The Rigveda's Word for Like Is Also Its Word for No

Seven Comparisons and a Word That Says No

RV 1.66 is a fire hymn with almost no fire in it. Parāśara Śāktya, to whom the tradition credits the whole run of Agni hymns from RV 1.65 to 1.73, opens by refusing to describe his god at all. He gives you seven other things instead.

rayir ná citrā sūro ná saṃdṛg āyur ná prāṇo nityo ná sūnuḥ takvā ná bhūrṇir vanā siṣakti payo ná dhenuḥ śucir vibhāvā

“Bright as wealth, as the sun is to look at; as life, as breath, as one’s own son; swift as a hunting bird he clings to the woods; as the milk of a cow, clear, blazing.”

RV 1.66.1. Literal rendering mine, following the standard reading.

Seven comparisons, no main clause worth the name, and one word repeated six times to hold the whole thing together: ná. It is the workhorse of Rigvedic imagery, the hinge every one of those pictures swings on.

It is also the ordinary Vedic word for “not”.

Not a homonym in some loose poetic sense, and not a pun. The two words are spelled identically, carry the same accent in the transmitted text, and are written the same way in the Padapāṭha, the word-by-word analysis the tradition built precisely so that nothing in the Rigveda could be lost or misread. A Vedic poet could write sūro ná and mean “like the sun”, or write sūro ná and mean “the sun does not”. Sometimes, in the same hymn, both.

English has nothing comparable. Imagine a language in which “no” and “like” were the same syllable and poets built their metaphors out of it. That is the situation, and the interesting part is not that the ambiguity exists but that it almost never bites. Vedic listeners resolved it instantly, in performance, at speed. What they were using to do that, and what the choice of ná did to the shape of Rigvedic poetry, is a question you can actually answer by counting.

Two Hymns, One Particle, Opposite Meanings

Put two verses side by side and the split is visible without any Sanskrit.

The first is from the Atri family’s Marut hymns in the fifth book. It is a blessing on the man the storm gods favour, and it piles up denials the way a charm piles up protections:

ná sa jīyate maruto ná hanyate ná sredhati ná vyathate ná riṣyati

“Never is he, O Maruts, slain or overcome, never doth he decay, ne’er is distressed or harmed.”

RV 5.54.7, tr. Griffith 1896.

Five ná in two lines, every one of them a “not”. Now the tenth book’s Marut hymn, a late composition and one of the strangest things in the collection, because it is built almost entirely out of comparisons:

“Like spokes of car-wheels in one nave united, ever victorious like heavenly Heroes, shedding their precious balm like youthful suitors, they raise their voice and chant their psalm as singers.”

RV 10.78.4, tr. Griffith 1896.

Four ná in two lines, every one of them a “like”. The hymn keeps it up for eight verses: twenty-one instances of the particle, the densest concentration in the Rigveda, and the Maruts are never once described directly. They are spokes, heroes, suitors, singers, stallions, charioteers, floods, Aṅgirases, all of it strung on the same two-letter word.

The tradition knew about the double life of ná and said so. Yāska’s Nirukta, the oldest surviving treatise on Vedic word-meaning, sorts particles into those used for comparison, those used for conjunction, and those that merely fill out a line, and it puts na in the comparison group alongside iva, cid and nu [1]. That is an early and deliberate piece of analysis: a grammarian looking at the word everyone knew meant “not” and recording that in the hymns it also means “like”. Two thousand years later Sāyaṇa, commenting in fourteenth-century Vijayanagara, regularly glosses the comparative ná with iva, which is the commentator’s practical way of telling a reader which of the two words is in front of them.

Aside. The particle iva is the other simile marker, and the one that won. In classical Sanskrit, iva and yáthā do all the work of comparison and the comparative ná has effectively vanished. Erica Biagetti’s study of iva traces how it broadened from marking a standard of comparison to marking approximation, the “sort of, as it were” function familiar from later literature [2]. The Rigvedic ná-simile is therefore a dead construction, preserved in the one text old enough to have used it heavily.

How the Ear Decides

If accent does not separate the two ná, something else must. The rule, stated in the reference grammars since the nineteenth century, is about position: comparative ná is postpositive and leans backward onto the word it compares, while negative ná leans forward onto the verb or clause it denies [3][4].

That is a rule with a testable consequence. If comparative ná must have something in front of it to attach to, it should never open a line. Negative ná, which introduces what it denies, should open lines constantly.

So I tested it. Taking the Saṃhitā text hymn by hymn, I pulled every free-standing ná, kept the verses containing exactly one of them, discarded any verse that also contained iva or yáthā (so that no other comparison marker could be responsible for the English), and then asked what Griffith’s 1896 translation does with each one: renders it as a comparison, as a negation, or neither. The point of using a full published translation as the classifier is that it is an independent judgment made verse by verse by someone who was not counting anything.

Of the 84 test verses whose single ná stands at the head of a hemistich, Griffith reads exactly one as a comparison. Sixty-four are negations, the rest ambiguous or paraphrased away. Of the 794 where the particle stands anywhere else in the line, comparisons outnumber negations by four to one.

A rule from a grammar book, confirmed by a translator who was not thinking about the rule, across nearly nine hundred independent cases. Nothing in Vedic philology is usually this clean.

Methods note. Three caveats, and they matter. First, the counts are of ná written as a separate word: Vedic sandhi fuses the particle to a following vowel often enough that the true total is higher than 1,584, and the invisible cases are not random. Second, before a vowel the enclitic naḥ “us” also appears as na, which is why the analysis above is restricted to the 1,584 instances followed by a consonant, where that confusion cannot arise. Third, Griffith renders freely, and my classifier looks for the English words “like” and “as” against “not”, “never”, “nor” and their relatives. A stricter pattern that ignores bare “as” gives 344 comparisons and 233 negations rather than 553 and 195, so the comparison share is somewhere between 60 and 74 per cent of decided cases depending on how generously you read the English. The direction of the result does not move.

graph TD
    A["A na in the line"] --> B{"Does it open the line?"}
    B -->|Yes| C["Negation, almost always"]
    B -->|No| D{"Is the word before it a noun?"}
    D -->|No| E["Negation: leans on the verb"]
    D -->|Yes| F{"Does that noun match a case in the clause?"}
    F -->|Yes| G["Comparison: X like a Y"]
    F -->|No| H["Genuinely open; context decides"]

Counting the Ambiguity

Across the whole Saṃhitā, 1,283 verses out of roughly 10,500 carry at least one unambiguous ná, and 581 of the 1,028 hymns use the word at least once. Four hundred and forty-seven hymns, in other words, run start to finish without needing either “not” or “like”.

Broken out by book, the pattern is not flat, and the shape of the variation is the interesting part.

Maṇḍala 9 is the outlier that makes sense the moment you see it. The Soma book is a thousand verses about a liquid moving through a filter, and a liquid in motion has almost nothing to be denied about it: there is no enemy to refuse, no rival poet to dismiss, no theological claim to negate. What there is, is the problem of describing a flow, which the Pavamāna poets solve by comparing it to every other moving thing they know. Of the ninety-eight verses tested in Maṇḍala 9, eighty-six are comparisons and four are negations. See the workshop of the pressing stones for what those poets were actually watching while they composed.

At the other end, Maṇḍala 8 and Maṇḍala 10 sit near or below half. Both are late, both contain argument rather than pure praise, and argument requires denial: the Rigveda’s one flat statement of disbelief, the dialogues, the riddles, the cosmological speculation. Once a text starts saying what is not the case, its share of ná shifts accordingly. The particle is a quiet genre indicator.

Maṇḍala 3’s low rate, 9.4 per hundred verses against 19.7 in Maṇḍala 4, is harder to explain and I will not pretend otherwise. Book-level stylistic differences among the family collections are real and well documented [5], but attributing this one to Viśvāmitra’s clan style rather than to accidents of subject matter would need the kind of hymn-by-hymn work this article has not done.

The Image Bank

The word before a comparative ná is the thing being compared to, which means the particle hands you a clean, mechanically extractable inventory of what the Rigvedic poets thought was worth thinking with. Not what they wrote hymns about, which is gods; what they reached for when they needed a picture.

Here is the top of that list, with the recurring comparanda grouped by domain.

Three things stand out.

The first is how small the world is. These are the comparanda of people who live with animals, move by cart and horse, cook with clarified butter, watch weather for a living and fear rivers. No simile is drawn from writing, because there was none; almost none from architecture, which matches how little the Rigveda says about its own houses; a few from fortifications, but as things Indra smashes rather than things anyone lives in.

The second is the direction of comparison. A Rigvedic god is regularly likened to a human being and to a domestic animal, not the reverse. Sūrya pursues Dawn the way a young man goes after a girl he has been watching. Hymns lick Indra the way mother-cows lick a calf, and the verb in RV 3.41.5 really is rihanti, “they lick”. The traffic of prestige in these images runs opposite to what a later, more theologically nervous literature would allow, and it is one of the reasons the Rigveda reads less like scripture than like a record of people paying close attention.

The third is that the image bank includes the unglamorous. One line, repeated in two widely separated hymns, is the bleakest thing in the collection:

“As rats eat weavers’ threads, cares are consuming me, thy singer, Śatakratu, me. Have mercy on us once, O Indra.”

RV 10.33.3, tr. Griffith 1896; the same line stands at RV 1.105.8.

mūṣo ná śiśnā vyadanti mādhyaḥ. A poet in the middle of a hymn to the most powerful god in his pantheon reaches for mice, and for a gnawing that never quite finishes. The image survives into two hymns because it was too good to lose, which is how repetition works in an oral corpus.

Where “Like” Might Have Come From “Not”

The obvious question is whether the two ná are one word or two, and if one, which came first.

The majority view runs from negation to comparison, by a route that sounds odd stated baldly and reasonable once you hear it in a line. Start with a construction of negative parallelism: “he is not a bull, he roars”. The denial is not really a denial; it sets up an identity the speaker then half withdraws, and what the listener takes away is “he is bull-like”. Repeat the pattern often enough in a tradition that prizes compression, let the negation bleach out of it, and the particle is left marking the comparison alone. Recent corpus work on Vedic comparison constructions takes this development as the standard account, with the ná-simile turning syntagmatic while iva specialises for figurative comparison [2][6].

The competing view is that Vedic simply inherited two particles of identical shape and that the resemblance is an accident of sound. This is not an unreasonable position, because languages do collide words this way constantly, and because the derivation above, once you try to pin it to actual Rigvedic passages, is hard to catch in the act. The passages that ought to be the bridge are exactly the ones every translator has to make a decision about, which makes them weak evidence for anything.

What can be said with confidence is narrower and more interesting than either story. In the Rigveda, comparative ná and negative ná are already fully distinct in behaviour: different position, different prosodic attachment, different distribution across books. Whatever their history, the poets were not punning. They were using two tools that happened to sound alike, the way an English speaker uses “left” without ever noticing the other “left”.

graph LR
    A["Negative parallelism"] --> B["Simile marker na"]
    B --> C["Rigveda: na and iva share the work"]
    C --> D["Middle Vedic: iva spreads"]
    D --> E["Classical: iva and yatha only"]
    B -.-> F["Rival view: two inherited particles"]

What English Does to It

Every translation of a ná-simile adds something the Sanskrit does not have. This is not a complaint about translators; it is a structural fact about the two languages, and it changes how the poetry feels.

The Vedic construction is a bare juxtaposition. Noun, particle, and then the clause the noun is supposed to illuminate, with no verb repeated, no “as when”, no relative clause. payo ná dhenuḥ: milk, ná, milk-cow. You are handed two objects and a signal that they belong together, and you do the work. English cannot do that without sounding broken, so a translator supplies the joints: “like the milk a cow yields”, “as ’twere a radiant-coloured cow”. Griffith is unusually visible in this, because he was working in verse and needed syllables; Jamison and Brereton’s 2014 version, which sets out to be readable as English poetry while staying literal, still has to choose a hinge for each one [7].

The result is that Rigvedic similes read as more explicit in English than they are in Sanskrit. Try it the other way: keep ná untranslated and see what the poet actually gave his listeners.

“Bright wealth ná, sun’s look ná, life ná, breath, one’s own son ná; a hunting bird ná, quick, he clings to the woods; milk ná, a cow, clear, blazing.”

RV 1.66.1, rendered with the particle left in place.

That is closer to the experience: a rapid series of nouns, each detonating a picture and then dropping it, and the god assembling himself in the gaps. The technique is much nearer to what a modern reader would call imagism than to the elaborate upamā of classical Sanskrit poetics, which is built on iva and on a theory of comparison the Rigveda does not have. The Rigvedic poets were craftsmen who talked about hymns as objects they built, and this is one of their tools: the smallest possible joint between two things.

It also explains a habit of the collection that puzzles first-time readers. Rigvedic hymns can seem to lurch between registers, a flood, then a suitor, then a spoke of a wheel, with nothing in between. The lurching is the form. When your comparison marker costs one syllable and needs no clause, you can afford a new image every half-line, and the poets spend freely. Compare what happens when the same tradition wants sameness rather than likeness: jāmi, the word for sameness and kinship, is a fault in ritual and a danger in poetry. Resemblance is productive; identity is a problem. The particle sits exactly on that line, which may be why the tradition was content to let it also mean “no”.

Reading With the Grain

The next time you open the Rigveda in translation, try a small exercise. Find the similes on a page, count them, and then ask what the poet gained by comparing rather than stating. In a Pavamāna hymn from Maṇḍala 9 you will find a comparison in nearly every verse and a god who is never described directly at all. In a dialogue hymn or a philosophical one you will find almost none, and a lot of denial: the Nāsadīya’s famous stack of negatives is the extreme case, a cosmogony written entirely out of what there was not.

Both are the same word. A tradition that had no script, no punctuation and no way to mark which sense was meant handed its poets a particle that could deny or compare depending on where it stood in the line, and the poets used it for likeness roughly twice as often as for denial without, so far as we can tell, ever confusing an audience. That is a fact about grammar. It is also a fact about how carefully these hymns were composed, and how much of their meaning was carried by shape rather than by vocabulary, which is the same thing that let the whole collection travel three thousand years by mouth without a written copy.

Start with RV 10.78 and read the eight verses aloud. It is the most concentrated piece of simile-writing in the book: no statement about the Maruts survives the hymn, only a chain of things they are like, and the chain holds. Then open RV 5.54 at verse 7 and hear the same syllable five times in two lines, meaning the opposite. The Rigveda expects you to know the difference. The remarkable part is that, within a line or two, you do.