The Lost River: On The Trail of Saraswati

by Michel Danino

  And what about the Sutlej, which the two Oldhams had also blamed for the Sarasvatī’s disappearance? Geologists (Gurdev Singh,11 for instance, in 1952) have long identified ‘a wide dry channel coming south from the spot near Ropar where the Satluj abruptly swings westward’; that palaeochannel meets the Ghaggar near Shatrana (Fig. 3.1), some 60 km south of Patiala, close to the point where the Sarsuti also joins the Ghaggar. It roughly follows the bed of the seasonal Patialewali. Remarkably, notes Valdiya, ‘at the point of confluence, the Ghaggar channel suddenly becomes 6-8 km wide—and remains unusually wide until it loses itself in the sand dunes of the Thar desert, west of Anupgarh’.12 This sudden broadening of the Ghaggar is the unmistakable sign that it once received some of the Sutlej’s waters at this point.

  Further downstream, the Wah and the three Naiwals represent more palaeochannels of the Sutlej in its westward migration. Then, shortly after the Sutlej finally moves away into Pakistan, another dry channel runs parallel to the international border on the Pakistani side until it joins the Hakra near Walhar. One more palaeochannel starts some 30 km northeast of Bahawalpur and proceeds southward to meet the Hakra again. And the list is not exhaustive: in fact, geologists have found ‘a large number of abandoned channels left by the ever-shifting Shatadru [Sutlej] in the Panjab plain’.13

  Clearly, then, the Sutlej has had a turbulent history. Something of its evolution is reflected in the ancient literature: named ‘Shutudrī’ or ‘swift-flowing’ in the Rig Veda, it became ‘Shatadru’ in post-Vedic literature, which means ‘of a hundred channels’, one more sign that ancient Indians were keenly observant and knew their geography; but rather than record it in scholarly accounts in the manner of ancient Greeks, they preferred the medium of ‘legends’. Let us hear one more.

  The Mahābhārata tells us how the great rishi Vasishtha, sorely distressed when he found that all his sons had been killed by his arch rival Vishvāmitra, wished to end his life. He tried various ways, but the elements always refused to cooperate; the sea or rivers into which he repeatedly hurled himself, bound with ropes or weighed with stones, stubbornly cast him back ashore. Thinking he was a ball of fire, the last river he plunged into ‘immediately flew in a hundred different directions, and has been known ever since by the name of the Shatadru, the river of a hundred courses’.14 Here again, the textual tradition is in accordance with what we find on the ground in the form of the Sutlej’s multiple channels.

  In 1983-85, an Indo-French mission explored an area of Haryana and Rajasthan between the Ghaggar and the Chautang; experts in geology, hydrology and archaeology were drawn from the Archaeological Survey of India and France’s CNRS. I will discuss in Chapter 11 its chief conclusions, which proposed a new perspective on the Harappan environment and agriculture, and challenged generally accepted views on the Vedic Sarasvatī. For the moment, I will only mention French geologist Marie-Agnès Courty’s findings, based on a microscopic study of the area’s sediments, of ‘true grey sands at a depth of over 8 m, identical to those of the Yamuna and the Sutlej’.15 This is similar to Raikes’s mention of greyish sand, suggestive of an ancient connection with the Yamunā, although Courty’s chronology differs from his: according to her, ‘mighty rivers with their sources in the Himalayas flowed at the end of the last ice age in the Ghaggar’s present basin’,16 but those ‘Yamuna-like rivers . . . stopped flowing in the study area well before the Protohistoric period’,17 that is to say, well before Harappan times (2600-1900 BCE).

  Explorations of riverbeds and their terraces in the Shivaliks have also yielded important results. Overlooking the plains, the west-flowing Markanda and the east-flowing Bata (Fig. 3.2), both of them relatively modest and now seasonal rivers, flow in disproportionately broad valleys, over 1 km for the former and reaching 6 km and more for the latter (before it joins the Yamuna at Paonta Sahib)—widths suggestive of far more copious flows in the remote past.

  V.M.K. Puri, a geologist and former director of the Geological Survey of India, took part in a worldwide listing of glaciers organized from Switzerland, and identified over 1500 of them in the Himalayas. In 1998, he and his colleague B.C. Verma published their findings based on a study of four terraces in the Markanda and Bata Valleys.18 In summary, they found that the higher, more ancient terraces abounded in metamorphic rocks and quartzites characteristic of the Higher Himalayas, not of the lower Shivalik Hills. This, they argued, proved that those rocks had been carried there by a river fed by glaciers ensconced in the inner Himalayas. Climbing to an altitude of 4000 to 5000 m, they identified three such glaciers in the Bandarpunchh massif; their meltwaters now meet at Naitwar, high up in Uttarkashi, and feed the Tons, the largest tributary of the Yamuna before the latter reaches the plains (the Tons is, in fact, larger and longer than the Yamuna upstream from their confluence). To explain the Himalayan deposits in the older terraces of the Bata and the Markanda, Puri and Verma assumed that the Tons and the Yamunā once flowed westward into the wide valley of the Bata, and onto the Markanda’s. To them, ‘all the evidences point to only one conclusion, that the present-day Tons was in fact Vedic Sarasvatī in its upper reaches’.19

  While the exact evolution of those Shivalik rivers remains to be confirmed, the work of Puri and Verma has shown, at the least, that larger glacier-fed—and therefore perennial—rivers once flowed through those wide valleys, a finding endorsed by Valdiya.20

  However, it is fair to point out that the small Sarsuti, born on the slopes of the Shivaliks, might never have answered that description, as there is no marked opening connecting it to the Markanda-Bata corridor above. It seems to me that the Markanda itself, with its much longer course in the Shivaliks and its broader valley, is a more suitable candidate to represent the upper course of the ancient Sarasvatī. Technically, in fact, the Markanda is not a tributary of the Sarsuti: it is the other way round. For some reason, perhaps one as prosaic as ease of access for worship, tradition may have transferred the origin of the Sarasvatī in the plains from Kala Amb (or Kalamb), where the Markanda flows down from the Shivaliks, to nearby Ad Badri, where the Sarsuti is said to emerge; the distance between the two being no more than 15 km.

  TECTONIC EVENTS

  Geologists have long suspected tectonic and seismic events—earthquakes, in plain language—to be responsible for some of the shifting of rivers. The reason is not far to seek: the entire belt formed by the Himalayas, including their foothills and piedmonts, has been seismically active ever since the Indian subcontinent, separating from the Gondwana supercontinent and cruising along in a northeasterly direction at the speed of 15 cm a year, collided with Eurasia some fifty million years ago; forced to slide beneath the Eurasian plate, it uplifted the latter higher and higher, somewhat as Varāha is said to have uplifted the earth. The Tibetan plateau and the Himalayas are the result of this prodigious impact—and the mountain range continues to rise, since India’s subduction goes on at the rate of almost 6 cm a year. The colossal friction between the landmass of India and that of Asia resulted in ripples (the Shivaliks are one of them) and numerous faults; inhabitants of not only the subcontinent’s northern parts, from Pakistan to Bangladesh, but also of the region as far south as Gujarat and Maharashtra, are all too familiar with the destructive earthquakes periodically witnessed along one or the other of those active faults.

  Earthquakes in the region that concerns us have left both geological and archaeological scars. Among the latter, Kalibangan, where Raikes explored the Ghaggar’s bed, displays a marked cleavage in its lower layers, proof of a strong earthquake dated about 2700 BCE, which is thought to have put a violent end to the city’s Early phase.21 Coincidentally, at the other end of the Sarasvatī basin, Dholavira, a fascinating Harappan site in the Rann of Kachchh to which we will return, suffered considerable damage, including the collapse of massive fortification walls, as a result of an earthquake during the same epoch.22 Bracketed between five faults, Kachchh has a long seismic history; many will remember how it was devastated by a powerful earthq
uake on 26 January 2001.

  With this context in mind, Puri and Verma concluded that a tectonic event must have been responsible for the opening of the ‘Yamuna tear’, as it is called (clearly visible in Fig. 3.2), through which the joined waters of the Tons, the Bata and the Yamunā escaped southward, robbing the Sarasvatī of its headwaters. Of course, if an earthquake did cause this tear, it would not have been instantly as broad as we now see it: a small opening and eastward tilt in the slopes would have been enough to funnel part of the glacier-fed rivers through the new gap; erosion would have done the rest in the course of time.

  Valdiya reaches a similar conclusion, quoting recent work23 to the effect that sometime after 1900 BCE, a major earthquake uplifted a riverine terrace near the Yamunā by 20 to 30 m. That earthquake struck along the fault that passes through the Paonta Sahib Valley, a fault still quite active today. Valdiya thus wonders, ‘Was this the tectonic event that caused the river to deflect abruptly from its previous westerly course and enter the channel of a river that flowed south . . . now known as the Yamuna?’24 It is certainly a good candidate, at least.

  SATELLITE PHOTOGRAPHY

  In the West, aerial photography from aircraft was first put to military uses during World War I; soon after the war ended, it was directed to more peaceful areas, geology and archaeology in particular. But only in the 1960s was aerial photography of the Haryana-Punjab region initiated, chiefly to make more precise topographical surveys; as far as I know, its considerable ability to identify potential archaeological sites was not exploited. But the next decade saw a radically new technology upset the field: remote sensing through satellite photography and imagery, which had been rapidly developing since the 1950s in the West, where it served such diverse purposes as mapping, search for oil and other resources, meteorology, or, notoriously, intelligence gathering.

  Photographs taken by satellites of the NASA’s LANDSAT series, followed by those of the French SPOT series, and more recently the IRS (Indian Remote Sensing) series, were used to study the Sarasvatī’s basin. For the first time, the dry bed of the Ghaggar-Hakra was revealed in dramatic fashion (Fig. 3.3). It was, of course, not the bed itself that appeared in the photographs, but the contrast created by the richer soil and vegetation found all along the river’s course. Processed by advanced digital enhancing techniques, satellite imagery vividly brought out the numerous palaeobeds that criss-cross the Sutlej-Yamuna watershed, most of which are invisible at ground level. The whole question now was to try and make sense of these Himalayan data, in both senses of the term.

  In 1980, four scientists, Yash Pal, Baldev Sahai, R.K. Sood and D.P. Agrawal, published a paper entitled ‘Remote Sensing of the “Lost” Sarasvatī River’, based on an analysis of many photographs of LANDSAT satellites; their work soon became something of a classic among such studies. They began by confirming ‘the sudden widening of the Ghaggar Valley about 25 km south of Patiala which is obviously a misfit if we take into account the considerably narrow bed of the Ghaggar upstream. This sudden widening can be explained only if a major tributary was joining the Ghaggar at this place. The satellite imagery does show a major palaeo-channel joining the Ghaggar here’25—the same channel that we mentioned earlier as coming straight from the Sutlej’s sharp bend near Rupar.

  They detected, in fact, not just one, but a ‘multitude of small channels into which the Satluj braided till it found its present channel’.26 Shatadru, or ‘flowing in a hundred channels’, is therefore a most apt designation for this capricious river! Indeed, the authors themselves remarked how ‘the braiding of the Satluj seems to have been echoed’ in the Mahābhārata’s legend.

  Our scientists attributed the Sutlej’s westward migration away from the Ghaggar to tectonic movements. Downstream, they found ‘a distinct paleo-channel which seems to suggest that the Satluj flowed through the Nara directly into the Rann of Kutch’, as C.F. Oldham had proposed.

  In the east, they traced three ancient beds of the Yamunā, indicative of a gradual eastward migration; one of them coincides with the Chautang or Drishadvatī.

  The paper’s conclusions, accompanied by a map (Fig. 3.4), are quite in tune with previous topographic explorations :

  The ancient bed of the Ghaggar has a constant width of about 6 to 8 km from Shatrana in Punjab to Marot in Pakistan. The bed stands out very clearly . . . The vast expanse of the Ghaggar bed can be explained only by assuming that some major tributaries were flowing into it in the past . . . Our studies thus show that the Satluj was the main tributary of the Ghaggar and that subsequently the tectonic movements may have forced the Satluj westward and the Ghaggar dried . . . The other major river system contributing waters to the Ghaggar may have been some prior channel of the Yamuna. [These two] main feeders were weaned away by the Indus and the Ganga, respectively.27

  Subsequent studies of satellite imagery have delineated more palaeochannels in the Sutlej-Yamuna watershed. As an example, a few years ago, three scientists led by A.S. Rajawat of ISRO,28 examined the area between Tanot and Kishangarh in the northwestern part of Jaisalmer district, Rajasthan, close to the international border. Enhanced photographs of the IRS 1-C satellite revealed, buried under the thick sand dunes of the desert, two important palaeochannels (Fig. 3.5), broadly oriented northeast to southwest; their width, ranging from 2 to 4 km, bears witness to the existence of a respectable river system in what is today a barren landscape of endless dunes. Since the area is hardly 30 km east of the Hakra, these paleochannels must have been connected to it.

  The existence of these palaeochannels has been confirmed by the most recent study in the field, an ambitious attempt to trace the entire drainage of the Sarasvatī. Three ISRO scientists, J.R. Sharma, A.K. Gupta and B.K. Bhadra, presented in 2006 the results of their research based on multi-spectral data from the new generation of IRS satellites.29 They identified five principal courses, numbered 1 to 5 in Fig. 3.7 (made clearer in Fig. 3.6).

  The first, 4 to 10 km wide, is more or less the Ghaggar-Hakra drawn by earlier researchers, except that the branch that runs past Fort Abbas and Marot in Pakistan dies out in the desert and the real course turns south just before the international border, meeting the Hakra a little farther south. A second difference is that its last stretch is not the Nara but a course 40 km east of it, although the authors cautiously suggest that this needs to be verified. The estuary is the Rann of Kachchh, in accordance with earlier studies.

  The second course, 4 to 6 km wide, roughly follows the international border up to the Jaisalmer district, where it precisely connects with the palaeochannels identified by A.S. Rajawat and his colleagues in the Tanot-Kishangarh area; it then turns due south up to the Rann.

  The third is a minor channel running west of the second. The fourth and fifth courses start south of the Chautang and hug the foothills of the Aravallis; they broadly correspond to the Luni’s basin. Those last three courses are fairly narrow—a few hundred metres at the most—in comparison with the first two. In fact, several experts have suggested that the Luni’s drainage could have been the most ancient course of the Sarasvatī, which would have drifted westward in stages, all the way to the Hakra.30 Our three ISRO scientists disagree and opt for the first course (strictly speaking, the second is inseparable from the first). But the two viewpoints are not wholly irreconcilable: the ISRO’s own map suggests that through the Chautang (the Hissar-Nohar stretch), a connection between the Sarasvatī and the Luni systems must have existed at some point, probably when some of the Yamunā’s waters flowed into the Chautang.

  Overall, the ISRO study confirms the existence of numerous palaeochannels and proposes the most likely courses for the Sarasvatī—not as neat as the single line we see on many maps (including mine in Fig. 1.1 or 3.1). It is a welcome reminder of the complexity of the region’s history.

  DATING ANCIENT WATERS

  The latest entrant in the field is nuclear physics, or rather one of its byproducts: a wide array of dating techniques, from which geology, oceanography
, archaeology and other disciplines have benefited immensely. Radiocarbon dating, based on the carbon-14 isotope (‘normal’ carbon being 12), is the best known of the lot; it is effective for carbon-based material such as wood, cloth or bone, which makes it the favourite dating tool of archaeologists. It even created something of a revolution, enabling for the first time an excavator to obtain absolute dates, instead of relative ones, based on comparison with other sites and ultimately on literary evidence. However, for materials like pottery, stone or metal, which hold little or no carbon, other dating techniques have been perfected.

  Water is what interests us here. In 1995, S.M. Rao and K.M. Kulkarni, two scientists from the Bhabha Atomic Research Centre (BARC), drew samples from wells in various parts of Rajasthan. They studied the proportions in isotopes of hydrogen (deuterium and tritium) and oxygen (18, while ‘normal’ oxygen is 16), in addition to radiocarbon from dissolved carbonate compounds (such as limestone). In the northwestern part of Jaisalmer district, precisely the area where A.S. Rajawat and his colleagues identified two important palaeochannels (Fig. 3.5), they found that ‘in spite of very low rainfall (less than 150 mm) and extreme conditions of the desert, groundwater is available at depth of about 50-60 m along the course of the defunct river* and a few dug wells do not dry up throughout the year.’ This groundwater is not a static water table; it actually flows subterraneously at a speed estimated at 20 m a year. Their analysis of the water samples taken from shallow wells (typical depth less than 50 m) showed that