-3000 BCE and is regarded as preceding the Early Harappan phase (which is why these sites are not included in the tables below). In addition, Mughal identified fourteen sites of the post-Harappan PGW culture, the same that Ghosh had first spotted on the Indian site of the valley.

  A similar windfall awaited explorers in Gujarat. In 1954, the Indian archaeologist S.R. Rao identified a few Harappan settlements, including the well-known port town of Lothal, some 60 kilometres southwest of Ahmedabad, which he proceeded to excavate the next year. J.P. Joshi surveyed Kachchh and Saurashtra for a few years from 1964, discovering numerous settlements, including Dholavira (in 1966) and Surkotada. More sites came to light in the region through surveys by P.P. Pandya, Gregory Possehl and Kuldeep Bhan, among others.

  The sum total of the explorations was simply prodigious and could be ranked among the most important archaeological discoveries of the twentieth century, even though we rarely hear about it—especially in India, where most school textbooks describe the ‘Indus Valley’ civilization as it was known in the 1930s!

  If we take the Sarasvatī basin to mean the Ghaggar-Hakra and its tributaries, as did Stein and Ghosh and many of their followers, almost 2400 settlements related to the Indus civilization have come to light there: 640 Early, 360 Mature, and almost 1400 Late Harappan (Table 6.1), according to figures recently compiled by the late Indian archaeologist S.P. Gupta and his team.35 No one could have foreseen such a density of settlements 1000 km away from Mohenjo-daro.

  Table 6.1. Distribution of Harappan sites in the Sarasvatī basin (adapted from a list compiled by S.P. Gupta, with inputs from G. Possehl and M. Rafique Mughal).36

  Sarasvatī basin (east to west) Early Harappan Mature Harappan Late Harappan Total

  Haryana 558 114 1168 1840

  Indian Punjab 24 41 160 225

  Rajasthan 18 31 0 49

  Cholistan (Pakistan) 40 174 50 264

  Total 640 360 1378 2378

  But that was not all. Further east, over forty sites, three-fourths of them of the Mature phase, have come to light in the doab (‘two rivers’) or the interfluvial region between the Yamuna and the Ganges, which is today the western part of Uttar Pradesh. No Harappan site has so far been spotted east of the Ganges. In other words, the Ganges is the eastern boundary of the Indus civilization, just as the Tapti marks its southern boundary. (There are a few Late Harappan sites in Maharashtra’s Godavari basin, such as Daimabad, but none of the Mature phase sites has been identified so far south.)

  Gujarat also turned out to be dotted with Harappan sites—over 500 of them (mostly in Kachchh and Saurashtra), with more than 300 of the Mature phase, and very few of the Early phase.

  Put together, the explorations outside the Indus Valley proper have vastly expanded our Harappan horizons. The overall picture is summarized in Table 6.2, with a grand total of over 3700 sites—a long way from the forty-odd at the time of Partition.

  Table 6.2. Overall distribution of Harappan sites.37

  Regions of the Subcontinent Early Harappan Mature Harappan Late Harappan Total

  Sarasvatī basin (Table 6.1) 640 360 1378 2378

  Uttar Pradesh 2 32 10 44

  Himachal, Jammu and Delhi 1 - 4 5

  Gujarat 11 310 198 519

  Pakistan’s Indus basin and western regions ¶ 385 438 12 835

  Total 1039 1140 1602 3781

  A few caveats are in order at this point. Tables 6.1 and 6.2 do not give us a count of separate geographical sites: because a given site appears twice if it has two phases (say, Early and Mature), and three times if it has all three phases, the actual count of geographical sites is much less, somewhere between 2000 and 2500. There may also be errors: possible duplications apart, most sites have been identified through the method of surface collection, or at best a trial pit; as a result, some of the sites detected as ‘Late’ may conceal earlier phases. In addition, a small proportion of the settlements in densely populated areas are temporary or ‘camp sites’, to use Mughal’s term in the case of Cholistan,38 which may or may not qualify as genuine settlements. Lastly, while I have tried to include recent finds from India and Pakistan, some may have been left out, and new sites keep coming to light all the time.39The tabl
es should therefore be viewed as broadly indicative; the distribution patterns they point to are unlikely to be altered much, and those are more important than the numbers themselves.

  So what conclusions can we draw from this ‘mushrooming’ of Harappan sites?

  THE ‘INDUS-SARASVATĪ CIVILIZATION’

  The first is a touchy question of terminology: unquestionably, the Indus civilization is no longer restricted to the ‘Indus Valley’. If we limit ourselves to the Mature phase, Baluchistan alone has 129 sites against 108 in Sind, where Mohenjo-daro is located;40 Gujarat has 310, while the Sarasvatī basin has 360—four times as many as Sind. Table 6.3 shows the relative concentration of Mature sites in these four regions:

  Table and Chart 6.3. Region-wise distribution of Mature Harappan sites.

  Region No. of Mature Sites Percentage

  Sarasvatī basin (Haryana, Indian Punjab, north Rajasthan, Cholistan) 360 32%

  Gujarat 310 28%

  Baluchistan 129 11%

  Sind 108 9%

  Pakistan’s Punjab41 60 5%

  Others 173 15%

  Total 1140 100%

  The last column of Table 6.3 is eloquent: almost one-third of all Mature Harappan sites are located in the Sarasvatī basin, and over one-fourth in Gujarat alone: together, these two regions hold 60 per cent of all Mature sites. Clearly, the designation of ‘Indus civilization’ is no longer quite apt—much less is that of ‘Indus Valley civilization’. An easy way out is to opt for ‘Harappan civilization’, after the first identified site (or ‘type site’), and indeed the term, though a bit dated, is still used by many archaeologists; but it hardly helps us to grasp the civilization’s geographical extent, which happens to be one of its specificities.

  In 1998, Kenoyer gave a lucid synthesis of what we have explored so far; he noted that apart from the Indus,

  another ancient river, the Saraswati or Ghaggar-Hakra had taken its course along the eastern edge of the plain. Numerous surveys in the deserts of Cholistan and Rajasthan made it clear that large numbers of settlements dating from the fourth to the first millennium B.C. were situated along the banks of this other major river system . . . Now that we know of the presence of the ancient Saraswati river (also known as the Ghaggar-Hakra along its central stretches), some scholars refer to this culture as the Indus-Sarasvatī civilization.42

  Among the scholars in question, S.P. Gupta was the first, in 1989, to propose this new term. More recently, Jane McIntosh, whom we met earlier in connection with the peaceful character of the Harappan society, commented on the discoveries of Mughal and others in these words:

  This work revealed an incredibly dense concentration of sites, along the dried-up course of a river that could be identified as the ‘Saraswati’. . . Suddenly it became apparent that the ‘Indus’ Civilization was a misnomer—although the Indus had played a major role in the rise and development of the civilization, the ‘lost Saraswati’ River, judging by the density of settlement along its banks, had contributed an equal or greater part to its prosperity.

  This led McIntosh to the following conclusion on the issue of terminology:

  Many people today refer to this Early state as the ‘Indus-Saraswati Civilization’ and continuing references [in her book] to the ‘Indus Civilization’ should be seen as an abbreviation in which the ‘Saraswati’ is implied.43

  Despite such plain assessments of the share of the Sarasvatī region in the Harappan world, the designation ‘Indus-Sarasvatī civilization’ has not caught on. There are several reasons for this, apart from a predictable reluctance to alter a time-honoured terminology.

  The first is that the new term still does not include Baluchistan’s or Gujarat’s numerous sites (nor also, to be precise, those east of the Yamuna). Even with this restriction, ‘Indus-Sarasvatī’ is certainly more comprehensive and closer to the mark than ‘Indus’ alone.

  A second reason is an objection raised by some archaeologists that the high concentration of sites along the Sarasvatī and the small number along the Indus might be something of an optical illusion. As an example, Shereen Ratnagar, a professor of archaeology who has particularly researched and written on the Indus civilization, asserts that ‘fewer Harappan sites lie along the banks of the Ghaggar-Hakra than is made out . . . Where the Sarasvati Valley sites are concerned, we find that many of them are sites of a local culture . . . some of them showing Harappan contact, and comparatively few are full-fledged Mature Harappan sites’.44 But such a statement is questionable, since the said ‘local cultures’ can just as well be viewed as regional variations of the Harappan culture : in Baluchistan or Gujarat, archaeologists have long noted such distinctive regional stamps, yet never sought to exclude sites of these two regions from the Harappan sphere.

  In reality, there is little scope for disputing the identification of Harappan sites in the Sarasvatī basin by Ghosh, Joshi, Mughal or Bhan, unless we are prepared to question their professional competence, which none of their colleagues, to my knowledge, has ever done. If anything, their judgement has been confirmed wherever their surveys have been followed by actual excavations: Rakhigarhi, Banawali, Kalibangan, Lothal, Surkotada and Dholavira are shining examples of their surveying skills. Moreover, even if we assumed that dozens of the sites surveyed were erroneously labelled as Harappan (which is very unlikely), it would hardly make a difference to the overall numbers.

  Ratnagar further tries to tip the scales by proposing that ‘many sites near the Indus may have been washed away when the river flooded or changed its course’.45 That is a better point, and we have already noted the whims of the Indus; indeed, it is also fairly certain that the ruins of some sites must have been buried under the river’s sediments. But in the absence of evidence, their number must remain conjectural, and even if tens or hundreds of them were destroyed by the Indus, that would in no way obliterate the Sarasvatī sites: the term ‘Indus-Sarasvatī’, proposed by Gupta, noted by Kenoyer and accepted by McIntosh, by no means excludes ‘Indus’.

  Kenoyer’s remark on the ‘large numbers of settlements’ in the Sarasvatī region figured in his 1998 Ancient Cities of the Indus Valley Civilization, one of the best introductions to the field and a fine synthesis of Harappan society, technology and crafts. More recently, we see him cast doubt on those ‘large numbers’; writing in 2006, he suggests that ‘most sites along the dry river channel are relatively small, and even the few large ones are not as large as the major cities on the Indus or its tributaries’.46 Coming from such a seasoned archaeologist, this statement is surprising: according to Possehl’s recent figures,47 the average size of Mature Harappan sites in the Sarasvatī region is 13.5 ha, while that of the sites in Sind is 8 ha. If anything, it is the Indus sites that are ‘relatively small’. Rakhigarhi in Haryana, in the Sarasvatī basin, for example, has seven mounds spreading together over 105 ha.

  Even prior to the Mature phase, the presence in the Sarasvatī basin of 640 Early Harappan sites (nearly 63 per cent of all known sites of that phase, see Table 6.2), with at least four of them in the range of 20 ha,48 shows that the region was not ‘colonized’ at the start of the urban phase, but was part of the vast process of convergence that culminated in the rise of Harappan urbanism over much of northwest India.

  So why not use the term ‘Indus-Sarasvatī civilization’? The real reason is that, much like Gangā losing her way in Shiva’s hair, the Sarasvatī found herself entangled in the Aryan controversy. We will have to wait till the last part of this book to see whether the legendary river manages to flow out of it.

  Until then, let us not miss the central point of the momentous findings of the post-Partition era: although ‘Indus-Sarasvatī civilization’ suits the archaeological record better than ‘Indus civilization’, the issue of terminology is secondary and will, in time, settle on its own. What matters is that this civilization had not one, but several heartlands : Baluchistan, the Indus basin, the Sarasvatī basin and Gujarat.

  THE VERDICT OF ARCHAEOLOGY


  In the 1820s, we saw Tod record a tradition that blamed the Ghaggar’s disappearance for the region’s ‘depopulation’, a word also used by Colvin; the two Oldhams made similar observations. Tessitori sensed that ‘the Ghaggar once bathed with its waters a florid and prosperous region’. Those early explorers would have been delighted to see the findings of archaeology endorse their view. According to M. Rafique Mughal:

  Archaeological evidence demonstrates that the Hakra flood plain was densely populated between the fourth and the second millennia B.C.49

  And such a population density, in what is today a perfectly arid region barring a few wells along the Hakra’s bed, could not have been sustained ‘had the river been dead during the lifetime of the culture’, as we saw A. Ghosh put it. V.N. Misra, a distinguished archaeologist and prehistorian with a life-long experience of Rajasthan, summed up the verdict of archaeology in these words:

  The large number of protohistoric settlements, dating from c. 4000 BC to 1500 BC, could have flourished along this river only if it was flowing perennially.51

  This conclusion is obvious to most other archaeologists. M. Rafique Mughal, after reminding his readers that the Ghaggar-Hakra is ‘often identified with the sacred Sarasvatī River of the Vedic Aryans’, finds it ‘certain that in ancient times the Ghaggar-Hakra was a mighty river, flowing independently [of the Indus] along the fringes of the Rann of Kutch’.52

  However, Mughal disagrees that the Ghaggar-Hakra was perennial till 1500 BCE. We can understand the evolution of this ‘mighty river’ if we study the distribution patterns of sites: a simple look at Figs 6.7 and 6.8 shows that while Cholistan’s Early sites stretch all the way to the border, Mature sites appear to migrate to the southwest, as though the Hakra were no longer flowing near the border. Mughal therefore observes: