Cartographers of the Vedas: Two Centuries of Scholarship on the Rigvedic River Names

The Nadistuti Sukta names ten rivers in verse 5 and another set in verse 6 (see the previous post for a verse-by-verse reading). The question this post takes up is the long one: how, beginning in the middle of the 19th century, did scholarship work out which modern rivers those Vedic names correspond to? Some identifications are now considered secure. A few are still contested. The methodological story behind the secure cases, and the residual uncertainty in the contested ones, is itself a window onto how Indology became a discipline.

This is the second post in a three-part series on the Nadistuti. The first treats the hymn as a text. This one treats the rivers as a research problem. The third returns to the landscape itself, with the tools of paleohydrology and climate science.

A research problem in four moves

Identifying a river that has been called by one name for three thousand years and another name for the last thousand is harder than it looks. Four independent lines of evidence have been brought to bear, in roughly historical sequence:

Each line constrains the others. A proposed identification has to satisfy not only the sound-change rules from Vedic Sanskrit to modern Indo-Aryan or modern Iranian, but also the geographical location implied by the hymn’s order, and (where archaeological data exist) the presence or absence of a habitable channel at the relevant time. When all four agree, the identification is taken as secure. When they disagree, the dispute can run for decades.

The Sutlej is a useful warm-up case. Vedic Śutudrī (Rigveda 10.75.5, with variants Śutudri, Śatadrū) corresponds to Greek Zaradros (Ptolemy, Geography VII.1) and to the modern Sutlej through a sequence of regular phonological changes: Sanskrit śu → s, the loss of internal -d- with vowel reduction, and the final -rī/-rā alternation typical of feminine river-names. There is no disagreement, no rival candidate, and the river has flowed in roughly the same valley throughout the historical period. The Sutlej is the kind of identification on which the whole enterprise depends.

Phase 1: the German philologists, 1840-1879

The serious work begins with two German scholars and one short interval. Rudolph von Roth, in Zur Litteratur und Geschichte des Weda (Stuttgart, 1846), worked out the basic principles of reading the Veda as a historical document. Christian Lassen, in Indische Alterthumskunde (Bonn, 4 vols., 1847-1862), produced the first systematic effort to place the named rivers of the Rigveda on a map of north-west India. [1]

Lassen’s method was straightforward and, on the whole, durable. He assembled the Sanskrit forms, set them next to the Greek transmissions preserved in Strabo, Arrian, Ptolemy and Pliny, then triangulated against the modern Punjab and trans-Indus hydronyms collected by 19th-century surveyors. The chain Sanskrit-Greek-Modern, where it could be set up, was nearly always decisive.

Aside. Lassen’s chain depended on the survival of Greek geographical texts of Alexander’s campaigns. The four Punjab rivers crossed by Alexander in 326 BCE (the Hydaspes, Akesines, Hydraotis, and Hyphasis) had Greek names that were direct transmissions of the Sanskrit. Without that single chain of intermediate testimony, Lassen would have been guessing at the Punjab toponyms. With it, he could anchor four of the five Punjab river identifications in print by 1847.

The Greek correspondences Lassen relied on are still the spine of the field:

Note that Vipāś is named in RV 3.33 rather than in the Nadistuti, but the chain runs the same way. The Beas is not in the Nadistuti list, an absence that has occasionally been read as evidence that the Beas may have been considered a tributary of the Sutlej at composition time.

Heinrich Zimmer’s Altindisches Leben (Berlin, 1879) added the social-historical dimension. Zimmer compiled what the Vedic texts say about how the rivers were used: which ones were forded, which were navigable, which had associated patrons or clans. The work is dated in places, but Zimmer’s chapter on the Sapta Sindhu (pp. 14-26) established the practical question that would orient much later work: not just where are these rivers, but what kind of river system did the Rigvedic people know? [2]

Phase 2: the classical synthesis, 1900-1955

The second phase consolidated. The summit work was A. A. Macdonell and A. B. Keith’s Vedic Index of Names and Subjects (2 vols., London: John Murray, 1912), still the standard reference encyclopedia for Vedic toponyms, ethnonyms, and personal names. The Index contains a substantial entry for every named river in the Rigveda. Each entry assembles the textual loci, summarises the identification history, and offers a guarded conclusion. [3]

Macdonell and Keith’s working method was traditional Indology at its best. They cite every textual occurrence by reference, give the Sanskrit form with the Padapāṭha analysis, set out the proposed modern identification, and note any rival proposals with a short evaluation. Their entry on Sarasvatī is exemplary: it accepts the consensus that the Rigvedic Sarasvatī ran between the Yamuna and the Sutlej (an inference from RV 6.61), but flags the difficulty of squaring that with the Brāhmaṇa tradition that the river disappeared at Vinaśana in the desert. The problem they identified in 1912 is the same one that satellite imagery began to address eighty years later.

Karl Friedrich Geldner’s Der Rig-Veda (Harvard Oriental Series, 4 vols., 1951) is the German philological summit. Geldner’s notes on individual hymns are dense with cross-references and his treatment of RV 10.75 supplies a comparative discussion of every translation that had appeared by the mid-20th century. [4]

Louis Renou’s Études védiques et pāṇinéennes (17 vols., 1955-69) carried the French tradition forward. Renou was less interested in geographical reconstruction than Geldner and more interested in the philological texture of the hymns, but his treatment of the river-vocabulary scattered across the corpus (especially vol. XV) is still cited.

Phase 3: the Harvard-Helsinki school, 1980 onward

The most consequential modern work has come from a network of scholars sometimes called the Harvard-Helsinki school, centred on Michael Witzel and his collaborators. Witzel’s series of papers, beginning with “Sur le chemin du ciel” (1984) and including the influential “Tracing the Vedic Dialects” (1989) and “Early Sanskritization” (Electronic Journal of Vedic Studies, 1995), reset the methodological standards for what could and could not be inferred from the Rigvedic geographical record. [5] [6]

Witzel’s contributions are of three kinds.

First, he refined the dialectal stratigraphy. By tracking phonological isoglosses across the Rigveda’s ten Mandalas, he showed that the family Mandalas (2-7) cluster around a more western geography (the upper Indus, the western tributaries) while Mandalas 1 and 10 expand the geographical horizon eastward into the upper Indo-Gangetic plain. [5] The Nadistuti’s eastern reach to the Ganga and Yamuna, on this analysis, is one of the markers of its late date.

Second, he integrated substrate evidence. A substantial number of Rigvedic river-names have no clear Indo-European etymology and have been candidates for borrowing from a pre-Indo-Aryan substrate, possibly related to the language of the Indus Valley Civilization. Kubhā (Kabul), Gomatī, and several of the Mandala 6 western tributary names are in this category. Witzel’s 1999 paper “Substrate Languages in Old Indo-Aryan” (Electronic Journal of Vedic Studies 5.1) developed the case in detail. [7]

Third, he constrained the chronology. Witzel’s work, together with the volume he edited with George Erdosy (The Indo-Aryans of Ancient South Asia, Berlin: de Gruyter, 1995), placed the latest Rigvedic stratum at approximately 1200-1000 BCE, with the family Mandalas a century or two earlier. Any geographical reconstruction has to fit within that window. [6]

Aside. A note on academic temperature. The dating of the Rigveda is, outside Indology proper, a contested subject. Within mainstream Indology and historical linguistics, the 1500-1000 BCE window (give or take a few centuries) is the consensus. The methodological case rests on linguistic stratigraphy, comparative Indo-European chronology, and the relative dating of post-Vedic texts. Alternative chronologies have been proposed; they have not found support in the standard historical-linguistic literature.

Phase 4: the geosciences, 1980 onward

The fourth phase, running in parallel with the Harvard-Helsinki philological refinements, brought remote sensing and paleohydrology into the conversation. The seminal paper is Yashpal, Bina Sahai, R. K. Sood and D. P. Agrawal, “Remote Sensing of the ‘Lost’ Saraswati River” (Proceedings of the Indian Academy of Sciences (Earth and Planetary Sciences), vol. 89, 1980), which used early Landsat imagery to trace a broad paleochannel through the modern Ghaggar bed. [8]

The Yashpal study set the template for everything that followed: remote-sensing evidence for a paleochannel, supplemented by sediment cores and isotope dating to constrain when the channel carried water. The Indian Space Research Organisation (ISRO) and the Central Ground Water Board extended the surveys through the 1990s and 2000s. K. S. Valdiya’s Saraswati: The River that Disappeared (Hyderabad: Universities Press, 2002) is the standard Indian summary of this work. [9]

The most rigorous recent contributions have come from sediment-dating studies in the international geosciences literature. Two papers in particular have reframed the discussion:

• Liviu Giosan et al., “Fluvial landscapes of the Harappan civilization,” Proceedings of the National Academy of Sciences 109 (2012): E1688-E1694. The team mapped Holocene channels under the Ghaggar-Hakra and across the Indus floodplain using a combination of satellite imagery and shallow drilling, and dated the major channel-abandonment phases with optically stimulated luminescence (OSL). The headline finding: the Ghaggar-Hakra had ceased to be a perennial river by approximately 4000 BP, in keeping with widespread monsoonal weakening, and the Harappan urban contraction tracks the channel decline. [10]

• Peter D. Clift et al., “U-Pb zircon dating evidence for a Pleistocene Sarasvati River and capture of the Yamuna River,” Geology 40 (2012): 211-214. By comparing detrital zircon age signatures in modern and paleo-channel sediments, the team argued that the Sutlej and Yamuna had already turned away from the Ghaggar-Hakra system in the Late Pleistocene, well before the Holocene Harappan period. On this account the Ghaggar-Hakra was a primarily monsoon-fed river through the entire Holocene, not a great glacier-fed Sarasvati. [11]

The two papers complicate the popular picture in opposite ways. Giosan strengthens the case that the channel was active and substantial during the Indus urban phase; Clift weakens the case that the channel was ever the kind of mountain-snow-fed perennial river the Rigveda describes. The current scholarly position is closer to a synthesis: the Ghaggar-Hakra was a real and substantial monsoon-fed river through the Holocene, carried significant water during the early-to-mid Harappan, and declined to ephemerality through the second millennium BCE, just as the Vedic textual tradition began to record its disappearance. The full debate is reviewed in our companion post, The Lost Sarasvati.

The current consensus, river by river

The following table reports the working consensus for each river named in RV 10.75.5 and 10.75.6, with a confidence rating reflecting how settled the identification is across the four lines of evidence. The grades are subjective summaries, not formal probabilities; they are intended as a guide for non-specialists reading the literature.

Eleven of the eighteen named rivers are now identified with very high or high confidence. Three (Sarasvatī, Rasā, Ārjīkīyā) remain genuinely disputed. The rest are insufficiently specified by the text to permit definitive identification with the methods available.

What is at stake methodologically

The Nadistuti identification project illustrates a wider point about how textual disciplines and physical sciences converge.

First, the philologists got most of the work right with the tools they had. Lassen in 1847 produced identifications that, with one or two adjustments, are still accepted in 2024. The Greek intermediate testimony was the single most important external check, and it remains the single most important external check.

Second, the physical sciences have, on the Sarasvati question, both confirmed and complicated what the texts indicated. The Brāhmaṇa tradition that the river dried at Vinaśana is corroborated by the OSL-dated paleochannel record. The Rigvedic description of the Sarasvatī as a mighty perennial flow is more difficult, given that the Clift et al. zircon work suggests the major Pleistocene flows had ceased well before the Vedic period. A residual perennial regime through the early-to-mid Holocene, consistent with both the text and the geosciences, is the synthesis most defensible on current evidence.

Third, the silent cases matter. The Nadistuti does not name the Indus delta, the Arabian Sea coast, the Ganga delta, the Vindhya range, or any river of the Deccan. The cartographic horizon of the hymn is the upper Indus catchment and the western Indo-Gangetic plain. This silence is itself a datum: it tells us where the late Rigvedic poets were not, and constrains the dispersal model of the Indo-Aryan languages within South Asia.

The third and final post in this series turns to that silence and asks what the river atlas implies, taken together with the paleohydrology and the Holocene climate record, about the ecology and the settlement geography of the Vedic Punjab.

Onward. Continue with Reading a River: Ecology, Settlement and Climate in the Late Rigvedic Northwest.