§ 1 Stratification of the Rgveda
In the rest of this review, the points I have made summarily in
sections 1-10 will be discussed at some length. Due to their importance to
the rest of his work, points 1 and 2 are outlined in special detail.
Like most ancient Indian works, the Rgveda is a heavily stratified
text composed by many authors living in different eras. Like many such
texts, whether Vedic, epic, Buddhist or classical, it has been arranged by
its redactors according to the length of its subsections.
In the case of the Rgveda, the obvious and well-known subdivisions
are those involving 10 maNDala (usually, though ahistorically, called
"books"). Each maNDala is further broken into a number of sUktas (hymns),
and each sUkta into a number of Rc (verses).
This is the well-known division into 10 books -- containing 191+
43+62+58+87+75+104+103+114+191 hymns and some 10,500-odd verses.
However, as soon as one searches beyond this obvious division,
which is clearly distinguished in the manuscripts (and in the still
available, superior traditional oral recitation), one discovers an older
arrangement underlying the maNDala-sUkta-Rc division. While to a large
degree it mirrors the present division, on closer inspection certain
deviations are evident. Obviously, books 1 and 10, which each contain 191
hymns, stand out from the generally ascending pattern found elsewhere in
the length of books (from 42 to 104 hymns in books 2-7).
It has been long noticed that book 10 is linguistically younger and
that it in part overlaps with sections of the atharvaveda (AV). Pushing
things still further back, it was already noticed some 130 years ago, by
Abel Bergaigne and Hermann Oldenberg, that the so-called family books
(2-7) form the old core of the RV. This finding has been taken over by T.
without comment, quoting as his only witness the summaries of earlier
research compiled by the promising, but ill-fated, Bh. K. Ghosh (who
received a Ph.D. from Munich and a D.Litt. from Paris in the 1930s but died
of leprosy sometime after returning to India).
If one prefers to add up the verse count for books 2-7 -- rather
than counting the hymns -- one gets, according to T., 429+617+589+727+841
verses. (Cf. Satvalekar's RV edition, p. 809-826, which gives 2006 +
429+617+589+727+ 841+ 1108+ 1754 verses for the whole text.)
All this is not new. However, on further analysis, first carried
out by Oldenberg in his groundbreaking Prolegomena in 1888, we find that
the family books (RV 2-7) contain other organizational factors that involve
the authors (RSi), deities (devatA) and meters (chandas) of the hymns. Even
today, all three are still uttered before any formal Vedic recitation of a
hymn. The result is that each book in the "family" collection -- from the
43 hymns of book 2 to the 104 of book 7 or (with some deviations, even the
114 of book 8) -- is internally arranged as I summarized it in 1997:
"...the RV is structured according to several clear principles best visible
in the family books (RV 2-8): (1) the number of hymns per book increases,
(2) the family books begin with a small saMhitA addressed to Agni, Indra
and other gods, all arranged according to decreasing total number of hymns
in each deity collection. (3) Inside a deity series the hymns progress from
longer to shorter ones. The meter decides further: jagatI, triSTubh hymns
precede those in anuSTubh, gAyatrI" (Witzel 1997).
Incidentally, similar arrangements are also seen in the Pali canon
of early Buddhist texts, and elsewhere in Indian texts. Analogous
principles are also found in the Zoroaster's gAthAs, pointing to formal
links between Vedic and Avestan traditions that invite further
Any deviation from this strict numerical arangement has to be
explained. The reason, as demonstrated again by Oldenberg, is that various
hymns or sections of hymns have at later points been interpolated into the
text. This is found especially often in hymns of unusual length: small
individual collections of 3 verses (tRcas) or 2 verses (pragAthas) were
added to certain hymns or were combined into a new hymn during the final
standard RV redaction. This was carried out by zAkalya in the late brAhmaNa
period -- in other words, shortly before the time of the Buddha (c. 500/400
All such additions result in hymns that are too long and deviate
from the strict pattern. Later on, after zAkalya, more hymns, such as the
zrIsUkta, were added to the text, some of them clearly reflecting medieval
ideas. They were gathered together in the Kashmir khila collection -- and
always stand out insofar as they are not found in zAkalya's padapATha and
reflect post-Rgvedic grammar and contents.
None of the well-known structural principles in the nucleus of the
RV -- found in books 2-7 (and in a wider sense 1.51-8.66; see Witzel 1995:
309, 1997) -- are discussed by or even mentioned by Talageri. This gives
him the freedom to propose his own additions, based, e.g., on some rather
secondary and late evidence (these involve parts of RV 3 discussed in AB
6.18, which is found in a late stratum of that text; this issue is
discussed below). Talageri also views as interpolations the vAlakhilya
hymns of 8.49-59 (although these are, in fact, included and analyzed in
zAkalya's padapATha), some hymns mentioning the "tRkSi dynasty" and a few
minor but unspecified additions. All the rest of the text T. considers to
be genuine creations of RV-time seers.
As already noted, the redaction history of the RV is well-known, but is
nowhere discussed in Talageri's book. Summarizing this history, we can
distinguish five major steps in the redaction of the text (for details, see
Witzel 1995, 1997, etc.):
1. Stage one involved the original collection of the so-called family
books, in the kuru or Mantra period, which were organized using the
numerical principles described by Oldenberg.
2 Stage two involved the addition of materials that now comprise books 8,
1, 9, and 10, which were added at several distinguishable moments (for
details, see Oldenberg 1888 and Witzel 1995, 1997).
3. Stage three involved individual additions of whole hymns and of many
tRcas and pragAthAs to various RV books. As again shown by Oldenberg 1888,
these are often identifiable by the violation of the numerical principles
found in the first redaction of the family books and/or on linguistic
4. Stage four involved the redaction and final ordering of the text by
zAkalya in his padapATha. (For simplicity, we can ignore some minor
phonological changes that were later made to zAkalya's text.) The work
ascribed to zAkalya occurred in the late brAhmaNa period, as is evident
from his Eastern style, his grammatical misunderstandings of some RV forms
(Witzel 1989, 1997), and from further evidence found in the ZB tradition
5. The final stage included the addition of RV Khila that do not appear in
Close analysis of these redaction stages demonstrates that the
composition of the RV occurred in complex layers -- not in the tidy
sequential patterns imagined by Talageri (cf. again the chart noted in §6).
Once these complexities are recognized, it becomes obvious that many hymns
can no longer be securely claimed to have been composed by the authors they
are attributed to in many late traditions. In fact, many interpolated hymns
(e.g., in the partial list given in Witzel 1995) clearly stand out as late
additions due to their late book-10-style and AV-like grammatical forms and
contents (e.g., sorcery hymns).
All of this is well known -- but not to Talageri (and apparently not to
his proclaimed Western helper, Dr. K. Elst, who did not alert Talageri to
the problem.) Even one look at Geldner's scholarly translation of the RV
(Vol. I, xiv-xix) would have done the trick.
I have briefly explained these principles (discussed in my 1995
paper used by T.; also in Witzel 1997) to Talageri via email, in the summer
of 2000, to no avail (rebroadcast in Indic Traditions, 11/12/2000, see:
http://groups.yahoo.com/group/indictraditions). Far from being outdated or
'Romantic', as another self-proclaimed specialist of Old Indian history,
N.S. Rajaram, in the Organiser, will have it, Oldenberg's analysis has not
been challenged or refuted so far. We may hope, however, that the English
translation of Oldenberg's 545 pp. book (Motilal Banarssidass, c. 2001),
which is finally making it to English over 100 years too late, will help
would-be historians like T. understand the structure of the RV.
The result of all this is that T.'s book is based on what is
essentially the wrong Rgveda text -- the late Vedic compilation by zAkalya,
which had already been subjected to several earlier redactions, and which
mixed up materials from several eras in each of the books. Talageri, unlike
all serious Vedic scholars after Oldenberg, makes no attempt to reconstruct
the more ancient text on which that compilation was based. The result is
that all the far-flung historical conclusions that he draws regarding the
time and location of individual books, their authors, etc., are totally
unreliable. Many of his individual items of "proofs" (such as the
designation of the gaGgA in RV 6, gAGgya) immediately fall off the board
as late, not as being part of the "earliest RV" as T. claims (see §7) .
Amusingly, T. does not even exclude from his RV evidence stanzas
that were added long after zAkalya's padapATha. This includes 7.59.12 -- a
tryambaka verse, a late interpolation to the already older interpolated
stanzas 7.59.7-11 -- and similar late additions found at 10.20.1;
10.121.10; and 10.190.1-3.
Instead, T. accepts as "major interpolations" (p. 74) only the well
known vAlakhilya hymns (RV 8. 49-59) and RV 3.21, 30. 34, 36, 38-39 (with
68 verses). His reasoning here follows from his observation of a
discrepancy in the ascending number of the sums of verses for each of the
RV family books 2-7, which he tells us (per his zAkalya and anukramaNI
sources) contain 429-617-589-727-765-841 verses (p. 73-74). Noticing that
book 3 has too high a number, T. explains this as an obvious result of an
It is important to note that his claims concerning which verses were stuck
in here do not come from any internal study of the text but from his
reading (or misreading) of a much later text, AB 6.18. According to
Talageri, AB 6.18 tells us that RV 3.21, 30, 34, 36, 38-39, together
numbering 68 verses, were "misappropriated by vAmadeva," the traditional
composer of RV 4.
In fact, however, it is far from clear that AB 6.18 makes any such
claim. The text only tells us that vizvAmitra "saw" the saMpAta hymns in
question first and that vAmadeva actually "created" them (in RV 4).
vizvAmitra (of RV 3) therefore "created" counter-saMpAta hymns. The whole
section has other poets contributing to this endeavor as well: bharadvAja
(of RV 6), vasiSTha (of RV 7), nodhas (of part of RV 1). AB 6.18 continues
by listing the beginnings of RV hymns 3.48, 3.34, 3.36, 3.30, 3.31, 3.38 (=
81 verses) -- not, as T. tells us, RV 3.21, 30, 34, 36, 38-39 (= 68
verses). The result is that the list in his own cited sources does not
agree with the contents of his proposed "major interpolation."
A little countercheck of T.'s actual data is always useful, as
shown further below. Dozens of cases that I have doublechecked show marked
discrepancies of the sort found here.
Nothing in the AB passage in question speaks about the actual verse
numbers of RV 3 or 4, as T. suggests. The discussion revolves instead about
which RSi actually contributed some verses, not to the RV but to the
hotrakas' recitations. The story told at AB 6.18 comes from the late
brAhmaNa period, from North Bihar, where zAkalya's codification of the RV
took place (see Witzel 1997) and where the early layers of kAtyAyana's
anukramaNI took shape (see below).
The underlying lesson to be learned from studying the
representations of the RV in such late sources -- a lesson missed by
Talageri -- is that the politics of later priests and competing Vedic
schools (zAkhAs) and redactors active at the Sanskritizing court of Videha
often skewed the historical evidence found in the original RV. Note, e.g.
the wrangling between the various types of Veda proponents in BAU 3.3
(Witzel 1987); and cf. the "adoption" schemes among certain poets' clans
(Witzel 1995) as well as a divergent Rgveda at ZB 126.96.36.199.
On the issue of interpolations, it is critical to note again that
Talageri never bothers to check his claims against the accepted findings of
a long line of Vedic scholars. The 68 verses that he excludes in his "major
interpolation" turn out not to be interpolations using Oldenberg's
well-tested method (cf. Witzel 1995 : 311) -- while different nearby verses
that T. assumes are authentic do turn out to be late additions (e.g., RV
3.26?-27?, 28, 29, 51?, 52-53, 62?). The importance of definitively
identifying late and early strata in individual RV books is obviously
essential to T.'s endeavor, since his announced aim is to unlock the
historical truths hidden in this "hoariest" of all documents.
In passing, T. does mention other "incidental" interpolations in
the text. Talageri writes (p. 68): "There are other actual or alleged cases
of interpolations in the Rigveda (all interpolations made during different
stages of compilation of the Rigveda before the ten-maNDala Rigveda was
finalized), but all of them are incidental ones pertaining to ritual hymns
or verses" (p. 68).
Talageri does not specify which verses he has in mind besides (p.
70) quoting a note culled from Griffith's "hoary" translation (as T. might
put it) claiming that RV 4.42.8-9 and 7.19 are later additions. Not
surprisingly, both hymns turn out not to be interpolations when judged by
Oldenberg's well-tested criteria. Evidence suggests instead that they
belonged to the original, late Rgvedic "kuru saMhitA" (Witzel 1997).
Even if T. were right in dismissing certain RV interpolations as
"incidental," standard scholarly practice would demand that he identify
those interpolations and explain how he identified them. Otherwise, he
would have a free hand in tossing out any verses in the text that
conflicted with his "revolutionary" conclusions concerning the place and
time of the composition of different RV books. This he has not done (cf.
pp. 68, 73-4).
This is just one of countless examples of the methodological
laxness that characterizes his entire book.
§ 2 The anukramaNIs: Garbage in, Garbage out
If all of this were not enough to discredit T.'s book, any
examination of his use of the anukramaNIs would do an equally good job.
Among his most startling claims is one found at the start of his book (p.
7), where we find that the anukramaNIs were "part and parcel of the
Rigvedic text from the most ancient times." Nothing, however, suggests that
the anukramaNIs (or any putative lost predecessors) existed in the RV
period. In fact, this possibility is immediately ruled out due to the way
that our present RV gradually developed out of a series of earlier
redactions (see prior section).
It should be pretty obvious to most readers that vAc ("Speech") is
not likely to be the author of RV 10.125, as the anukramaNIs claim --
although the hymn does deal with speech (vAc). Nor is it very likely that
all the hymns of book 4 (which includes known late hymns) were composed by
one seer, vAmadeva -- as they also tell us.
The anukramaNIs also give us the names of ancient poets, deities,
and meters for many further hymns that are known to Vedic scholars (but not
to Talageri) to be late. These include RV 6.75, a long AV-like sorcery hymn
intended to make one's weapons victorious; the version of the anukramaNIs
used by Talageri identifies the hymn's deities as "Bow, Arrows," etc. !
It is important to point out -- as Talageri does not -- that
different anukramaNIs attached to the RV, SV, YV, and AV often disagree
concerning the poets of the very same hymns or verses. This fact is more
than enough to demonstrate the absurdity of his absolute reliance on these
late- and post-Vedic texts, which he tells us (p. 4) must be the "very
basis" of any analysis of the RV.
Remarkably, in his long discussion of these lists (pp. 3-20),
Talageri does not bother to mention which version of the anukramaNIs he is
following. Analysis of his book, however, shows that he is using the least
ancient of the two extant versions of the RV anukramaNI text, which we will
later see was compiled in the early Middle Ages!
What the anukramaNIs actually preserve, especially for the family
books, are traditional attributions to a clan, or to individual poets (real
or mythical), or to poets possibly assuming the name/title of a real or
mythical ancestor. In many sections of the family books, it is only the
relationships of certain hymns to each other that are correctly portrayed
by the AnukramaNI while the names of the poets involved are altogether
unknown. This usually involves a (group of) hymn(s) by the same (small
family of) known or frequently, by unknown author(s). Their names may
occasionally be preserved inside the hymns themselves (or they may have
artificially been derived by the anukramaNI from the text).
It is well known that traditional attributions of this sort may
change over time: from the oldest surviving in Sumerian texts to modern
Polynesian lists of chieftains, we can observe that there is no such thing
as a fixed list. Changes occur in part when texts originating in one
school or tradition are co-opted by late-coming rivals. This makes use of
attributions like this an extraordinarily tricky business, as has again
been known (but not to Talageri) since the 19th century.
In using this evidence, each case has to be carefully evaluated on
the grounds of internal evidence before we can assume that poet X mentioned
in the anukramaNIs was indeed the real-life poet of ascribed hymn Y. This
careful approach is followed by Geldner in his translation of the RV (1951)
and is further noted in my work (1995, 1997).
T., however, is quite innocent of any principles of the
philological enterprise: he does not mind using a later text as primary
source to explain an older one in front of him, just as he does not use the
archaic Sanskrit text of the RV front of him but only a Victorian English
translation three thousand years more recent. But, all later texts always
view older ones through the lenses of their own time (such as T. through
his own peculiar ones -- of which color, we shall see by the end of this
Instead of depending on late sources like the anukramaNIs --with
the motives of their composers remaining unstudied-- T. should have first
carefully collected the poet/clan names mentioned in the hymns themselves.
Of course, to get this job done, he would need to study the original Old
Vedic text -- not Griffith's badly outdated Victorian translation. In the
original, he would not only find references to authors who do not appear in
his Late Vedic list, but would also discover apparent poet or clan names
hidden away in Sanskrit anagrams (see, e.g., RV 10.24.2, where we find vi
... made for "Vimada").
Despite his enthusiastic claims for the anukramaNIs, it should be
noted that Talageri does not hesitate to change them when he is embarrassed
by their contents (see pp. 19 ff.). We hence find that his list of supposed
RV authors (pp. 7-19) varies considerably from what is found in any
anukramaNI. Talageri writes: "There are obviously corruptions in the
anukramaNIs in the form of ascriptions to fictitious composers .... we
have replaced the fictitious names in the anukramaNIs with the names of the
actual composers, ... the RSi of the hymn or the RSi of the maNDala."
Talageri's free-form adjustments of such evidence, late as it may
be, is still another distinctive mark of his book. The method by which he
determines the "actual composers" in such cases remains a mystery.
Again, for unclear reasons he also wants to limit the clans
involved in the composition of the Rgvedic hymns to ten families: "The
composers of the Rigveda are divided into ten families. These ten families
are identified on the basis of the fact that each family has its own
AprI-sUkta" (read AprI-! - p. 21). However, the attribution of at least
some of them to certain families depends on --what else could it be-- the
anukramaNI; and there are more AprI hymns found elsewhere, such as in the
Atharvaveda. Thus, the AprI hymn RV 1.142 is by an unknown poet, but
traditionally attributed to the AGgirasa dIrghatamas, and 1.188 is
attributed to agastya. There are two AprI hymns in book 10: RV 10.70 is
attributed to one sumitra, and 10.110 is attributed to jamadagni or his
son rAma. Obviously, T.'s statements are inconsistent and are a clear
indication of how cavalierly he establishes his divisions of the RV.
Incidentally, no study of the AprIs is mentioned, neither that by K.R.
Potdar (1945) nor the last one by van den Bosch (1985).
I have already mentioned, as T. does not, that not one but a number
of anukramaNIs are extant. Some particulars concerning these texts are
found in A.A. Macdonell's specialized study from 1886 and in his succinct
summary on pp. 272-4 of his History of Sanskrit Literature.
The best-known anukramaNI (the one used by T., though he does not
mention it by name) is the sarvANukramaNI attributed to kAtyAyana, the
alleged author of the White Yajurveda zrautasUtra (the kAtyAyana ZS) -- one
of the latest texts of this genre (cf. also kAtyAyanI in BAU and kAtyAyana,
the author of the pANinean vArttikas). These attributions suggest that
kAtyAyana was a late- or early post-Vedic Eastern figure, implying that
Eastern influences might be looked for in his anukramaNI as well. And
indeed, we find that the Eastern countries of aGga (1.116, in S.E. Bihar,
at the bend of the Ganges) and kASi (10.179, the Benares area) are
prominently mentioned in it, as well as a poet aGga aurava, 10.138. The
language of the text (displaying late compounds, use of perfect, etc.) is
certainly not Rgvedic, not even upaniSadic, but follows a terse sUtra
style. (In other anukramaNIs, zlokas even are the norm.) All this points to
late/post-Vedic and Eastern origins as well. (For discussion, see H.
Oldenberg, Zur Geschichte des zloka, 1987: 1188-1215; Horsch 1966; and
Note, for comparison, the typical difference between the prose
dharmasUtras (early) and the metrical smRtis (later), or the older prose
and the metrical middle upaniSads. The ArSAnukramaNI is a text of some 300
zlokas (Macdonell, 1886 p. v-ix). A hint about the general age of such
texts may be contained in the fact that the anuvAkAnukramaNI of zaunaka is
quoted (Macdonell 1886: vii) by Apastamba dharmasUtra 188.8.131.52 -- which P.
Olivelle (1999: xxxiv) now wants to date to the beginning of the third
Since the anukramaNI used in Talageri's book knows of a king of
kASi, Talageri -- who does not mention the country of aGga -- provides us
with this curious analysis (p.118):
".... the anukramaNIs provide us with a priceless [geographical] clue:
hymns IX.96 and X.179.2 are composed by a late bharata RSi who ...
attributes his compositions to his remote ancestor, pratardana. He
accordingly uses the epithets of his ancestors: in ... X.179.2, the epithet
is kASIrAja (King of kASI) [read kASirAja, kASi in Vedic! --MW]. pratardana
was a king of kASI, which is in eastern Uttar Pradesh. This can only mean
that the bharata Kings of the Early Period of the Rigveda were Kings of
kASI; and, in the light of the other information in the Rigveda, the land
of the bharatas extended from kASI in the east to kurukSetra in the west."
Here we have T.'s historical logic in full bloom.
Unfortunately for his views, kASi and other Gangetic lands mentioned here
do not show up in the RV at all. When the term kASi first occurs, in AV
5.22 (Witzel 1980, 1987, 1995, 1997 n. 259), the kASi tribals were still
regarded as despised outsiders to whom one sends illnesses. Even in the
late Vedic ZB 184.108.40.206, they still are not regarded highly. The rise of
kASI comes only in much later periods. In Vedic times, they remained a
small tribal area (which was conquered, by the time of the Pali texts,
first by kosala and then by magadha).
Predictably, T., makes much use of this late reference to the kASi
in the anukramaNI to argue that the oldest parts of the RV are Eastern in
origin. Here we get more vintage Talageri (my italics added):
"The above conclusion is inescapable: the information in the anukramaNIs
cannot be rejected on any logical ground (short of suggesting a conspiracy
theory), and it fits in with all the other evidence: ... [purANic and] even
the rest of the Vedic literature. The geography even of the yajurveda is
clearly an Uttar Pradesh centred geography [-- however, as generally
understood, the center of gravity shifted there in post-RV times -MW].
That the geography of the Rigveda is also the same has escaped the
recognition of the scholars purely and simply because these scholars are so
mesmerised by the Aryan invasion theory, and so obsessed with the vital
need to locate the Rigveda in the northwest and the Punjab for the sheer
survival of the theory, that their ideas and conclusions about the
geography of the Rigveda are based on the tenets of this theory rather than
on the material within the hymns of the text" (p. 119).
Conspiracy theories created by whom? Ironically, of course,
the text where the "early" kASi appear is not in the RV itself, but in
Talageri's late- or post-Vedic list, which he wrongly assumes is as old as
the RV itself.
In order to evaluate the information in the anukramaNI, given
its late/post-Vedic nature, it has to be compared with the various
late-Vedic lists of gotras and pravaras in the zrautasUtras, which trace
the origin of gotra names to many of the same RSis who show up in the
anukramaNIs (cf. J. Brough 1953, not mentioned by T). When making such
comparisons, one can deduce some of the political and social reasons why
such lists appear, and why they differ substantially among themselves.
More often than not, we find again that Late-Vedic Brahmanical
rivalries underlie specific attributions -- as I have pointed out in
relation to political "adoption" schemes (Witzel 1995, 1997; cf. Thapar
1984). Similar transformations can also be traced in the "historical"
sections of the purANas. It is therefore no surprise that genealogies
suggested in different anukramaNIs agree with those found in the similarly
late zrautasUtras or even later purANas. But all this has little bearing on
genuine RV genealogies, since traditions obviously repeatedly shifted in
the long centuries between the RV and late-Vedic periods (Witzel 1995: 339
Against this background it hardly surprises us that Talageri finds
in the anukramaNIs exactly what he hoped to find there: confirmation of
Epic-purANic data! By contrast, my own RSi lists in my 1995 paper were
based on the RV itself, not on such outside information: the anukramaNI was
cited there only as one, actually the last of the several means to assess
To crown it all, as earlier suggested, the sarvAnukramaNI of
kAtyAyana that T. uses throughout -- without identifying it by name or
further discussion -- is the younger one of our two preserved versions of
that text. This was shown way back in 1922 by Isidor Scheftelowitz
(Zeitschrift fuer Indologie und Iranistik, 1, 1922, p. 89-90).
Scheftelowitz also gave us the first edition of the RV Khilas, which is
preserved today only in Kashmir. In his 1922 study, he demonstrated that
the Kashmir version of the sarvAnukramaNI as well preserves a version of
kAtyAyana's text that is shorter and much older than the normal, received
version. Scheftelowitz also provided evidence that the longer version used
by Talageri --which the latter fantasizes goes back to RV times-- may date
no earlier than the middle of the first millennium CE! (On this see further
M. Tokunaga 1997: xv, xliv, lii.)
§ 3 Victorian Sanskrit?
I noted at the start of this review that Talageri shows no evidence
that he possesses anything remotely approaching an adequate knowledge of
Sanskrit -- not to speak of the archaic and enigmatic forms of Rgvedic
Sanskrit. He depends instead on a dated English translation and a modern
Sanskrit word list -- scarcely adequate tools to approach India's most
ancient text. Talageri does not admit his linguistic deficiencies, of
course, but they are nonetheless immediately evident in his frequent
misreporting of Rgvedic phrases. Thus his book consistently gives us the
Sandhi variant vara A pRthivyA (RV 3.23.4) instead of the correct
non-Sandhi form vara A pRthivyAH ("at the best place on earth"), the way
the expression is reported at RV 3.53.11-- but not by Talageri (cf. p.
115, 136, 210, etc.).
Similar mistakes are made (p. 117) in regard to nAbhA pRthivyA at
1.143.4, etc. for the correct nAbhA pRthivyAH -- which again occurs in
non-Sandhi form in 3.5.9. Even more glaring are Talageri's chanda for
chandas "meter" (frequently, e.g. on p. 3), and his misreporting of
aprI-sUkta for AprI-sUkta (p. 21 ff.). And so on throughout his book.
The translation that Talageri adopts as his authoritative text, as
already noted, is Griffith's Victorian version, which was first published
in 1889. Next to the even more antiquarian one by Wilson, Griffith's is the
only complete version of the work readily available to English speaking
readers without Sanskrit. Serious researchers would be expected to consult
the far more accurate scholarly translations made by K.F. Geldner (1951,
German), L. Renou (1955-1969, French) and now T. Elizarenkova (1989-99,
Russian). Talageri has not consulted any of these, nor (lacking German)
Oldenberg's still unsurpassed Noten (1909-12), which deal with each hymn
and verse in the book.
Every legitimate scholar knows that blindly using any translation
-- let alone one as inadequate as Griffith's -- can easily lead one astray.
Talageri, however, defensive about his dependence on the text, goes out of
his way to praise it as "the best, most complete, and most reasonably
honest English translation to this day" (p. 339).
As noted earlier, how someone who is incapable of reading an
ancient text in the original is capable of making such judgments remains a
mystery. Pace Talageri, the RV is a highly technical text composed in an
archaic literary tradition that is still poorly understood -- and whose
poetic forms are very imperfectly captured by Griffith. The forms of the RV
are not those of the later kAvya style but of those prevalent in the
preceding Indo-Iranian and Indo-European periods. To date, those forms have
not been adequately described as a complete system. Typically enough,
Talageri ignores all the detailed work that has been conducted over the
past 200 years in this direction. He dismisses all such research as the
disdained product of "the scholars" -- the academic philologists and
linguists who serve as straw men throughout his book. Meanwhile he pretends
to proceed as his own man -- despite his dependence on Griffith's ancient
translation! -- "invincible" in interpretation, as he often suggests.
§ 4 Failures in the 'Petty Conjectural Pseudo-Science' of Linguistics
Serious Vedic scholars need a thorough knowledge of the archaic
forms of Sanskrit (Old Vedic) and closely related languages in order to
correctly interpret obscure words, metaphors and similes, grammatical
forms, etc., unknown in later Vedic texts. Frequently by comparing the RV
with kindred documents, like the Avesta, or with other old Indo-European
texts, new insights arise concerning its more obscure passages.
While Talageri obviously is no Old Vedic scholar, he is not
altogether anti-linguistically minded (p. 309, 412, 415) -- unlike many
of his colleagues in what he refers to as the "the Voice of India family of
scholars." Rajaram, as noted before, denounces linguistics outright as a
"petty conjectural pseudo-science," without understanding its theoretical
basis and certainly without any knowledge of its procedures.
Linguistics is a "hard science, certain so far as sounds are
concerned (the linguistic subfield of phonetics). Sounds are, after all,
produced by physical instruments at various positions in the throat, mouth
and nose. Grammatical formations and the syntax of particular languages are
for the most part produced subconsciously by native speakers, but they
follow certain abstract rules (so well defined by pANini for Classical
Sanskrit). It is strange that many Hindutvavadins want us to believe that
linguistics (minus pANini of course!) is not a science. Deciding linguistic
questions becomes more difficult when it comes to items of meaning and
when linguistic reconstruction of pre-historic meanings are concerned. But
even here, we can argue within the realm of probabilities and do not have
to resort to fantasies of the sort found in Talageri, which are discussed
While Talageri's attitudes towards linguistics are more favorable
than Rajaram's, he too has little theoretical grasp of its principles and
is inept in carrying out linguistic investigations. His lack of linguistic
training is devastating, since linguistics is vitally important to nearly
all the questions that he treats in his book (see below on river, place and
Let us look here at a single example, involving his interpretation
of the female name "jahnAvI" (pp. 99-100; cf. also pp. 111-112), which he
wants us to believe is another name for the Ganges river. Talageri's object
is to use the word to support his claim that the RV originated in the East,
in the Ganges region. This is one of the few places where he disagrees with
Griffith's translation, which clearly conflicts with that view.
Talageri's discussion here is so revealing -- and convoluted --
that it is worth quoting in detail; my comments are added between brackets:
"jahnAvI, which is clearly another name of the gaGgA, is named in two
hymns; and in both of them, it is translated by the scholars as something
other than the name of a river: Griffith translates it as "jahnu's
children" (I.116.19) and "the house of jahnu" (III.58.6). The evidence,
however, admits of only one interpretation:
a. jahnAvI is clearly the earlier Rigvedic form of the later word
jAhnavI: the former word is not found after the Rigveda, and the latter
word is not found in the Rigveda.
The word clearly belongs to a class of words in the Rigveda which
underwent a particular phonetic change in the course of time: jahnAvI in
the Rigveda becomes jAhnavI after the Rigveda; brahmANa becomes brAhmaNa in
the Rigveda itself (both words are found in the Rigveda while only the
latter is found after the Rigveda) [I fail to find brahmANa in the RV; T.
or his source apparently misunderstood a plural form of brahman]...; and
the word pavAka has already become pAvaka in the course of compilation of
the Rigveda... the actual pronunciation of the word pAvaka must have been
pavAka in the Rigvedic age [an old chestnut: see the 19th cent. discussion
of RV diaskeuasis].
b. The word jAhnavI and therefore also the word jahnAvI ...
literally means "daughter of jahnu", and not "jahnu's children" or "the
house of jahnu" [but see below!] and ... has only one connotation in the
entire length and breadth of Sanskrit literature: it is a name of the
c. One of the two references to the jahnAvI in the Rigveda provides
a strong clue to the identity of this word: jahnAvI (I. 116.19) is
associated with the ziMzumAra (I.116.18) or the Gangetic dolphin. The
dolphin is not referred to anywhere else in the Rigveda. [But T.'s Gangetic
dolphin is also found in the Indus river! And RV 1.116.18-19 are not as
closely connected as T. wants us to believe; this is part of along 25-verse
list of the miracles of the Azvins.] "
The three different examples given by T. have three different
linguistic explanations and cannot be used to support his claims: his claim
that RV jahnAvI > Post-RV jAhnavI is demonstrably false (see below); his
second example involving brahmANa is fatuous, since that word is not found
in the RV at all; and his third case, involving the phonetic change pavAka
> pAvaka is a peculiarity of recitation schools. T.'s "linguistic
explanation" of the word jahnAvI thus is empty.
The meaning of that word can, however, be explained along
simple linguistic and grammatical lines as follows: female derivatives of
masculine names often have vRddhi in the second last syllable; thus manu :
manAvI, agni : agnAyI -- and consequently, jahnu : jahnAvI (cf. also
analogous formations, such as indra : indrANI, varuNa: varuNAnI, etc.)
That is all there is to it. Consequently, "the scholars" who followed older
translators or even sAyaNa were closer to the truth than T. jahnAvI was
the wife or a female relation of jahnu or otherwise connected to him or his
clan. The "ancient home" (purANam okaH) specified here, which T. thinks is
the "Ganges," is the territory of the jahnu clan, whose location is
unknown, that also figures prominently in later post-RV Vedic texts (AB,
To turn the word jahnAvI into a name for the Ganges can be done
only by retro-fitting the RV evidence to Epic-purANic concepts or to
Talagerian conceits of a Gangetic (Uttar Pradesh) homeland of the RV and of
the Aryans/Indo-Europeans (T., 1993). In short, jahnAvI "Ganges" is not
found in the RV. This robs T. of one of his important pieces of "evidence"
for a Gangetic home of the RV (for more, see below, s.v. pramaganda).
In sum, purANic preconceptions, coupled with an obvious lack of
grammatical and linguistic expertise, deliver for T. what he promised that
he would deliver in the book's preface: "this detailed analysis of the
Rigveda emphatically confirms our theory."
Not in this case, and not in the important ones mentioned above and
still further below.
§ 5 Trita's View From Inside the Well? Other Missing Sciences
The quality of Talageri's research is no better when it comes to
discussions of other fields that should have been included or consulted
when he prepared his book.
Little use is made of the secondary literature, even with the rich
Bombay University library at hand. T., however, mostly used what was
readily available and -- as he tells us at the start of his book -- what
his Western helpers, such as K. Elst, sent him. It is clear that he used a
very haphazard selection of sources. As was early noted, those sources
excluded the most important works for the present undertaking, including
studies of the structure of the RV by Oldenberg (1888) and of the various
anukramaNIs by Macdonell (1886) and Scheftelowitz (1922), not to speak of
more recent ones such as Tokunaga (1997). Nor did he consult any
post-Victorian translations of the RV -- not even the inadequate partial
one by O'Flaherty 1981 (Penguin), or even any one available in various
In fact, leaving aside his sketchy and highly selective discussions
in Chapter 8, the history of RV research does not exist for T. (In chapter
8, 'Misinterpretations of Rigvedic history', p. 335-424, he discusses a
number of opinions, but the chapter is primarily arranged by figure or
supposed "school" -- mainly focusing on supposed "invasionists" -- and not
by subject, and is hardly comprehensive).
T. usually refers derisively to past and present research, blanket
style, as the work of "the scholars" (implying that he is not one?!). When
he actually mentions the work of earlier scholars he tends to jumble their
research together with very recent work, as if the state of the art and the
opinions of the 19th century were identical to those of the year 2000. This
is a favorite tactic of the present rewriters of old Indian history --
meant to demonstrate "contradictions" in Indology.
In Talageri's discussions, large sections of relevant fields of the
humanities and sciences remain untouched, although they have a direct
bearing on the subject under discussion. A few examples here will
illustrate the ways that T.'s judgment is hampered by a lack of information.
His discussion of the "kingdoms" of his time frame of 3500-1500 BCE
is not tempered by discussion (or apparent knowledge) of semi-nomadic,
transhumance life style or the workings of early pre-state tribal
Nor does he display any familiarity with the realia needed to
interpret a difficult text such as the RV. As noted earlier, all notions of
RV poetics and their impact on interpretation are absent, since T. is only
familiar with Griffith's Victorian translation. He does not know about
geographical facts such as the nature of Panjab rivers and their constant
building of natural dams (RV 7.18). In his interpretation of jahnAvI (p.
100, RV 1.116.19) as a supposed name of the Ganges, he thinks that the
matter is clinched by the fact that a dolphin is mentioned in an
adjacent(!) stanza, RV 1.116.18, given in a long list of miracles of the
azvins (see above). However, the river dolphin (Platanista gangetica) is
not just found in the Ganges but also in the Indus river, as a simple check
of any encyclopedia would show (see the Encyclopedia Britannica, which he
actually lists in his references), s.v. dolphin, susu, e.g.,
Archaeology is largely ignored by Talageri as well. If he had
consulted any standard studies he would have found that all through the
time frame that he assigns to the RV (3500-1500 BCE) his supposed Aryan
homeland around the Ganges was exclusively inhabited by hunters and
gatherers and by some scattered chalcolithic agriculturists -- with no sign
of great pUru and bharata "kingdoms." The same applies to the absence of
horse and chariot in the Gangetic basin during this period -- a topic that
Talageri wisely never brings up. Nor is the Indus civilization (2600-1900
BCE) discussed at length. It covered at least the Western extremities of
his imaginary "Westward march" of the Vedic tribes during T.'s late RV
period, 2100-1500 BCE. Since this does not fit, we are informed that the
Indus people were anu, -- Iranians, in T.'s opinion (p. 41), as we shall
The section on comparative mythology, p.478 sqq. is seriously
misinformed as well. Discussions of the problems in it would lead us far
astray here, but the topic may be taken up at a later occasion. The list of
such cases could be easily amplified.
Finally, there is the silly but infuriating use of irregular
abbreviations of book titles which have to be learnt and re-learnt on any
use of the book. This leads to obvious problems for anyone who wants to
track down Talageri's amateurish uses of sources. What is ZTR or ZTH ? ZTH
is R. Gnoli's "Zoroaster's time and homeland".... The same annoying baffle
gab is also found, for example, in S.S. Misra's book "The Aryan Problem"
(1992). In the age of computers, these acronyms could easily have been
converted into something more recognizable, throughout the whole book, and
within minutes. Amusingly, HINDUTVA (given just like this -- all in
caps!), a book by the nationalist politician V.D. Sarvarkar, who closely
worked with Italian and German fascists, is not further abbreviated in T.'s
bibliography -- indicating, in an almost Freudian way, the bent of mind of
the author under discussion here. Nomen plenum est omen.
A short list of obvious omissions relevant to study of the RV, a
few of which we have already noted, includes the following. The list could
easily be expanded:
* The dates provided by the Old Indo-Aryan words in the Mitanni
documents of c. 1400 BCE are not mentioned anywhere in Talageri's book; the
forms of these words are slightly *older* than the corresponding forms in
the RV (ma-ash-da [mazd[h]a] for medhA, vaj'hana > vashana- [vazhhana]
for vAhana. They certainly do not support the fantastically "hoary"
chronology developed in Talageri's book.
* The evidence of archaeology is neglected for areas inside and
outside the Indus Civilization. As mentioned, it is certainly difficult to
picture "Vedic" hunter-gatherers and early chalcolithic agriculturists in
an Gangetic pUru "kingdom"! Talageri nowhere mentions the obvious
conflicts in his work with accepted archaeological evidence.
* Much other critical zoological and archaeological evidence is not
discussed: horse and two-wheeled chariot are prominent at all levels of
the RV, but are not found in South Asia before c. 2000/1700 BCE, more than
1000 years after T.'s early RV; river dolphins in the Indus are unknown to
T.; and so on.
* There is complete absence of discussion of social questions in the
RV: "kings" in the semi-nomadic (only very partially village-based) society
portrayed in early strata of the RV? Vedic "dynasties" reigning for
millennia? What kind of "state" is represented by the RV hymns? Any search
of the vast comparative literature on semi-nomadic peoples would have ruled
out much of Talageri's fantasies.
* There is no discussion of climate in the book. All of the RV
indicates the presence of cold winters, a prominence of long dawns, and
river flooding due to snow melt. All these conditions are typical of
conditions in the Panjab, not of the more southern and warmer Gangetic
plains under the heavy influence of the monsoon.
§ 6 Imaginary Chronologies: RSis of the Kali Yuga?
Any evaluation of T.'s grand effort becomes even more devastating
once the results of his discussions are pulled together and it is shown how
procedurally erratic they are.
We can begin by looking at T.'s conclusions as to the time period
of the RV. Based on his analysis of the post-Vedic anukramaNIs and related
data, T. attempts to interpret each maNDala as a unified text (with the
exception of book 1, which he subdivides into three phases, and to some
extent book 10). Each book, on his view, can therefore be studied as an
individual corpus of one or more clans belonging to one time period.
References from these 10 books are then taken to establish the time and the
place of the author(s) of the individual hymns and of the whole book. All
of them taken together are used to delimit the area of the RV and of the
relative time frame of the whole corpus and its parts.
It should be clear even to a superficial reader that T.'s rigid
division of the RV into 10 books belonging to unique time periods
contradicts well-known evidence, some of which we noted earlier.
Further, several maNDalas overlap in time, as illustrated by the
fact that they often mention the same chieftains (and sometimes, the same
One is due for a surprise when one examines T.'s treatment of these
facts. The "family reminiscences," such as RV 3. 53, and the similar late,
additional family hymns (as per Oldenberg 1888) RV 6.47 and 7.33, clearly
show historical awareness in such clan traditions. However, the poets often
do not say more than "we compose in the way of the atri (atri-vat)," or
they may refer back to poets of the mythical past (aGgiras, uzanas, etc.).
While we would expect, based on the internal evidence of the RV
(Witzel 1995), a period for each family book that extends over roughly five
generations -- with a few references thrown in to ancient predecessors and
a few late (post-RV) additions -- T.'s schemes calculate the RV books each
in term of many hundreds or even thousands of years (p.75 sqq.) :
"It is clear that the Rigveda was not composed in one sitting, or in a
series of sittings, by a conference of RSis [echoes of my 1997 paper - MW]:
the text is clearly the result of many centuries of composition. The
question is: just how many centuries?
The Western scholars measure the periods of the various maNDalas in
terms of decades, while some Indian scholars go to the other extreme and
measure them in terms of millenniums and decamillenniums.
A more rational, but still conservative [sic!- MW], estimate would
be as follows:
1. There should be, at a very conservative estimate, a minimum of
at least six centuries between the completion of the first nine maNDalas of
the Rigveda and the completion of the tenth.
2. The period of the Late maNDalas and upa-maNDalas (V, VIII, IX,
and the corresponding parts of maNDala I) should together comprise a
minimum of three to four centuries.
3. The period of the Middle maNDalas and upa-maNDalas (IV, II, and
the corresponding parts of maNDala I) and the gap which must have separated
them from the period of the Late maNDalas, should likewise comprise a
minimum of another three to four centuries.
4. The period of maNDalas III and VII and the early upa-maNDalas of
maNDala I, beginning around the period of sudAs, should comprise at least
5. The period of maNDala VI, from its beginnings in the remote past
and covering its period of composition right upto the time of sudAs, must
again cover a minimum of at least six centuries.
Thus, by a conservative estimate, the total period of composition
of the Rigveda must have covered a period of at least two millenniums
It is, incidentally, altogether unclear how T. arrives at
these estimates. If the composition of the RV stretched out over more than
two millennia, the text would be in the same language (except for some
innovations in the late book 10). This is a virtual impossibility as far as
any living language is concerned. Since T. assumes "at least 600 years"
for the period of composition between RV 9 and the end of 10 (p. 77), we
thus would arrive at 3500-2100 BCE for the bulk of RV composition.
See here http://www.safarmer.com/pico/talageri.html, put
together by Steve Farmer. Farmer attaches Talageri's own dates, provided in
an online exchange on 21 July 2000 (for Talageri's own words, see the
bottom, of the webpage) to an originally dateless chart that Talageri gives
in his book. Here are the "conservative" dates and "lower limits" --
outrageous by normal historical standards -- that Talageri provided in that
Mandala 6: 3500-2900 BC,
Mandalas 3, 7, early 1: 2900-2700 BC,
Mandalas 4, 2, middle 1: 2700-2400 BC,
Mandalas 5, 8, 9, late 1: 2400-2100 BC,
Mandala 10: 2100-1500 BC.
In other words, the RV would be contemporaneous with the early
village-like predecessors of the Indus civilization at Harappa itself. As
far as T.'s U.P./Bihar "homeland of the RV" is concerned, the text would
have evolved right among the hunter-gatherer bands of that area. That
country did not see forts (pur), chariots, and horses for another 2000
years -- while the RV is full of them.
Further, by 3500 BCE (the date for his oldest book, RV 6)
horse-drawn chariots were not even invented (even the bullock cart was only
first appearing in Mesopotamia about that time). Chariots only appear more
than a thousand years later both in Mesopotamia and in the S. Russian-W.
Siberian steppes. There also was no horse in S. Asia then either -- the
"indigenous" Siwalik horse (Equus sivalensis) by that time had been long
extinct. The modern horse (Equus caballus) first appeared, imported from
Central Asia (Bokonyi 1997; Meadow and Patel 1997), around 1700 BCE (Pirak,
Kachi plain in easternmost Baluchistan).
But these glaring anachronisms do not pose a problem for T. All
such historical, technological, zoological, and archaeological details do
not detain him, since they are simply disturbing little facts; all of this
will be discussed in further detail further below.
In the same vein, T.'s conclusions on "Kings and RSis" (p. 59
sqq.) must be regarded with utmost caution, based as they are on his
impossible view of the RV's internal chronology.
For example, as soon as one applies Oldenberg's well-known
principles in detecting late interpolations in the RV (i.e., looking for
late violations of the original numerical order of the family books: see §
1 supra and my letter in the IndicTraditions list on 11/12/2000), one can
simply forget the purANa-assigned "anu kings" of 6.45-46 -- where they
appear in suspicious, additional hymns. (For T.'s purANic mindset, cf. p.
138 sqq.). The same applies to the yadu "king" vItahavya(?) at 6.15.2-3,
where he is not a chieftain but the poet of the hymn, and in 7.19.3 (T.
has a misprint) where the word does not seem to designate a "King" but is
an adjective referring to the Great Chieftain sudAs.
Not surprisingly by now, T. arranges his list of "Kings" by
following the list found in the purANas (cf. Morton Smith 1973: 504).
Talageri tells us: "The names of these Kings are given above in order of
their relative positions in the dynastic list" (p.60). Then, he sets out to
prove that the purANas, after all, are "right" as per his 1993 book.
Tacitly following my 1995 historical paper, the close connection of sudAs
with books 3 and 7 is accepted, as well as the tenor of book 6 where the
prominence of sudAs' father divodAsa points to a slightly earlier time
frame. However, the conclusion that Talageri draws from these data (p. 62)
-- that all maNDalas after 6, 3, and 7 are "post-sudAs" is again
There is no way that any RV book can be that easily declared
"post-sudAs," since each book contains heterogeneous material from several
or even many generations of poets. Only after the multi-axial
investigations mentioned early in this paper (Witzel 1995) will have been
carried out, will statements of this sort have a chance to be
Once again the principle "garbage in, garbage out" applies to
Talageri's work. The chieftains of the RV are closely linked to certain
priests' and poets' clans, according to T.'s traditional sources, and the
hymns of these clans are concentrated in certain books. Therefore, the
moment one (arbitrarily) orders these books in a fixed temporal pattern,
guided by the anukramaNIs and purANas, etc., the "kings" and poets must
come out in the same order as that of the books. In consequence, T.'s whole
"historical analysis" is wrong, based, as it is, on the use of such late
sources and his rigid maNDala scheme, which ignores the temporal layering
of individual texts (see above). Someone else has to do this exercise all
over again, taking into account not just traditional RSi names and maNDalas
set in stone, but a multitude of parameters (Witzel 1995).
Not surprisingly, T. ends up with a list of kinds that more-or-less
agrees with those found in the purANas (p. 63). A little fudging helps, as
a check of sudAs' ancestors in T.'s list (p. 63) indicates (cf. Witzel
1995). Morton Smith who worked in the footsteps of Pargiter's purANic
studies, often comes to conclusions radically different from the ones in
T. Smith's comparison of epic, purANic and Vedic sources (1973) is
never mentioned by T. It is instructive to observe how T. has to twist and
turn in order to explain, within his own erroneous framework, "the tRkSi
dynasty" (p. 66-72). Facts at all times in Talageri's work have to agree
with his theory, not the theory with the facts!
§ 7 Talageri's Geography: A Moveable Feast
T.'s theory places Indo-Aryans impossibly in the Gangetic plains
before they moved Westwards into the Panjab and beyond. It has already been
shown (§ 4) that some of his "Eastern" data in the RV (jahnAvI as Ganges)
evaporate as soon as we take a close look at his historical ordering of the
text and his faulty grammatical analyses. The same applies to other river
names. I restrict myself here to the ones found in what T. claims, on
flimsy grounds, is the RV's "oldest book," RV 6.
One can immediately throw out the reference to the Ganges that
appears at RV 6.45.31 (gAGgya). This reference occurs in a series of long
hymns to Indra 6.44-46 which follow a shorter Indra hymn (6.43 of 4
stanzas) and precede the appendix hymn 6.47. Applying the principles
pioneered by Oldenberg, RV 6.45 can be shown to be a composite hymn built
out of tRcas at an uncertain period. The ordering principle of the old
family books clearly points to the addition of all these hymns in mixed
meters at the end of an Indra series. Such late additions must not be used
as an argument for the age of the bulk of book 6 (see my letter in
IndicTraditions from 11/12/2000).
Incidentally, it should be noted that such tRca and pragAthA hymns
are not listed in the short list of additions that I gave in my 1995 paper.
As I noted in the paper, however, they are prominently discussed by
Oldenberg (1888). T. again falls prey to his lack of knowledge of
Indological research -- not understanding my reference to Oldenberg -- when
he accused me of inconsistencies in the nature of that list.
The River sarasvatI found in book 6 (T., p. 102) may be discarded
just like T.'s Gangetic jahnAvI. In the hymns 6.49, 50, 52, 61, the order
of arrangement is disturbed and especially the group 6.49-52 is very
suspicious. According to the RV ordering principles, we expect 5 hymns for
sarasvatI, and do not obtain them even by a dissection of hymns 51 and 52.
All this points to an addition of materials at an unknown time. Therefore,
the Haryana River sarasvatI (mod. Sarsuti) is not found in the old parts of
book 6. Incidentally, it is entirely unclear that the physical river
sarasvatI is meant in some of these spurious hymns: in 6.49.7 the sarasvatI
is a woman and in 50.12 a deity, not necessarily the river (Witzel 1984).
(At 52.6, however it is a river, and in 61.1-7 both a river and a deity --
which can be located anywhere from the Arachosian sarasvatI to the Night
time sky, with no clear localization).
Interestingly, however, the other -- i.e. Western -- rivers
remain in the older parts of RV 6, such as the yavyAvatI and hariyUpIyA:
they still point to Eastern Afghanistan, to the river Zhob, and (perhaps)
the Hali[-Ab], as the location of these parts of the book.
It would lead us too far afield here to discuss all of T.'s
identifications of rivers. Among the more ludicrous ones are precisely the
last ones, the hariyUpIyA and yavyAvatI. Talageri writes (p.98 sq.):
"hariyUpIyA is another name of the dRSadvatI [in eastern kurukSetra, just
west of Delhi - MW]: the river is known as raupyA in the mahAbhArata, and
the name is clearly a derivative of hariyUpIyA. The yavyAvatI is named in
the same hymn and context as the hariyUpIyA, and almost all the scholars
agree that both the names refers to the same river."
The only item agreeing here is hari "tawny, etc." = raupya
"golden." RV 6.27 mentions the hariyUpIyA in the context of the vRcIvants
and the pArthava, -- uncommon Western names for the Rgvedic E.
Panjab/Haryana. Of course, hariyUpIyA cannot be, as it is often alleged,
the origin of the name of Harappa: medial -p- should have long
disappeared, via -v- and zero, and cannot have resulted in double p.
Mercifully, T. does not mention this. In fact, Harappan
Civilization is almost completely absent in his book, only noted in passing
in discussing some other scholars' views. According to his time table of
the RV, however, the Rgvedic period overlaps exactly with the later parts
of the Harappan civilization -- which he fantastically describes in one of
his few references to it as "a joint civilization of the anus (Aryans
belonging to the same linguistic stock as the latter-day Iranians and some
other Indo-European groups)... and the pUrus (the post-Rigvedic Vedic
Aryans), even perhaps more anu than pUru, at least in the case of the more
well-known western sites" (p. 419). This neatly shifts the evidence away
from the RV -- according to T. a pUru text of the Gangetic Plains.
Unfortunately for Talageri, the mature Indus civilization
covered all of RV territory, as it extended from E. Afghanistan to Haryana
and Western U.P, and from the Himalayas down to the Indus delta and
Kathiawar in Gujarat. His knowledge of the Indus Valley Civilization is
hence as misinformed as his knowledge of RV culture.
A few more curious points from his discussions of river names:
prayiyu and vayiyu in 8.19.37 actually are (non-Indo-Aryan) men of the
country of suvAstu (mod. Swat, just east of Afghanistan) -- not rivers at
all, as Talageri assures us (p. 102)! For zveti RV 10.75.6 (in T. p. 103),
read zvetyA! Talageri lists azmanvatI as an "Eastern" river (p. 98), but
in RV 10.53.8 (p. 103) it is probably not a river on earth at all but a
river in the night time sky (Witzel 1984).
But what casual reader of Talageri's book could be expected to pick
up on these points without spending weeks or months tracking down
Talageri's spurious "evidence."
Incidentally, the Avestan river "haroiiu" (map, opposite p. 104;
for details on these Afghani rivers, see Witzel 2000) is a nonce word, a
pseudo-stem, derived by T. from the Acc. harOiiUm; the stem would be
*haraEuua-, but to know that you would need some linguistics! The map
opposite p.120 gives a few more Avestan names, often in wrong form: read
margu for mourv, haraEuua for harOiva, airiian@m vaEjah for airyana vaEjah;
the latter Talageri improbably locates in Kashmir while the Avestan texts
make it the central land of E. Iran -- i.e. the cool, central mountain
pastures of Afghanistan (Witzel 2000). Also, note that kIkaTa, following
19th c. guesses, is located by T. in Bihar, while the RV gandhAri -- here
with correct spelling -- are located in Arachosia! However, on p.131
Talageri quotes the AV as identifying the gandhAri with the mUjavants,
which is not correct as these two (and the kAzi, aGga) are mentioned in
parallel fashion as distant tribes, seen from the point of view of the
central land of the kuru (Witzel 1980).
T. thinks that such linguistic distinctions as mentioned just now
are mere nitpicking -- again to his undoing. Speaking about Northwestern
rivers, he tells us that "three of them, kosala, zutudrI and kubhA are
clearly Indo-European names ("the hairsplitting about the letter -s- in
kosala is a typical "linguistic" ploy....)" (p.248). However, even a brief
look at the discussion of these words in Mayrhofer's etymological
dictionaries (1956-1976, 1985-, unknown to T.!) and or my detailed study of
Vedic river names (Witzel 1999) would have told him otherwise. There is
nothing "clearly Indo-European" about them and words with is, us, es, os
etc. are extremely rare in Vedic and stand out as loan words.
The summary of river names and their location given on pp. 103-112,
and the inherent "Westward expansion" of the Indo-Aryans in RV times is of
course based on the philological and linguistic mistakes criticized above
(§1-4) T.'s conclusions (exemplified in his table and insert on p.104) are
therefore moot. We need a fresh investigation of references to rivers in
the RV. The results should be close to those found in Witzel 1995,1999
which was based on text-internal evidence.
T.'s lack of a linguistic background becomes painfully obvious
when we consider his absurd analyses of RV place names. Since T. believes
that all through the early Vedic period the Proto-Indo-Europeans were
present in northern India, the Mundas in eastern India, and the Dravidians
in southern India (but see Kuiper 1991, Witzel 1999a), he is not surprised
that the northwestern part of the subcontinent shows predominantly
Indo-Aryan names during the Vedic period. Relying solely on my short 1995
paper, which provided only a terse summary of this topic, he takes the
prominence of Indo-Aryan names in North India as a proof for an Indo-Aryan
(and Indo-European) origin there and for their early settlement in northern
India. Along the way, he excoriates me for failing to note what he
considers to be that obvious conclusion (p.248). However, like most of my
1995 paper, the views presented here were just a short summary of a broader
base of evidence. A much fuller treatment of Vedic place names can be found
in my 1999 paper, which came out one year before T.'s book.
The evidence assembled there suggests exactly the opposite of what
T. assumes: Although the Northwest generally abounds in Indo-Iranian names,
the remaining indigenous, non-Indo-Aryan names are of special interest.
Their sounds and their word formations clearly show that they belong to a
non-Indo-European, Greater Panjab substrate (Kuiper 1991, Witzel 1999a).
This evidence renders all talk of an original Indo-European settlement of
northwestern/northern India moot.
In addition, the recent discussion of the substrate words common to
both Indo-Aryan and Iranian (Witzel 1999a, Lubotsky forthc.) adds
substantial new evidence for this view. Such common non-Indo-Iranian words
differ from the typical Rgvedic and post-Rgvedic substrate and indicate
that both the Proto-Indo-Aryans and Proto-Iranians, perhaps even the
speakers of Proto-Indo-Iranian, entered a Central Asian/Afghan territory
that was already occupied by a previous population speaking
non-Indo-European language(s) (pace J. Nichols!) -- most probably the
language(s) of the Bactria-Margiana Archaeological Complex (BMAC).
(Continued with part 3)
Department of Sanskrit Indian Studies, Harvard University
2 Divinity Avenue, Cambridge MA 02138, USA
ph. 1- 617-496 2990 (also messages)
home page: http://www.fas.harvard.edu/~witzel/mwpage.htm
Elect. Journ. of Vedic Studies: http://www1.shore.net/~india/ejvs