How many gods are there?' Yajnavalkya answered in public, and Max Muller had to invent 'kathenotheism

Vidagdha Shakalya put the question to Yajnavalkya in front of King Janaka's assembled court, and it was not a friendly question. How many gods are there? The answer came back as a number, then a smaller number, then a smaller one, until there was nothing left to divide. Three hundred and three, and three thousand and three. Then thirty-three. Then six, then three, then two, then one and a half, then one.

That exchange is Brihadaranyaka Upanishad 3.9, and it sits in the oldest layer of Upanishadic thought, centuries before anyone in Europe had a category for what the Veda was doing. The tradition asked itself the counting question first, in public, and answered it by taking the count apart on the spot.

So the short reply to "how many gods does Hinduism have" is that the tradition answered the question itself, in public, close to three thousand years ago, and answered it more completely than the modern versions of the question ever manage. The question wants a census, and the tradition was never running one. What it was running instead is hard to name in English, which is why a nineteenth-century philologist reached for a word nobody had needed before.

The question, asked by a hostile expert

Book three of the Brihadaranyaka is a public contest. Janaka of Videha has put up a prize of a thousand cows with gold fastened to their horns, Yajnavalkya has claimed it before answering anything, and the assembled brahmins take turns trying to unseat him. Shakalya is the last of them, and by then the mood in the hall is not collegial.

Shakalya opens with the counting question, and Yajnavalkya's first answer is not his own. He quotes: as many as are named in the nivid, the invocation formula belonging to the hymn of the All-Gods. Three and three hundred, three and three thousand, which adds to three thousand three hundred and six.

Shakalya says yes, and asks again. How many gods are there really? Thirty-three. Yes, and how many really? Six. Yes, and really? Three. Two. One and a half. One.

Each rung gets an identification as the brahmana proceeds. The thirty-three are named at 3.9.2. The six are fire, earth, wind, the intermediate region, the sun and the sky. The three are the three worlds. The two are food and breath. The one and a half is the wind that blows, and the text accounts for the awkward half with a play on words rather than an argument. The one is prana, breath, which Yajnavalkya says is Brahman, and that they call it Tyat, "That."

Notice what Yajnavalkya does not do. He never says the three thousand three hundred and six were an error, a superstition, or somebody else's religion. He accepts the number, then shows it is the same thing counted at a coarser resolution. Every rung is true at its own level. The descent is not a correction. It is a change of magnification.

Book three closes badly for Shakalya. Yajnavalkya turns the questioning around, Shakalya cannot answer, and the text says at 3.9.26 that his head shattered and robbers carried off his bones, having mistaken them for something else. The tradition never softened that ending. This was a knowledge contest with real stakes, not a devotional set piece.

Thirty-three is a structure, not a roll of names

Thirty-three is the number the Vedic corpus keeps returning to. Rigveda 1.139.11 addresses eleven gods whose home is heaven, eleven who dwell on earth, and eleven who live in the waters. Three registers, eleven each. The number is doing architectural work, mapping the gods onto a three-storey cosmos rather than listing personalities. The best evidence that this is a structure rather than a census is that the texts cannot agree on who fills it.

Brihadaranyaka 3.9.2 gives eight Vasus, eleven Rudras and twelve Adityas, which comes to thirty-one, and then adds Indra and Prajapati to reach thirty-three. The Shatapatha Brahmana, at 4.5.7.2, has the same eight, eleven and twelve, and then fills the last two places with Dyaus and Prithivi, Heaven and Earth. Yaska's Nirukta, the oldest surviving work of Vedic etymology, puts the two Ashvins in those places instead.

The total is fixed and the roster is not. Three respected sources agree on thirty-one of the thirty-three and disagree about the final two. That is not what a list of gods looks like. It is what a numbered slot in a cosmology looks like.

The Brihadaranyaka then does something stranger still. Asked what the eight Vasus are, it answers with fire, earth, wind, the intermediate region, the sun, the sky, the moon and the stars. The eleven Rudras turn out to be the ten faculties in a human body with the mind as the eleventh. The twelve Adityas are the twelve months of the year. The gods are being read as the furniture of the cosmos, the equipment of a person, and the divisions of time. Read in the tradition's own direction, the identification is expansive rather than reductive. The divine is not being shrunk into the world. The world, the person and the year are being disclosed as divine.

The count was never stable even inside the Rigveda. At 3.9.9, and again at 10.52.6, three thousand three hundred and thirty-nine gods attend on Agni. That is neither thirty-three nor the three thousand three hundred and six Yajnavalkya quotes. Nobody appears to have treated the discrepancy as damage needing repair, and that composure is itself the tell. A tradition that thought the numbers were the point would have reconciled them. The Rigveda keeps multiple accounts of creation standing side by side in the same untroubled way, a pattern examined in the four creation hymns.

Thirty-three crore, and the misreading underneath it

The number most people have actually heard is not thirty-three. It is thirty-three crore, three hundred and thirty million, quoted at Hindus as often as by them. The figure rests on one word, and the word does not bear it.

The phrase in question is trayastrimsati koti, and reading koti as crore is the misreading. Monier-Williams gives koti as the point or edge of a sword, the curved end of a bow, the horns of the moon, the highest point, eminence or excellence, a side in an argument, and a crore. "Highest point" and "eminence" are genuinely there, and "the thirty-three pre-eminent ones" is a defensible reading. The familiar defence overshoots only when it glosses koti as "type" or "category", a sense the lexicon does not support, and it never needed the stretch. The classes are already in the texts themselves: eight Vasus, eleven Rudras, twelve Adityas and two more, the same architecture of thirty-three the Veda keeps returning to.

There is a second point worth being straight about, because precision here is the tradition's friend. The phrase circulates without a locus. It gets attributed to the Atharvaveda, the Yajurveda and the Shatapatha Brahmana, usually with no verse attached, and this article is not going to invent one either. What can be verified is the decisive half: the Vedic counts that do exist give thirty-three, and three hundred and thirty million is not among them. The scriptures never made the claim the jibe attacks.

A note on where these numbers can and cannot be checked. Brihadaranyaka 3.9, Rigveda 1.139.11, Rigveda 3.9.9, Rigveda 10.52.6 and Shatapatha Brahmana 4.5.7.2 are all locatable and quotable; the working translation behind the Upanishadic material here is Patrick Olivelle's The Early Upanisads (Oxford, 1998). The phrase trayastrimsati koti is a different matter. Treat any citation of it as unverified until someone shows you the passage.

Now the part that matters. Winning the arithmetic is not the whole answer, because "you counted wrong, the correct total is smaller" still accepts the premise that a religion's seriousness is a function of how few gods it has. The tradition never accepted that premise, and it did not wait for the modern quarrel to refuse it.

Shakalya asked for a number and got seven of them, each true at its own resolution. That is the older and better answer, and it was on record millennia before the taunt it settles.

Muller's problem, and the word he actually coined

By the middle of the nineteenth century, European scholars reading the Rigveda closely had run into a pattern their vocabulary could not hold. Hymn after hymn addressed a single deity in language of unqualified supremacy: this one made the worlds, holds up the sky, is beyond all rivals. Then the next hymn did the same for a different deity, in the same terms.

Polytheism did not describe it, because in a polytheistic system the gods are ranked and a supreme god sits at the top. Monotheism did not describe it either, because the other gods were not denied or demoted. They were simply not in view while the poet was addressing this one.

Friedrich Max Muller worked at this problem for decades and set it out at length in the sixth of his Hibbert Lectures, delivered at Westminster Abbey in 1878 and published as Lectures on the Origin and Growth of Religion as Illustrated by the Religions of India. The lecture is titled "On Henotheism, Polytheism, Monotheism, and Atheism." What he was naming, in his own summary phrases, was a belief in single gods, a worship of one god after another, in which each deity is to the worshipper at the time a real divinity, supreme and absolute, and not bounded by the powers of any other.

Here the popular version of the story needs correcting, and the correction is not small. Muller did not invent the word henotheism. Friedrich Schelling coined Henotheismus, from Greek hen, one, plus theism, in his lectures on the philosophy of mythology, printed after his death. Muller had every reason to know the term: he went to Berlin in 1844 specifically to study under Schelling, and translated passages of the Upanishads at Schelling's request. The word Muller minted himself, some years before the Hibbert Lectures, is the more exact one, kathenotheism, from kath' hena theon, one god at a time. He lost that battle too. By the early 1880s he was conceding that the shorter word had won out in general use, because what readers wanted was the contrast with monos, one only, rather than a description of sequence.

Rigveda 1.164.46, the verse most often produced in this argument, in Griffith's rendering: "They call him Indra, Mitra, Varuna, Agni, and he is heavenly nobly-winged Garutman. To what is One, sages give many a title: they call it Agni, Yama, Matarisvan."

That verse sits inside Rigveda 1.164, the long riddle hymn attributed to Dirghatamas, a sustained sequence of puzzles about the sacrifice, the year and the sun. It is poetry rather than a doctrinal article, and what it says is exactly what it is quoted for. Ekam sat, the real is one, and the wise speak of it in many ways. The line is not a modern gloss laid over the hymns from outside. It is the hymns describing themselves, from inside the oldest text the tradition has.

One honesty check on Muller. He was building an evolutionary theory of religion in which henotheism was an early stage later forms grew out of, and that developmental frame is his rather than the Veda's. Kathenotheism describes what Vedic poets do. It is not a theology, and the poets were not groping towards a monotheism they had failed to reach. They were already addressing the One through whichever door stood open, and had said so in their own words. Muller's category was a European vocabulary catching up with a practice the Veda had carried, complete, for millennia.

The same pattern, in an ordinary house

None of this is confined to a seminar room. A typical Hindu household holds several allegiances at once and does not experience them as competing. There is often a kula-devata, the deity of the lineage, which nobody chose and which arrived with the family. There is frequently a grama-devata attached to a place rather than a bloodline. And there is the ishta-devata, the chosen deity, the one an individual actually turns towards, which may be neither of the others and may change across a lifetime.

Standing at that shrine, a person will use exactly the language Muller noticed in the Rigveda. Lord of all, without a second, refuge of everything. Nobody present hears that as a claim that the deity in the next shrine is fictional. Walk twenty steps to somebody else's ishta-devata and the same person will fold their hands and mean it. An observer counting deities in that house gets a number and learns nothing. Address is singular, existence is plural, and the tradition has never found the combination difficult, because the One being addressed does not change when the door does. What actually happens at such a shrine is its own subject, taken up in what murti puja really is.

Where the schools genuinely disagree

The tempting shortcut at this point is to say that Hinduism is "really" monotheistic, or "really" monistic, with the many gods as faces of one reality. That answer is real and central. It is the Advaita Vedanta reading, and Shankara defended it better than anyone. But the tradition honours the question enough to have produced more than one worked-out answer, built by minds of the same rank, and stating them accurately shows the depth of the house rather than a crack in it.

Advaita Vedanta, following Shankara, does read Brihadaranyaka 3.9 as a descent towards the nirguna, the Brahman without attributes, with the named deities holding real but provisional status at the level of ordinary transaction. One is the true answer; thirty-three is a concession to where the questioner is standing.

Dvaita Vedanta, following Madhva, holds the opposite and holds it deliberately. Madhva's pancha-bheda names five differences that are real and that liberation never dissolves: between God and soul, God and matter, soul and matter, one soul and another, and one material thing and another. He adds taratamya, a genuine gradation among divine beings, headed by Vishnu, with Lakshmi next and Vayu the highest of the dependent souls. The tradition's slogan is Hari sarvottama, Vayu jivottama. For Madhva, "all the gods are one" is not a deeper truth. It is false.

Shakta traditions refuse a different flattening. In the Devi Mahatmya the goddess is not a form of a prior neuter absolute; she is the ground on which the gods stand, the power they draw on, and the one who withdraws so that Vishnu can wake. Her most quoted self-declaration in that text, delivered to Shumbha, is that she alone is here and there is no second beside her. Subsuming Devi under a Brahman conceived elsewhere is the move those texts were written to block.

Purva Mimamsa puts the sharpest question of the four, and is the school least often mentioned in these conversations. Mimamsa is the oldest professional discipline of Vedic interpretation, running from Jaimini through Sabara to Kumarila Bhatta, and it has no supreme creator God at all. Kumarila argued against Ishvara at length in the Slokavarttika while defending Vedic authority absolutely. For Mimamsa, the Veda is authorless and needs no author; the deity named in a mantra has no body, does not consume the offering, and does not dispense the result; the rite produces its own effect. Ask a Mimamsaka how many gods there are and the answer is not a number and not "one." It is that you are counting the wrong thing.

Vishishtadvaita, Kashmir Shaiva and Shaiva Siddhanta each answer differently again.

Many doors, one reality

The disagreement between the schools is real, in writing, and centuries deep. Advaitins and Dvaitins have been publishing against each other on Brihadaranyaka 3.9 since the thirteenth century and neither side has conceded. Madhva, the Shakta acharyas and the Mimamsakas were not failing to reach the conclusion Shankara reached. They were engaging it with arguments, and the arguments are part of the inheritance. A tradition that keeps its sharpest minds debating its deepest question for seven hundred years, all of them working from the same texts, is not confused about its gods. It is doing theology at a depth the counting question cannot reach, and the debate is a form of reverence, not a symptom of disorder.

That is also the soft answer to the critic of polytheism. The critic looks at the many shrines and sees a crowd. The tradition looked at the same many, close to three thousand years ago, in public and under hostile examination, and walked the count down rung by rung without calling a single rung false. "How many gods are there" presupposes that deities are discrete countable entities of one kind, so that a total settles something. Brihadaranyaka 3.9 does not so much reject that frame as walk through it and out the other side. Nothing in the modern objection is new, and nothing in it was left unanswered.

The names are many because the doors are many. What they open onto, the Rigveda states in its own words at 1.164.46: the real is one, and the wise speak of it in many ways. Yajnavalkya gave the same answer in Janaka's court, from three thousand three hundred and six down to breath, to Brahman, to That. Holding the many and the One together, each fully honoured, is not a muddle waiting for a tidier religion to resolve it. It is the theological sophistication that has sat at the heart of Sanatan Dharma from its oldest layer. Asked how many gods there are, the tradition answered before the question had learned to be an objection. The answer was never a census. The answer is One.