Veḷi

Evidence page · geography, counted

Two rivers carry the corpus: the Indus and the Sarasvatī.

Every named river in the Rigveda, counted by maṇḍala. Sindhu 204 and Sarasvatī 75 dominate everything else by an order of magnitude. The Ganges appears once, in the last book. And the Afghan rivers, which are routinely over-weighted, occur twice each.

Rivers by maṇḍala

FAMILY BOOKS later books 6 3 7 4 2 5 1 8 9 10 Sindhu (Indus) 204 6 9 8 16 8 9 57 23 34 34 Sarasvatī 75 14 3 16 7 4 9 3 3 16 Rasā 10 1 2 1 1 1 4 Vipāś (Beas) 3 2 1 Śutudrī (Sutlej) 2 1 1 Vitastā (Jhelum) 1 1 Yamunā 3 1 1 1 Gaṅgā 1 1 Kubhā (Kabul) 2 1 1 Krumu (Kurram) 2 1 1 Gomatī (Gomal) 2 1 1 Suvāstu (Swat) 1 1 occurrences per maṇḍala · orange Indus · gold Sarasvatī · green Punjab · blue east · red Afghan

Exact lemma counts per book. Family books first.

What the book distribution shows

Sarasvatī is the river of the family books. 14 in Book 6, 16 in Book 7, 7 in Book 2 — and it holds up right through to 16 in Book 10. It is not a late addition and it is not marginal.

Sindhu dominates the later books — 57 in Book 1, 34 each in 9 and 10 — while being present but modest in the family books.

The Punjab rivers are nearly invisible. Vipāś 3, Śutudrī 2, Vitastā 1 — and Vipāś and Śutudrī appear together in one hymn, RV 3.33, Viśvāmitra's river crossing. They are named because a specific event happened at them, not because the corpus lives there.

The Afghan rivers are 2, 2, 2 and 1 — Kubhā, Krumu, Gomatī, Suvāstu — and every one of them appears in Book 5, 8 or 10, not in the older family books. Strata and book order disagree here. The corpus does not know Afghanistan in its oldest layer by book; the Archaic stratum attribution came from tokens inside later books.

The Rigvedic world is the Indus and the Sarasvatī. Everything else is a mention.

The single most consequential number on this page

Gaṅgā: one occurrence, in Book 10, the latest stratum. Yamunā: three. Sarayū: three. Himavant — the Himālaya — one occurrence, Book 10, Popular stratum.

Sindhu: 204. Sarasvatī: roughly 70.

The Rigveda is not a text of the Gangetic plain. It is a text of the Indus basin and the Punjab, and it reaches the Ganges only at the very end of its composition.

The northwest is in the oldest layer

Archaic Strophic Normal Popular NORTHWEST rivers Kubhā · Krumu · Gomatī 3 PUNJAB rivers Vipāś · Śutudrī · Dṛṣadvatī · Marudvṛdhā 1 5 EAST Gaṅgā · Yamunā · Sarayū 2 1 1 HIMĀLAYA Himavant 1 metrical strata, oldest → youngest →

Occurrences by Arnold metrical stratum, oldest to youngest.

RiverModernCountWhere in the corpus
KubhāKabul river2Archaic stratum — Books 5 and 10
KrumuKurram2Archaic
GomatīGomal2Archaic
Vipāś · ŚutudrīBeas · Sutlej3 · 2Book 3 — Viśvāmitra
Dṛṣadvatī · ĀpayāHaryana1 · 1Book 3
GaṅgāGanges1Book 10 only
HimavantHimālaya1Book 10, Popular stratum
What the distribution says

The rivers of eastern Afghanistan — Kabul, Kurram, Gomal — are in the Archaic stratum. They are known to the oldest layer of the corpus.

The Ganges and the Himālaya are in the newest.

The corpus knows Afghanistan before it knows the Ganges.

This is a directional result and it is independent of any genetic or archaeological argument. It is a count of words in a text, by internal date, and it points one way.

Where the river names come from

Inherited, coined, or unaccounted for

RivernEtymologyClass
Sarasvatī75sáras "pool" + -vatī. Avestan Haraxᵛaitī; and Greek hélos "marsh" < PIE *selos-, by the regular Greek s→h shiftINDO-EUROPEAN
Greek cognate
Rasā10Avestan Raŋhā; and Latin rōs, Lithuanian rasà, Old Church Slavonic rosa — all "dew"INDO-EUROPEAN
Italic, Baltic, Slavic
Gomatī2go-matī — "cow-possessing." Transparent SanskritCOINED — a Sanskrit name for the Gomal
Suvāstu1su-vāstu — "good dwelling." Transparent SanskritCOINED — the Swat
Asiknīasikní — "the dark one," from ásita "black"COINED — a description, not a name
Sindhu204syand "to flow" is a traditional gloss; Mayrhofer marks it uncertain. Dravidian cintu "to spill, trickle" — DEDR 1546, see belowNO ACCEPTED INDO-EUROPEAN DERIVATION
Dravidian candidate
Gaṅgā1The reduplicated √gam "to go" derivation is weak and widely doubted. Frequently proposed as substrateNO ACCEPTED INDO-EUROPEAN DERIVATION
substrate candidate
Yamunā3Connection to yama "twin" is unprovenNO ACCEPTED INDO-EUROPEAN DERIVATION
Vipāś3vi-pāś "unfettered" is a nirvacana — a folk etymology attached to the Vasiṣṭha suicide storyNO ACCEPTED INDO-EUROPEAN DERIVATION
Śutudrī2No accepted derivation. Later Śatadru "hundred-flowing" is a re-etymologisation of a name already unparseableNO ACCEPTED INDO-EUROPEAN DERIVATION
ParuṣṇīNo Sanskrit derivation. Dravidian pari "to flow", parivu "current" — DEDR 3963NO ACCEPTED INDO-EUROPEAN DERIVATION
Dravidian candidate
Vitastā1No accepted derivationNO ACCEPTED INDO-EUROPEAN DERIVATION
Kubhā · Krumu2 · 2No Sanskrit parse. Kubhā is the Kabul, Krumu the KurramFOREIGN
The pattern is the finding

Sarasvatī and Rasā are the only two names that appear on both sides of the mountains. The Avestan forms correspond by the s→h shift that runs through the whole Iranian lexicon — soma/haoma, sapta/hapta, sindhu/hindu. That regularity shows the words were already there before Indic and Iranian separated. It says nothing about where they came from before that.

An Avestan correspondence alone could not settle this — a word borrowed into the common ancestor would show the same regular reflexes as one inherited. But both words have cognates outside Indo-Iranian, and that does settle it.

Rasā: Latin rōs, Lithuanian rasà, Old Church Slavonic rosa — all meaning "dew." Italic, Baltic and Slavic, three branches with no contact with India.

Sáras: Greek hélos "marsh," from PIE *selos-, by the same s→h shift that operates in Greek independently of Iranian.

These are Indo-European words. The evidence is the same kind demanded of any Dravidian proposal on this platform — regular correspondences across branches that had no opportunity to borrow from each other — and here it exists.

What the correspondence does show is portability. The same name is attached to the Helmand in Iran and to a river in Haryana. A name applied to two rivers two thousand kilometres apart is a name that travelled with people, whatever its ultimate source.

Every river that is unambiguously in India has a name with no accepted Indo-European derivation: Sindhu, Gaṅgā, Yamunā, Vipāś, Śutudrī, Vitastā, Paruṣṇī. Not "uncertain" — none of them has a derivation the etymological literature accepts. Where one is offered it is a nirvacana, a gloss invented after the fact: Śatadru "hundred-flowing" for a name nobody could parse, vi-pāś "unfettered" built onto the Vasiṣṭha story.

And that includes Sindhu — 204 occurrences, the most-named river in the corpus, the one the country is named after. syand "to flow" is a traditional gloss, not an established etymology, and Mayrhofer marks it uncertain.

And where a river IS given a transparent Sanskrit name — Gomatī "cow-possessing", Suvāstu "good dwelling", Asiknī "the dark one" — those are descriptions, and they are exactly the kind of name arriving speakers coin.

Two river names are demonstrably Indo-European. Neither is a river in India — one is in Iran and one is mythical. The rivers actually in India have names nobody can parse.

A population that named the Indus, the Ganges and the Sutlej itself would be able to parse those names. It cannot.

Searching the Dravidian dictionary for the river names

DEDR, twenty-eight languages, 54,290 entries

DEDR 1546 — Sindhu

Tamil cintu — "to be strewn, spilled, to trickle"; Malayalam cintuka "to be spilt"; Telugu cindu "to be spilt or shed, run out of a vessel"; Kannada cedaṟu; Kurukh kidgnā "to drizzle, to rain"; Toda kïd- "to sprinkle."

A flow-and-spill root attested across Tamil, Malayalam, Kannada, Tulu, Telugu, Kuwi, Kurukh, Kota and Toda — the south, the centre and the northeast of the family. Sanskrit ś regularly renders Dravidian c: cintusindhu is a phonologically unremarkable adaptation.

Set against the Sanskrit side:syand "to flow" is a traditional gloss and Mayrhofer marks the etymology uncertain.

The most-named river in the Rigveda, and the name the country carries, has a better-attested Dravidian comparandum than a Sanskrit one.

DEDR 3963 — Paruṣṇī

Tamil pari "to run"; Kannada pari, hari — "to run, to FLOW", parike / pariya "moving, running, stream", parivu "flowing, current"; Tulu pariyuni "to run, flow", parapuni "to flow, as water"; Telugu parapu "to cause to flow"; Malto pārap- "to flow."

A root meaning "to flow", with derived nouns meaning "stream" and "current", across six languages including Malto in the northeast. Paruṣṇī is the river the Battle of the Ten Kings is fought at — and it has no accepted Sanskrit derivation at all.

What these are and are not

These are attested roots with the right meaning and the right shape, distributed across the whole family — not single-language look-alikes. That distribution matters: a root present in Tamil, Kannada, Telugu, Kurukh and Toda was in the proto-language, not borrowed from Sanskrit into one branch.

What has not been done for either is the demonstration of regular correspondences across a set of river names — the standard that would move these from candidate to established. Nobody has done that work in either direction, and the Sanskrit derivations offered for these names do not meet it either.

The asymmetry worth naming: syand for Sindhu is printed in handbooks as though settled, and cintu is filed as speculation. Both are proposals. Only one gets the benefit of the doubt.

"Minority position" is a fact about who holds chairs, not about evidence

Filing the Dravidian derivations as "minority positions" treats a sociological fact as a verdict. It is a description of the distribution of scholarly opinion, and that distribution has a cause that is not evidential.

Indology was built on Indo-European comparative philology. Its practitioners are trained in Sanskrit, Avestan, Greek and Latin. Proposing a Dravidian etymology requires working competence in a different family and its comparative literature — which most Sanskritists do not have, and are not expected to have.

A proposal nobody in the room is equipped to evaluate does not become a minority position because it was weighed and found wanting. It becomes one because it was not weighed. The two outcomes are indistinguishable in a citation count and completely different in what they mean.

A consensus assembled by people who can only read one of the two languages is not a consensus about which language it came from.

Sindhu ← cintu, worked through

The correspondences, and the one place it does not close

ElementDravidianSanskritStatus
InitialProto-Dravidian *c- → Tamil, Malayalam, Telugu c-; Kannada c-/k-; Kurukh and Toda k-s-REGULAR. Dravidian c- is standardly rendered by Sanskrit s- or ś-. And the c-/k- alternation inside Dravidian is itself a known regular split
Vowel-i- (cintu, cindu)-i- (sindhu)EXACT
Nasal + stop-nt- — a canonical Dravidian NC cluster, voiced after nasal-ndh-PARTIAL. The nasal and the voicing carry over. The aspiration does not, and Dravidian has no aspirates at all
Meaningto spill, trickle, be shed, run out of a vessel; to drizzle, to rainriver; the Indus; flood, streamCLOSE. A flow-and-spill verb becoming a river name is an ordinary derivation
Where it does not close

Dravidian has no aspirated consonants. Sanskrit sindhu has -dh-. A direct borrowing would have to acquire aspiration in the process.

That is a normal outcome of Sanskrit nativisation — Sanskrit adds aspiration freely, and the same objection applies to every proposed Dravidian loan into Sanskrit, including ones that are widely accepted. But it is a gap, and it should be stated by whoever is making the proposal rather than discovered by whoever is attacking it.

Set the two side by side.syand requires an unexplained loss of the -y- and gives no account of why the noun means specifically this river; Mayrhofer marks it uncertain. cintu matches the vowel exactly, the initial regularly, the nasal-stop partially, and the meaning closely — and fails on aspiration.

Neither is demonstrated. One of them is printed without a caveat.

Kuyava — what the text actually puts near him

No barley, no harvest, no field

The parse ku-yava "bad barley" is available and is often repeated. The collocations do not support it. Across all six stanzas where Kuyava appears:

Collocate×exp
aśúṣa — voracious753.7
Śuṣṇa231.4
Kutsa — the human patron195.4
Śambara167.5
randh — to subdue85.8
cakrá — wheel64.0
púr — fort42.9
yáva — barleyZERO. Never co-occurs.
urvárā / kṛṣí — field / cultivationZERO in all six stanzas

Kuyava keeps company with other named enemies, with a fort, with a wheel, and with the patron he is destroyed for. He belongs to the conquest material, not to a harvest.

Which reopens the name. DEDR 1762: Tamil kuyavaṉ, Malayalam kuyavan — POTTER; kuyam — the potter caste. Phonetically that is kuyava against kuyava-(ṉ), with the ordinary Tamil masculine suffix.

This is a lexical match, not a demonstrated etymology — no regular correspondences have been established, and resemblance alone is the error this platform documents elsewhere. But it is a candidate, the Sanskrit parse has no contextual support, and the civilisation in question is known above all by its pottery.

Cumuri

DEDR 2677: Tamil cumai "burden, load", cumaiyaṉ "porter", cumattu "to burden"; with reflexes in Kannada, Telugu, Kurukh and Brahui (cumbaṭa-, a carrying pad). A root meaning "to carry", distributed across the whole family including the northwestern outlier. Moderate.

Mountains, forests, and what is missing

TermCountNote
párvata — mountain122Common, but generic. Often a cloud or a rock in myth rather than a place
girí — hill43Generic
himávant1The only named mountain in the corpus, in the last book
áraṇya — wilderness3The forest barely exists. Two of the three are in the Popular stratum. The Āraṇyakas, the "forest books," come later — and the forest ascetic is a later figure still
vána — wood, tree114Mostly material — timber, the wooden pressing-boards — not landscape
samudrá141Sea or confluence. Whether the composers knew the ocean is disputed; the word also covers a large body of water generally
The forest is a later idea

Three occurrences of áraṇya in 1,028 hymns, mostly late. And araṇyā́nī — the goddess of the forest, subject of the beautiful hymn RV 10.146 — appears only in the Popular stratum, the youngest.

The renunciant who withdraws to the forest, the vānaprastha stage, the Āraṇyaka texts — none of this belongs to the world of the family books. Those hymns are about cattle, rivers, chariots and fortified places. The forest becomes a religious category only once the population is settled enough for leaving to mean something.

The named peoples — and where they cluster

PeopleCountDistribution
pūrú214*Throughout — but heavily contaminated; see limits
bhárata17Books 2–7. The dynasty of the family books
turváśa20Spread
yádu15Spread
tṛ́tsu9ALL NINE IN BOOK 7
drúhyu6Mostly Book 7
śigru · yakṣu · vaikarṇá · viṣāṇin · bhalānas · alína · matsyá1–2 eachALMOST ALL IN BOOK 7
gandhārí1Book 1 — and it is about their sheep
kīkaṭa1Book 3 — a people who "do not press soma"
The ethnography of the Rigveda is essentially one battle's roster

Śigru, Yakṣu, Vaikarṇa, Viṣāṇin, Bhalānas, Alina, Matsya, Pakthas, Tṛtsu — the "tribes of the Rigveda" that fill textbooks and maps — occur once or twice each, and overwhelmingly in Book 7, in and around the Battle of the Ten Kings.

They are the participant list of a single conflict, recorded by the priest of the winning side. Not a survey of the peoples of northwest India. One hymn's cast, promoted to an ethnographic map by later readers who had nothing else.

The enemies, and where they are

TermTotalArchaicNote
vṛtrá18734Overwhelmingly the cosmic serpent, not a people
dāsá13136Heavily Archaic. Iranian cognates: daŋhu-, dahyu-, "land, people"
ásura7712Means LORD in the family books. The demotion is a Book 10 development
dásyu6717Archaic-weighted
paṇí5313Cattle-withholders, misers
śúṣṇa · śámbara · pípru · námuci · dhúni · cúmuri4–38Mostly Archaic
nāgá0The Nāgas are absent from the Rigveda entirely. sarpá occurs once, Book 10
Two findings that cut against easy readings

The Nāgas are not in the Rigveda. Not once. Every Nāga narrative — the serpent peoples, Takṣaka, the snake sacrifice — is later material. They cannot be the Rigvedic enemy, because the Rigveda does not mention them.

And dāsa may not mean an indigenous population. Its Iranian cognates mean "land" and "people" — Old Persian dahyu, and the ethnonym of the Dahae. A dāsa may be a rival Indo-Iranian group rather than a local one. The named chiefs — Śambara, Śuṣṇa, Pipru, Namuci — cluster in the Archaic stratum and none is described in terms that identify them geographically.

Limits

Three contaminations in the table above, named

ánu — 481. Almost all of these are the preposition "along, after," not the tribe. The figure is useless as it stands and is excluded from the ethnonym discussion.

pūrú — 214 and pṛ́thu — 46 similarly merge a proper name with a common adjective (pṛthu, "broad"). Treat both as upper bounds.

śiva — 51 is the adjective "auspicious," not a people or the god.

The rivers, mountains and enemy names are not affected — those lemmas are distinctive. The tribal counts are the weakest column on this page, and the Book 7 clustering is visible despite the noise rather than because of it.

Sarasvatī's figure is from a string count, not the lemma table, because the lemma form used here missed it. Roughly 70; treat as approximate.