Purusha Sukta: The Four Varnas From One Body, and the Two Ways That Hymn Has Been Read

Rigveda 10.90.12, in Ralph Griffith's 1896 translation, reads: "The Brahman was his mouth, of both his arms was the Rajanya made. His thighs became the Vaisya, from his feet the Sudra was produced."

That is the verse critics of caste actually cite, and they are right to cite it rather than anything in the Gita, because it is the older and the harder passage. Two readings of it have been in circulation for a very long time. One takes the hymn as a description of an organism whose parts do different work, and notes that the verse never ranks the four, never issues an instruction, and does not even use the word varna. The other takes it as a ladder running from head to foot with divine authorship stamped on it. The second reading is not a colonial invention. Hindu law-books, Puranas and centuries of practice read it exactly that way, and this piece concedes that before it argues anything.

The verse, and the verses around it

Two verses have to be read together, because verse 11 is a question and verse 12 is its answer.

"When they divided Purusa how many portions did they make? What do they call his mouth, his arms? What do they call his thighs and feet?

The Brahman was his mouth, of both his arms was the Rajanya made. His thighs became the Vaisya, from his feet the Sudra was produced."

Rigveda 10.90.11 and 10.90.12, translated by Ralph T. H. Griffith

Note what the Sanskrit says and does not say. The second group is named rajanya, roughly royal or of the ruling kind, not kshatriya, the term every later text uses. The word varna does not occur in the line at all. It runs brahmano 'sya mukham asid bahu rajanyah kritah, uru tad asya yad vaishyah padbhyam shudro ajayata, and there is no term in it for class, rank or order. The four names sit inside a hymn of sixteen verses whose subject is not society but a sacrifice.

The hymn opens with Purusha, a being of a thousand heads, a thousand eyes and a thousand feet, who pervades the earth on every side and fills a space ten fingers wide. The gods then perform a sacrifice in which Purusha himself is the victim, and what follows is a long inventory of what that sacrifice produced. Verse 8 yields the creatures of the air and animals both wild and tame. Verse 9 yields the Rig verses and the Saman chants, then spells and charms, then the Yajus. Verse 10 yields horses, cattle, goats and sheep. Verse 11 asks the question above. Verse 12 answers it in a single line. Verses 13 and 14 continue the same inventory without a pause. Verse 15 returns to the mechanics of the rite, and the closing verse calls what the gods did there the earliest of the holy ordinances.

The four names, in other words, appear well down a list that also holds the hymns of the Veda, livestock, the moon and the sun. That is not a defence of anything yet. It is a fact about the passage's position that most people quoting it have never seen, because the quotation almost always begins at verse 12 and ends there.

Where the hymn sits, and what lateness does not mean

Vedic philology has treated Mandala 10, the tenth book, as the youngest layer of the Rigveda since Hermann Oldenberg's Prolegomena of 1888. The grounds are linguistic and structural rather than doctrinal: the language of the tenth book is later than that of the family books, Mandalas 2 to 7, and the arrangement of the collection shows the tenth appended to a core already ordered without it.

Stephanie Jamison and Joel Brereton, whose 2014 Oxford translation is the current scholarly standard in English, place the verses on the social estates in the latest layer. About the collection as a whole they are blunt: there is, they write, no evidence in the Rigveda for an elaborate, much-subdivided and overarching caste system, and the varna system appears there in embryo, a social ideal rather than a social reality.

This matters, but not in the way people usually want it to. Lateness is a claim about when something was composed relative to the material around it, not a demotion. Mandala 10 also contains the Nasadiya Sukta and the Hiranyagarbha Sukta, and nobody treats those as second-class material because of where they sit. A hymn does not become less scripture by being younger than its neighbours.

There is a stronger claim available, and this article declines it. A serious line of scholarship holds that the Purusha Sukta, or the social verses within it, are an interpolation inserted to give divine backing to an order that was already hardening. Suvira Jaiswal argues it. Vijaya Nagarajan argues it. B. R. Ambedkar argued it at length in Who Were the Shudras?, resting partly on the hymn's absence from Samhitas of the Black Yajurveda and partly on the observation that the Shatapatha Brahmana knows only three groups.

That is a real argument made by serious people. It is also contested, it cannot be settled here, and nothing below depends on it. The verse is in the Rigveda and has been recited as Rigveda for as long as anyone can trace, and calling it somebody's forgery is a shortcut that collapses the moment a critic asks for manuscript evidence of the insertion. Better to take it at full strength and argue anyway.

A body distributes function, and the hymn keeps distributing

Read verses 13 and 14 and the anatomy stops looking like a rank order.

The same body that yields four social names in verse 12 yields the moon from its mind and the sun from its eye in verse 13, and in that same verse Indra and Agni are born from the mouth. Verse 14 has mid-air from the navel, the sky from the head, the earth from the feet and the regions from the ear. Two of the most invoked gods in the Rigveda come out of the same mouth as the brahmana, and the earth itself comes out of the same feet as the shudra.

If the anatomy is a ladder, then Agni ranks alongside the priest and the ground under everyone's feet ranks with the servant. No school of Indian thought argues that, which tells you how much weight the anatomy was ever meant to carry on its own. The hymn uses one figure, the division of a single body, to say that many different things came from one source, and applying it to society is one application among several the hymn itself makes.

The second observation is about what is missing. The verse contains no imperative. There is no rule about who may marry whom, none about occupation, inheritance, purity, education or contact, and nothing saying the arrangement is permanent or heritable. Every one of those rules exists in Hindu literature, and every one was written centuries later, by named authors, in a different genre. To get from verse 12 to a social code you have to supply premises the verse does not contain.

An honest caveat belongs here. The organic reading is a reading. Its advantage is that it adds nothing to the words, where the hierarchical reading adds a great deal. But the hymn does not say the four are equal either. It says almost nothing. Anyone claiming the text obviously teaches equality is performing the same operation as the critic, in the opposite direction.

The half that cannot be skipped

Open the Manusmriti at its first chapter and the argument above meets resistance from inside the tradition.

Verse 1.31 restates the image almost word for word, with one telling change. Where the hymn says rajanya, the law-book says kshatriya: for the sake of the prosperity of the worlds, Manu has it, he caused the Brahmana, the Kshatriya, the Vaishya and the Shudra to proceed from his mouth, arms, thighs and feet. Verse 1.87 then says that in order to protect this universe the most resplendent one assigned separate duties and occupations to those who had sprung from those four parts. The four verses that follow set those duties out one at a time: teaching and study and sacrifice for the brahmana, protection of the people for the kshatriya, cattle and trade and agriculture for the vaishya, and then 1.91, which allots the shudra one occupation only, to serve the other three meekly.

Look at what has happened between the hymn and the law-book. A cosmogony that named four groups and moved on to the moon has become an allocation of work, with a stated purpose, a ranked list of duties and a floor. Nobody smuggled it in. Manu performs the move in the open, across consecutive verses, and the entire point of the passage is to derive the duties from the anatomy. The Bhagavata Purana at 11.5.2 restates the anatomy again, centuries later still.

The hierarchical reading of Rigveda 10.90 is the tradition's own. It is written into a Hindu law-book more than a thousand years before any European set foot in India with an opinion about Hindu society. A defence of Hinduism that starts by blaming outsiders for the ranking is not a defence. It is an evasion, and it does not survive one person opening the first chapter of the Manusmriti.

This is why the argument that critics misread the hymn fails. Ambedkar read 10.90 the way the Dharmashastra had read it, and he was right about how it had been read. His case rested on the plain sense the law-books, the Puranas and the social order had already assigned to the verse, and its force came precisely from the fact that he did not have to invent anything.

Ambedkar made two different arguments about this hymn and they should not be blurred. One concerns its date and its authenticity, discussed above and not adopted here. The other concerns its use: that for as long as caste is believed to carry divine sanction, its consequences follow from the belief, and the moral authority of the texts supplying that sanction is what has to be confronted. The second argument does not depend on the first, and it is the one that has proved hardest to answer.

The rest of the corpus, which was there the whole time

The same body of literature that produced Manu produced other things.

In the fourth chapter of the Chandogya Upanishad, a boy named Satyakama wants to study. He asks his mother Jabala what family he belongs to. She tells him she moved about a good deal in her youth, serving in many places, and does not know. He goes to the teacher Haridrumata Gautama, is asked his lineage, and repeats his mother's answer without softening it. The teacher accepts him, and the reason he gives is that no one but a true brahmana could speak that plainly.

Be precise about what this establishes. It does not abolish varna, it is one episode, and the teacher's compliment still uses brahmana as the word of approval. What it does show is a text at the centre of the Vedanta corpus in which a teacher admits a student of unknown parentage and the narrative praises him for it.

In the Mahabharata, during the Ajagara episode in the Vana Parva, Yudhishthira is questioned by Nahusha, who has been cursed into the body of a serpent and has Bhima in his coils. Asked what makes a brahmana, Yudhishthira answers, in Kisari Mohan Ganguli's translation, that the wise assert a brahmana to be one "in whom are seen truth, charity, forgiveness, good conduct, benevolence, observance of the rites of his order and mercy."

Nahusha objects at once, and the objection is the sharp one: those qualities are seen in the shudra as well. Yudhishthira does not retreat. Birth alone, he answers, makes neither the one nor the other. "He, it is said by the wise, in whom are seen those virtues is a Brahmana. And people term him a Sudra in whom those qualities do not exist, even though he be a Brahmana by birth." He then adds a second argument that gets quoted far less and is harder to dismiss: that descent cannot in practice be established at all, given how thoroughly the four orders have mixed. The epic gives the objection to the serpent, lets the answer stand, and has him release Bhima.

Then there is the Bhagavata Purana verse cited above. Its line has the four varnas born from the mouth, arms, thighs and feet of the Purusha, together with the four ashramas, and distinguished by the gunas, the qualities of nature. The anatomy is kept and the gunas are named in the same breath. It should not be oversold: the verse does not say birth is irrelevant. What it shows is a tradition reaching for a criterion other than descent while still transmitting the image that had been used to fix descent in place.

Which is also the argument of Bhagavad Gita 4.13, where Krishna says the fourfold order was created by him according to guna and karma, quality and action. That verse is the defender's usual first move, and it needs a treatment of its own. It is not the centre of this article because leading with it is how the argument gets lost. Answer 10.90 with 4.13 and you have swapped one quotable verse for another, and the person you are talking to will notice.

What a corpus is for

There is a method question underneath all of this, and it cuts both ways.

If a body of literature can be settled by one passage, the Rigveda alone will yield half a dozen incompatible verdicts. The same collection that contains 10.90.12 contains, thirty-nine hymns later, the Nasadiya Sukta, which places the gods after the world's production and says nobody knows how it began, the one who surveys it from the highest heaven possibly included. A method that lets you pick your verse lets everyone pick theirs. What works instead is to read what a corpus contains, mark where it disagrees with itself, and note which of its own positions it kept arguing about rather than dropping.

The rule cuts against the defender just as hard. Quoting Gita 4.13 and stopping is the identical operation to quoting Rigveda 10.90.12 and stopping. A tradition that wants to be judged on its whole corpus has to actually produce the whole corpus, including the Manusmriti's first chapter.

And the corpus argument is not an acquittal. The temptation to stop one sentence early is strong, so say it plainly. Yudhishthira's answer to Nahusha did not prevent the Dharmashastra. Satyakama's admission to Gautama's school did not become a rule of admission anywhere. The counter-texts sat in the corpus, preserved and recited and commented on, while a birth-based order operated for centuries and caused suffering that is documented and not in dispute. That the better reading was available the whole time is not a mitigation. It is closer to the opposite.

Whoever composed verse 11 asked what they called the mouth, the arms, the thighs and the feet. It is a question, and the hymn answers it in a single line before moving straight on to the moon, the sun and the making of the earth. Everything built on top of those two lines was built later, by people with names, in books that can be dated, and they signed their work.