The Two Mitras and the Two Varuṇas: The Rig Veda's Grammar of Pairs

Which of You Two Was First?

A poet in the first book of the Rigveda is looking at the sky and the ground, the two largest things he knows, and he wants to establish their birth order. He fails, and says so in the opening line of the hymn.

katarā́ pū́rvā katarā́parāyoḥ kathā́ jāte kavayaḥ ko vi veda |

víśvaṃ tmánā bibhṛto yád dha nā́ma ví vartete áhanī cakríyeva

“Which of the two is the earlier, which the later of the pair? How were they born, poets? Who understands it? By themselves the two carry everything that has a name. Day and night roll on like two wheels.”

Rigveda 1.185.1, rendering mine from the Saṃhitā text.

Read that last clause again in the Sanskrit. The words for “day and night” are one word, áhanī, which is not a compound and does not mean “day and night” in any dictionary sense. It is simply the word for “day”, put into a grammatical form that Vedic Sanskrit has and English does not: a form used when there are exactly two of something. Say áhanī and you have said “the day-pair”, and a Vedic listener knew without being told that the second member was night.

This is a small thing until you notice how far it goes. Vedic has three numbers, not two. Singular, dual, plural. The dual is not optional or literary; it is compulsory, in nouns, adjectives, pronouns and verbs, and using a plural for two things is as wrong as saying “three god” in English. A language with that equipment tends to notice pairs. And the Rigveda, which is the oldest surviving document composed in such a language, took the habit further than mere grammar required: it built a theology out of it.

Fifty-seven of its hymns are addressed not to a god but to a couple. Heaven and Earth are asked for protection as a unit. Indra and Agni are told they are twins. Soma and Rudra are asked jointly to drive sickness out of the house. And the most famous of these couples, Mitra and Varuṇa, carry a joint name whose literal grammatical content is stranger than any translation lets on. Mitrā́váruṇā does not say “Mitra and Varuṇa”. It says, syllable by syllable, “the two Mitras, the two Varuṇas.”

What follows is an attempt to take that oddity seriously: where the form comes from, how far back it goes, what the poets did with it, and why a grammatical category that dropped out of India’s spoken languages long ago left its name on a philosophical idea that is still in circulation.

A Number for Two

English speakers have one surviving fossil of this system and use it without noticing: both, either, neither, between. We feel a difference between “between the two” and “among the three”, and that faint pressure is the ghost of a category Vedic had in full.

Vedic marked it everywhere. One god is deváḥ, two gods are devaú, many gods are devā́ḥ. One of you is tvám, two of you are yuvám, more of you are yūyám. He carries, bhárati; the two of them carry, bhárataḥ; they all carry, bháranti. Imperatives addressed to two people have their own ending, which is why so many Rigvedic lines end in that characteristic -tam: pibatam, “drink, you two”; avatam, “help, you two.”

Nothing here is exotic. Ancient Greek had a dual, Old English had one in its pronouns, Slovene still uses one daily, and Arabic and Hebrew keep theirs. What is unusual about Vedic is not that the category exists but that the poets found it theologically interesting. In the Rigveda the dual is not just a way of counting. It is a way of asserting that two things belong together so tightly that they take one verb, one epithet and, eventually, one name.

Methods note. The counts in this article come from the Saṃhitā text in Devanagari and Roman transliteration, searched as strings, verse by verse, rather than from a parsed and lemmatised corpus. Two consequences follow. Numbers of the type “91 verses contain a form of mitrā́váruṇā” are reliable, because that string is unambiguous. Numbers for short forms like mitrā́ on its own are not, because without accents the dual mitrā́ and the vocative singular mitra are hard to separate mechanically. Where the distinction matters below, I lean on the philological literature rather than on my own grep.

The Word That Means “X and His Partner”

Start with the simplest case, because it is also the oldest. Vedic can put a single noun in the dual and let it stand for two different things, one named and one merely implied. Grammarians call this the elliptic dual, and Samuel Grant Oliphant devoted a whole instalment of his long study of the Vedic dual to it in 1912.[1]

The commonest example is domestic. Pitárā is the dual of “father”, so it ought to mean “two fathers”. It means father and mother, that is, parents. It turns up in roughly nineteen verses, often of Heaven and Earth as the parents of everything, as in the Dyāvāpṛthivī hymn where the poet muses on “the mind of the father who does not deceive and the great inherent power of the mother” and then calls them, in one word, suretásā pitárā, “the two well-seeding parents” (RV 1.159.2).

The same trick runs through the vocabulary.

That last line is the load-bearing one. In Vedic you can say mitrā́ and mean Mitra together with his inseparable colleague Varuṇa, exactly as you say pitárā and mean father together with mother. The named member stands for the pair; the unnamed member is understood, the way a listener understands who the other parent is.

Jacob Wackernagel noticed in 1877 that Homer does the same thing. The Iliad’s Aíante, a dual, is usually translated “the two Ajaxes”, meaning Ajax son of Telamon and Ajax son of Oileus. Wackernagel argued that the original sense was Ajax and his brother Teucer, the named hero standing for himself plus his constant companion, which is the Vedic pattern precisely.[2] The construction is therefore not a Vedic mannerism. It is inherited, and it is older than any of the traditions that use it.

The Iranian evidence tightens the case. Avestan, the Rigveda’s closest linguistic cousin, has the archaic formula miθra ahura bərəzanta, “Mithra and Ahura, the two lofty ones”, and the compounds miθra.ahura and ahura.miθra in the Mihr Yašt.[3] Two priesthoods separated by centuries and a mountain range, one in the Punjab and one on the Iranian plateau, were addressing the same pair of gods with the same grammatical device. Whatever this pairing is, it predates both.

What Mitrā́váruṇā Literally Says

Now put two elliptic duals side by side and you get the Rigveda’s characteristic divine name. Mitrā́ is the dual of Mitra. Váruṇā is the dual of Varuṇa. Join them and you have a compound in which both halves are separately in the dual: “the two Mitras, the two Varuṇas”, meaning one Mitra and one Varuṇa. Sanskrit grammarians call the type devatā-dvandva, a “divinity co-compound”, and it is the only class of Sanskrit compound built this way.

Two further features mark it out as something not quite a single word.

The first is accent. Ordinary Sanskrit compounds have one accent, because a word has one accent. In the Rigveda, mitrā́váruṇā has two, one on each member, which is exactly what you expect if the sequence is really two words that have been standing next to each other for a very long time. Paul Kiparsky, who devoted an article in Language to the problem, described these compounds as being “in some respects like asyndetically coordinated syntactic phrases, in others like single compound words”, each element carrying its own lexical accent.[4]

The second is separability. Words cannot normally be cut in half by other words, and the fact that these can is, as Tamisha Tan puts it in a recent study of the type, the most compelling evidence available for two-word status.[5] The Rigveda supplies examples. In a hymn of the fifth book the poet says:

adhārayatam pṛthivī́m utá dyā́m mitrā́ rā́jānā varuṇā máhobhiḥ

“You two upheld the earth and also the sky, Mitra, you two kings, Varuṇa, by your greatnesses.”

Rigveda 5.62.3, rendering mine. Griffith (1896) has “O Mitra-Varuṇa, ye by your greatness, both Kings, have firmly stablished earth and heaven.”[6]

Griffith reassembles into a hyphenated unit what the Sanskrit deliberately holds apart. In the original the phrase “the two kings” sits between the two names, and the sentence still parses, because the two names were never a single word to begin with. They were a formula: two nouns, each in the dual, habitually adjacent, on their way to becoming a compound but not there yet.

They did arrive eventually. Pāṇini, codifying the language perhaps a thousand years later, has to write special rules for these divine pairs, because by his time they no longer behave like anything else: ānaṅ ṛto dvandve and devatādvandve ca (Aṣṭādhyāyī 6.3.25 and 6.3.26) handle the peculiar shape the first member takes in exactly this class.[7] A grammar famous for its economy spends rules on a construction that had already become a museum piece.

graph TD
    A["Noun in the dual"] --> B["Elliptic dual: mitra = Mitra + partner"]
    B --> C["Two elliptic duals side by side"]
    C --> D["Split phrase: mitra rajana varuna"]
    C --> E["Fixed formula, two accents"]
    E --> F["Devata-dvandva compound"]
    F --> G["Panini writes special rules"]
    F --> H["Classical dvandva: one accent, plural"]
    H --> I["Dvandva as 'pair of opposites'"]

Not all the pairs are built alike, which is itself informative. Tamara Ditrich, surveying the Rigvedic material, distinguishes several structural types that look like different stages of the same process.[8] Mitrā́váruṇā has both members in the dual. Índravāyū, “Indra and Vāyu”, which the text uses seventeen times, has only the second member marked, the first standing in a bare stem form. Agnī́ṣómā, “Agni and Soma”, shows the internal sound change that Sanskrit applies inside words rather than between them, which suggests the two halves had fused more tightly; Tan counts twenty-one occurrences, all of them with that change applied.[5]

Aside. Why be confident about accents in a text transmitted by mouth? Because the Vedic schools preserved the tonal accent in recitation and in the analytic padapāṭha, the word-by-word recitation that breaks every compound apart and shows each element as the tradition understood it. That apparatus, described in more detail in the piece on oral transmission and the padapāṭha, is why an argument about the accentuation of a Bronze Age compound can be made at all.

Who Gets Paired With Whom

The pairing is not a fixed list handed down from some earlier pantheon. It is a live template, and Rigvedic poets applied it freely.

Counts follow the traditional ascriptions as given in Griffith’s headings, which reproduce the Anukramaṇī; a few hymns are shared or disputed, so treat the totals as close rather than exact.

Notice who dominates the left column. Indra, the most sung god in the collection, is also the most sociable: he is paired with Agni, Varuṇa, Soma, Viṣṇu, Bṛhaspati, Pūṣan and Vāyu, and beyond the hymn dedications the text also produces índrāparvatā, “Indra and the Mountain” (RV 1.122.3, 3.53.1) and even índrākutsā, “Indra and Kutsa”, where the partner is a human hero riding in the chariot beside him (RV 5.31.9). The mould is available for anything that comes in twos.

That includes things nobody would call gods in a systematic theology. Sūryā́māsā, “Sun and Moon”. Parjányavātā, “Rain-cloud and Wind”. And the pair whose two names appear in both orders, uṣā́sānáktā and náktoṣā́sā, dawn-and-night and night-and-dawn, depending on which one the poet wants first (RV 1.122.2; RV 1.13.7). The reversibility is a good indication that these were never frozen titles. They were a way of speaking, and speakers improvised in it.

At its most compressed the pattern can stack. One hymn of the seventh book opens by asking for well-being from four different divine couples in four consecutive lines:

śáṃ na indrāgnī́ bhavatām ávobhiḥ śáṃ na indrā́váruṇā rātáhavyā |

śám indrā́sómā suvitā́ya śáṃ yóḥ śáṃ na indrā́pūṣáṇā vā́jasātau

“Well-being to us from Indra-and-Agni with their help; well-being from Indra-and-Varuṇa, to whom the oblation is poured; well-being and health from Indra-and-Soma, for good going; well-being from Indra-and-Pūṣan in the winning of prizes.”

Rigveda 7.35.1, rendering mine.

Four pairs, one after another, built to the same pattern by a poet who could clearly generate more of them at will.

What the Second God Is For

A reasonable objection at this point: perhaps the pairs are just an economy of praise, two gods invoked at once to save the poet a hymn. The text argues otherwise, because in a pair hymn the poets keep doing something they do nowhere else. They divide the labour.

“Praise those Twain Gods for powers that merit worship, Indra and Varuṇa, for bliss, the joyous. One with his might and thunderbolt slays Vṛtra; the other as a Sage stands near in troubles.”

Rigveda 6.68.3, translated by Ralph T. H. Griffith (1896).[6]

The Sanskrit is blunter than the English: anyáḥ … anyáḥ, “the one … the other”. Within a single verse addressed to a couple, the poet distinguishes the violent partner from the wise one, and then goes on addressing them jointly. This is the pattern that runs through the Mitra and Varuṇa material as well. Mitra is the god of the agreement, the alliance, the thing that holds because both parties want it to; Varuṇa is the god of the oath and its enforcement, the one who watches and binds. Neither is complete. Together they are how order actually works: consent plus sanction.

Jan Gonda gave the phenomenon a four-hundred-page monograph in 1974, and his conclusion was that dual deities are not an addition sum.[9] The pair is a unit with its own character, invoked as a unit, offered to as a unit, and doing things that neither member does alone. Gonda followed it two years later with a companion volume on triads, which tells you something about how systematically the Vedic ritualists thought in small numbers.[10]

Sometimes the joint identity is asserted outright:

“Thus, Indra-Agni, verily your greatness merits loftiest praise, Sprung from one common Father, brothers, twins are ye; your Mother is in every place.”

Rigveda 6.59.2, translated by Griffith (1896).[6]

Consider what that claims. Agni is fire on the ground, the priest and the mouth of the offering. Indra is the storm champion who breaks Vṛtra and lets the waters out. Their functions barely overlap, and the poets who sing to them separately treat them as entirely different powers. Yet here they are declared yamáu, twins, with one begetter and mothers everywhere. The pairing has produced a kinship claim that nothing else in the mythology supports. The form came first and the theology followed it.

Aside. Georges Dumézil built an influential comparative theory on Mitra and Varuṇa as the two faces of Indo-European sovereignty, the jurist and the magician, and the reading has real explanatory power.[11] Two cautions. First, Dumézil’s scheme is a hypothesis about inherited structure, not a description of what any Rigvedic poet believed. Second, it fits Mitra and Varuṇa beautifully and the other twenty-odd pairs much less well; a theory of complementary sovereign functions does not obviously predict Soma-and-Pūṣan, and it certainly does not predict Indra-and-the-Mountain. Joel Brereton’s study of the Ādityas is a good corrective, staying close to what the hymns actually say about these gods.[12]

The limit case is the Aśvins, who receive fifty-three hymns of their own and who are never separated at all. They ride one chariot, share every epithet, and act in unison in every story about them. In the Rigveda neither of them has a name of his own; they are simply the two, the Nāsatyas, and no verse tells you what to call each. A pair with no separable members is the natural end point of a system that had a grammatical form for exactly two.

Heaven and Earth, and the Question That Cannot Be Answered

The oldest pair of all is the one the hymn quoted at the start could not sort into first and second. Dyā́vāpṛthivī́, Heaven and Earth, appears in seventy-eight verses under that name, plus a second joint name dyā́vākṣā́mā, plus the word ródasī, “the two world-halves”, which turns up in about a hundred and ninety verses. That is a great deal of attention for a pair with only six dedicated hymns, and the reason is that Heaven and Earth are not one item in the pantheon among others. They are the frame everything else happens inside.

The hymns treat them as a married couple, as parents, and as a mechanism. They are pitā́ mātā́ ca, father and mother, keeping creatures safe (RV 1.160.2). They are motionless and footless, and they hold up a moving, footed world (RV 1.185.2). They are propped apart by Varuṇa’s ordinance and cannot age (RV 6.70.1). And their relation is symmetrical enough that the poet of RV 1.185 cannot say which of them came first, a question he puts to other poets rather than to the gods.

That question deserves a moment. It is not rhetorical piety. It is the same intellectual move that produces the Rigveda’s most famous act of self-doubt, the creation hymn that ends by wondering whether even the highest overseer knows how it all began. The Nāsadīya hymn is usually read as a late and unusual outbreak of scepticism. RV 1.185 shows the same instinct applied to a much more ordinary object: two things so evenly matched that priority between them is undecidable, and a poet honest enough to say he does not know. The dual gave him a way to hold the two in view at once without ranking them, which may be precisely why the question could be posed at all.

Notice also what the pairing does to translation. Griffith renders dyā́vāpṛthivī́ as “Heaven and Earth”, which is correct and loses the point: in Sanskrit it is one word taking one verb in the dual, and English has no way to say “these two, considered as one thing”. When you read “Mitra-Varuṇa, ye Kings” in an English Rigveda, the hyphen is doing work that Sanskrit did with morphology.

The Number That Died

Here is the strange ending. The dual, so central to Vedic that its gods were built out of it, did not survive in India. Middle Indo-Aryan lost it: Pali and the Prakrits have no dual, using the plural for two, with only frozen relics like the numeral “two” itself. Modern Hindi, Bengali, Marathi and Punjabi have nothing left of it. A grammatical category that had been obligatory for every noun, adjective, pronoun and verb in the language simply went away, and with it the automatic reflex of treating two-ness as different in kind from many-ness.

What survived is the word. Dvandva (Sanskrit: द्वन्द्व) is built from dvaṃ dvam, “two by two”, and it began as a technical grammarians’ label for the co-compound, the type Pāṇini needed rules for. It then moved out of grammar and into ordinary and philosophical Sanskrit, where a dvandva is a pair of opposites: heat and cold, pleasure and pain, honour and dishonour. That is the sense the Bhagavad Gītā uses when it tells the practitioner to become nirdvandva, free of the pairs, and it is the sense in dvandva-yuddha, single combat, a fight between exactly two. A term coined to describe a shape in the language ended up describing a shape in experience. Anyone tracing how Vedic vocabulary turned into later Indian philosophy can watch the whole path here, from morphology to metaphysics, in one word. The map of Vedas, Upaniṣads and Gītā shows the textual side of that journey; this is its grammatical shadow.

There is a temptation to make more of this than the evidence supports, and it should be resisted. Vedic poets did not have a doctrine of complementarity because their language had a dual, any more than English speakers lack one because ours does not. Languages with duals are common and their speakers are not visibly more attentive to couples. What the Rigveda shows is narrower and more interesting: a body of poets who found a grammatical resource lying in their language and used it as a compositional device, generating gods in pairs the way they generated metres and refrains, until the pairing itself became something they had opinions about. That is a choice, not a consequence.

Two observations to take away from it.

The first is about reading. When a Rigvedic hymn addresses “you two”, it is not an artifact of translation but a live grammatical form on every verb in the verse, and the hymn is constructed around it. Griffith’s “ye” and Jamison and Brereton’s “you two” are both trying to render something English cannot mark, and the reader who keeps that in mind will find the pair hymns much less repetitive than they look.[13]

The second is about the shape of the pantheon. We tend to describe Vedic religion as a list of gods with a ranking, so many hymns to Indra, so many to Agni, and to argue about who was chief. The dual material suggests the poets were not always working with a list. Sometimes they were working with a set of relations, and a relation needs two terms. Fifty-seven hymns is more than the Rigveda gives to Varuṇa, Rudra and Viṣṇu combined when each is counted alone. On the evidence of what the poets actually chose to sing, the couple was a category of divinity in its own right, and not a marginal one.

Open RV 1.185 and read it once for the argument and once for the grammar. Count how many times a single verb, a single adjective, a single request is aimed at two things at once, and how rarely the poet feels any need to say which of them is doing what. Then try to reproduce the effect in English. The difficulty of doing so is the measure of what the form was worth.