Three Snakes, a Monkey and a Goddess Named Faith: The Rigveda's Impossible Authors

Name the Seer, the Metre, the God

There is a rule in traditional Vedic practice that sounds like bureaucracy and turns out to be metaphysics. Before you use a mantra, you state three things about it: its ṛṣi, the seer; its devatā, the deity addressed; its chandas, the metre. The declaration is still made before recitation in households that keep the morning rite. It is a kind of chain of custody. You are about to speak words that are not yours, and the protocol requires you to say whose mouth they came out of first.

Which means somebody had to know the answer for all 1,028 hymns.

Open the entry for RV 10.94 in the standard index and you get the usual three fields filled in with complete confidence. The deity is the pressing stones, which is already unusual: the hymn is addressed to lumps of rock. The seer is Arbuda Kādraveya, which is more unusual still, because Arbuda is a snake, a son of Kadrū, who in later Sanskrit mythology is the mother of all serpents. Check the only other two hymns in the collection addressed to those stones, RV 10.76 and RV 10.175, and the pattern holds: two more serpent seers, named and placed in their lineages.[15]

It does not stop at snakes. A hymn in the tenth book is credited jointly to Indra, his wife Indrāṇī and Vṛṣākapi, who is an ape. The hymn in praise of faith is credited to Faith, daughter of Desire. Three consecutive hymns list Indra as the poet of his own praise. One hymn’s seer is called Śiśu, which means “baby”. Another’s is called Laba, which may be a quail.

The reflex is to call this a muddle, or a pious fiction, and move on to the real philology. That reflex is a mistake, and an expensive one. The attribution lists are the oldest surviving attempt to do systematic scholarship on the Rigveda, made by people closer in time to the hymns than we are to Chaucer, working with a text whose authors had left no signatures. Their failures are not noise. They encode a working theory of what the author of a hymn even is, and that theory is stranger and more interesting than the one we use.

The Oldest Rigveda Scholarship Is a Filing System

The genre is called anukramaṇī, “enumeration”, and it is exactly what the name says: a catalogue. Six of them are traditionally ascribed to Śaunaka, each indexing a different axis of the text. One listed the seers, one the metres, one the deities, one the hymn divisions, one the opening words. Of those six, only the Anuvākānukramaṇī survives as a text; the rest are known from quotations in a twelfth-century commentator, Ṣaḍguruśiṣya, who had copies we do not.[2]

Kātyāyana’s Sarvānukramaṇī is the one that mattered. It folds the separate lists into a single compressed register: for each of the 1,028 hymns, the first word, the number of verses, the poet’s name and family, the deity, the metre.[2] It is a database schema, and like any schema it forces a value into every field. A hymn with no known author does not get a blank. It gets a guess, recorded in the same flat tone as a fact.

Macdonell edited the Sarvānukramaṇī in 1886, with extracts from Ṣaḍguruśiṣya’s commentary, as his Leipzig doctoral work, and the edition is still the one scholars reach for.[2] Eighteen years later he edited and translated the Bṛhaddevatā, which is the next step up in ambition: not just a list of deities but the legends that explain how each hymn came to be spoken.[3] Laurie Patton’s study of that text reads it not as credulous myth-collecting but as argument, a commentary that works by narrative rather than by gloss: tell the story in which these words were first said, and the words become usable in ritual.[4]

Methods note. The usual date given for Kātyāyana’s index is around the fourth century BCE. Treat that as a tradition-internal estimate, not a datum; Ṣaḍguruśiṣya’s commentary is twelfth century, the manuscripts are later still, and the compositional layers inside this literature are genuinely hard to separate. What is not in doubt is the direction of travel: the indexes are centuries younger than the hymns they index, and they were compiled by people who already found parts of the Rigveda obscure.[6][13]

Three Snakes and a Heap of Stones

Start with the serpents, because they are the hardest case and the one that shows the method working.

Three hymns in the tenth book are addressed to the grāvāṇaḥ, the stones that crush the soma stalks. All three are credited to seers explicitly marked as sarpa, snakes, each with a lineage: Jaratkarṇa Airāvata for RV 10.76, Arbuda Kādraveya for RV 10.94, Ūrdhvagrāvan Ārbudi for RV 10.175.[15]

Read the hymns and the choice starts to feel less arbitrary. These are poems about objects that make noise and do not die. The stones roar like a thousand men, they drink before the priest does, they go green at the mouth from the juice. The eleventh verse of RV 10.94 is a list of things the stones are not:

tṛdilā atṛdilāso adrayo aśramaṇā aśṛthitā amṛtyavaḥ

“Bored deep, but not pierced through with holes, are ye, O Stones, not loosened, never weary, and exempt from death, eternal, undiseased, moving in sundry ways, unthirsting, full of fatness, void of all desire.”

Rigveda 10.94, verse 11. Translation Griffith (1896) [11], romanisation from the Saṃhitā with accents omitted.

A deathless thing that lives in the ground, sheds nothing, tires of nothing, and speaks. If you were an indexer looking for a plausible mouth for that poem, a snake is not a joke. And there is a further point that a modern reader is likely to miss: serpent lore was a recognised branch of knowledge in this culture, not a folk superstition. When Nārada lists his accomplishments in the Chāndogya Upaniṣad, somewhere between astronomy and the fine arts he claims sarpavidyā, the science of snakes.[19] Snakes had experts. Experts can have hymns.

Three readings are on the table and no one can presently choose between them. The first is acoustic and thematic, as above: the hymn sounds like what a snake would say. The second is sociological: sarpa may name a human group or a priestly specialism whose traditions were marginal to the brahmin families that dominate the rest of the collection, and whose hymns to the ritual hardware were filed under their own names.[15] The third is the deflationary one, and it should not be dismissed. These are late hymns in the most heterogeneous book of the Rigveda, exactly where the indexers had the least family tradition to draw on; a strange hymn invited a strange author.

What the pattern does rule out is improvisation. All three stone hymns drew snakes. Whatever principle was operating, it was operating consistently, which is the signature of a method rather than a shrug. For what the stones were actually doing, and the engineering that went into making them louder, see the soma workshop.

Whoever Speaks, Sees

Most of the impossible authors dissolve once you notice the rule the indexers were following. They were not asking who composed the hymn. They were asking who is talking in it.

This works because a surprising number of Rigvedic hymns are scripts. Oldenberg argued in 1885 that the dialogue hymns, the ākhyāna hymns, are the surviving verse skeleton of a narrative form that mixed prose and poetry: the characters’ speeches were fixed in metre and memorised, the connective story was told in prose and was never canonised, so it evaporated.[7] What remains looks like a play with the stage directions torn out. Hand that to an indexer whose job is to fill in a seer field, and the cast list is the obvious place to look.

So the hymn in which Yamī propositions her twin brother is entered under Yama Vaivasvata as seer with Yamī as deity. The hymn in which Indra’s messenger bitch Saramā negotiates with the Paṇis across the river Rasā is entered under the Paṇis, who are demons, as seers, with Saramā as the deity: the thieves get the byline for the poem in which they refuse to return the cattle. (The hymn itself is worth reading on its own terms; see Saramā and the Paṇis.)

And then there is RV 10.86, the Vṛṣākapi hymn, which the index credits to Vṛṣākapi Aindra, Indrāṇī and Indra jointly. Vṛṣākapi is a male monkey, Indra’s favourite, who has fouled the offerings; Indrāṇī wants his head; Indra takes the monkey’s side. The verse where Indrāṇī makes her case is as blunt as anything in the collection:

“No Dame hath ampler charms than I, or greater wealth of love’s delights. None with more ardour offers all her beauty to her lord’s embrace.”

Rigveda 10.86, verse 6. Translation Griffith (1896) [11], who rendered the frankest lines with some restraint.

Three speakers, three names in the seer field, one of them an ape. It is not credulity. It is a filing convention applied without flinching.

The convention extends to ātmastuti, self-praise, which the tradition treats as a recognised type. RV 10.125 is spoken by Vāc, Speech, in the first person, and is entered under Vāc Āmbhṛṇī; the hymn argues its own case so completely that it needs no external poet at all, which is the subject of its own close reading. Hymns 10.48 to 10.50 are Indra talking about Indra, and the index obliges:

“Indra am I: none ever wins my wealth from me, never at any time am I a thrall to death. Pressing the Soma, ask riches from me alone: ye, Pūrus, in my friendship shall not suffer harm.”

Rigveda 10.48, verse 5. Translation Griffith (1896) [11].

Then RV 10.119, the most purely enjoyable hymn in the Rigveda, thirteen verses of a soma-drinker taking inventory of his own enlargement, each closing on the same refrain, kuvit somasyāpām iti:

“The draughts I drank have borne me up, as fleet-foot horses draw a car. Have I not drunk of Soma juice?”

“Aha! this spacious earth will I deposit either here or there. Have I not drunk of Soma juice?”

Rigveda 10.119, verses 3 and 9. Translation Griffith (1896) [11].

The seer, per the index, is Laba Aindra.

Aside. Nobody knows who or what Laba is. The name is opaque, laba in later Sanskrit is a word for a kind of quail, and the epithet Aindra ties the figure to Indra, who is plainly the speaker of the hymn. Whether the indexers meant a bird, a man with a bird’s name, or a figure from a story they knew and we have lost is not recoverable from the entry. I include the quail because it is the standard suggestion, not because the evidence settles it. The honest reading is that the Sarvānukramaṇī preserves a name whose referent was already gone.

Names Lifted Off the Page

The second method is cruder and easier to catch in the act. When no speaker presented itself, the indexers reached into the hymn’s own vocabulary and made a name out of it.

The great river catalogue at RV 10.75 is credited to Sindhukṣit Priyamedha, and sindhu-kṣit means “dweller on the Sindhu”, which is a job description extracted from the subject matter rather than a person (the hymn gets a full treatment in the Nadīstuti close reading). The battle hymn at RV 10.103 is credited to Apratiratha Aindra, “he whose chariot has no rival”, a word that belongs to the hymn’s own martial vocabulary; the hymn became battle liturgy for the king’s priest, as the weapons hymn piece describes. The hymn about the flying bird that carries speech, RV 10.177, is credited to Pataṅga Prājāpatya, and pataṅga means “flier”.[2]

The clearest specimen is the five-verse hymn to Faith, RV 10.151, whose seer is Śraddhā Kāmāyanī, Faith the daughter of Desire. Faith is the deity of the hymn, the subject of every verse, and, per the index, its author:

śraddhayāgniḥ samidhyate śraddhayā hūyate haviḥ

“By faith the fire is kindled, by faith the oblation is poured.”

Rigveda 10.151, verse 1, first line. Translation mine from the Saṃhitā text, accents omitted; compare Griffith (1896) [11].

“Guarded by Vāyu, Gods and men who sacrifice draw near to Faith. Man winneth Faith by yearnings of the heart, and opulence by Faith.”

Rigveda 10.151, verse 4. Translation Griffith (1896) [11].

Faith as its own author is not a slip of the pen. It is the logical end of the speaker-is-seer rule applied to a hymn with no human in it. The same instinct puts Hiraṇyagarbha, the golden embryo, down as the seer of the golden-embryo hymn at RV 10.121, and Prajāpati Parameṣṭhin as the seer of the creation hymn at RV 10.129. The poem’s content becomes the poem’s origin. Other abstractions got the same treatment across the late hymns, and the Rigveda’s abstract deities shows how quickly that family of nouns was hardening into persons.

My favourite entry in the whole index is quieter. RV 9.112 is the hymn about how everyone wants a different thing, famous for a verse that reads like a census return:

kārur ahaṃ tato bhiṣag upalaprakṣiṇī nanā

“A bard am I, my dad’s a leech, mammy lays corn upon the stones. Striving for wealth, with varied plans, we follow our desires like kine.”

Rigveda 9.112, verse 3. Translation Griffith (1896) [11], whose nursery diction catches the tone better than the dignified versions do.

The seer the index assigns is Śiśu Āṅgirasa. Śiśu means child, infant, baby. Somebody read a verse in which the speaker introduces his father and his mother, and concluded, reasonably enough, that the speaker was the kid.

This is circular, and the circularity is not hidden. Nothing in the Sarvānukramaṇī claims to be reporting a discovery; an entry is a label, assigned so the hymn can be filed, taught and used.

What the Lists Get Right

Having watched the index invent authors, it would be easy to throw it away. That would be a worse error than trusting it.

Jamison and Brereton, who had to make a judgement about this for every one of the 1,028 introductions in their translation, come down firmly on the side of use. The indexes are late, they say, and still invaluable: their identifications are by and large plausible, their groupings track the actual organisation of the collection, and the great majority of the roughly five hundred poets they name turn up independently in the pravara lists of brahmin ancestors, the lineage registers that priestly families kept for their own purposes.[1] That overlap is the strongest single argument that something real is being transmitted. You do not accidentally agree with an independent genealogical tradition five hundred times.

The hard limit was drawn by Oldenberg in 1888 and has held. He concluded that the compilers knew which family owned which of the family books, Maṇḍalas 2 through 7, and that beyond that they betray no sign of a genuine tradition about individual hymn authorship; Mayrhofer, surveying Rigvedic personal names a century later, endorsed the verdict.[6][9] That is a precise and useful finding. It means the attributions are reliable at the level of the clan, soft at the level of the hymn, and in places frankly manufactured, and it tells you which of the three you are looking at: family books, firm; Maṇḍalas 1 and 10, handle with gloves.

The family-level signal is not a matter of faith, either. The family books carry internal evidence of ownership: shared refrains, recurring sign-off verses, favoured metres, distinctive god-preferences, even house styles of simile. The Vasiṣṭhas sound like Vasiṣṭhas. I have written about those family signatures and refrains separately, and they line up with the index’s clan assignments well enough that the index can be used as a first approximation and then checked. Witzel built a political and dialect geography of the Rigvedic period partly on that foundation, cross-reading poet families against the chieftains they praise and the regions their dialects point to.[13][14]

graph TD
    A[Hymn needs a seer] --> B{Family book,<br/>clan markers present?}
    B -->|yes| C[Assign clan poet<br/>usually sound]
    B -->|no| D{Someone speaking<br/>in the hymn?}
    D -->|yes| E[Speaker becomes seer<br/>gods, demons, animals]
    D -->|no| F[Build a name from<br/>the hymn's own words]
    C --> G[Entry recorded<br/>in the same flat tone]
    E --> G
    F --> G

Authorship Before There Were Authors

Here is where the impossible authors stop being a curiosity and start being an argument.

Ṛṣi does not mean composer. The tradition glosses it as one who sees, and the hymns agree: Vedic poets describe their poems as found, seen, won, or fashioned out of materials they did not make, the way a wright fashions a chariot (the metaphor is the Rigveda’s own, and the subject of the poet as carpenter). Gonda’s long study of the vocabulary concluded that dhī and its relatives name a kind of visionary seeing rather than invention, and that the poet’s claim is to privileged perception, not to originality.[8] What the poets ask for is not inspiration in our sense but better eyesight.

Follow that out. If a hymn is something seen rather than something owned, then the question “who wrote this?” is malformed, and the question the indexers actually answered, “in whose mouth did these words first stand?”, is the right one. A monkey can be in a hymn’s mouth. So can Faith, if Faith is the one talking. The entries that look like category errors to us are category errors only inside a theory of authorship that nobody in this tradition held.

Notice what that costs. A culture with this theory cannot produce a literary history of individual genius, and it did not try to; what it produced instead is a transmission system of unmatched fidelity, in which the integrity of the words mattered more than the identity of the person, and in which the machinery of memorisation carried a corpus across three millennia with its accents intact. The trade was deliberate. You cannot have both.

Three cautions for the modern reader.

First, an attribution printed at the head of a hymn in any translation, including the best ones, is a quotation from a much later index, not a datum from the hymn. Read it as you would read a library classification: informative about the librarians.

Second, the absurd entries are not evidence that the tradition was gullible. They are evidence that it was systematic, which is rarer. A gullible tradition leaves gaps and contradictions; this one leaves a rule applied past the point of comfort.

Third, resist the inversion that says the strange attributions prove the hymns are strange. The Vṛṣākapi hymn is odd for its own reasons, and the pressing-stone hymns are not odd at all once you have watched soma being made. The oddity belongs to the index.

Read the Header First

The next time you open a Rigvedic hymn in translation, read the header before the poem. That one line of apparatus is a fossil of the moment when a tradition looked at a text it revered, found that the text had come to it with no authors attached, and refused to leave the field empty. It filled it from the only sources it had: the clan registers, the voices inside the poem, and the poem’s own words. The result is wrong in the places where no answer existed and right in the places where a real memory survived, and it is honest enough about both that we can still tell which is which, three thousand years later.

That is a better record than most ancient scholarship manages. The Rigveda’s own habit of asking who, of refusing to name what it does not know, from the creation hymn’s final shrug to the poets who openly wondered whether Indra existed at all, seems to have survived into the work of its first cataloguers, who were willing to write “a snake composed this” rather than write nothing.

Open RV 9.112 and read it with the header showing: a hymn about a poet, a healer and a woman grinding corn, signed by a baby. Then read the three pressing-stone hymns in a row, out loud and fast, and decide for yourself whether a serpent is really the wrong author for them. If you want the hymns whose authorship the tradition got demonstrably right, go to the family books and the poet lineages instead, where the index and the internal evidence shake hands.