Somewhere near the end of the twelfth century, a poet in the east of India set the ten descents of Viṣṇu to music. Jayadeva’s Gītagovinda became one of the most beloved devotional poems in Sanskrit, and its opening song gives one verse to each avatāra, each divine descent, in turn. When it reaches the ninth, it addresses the Buddha:

“Moved by deep compassion, you condemn the Vedic way
That ordains animal slaughter in rites of sacrifice.
You take form as the enlightened Buddha, Krishna.
Triumph, Hari, Lord of the World!”

This is a Vaiṣṇava hymn praising Viṣṇu for condemning the Veda. It is worth pausing on how strange that is.

Now open the Agni Purāṇa, chapter sixteen, and meet the same figure in the same slot:

“Formerly in the war between gods and demons the former were defeated by the latter. Saying ‘Save us, Save us!’ they sought refuge with Iswara. Being incarnation of illusion and infatuation he was born as the son of Suddhodana. He infatuated the Daityas and made them relinquish the religion of the Veda. They then became followers of Buddha and induced others to forsake Vedic religion.”

It is the same tradition, and the same ninth slot right before Kalki, but the two Buddhas are opposites. In Jayadeva’s verse God comes down out of mercy for slaughtered animals. In the Agni Purāṇa God builds a weapon of mass apostasy and points it at the enemies of the gods, and the historical Buddha’s own father, Śuddhodana, gets written into the operation.

I cannot think of a clean parallel for this anywhere else in the history of religions, a tradition reaching out and placing the founder of its rival inside its own supreme god. This essay is about how that happened, and it draws almost entirely on the primary texts and on the handful of scholars who did the philological digging, Adalbert Gail above all, then Paul Hacker, Wendy Doniger, R. C. Hazra, John Holt, Edwin Bryant, Simon Brodbeck, Johannes Schneider.

What the oldest texts refuse to say

The place to start is with an absence. The critical edition of the Mahābhārata, the text scholars reconstructed from hundreds of manuscripts, contains no Buddha avatāra, and no standard list of ten avatāras either. Bhandarkar noted this a century ago, and Simon Brodbeck’s recent survey of divine descent in the epic confirms it. The one place the epic corpus does register the Buddha’s religion is a Harivaṁśa passage about the end of the age, and there he is a symptom of decline, not a saviour: śūdras, men of the labouring order, wearing ochre robes, “who follow the Buddha of the Śākyas,” keeping their own dharma while the world runs down. The Buddha’s people are one more sign that the age is rotting, which is roughly where the Sakas sit in the same literature.

Kalki, by contrast, the epic already knows in full dress. Mārkaṇḍeya’s prophecy hands him his name, Kalki Viṣṇuyaśas the brahmin, his war on the barbarians, his restoration of the golden age. That ordering is the interesting part, and Doniger states the consequence flatly: the Buddha avatāra “may well have been inspired by the Kalkin avatar.” The tenth chair was filled before the ninth was built. So when the Buddha does finally arrive in the lists, he takes a seat directly in front of a Kalki who had been waiting there for centuries, and in several of the older lists he simply occupies the slot that used to belong to Kṛṣṇa.

One epic manuscript tradition breaks the silence. The southern Kumbhakonam edition of the Śānti Parvan carries a verse: “Matsya, Kūrma, Varāha, Narasiṁha, then Vāmana, Rāma, Rāma and Rāma, Buddha, Kalki: these are the ten.” No other major edition has it. The fuller interpolation around that verse, in Doniger’s summary, already carries the whole program: “At the beginning of the Kali Age, Visnu will become the Buddha, son of Suddhodana, and he will be a bald preacher who will speak in the Magadha dialect and delude men. All men will become bald and take the ochre robe, and priests will cease to offer oblations or recite the Vedas.”

The delusion machine

Nobody sat down and invented this doctrine in one piece; it was put together over centuries, and in 1969 Adalbert Gail traced the assembly through four steps that still hold up. It begins in sorcery. In the old Raji myth the gods lose heaven to the sons of the king Raji, and Bṛhaspati, their guru, wins it back not by force but by ritual crime: he “daily offered a sorcery sacrifice to confuse the minds of Raji’s sons,” until they became “brahmin haters, abandoned the dharma, and acted contrary to Vedic prescriptions,” at which point Indra could kill them. There is no heresy here yet. There is only black magic worked through the fire, a way of making your enemy abandon his own defences.

The second step turns the spell into a book. In a Harivaṁśa insertion Bṛhaspati composes “a treatise of nāstika doctrine, utterly hostile to dharma.” Nāstika is the tradition’s word for a Veda denier. The Matsya Purāṇa names what he wrote: Bṛhaspati, “the knower of the Veda, promulgated the Jina doctrine foreign to the Veda,” Jina being the Jain teacher. So by this stage the weapon is no longer a rite performed over a fire, it is a whole religion, drafted by the gods’ own chaplain, and whoever takes it up loses the protection of the Veda.

The third step is the one everything afterward depends on, and it sits in the third book of the Viṣṇu Purāṇa. The setting is a war the gods have been losing. The demons keep their svadharma, their proper duty, and perform Vedic worship, and the text says plainly that as long as they do this they cannot be killed. Viṣṇu’s answer is to emit from his own body a being called Māyāmoha, delusion made out of māyā, out of divine illusion. Māyāmoha goes to the demons naked, as a Jain teacher. Then he changes into red robes and starts saying things like “understand that this all is nothing but consciousness,” which any reader in the fifth century would have heard as the talk of a Buddhist monk. The demons drop the Veda. In Doniger’s rendering of what follows, “the gods attacked them and killed them.” H. H. Wilson, who translated the Purāṇa in 1840, gave the moral of the episode in a sentence I have not been able to improve on:

“The armour of religion, which had formerly protected the Daityas, had been discarded by them and upon its abandonment followed their destruction.”

Two details of this episode are easy to read past, and both matter. Māyāmoha is a phantom, emitted for one job, and the text never calls him an avatāra. And nobody in the Viṣṇu Purāṇa worships him. The fullest telling of the delusion story treats the deluder as a tool.

The promotion comes in the Bhāgavata Purāṇa, probably the most influential Vaiṣṇava book ever written. Its first canto announces:

“Then, at the onset of the Kali age, he will be born, for the deluding of the enemies of the gods, as the son of Ajana, named Buddha, among the Kīkaṭas.”

Kīkaṭa is the Gayā country, where Siddhārtha Gautama sat under his tree. The father’s name comes down unstably, Ajana in some editions and Añjana in others, a small wobble on which, as we will see, the Gauḍīyas later hung a whole doctrine. The Bhāgavata repeats the Buddha in two more of its avatāra catalogues, and the Pāñcarātra ritual manuals, the Sāttvata Saṁhitā and the Lakṣmī Tantra, boil the entire biography down to one epithet, mohana, the bewilderer.

Gail pointed out that the tradition had promoted a deception once before. Mohinī, the enchantress Viṣṇu becomes to cheat the demons out of the nectar of immortality, is a passing trick in the Viṣṇu Purāṇa and a numbered avatāra in the Bhāgavata. The precedent was there. A delusion that worked could end up in the list.

How old is it? The honest answer

On the dating, the scholars simply do not agree, and I would rather show the disagreement than average it. One piece of background first, from Paul Hacker’s study of the avatāra idea itself: the word is late. Older texts call Viṣṇu’s appearances prādurbhāva, manifestations. Avatāra, descent, only settles in as the standard term around the sixth century, so the word and the ninth seat are roughly the same age.

R. C. Hazra, whose 1940 study of Purāṇic strata is still the grid everyone argues inside, put the Viṣṇu Purāṇa very early, in the late third or early fourth century. An early date left no room for the Māyāmoha chapters, so he cut them loose and called them a later insertion. Doniger preferred the fifth century and called the Purāṇa “the first text to describe the Buddha avatar.” Gail argued for the sixth, and then made the move I keep coming back to whenever I reread his monograph. Suppose, with Hazra, that the episode were a late insertion. A late forger would have written it to give a backstory to the Buddha avatāra his readers already knew. But this episode knows no Buddha avatāra. Māyāmoha carries no list number and gets no worship, and a forger writing after the promotion would not have left those out. The episode is old, Gail concluded, precisely because it has not heard of the doctrine it was later used to explain.

From there his chain runs in order. The Viṣṇu Purāṇa carries the preparatory myth in the sixth century. The first datable daśāvatāra list with the Buddha in it, the canonical ten descents, is carved in Pallava country late in the seventh, in wording that is essentially the Kumbhakonam verse quoted above. The Bhāgavata follows around the middle of the eighth, with the Buddha seated in three of its lists.

Then there is the witness nobody counted until recently, and he changes the arithmetic. The philologist Johannes Schneider has shown that Bhavya, a Buddhist doxographer writing between roughly 500 and 570, already reports the Vaiṣṇava ten in his Tarkajvālā, with the Buddha standing ninth. So the earliest firmly datable list of ten avatāras containing the Buddha comes down to us from the rival’s own pen, a full century before the Pallava inscription. Buddhist scholars had catalogued their own annexation almost before the annexing texts circulated, and Schneider notes that this leaves the older datings of the doctrine’s rise looking too late.

The chain has one more loose link, and honesty requires showing it. Edwin Bryant has argued from the Vaikuṇṭha Perumāḷ temple at Kāñcīpuram, built around 770, whose relief panels follow the Bhāgavata’s narrative sequence, that the Bhāgavata must be considerably older than the eighth century, perhaps Gupta era. If Bryant is right, the last link of Gail’s chain moves back in time, and the Bhāgavata’s Buddha verses with it. The question is live. But the floor is now firm: by the middle of the sixth century the ten with the Buddha ninth existed, and even the Buddhists knew it. Everything earlier is inference from texts that resist dating.

The eighth century coup

Set the theology beside the throne room now, because the timing of the doctrine follows the timing of political power more closely than is comfortable.

The historian Ronald Inden documented a transformation in Indian kingship: before the eighth century, the Buddha functioned as the empire’s supreme deity, and a king claimed imperial status through grand donations to the monasteries and the installation of the Buddha in a stūpa. Then, in Inden’s words quoted by John Holt, “this pattern changed in the eighth century. The Buddha was replaced as the supreme, imperial deity by one of the Hindu gods.” The dynasties that made the switch were the Karkoṭas in Kashmir, the Gurjara Pratīhāras at Kanyākubja, the Rāṣṭrakūṭas in the Deccan, and the Pallavas in the south. The one great exception: the Pālas of eastern India, in the Buddha’s own homeland, who kept him imperial.

Now set that date against the doctrine. In the same century the Buddha loses his throne as the imperial god, he is, by Holt’s count, declared an avatāra of Viṣṇu in at least four Purāṇas, and inscriptions at Mahabalipuram, in Pallava country, name him the ninth of Viṣṇu’s ten. The court circle that produced the first datable Buddha daśāvatāra inscription is the same one that built the Vaikuṇṭha Perumāḷ temple. Holt reads the coincidence without any romance about it: the avatāra doctrine “was an apt and convenient means to give space to the Buddha, or better, to put the Buddha in his Brahmanical place.” Behind that, drawing on P. V. Kane, he sees a double move, absorb the thing that had become religious common sense and subordinate the institution that had taught it. The ninth avatāra was a tool of kingship first. Its reputation as a gesture of tolerance came much later, and from other people.

The deluder at work

Once he was installed, the deluder Buddha was put to work, and the texts are not shy about the work he does.

The Padma Purāṇa uses him as a siege engine. In its version of the destruction of the demons’ Triple City, Śiva gives the credit for victory to a softening operation that came first: “The glorious one, taking up the form of Buddha, deluded his enemies. Those who were deluded by that teaching gave up all religious practices. (These) enemies of gods were killed by me with Visnu’s missile.” The delusion has to come before the weapon, because the armour of religion has to be talked off a demon before anything can land on him.

The same Purāṇa then carries the operation forward into the history of philosophy itself, in one of the most consequential polemical passages Hinduism produced. Śiva tells Pārvatī: “Visnu of the form of Buddha proclaimed the false Buddhist doctrine... for the destruction of the demons. The doctrine of Maya (illusion) is a wicked doctrine and said to be pseudo-Buddhist. I myself, of the form of a brahmana, proclaimed it in Kali (age).” The “doctrine of Māyā” is Advaita Vedānta, and the tradition has always taken that brahmin to be Śaṅkara. The charge is pracchanna bauddha, Buddhism in disguise, and the theology here is startling: Śaṅkara’s philosophy is being framed as the second phase of the same divine deception, the deluder’s campaign carried on by other means. Rival schools would spend the next several centuries throwing that accusation at the Advaitins.

Even the liturgy kept the memory of the job. A medieval formula of homage runs: “namo Buddhāya śuddhāya daitya dānava mohine,” adoration to the Enlightened One, the Pure One, who deluded the daityas and the dānavas, the two great demon clans. The homage and the accusation share a single line.

Later texts stretch the operation in both directions of time. The Bhaviṣya Purāṇa historicizes it without ceremony: “the god Visnu became born as Gautama, the Sakyamuni, and taught the Buddhist dharma for ten years... the path of the Vedas was destroyed and all men became Buddhists.” The Śiva Purāṇa, meanwhile, gives the deluder his orders in advance. Once the Triple City has fallen, Viṣṇu tells the man of delusion: “After spreading the dharma of darkness and destroying the Triple City, go to the wilderness and maintain your svadharma there until the beginning of the Kali Age. Then reveal your dharma and cause your disciples to spread it.” The heresy is being scheduled ages ahead of its own arrival, its founder held in reserve until the appointed time.

The goat becomes the gardener

Then something strange happens inside the texts, and Gail gave it an unimprovable name: der Bock wird zum Gärtner gemacht, the goat is made the gardener.

The Bhāgavata, in its great protective litany, prays: “may Buddha protect from derangement by the heretic hosts.” The Garuḍa Purāṇa echoes it: “may Buddha protect from the mass of heretics.” The avatāra whose one job had been to manufacture heretics is now being asked to defend the faithful against them. Gail’s explanation is mechanical rather than mystical: sitting next to Kalki in the list, the Buddha absorbed his neighbour’s protective role. The Padma’s litany manages both directions at once, praising the Buddha who “deludes the wicked” and “has preserved the meaning of the Vedas.”

The portrait softens along with the function. The Agni Purāṇa, the same text whose delusion narrative opened this essay, prescribes his image for worship: fair, lotus seated, long eared, “the calmness of the heart within should be made to reflect in his countenance, as he is the bestower of all blessings, and the protector of all beings from evils and harms.” The Matsya Purāṇa goes further and produces a third Buddha altogether, neither deluder nor yet compassionate: ninth in a list of sacrificial avatāras, “of Divine splendour, with His Eyes as beautiful as lotus,” born “for the establishment of righteousness and the destruction of Asuras, through asceticism,” with Vyāsa himself as officiating priest. This Buddha has been Vedicized so thoroughly that even his priest is a sage of the Brahmanical tradition.

Last of all comes the compassionate Buddha. The Padma Purāṇa’s Kriyāyogasāra salutes “that Buddha who, seeing the killing of beasts, censured the Vedas with compassion.” The Devībhāgavata honours the descent “in order to stop the slaughter of animals.” The eleventh century Kashmiri poet Kṣemendra retells the Buddha’s whole life as a straight heroic story, drawn from the Buddhist tradition itself. This warm lineage is the one Jayadeva inherits when he sings the verse this essay began with.

There is a disagreement about when this warmth sets in, and I find it a revealing one. Doniger points to Kashmir, where sympathetic treatments of the Buddha turn up already in the tenth and eleventh centuries, at a time when Buddhism was still a living presence there and Kṣemendra was at work. Holt puts the real flowering later, in the twelfth and thirteenth centuries, by which point Buddhism was institutionally dead across India and praising it cost a Hindu nothing. My own sense is that they are describing two phases of one thing. In Kashmir the respect is paid while the neighbour is still alive to see it; by Jayadeva’s time the praise is being sung over a grave.

Who salutes, and who refuses

Put the question to the living Vaiṣṇava schools and they answer three different ways. That they cannot agree is itself part of the evidence.

The Śrī Vaiṣṇavas, the school of Rāmānuja, whose great shrine at Śrīraṅgam contains a sanctum for the ten avatāras, never admitted the Buddha at all. Vedānta Deśika’s Daśāvatāra Stotra, composed at that very shrine, runs Matsya, Kūrma, Varāha, Nṛsiṁha, Vāmana, Paraśurāma, Rāma, Balarāma, Kṛṣṇa, Kalki. Balarāma eighth, Kṛṣṇa ninth, no Buddha anywhere. The icon set of the Śrīraṅgam sanctum matches the poem. And this is the same Deśika whose Śatadūṣaṇī calls the faithful to war against the pracchanna bauddhas, the crypto Buddhists, by which he means the Advaitins. His school kept the insult and refused the incarnation.

The Gauḍīyas, the school of Caitanya, accepted him, and their greatest modern voice states the old doctrine without blinking. Prabhupāda’s commentary on the Bhāgavata verse explains that the Buddha’s denial of the Veda “is an act of camouflage by the Lord.” But acceptance created a problem the school solved with surgery: the avatāra Buddha, son of Añjanā (there is the Bhāgavata’s wobbling name again), born at Gayā, is declared a different person from the historical Śākyamuni, son of Māyādevī, born at Kapilavastu. Two Buddhas, one divine and adorable, one human and mistaken. The Gauḍīya tradition credits the underlying framework, a plurality of Buddhas across the cosmic ages, to Jīva Gosvāmī, though the attribution rests on Prabhupāda’s report and Jīva’s own wording remains to be verified from his Sanskrit; what is certain is that the two Buddhas solution was already being improvised by Sir William Jones in 1790, when he noticed the contradiction between Jayadeva’s praise and the Bhāgavata’s demonology and, in Doniger’s phrase, proposed “a bold but indefensible example of dualistic ‘splitting.’”

Madhva’s Dvaita school took the deflationary road. His Bhāgavata Tātparya glosses the Buddha verse through a Brahmāṇḍa Purāṇa citation in which Viṣṇu merely poses as the child of a demon named Jina in order to delude demons with talk of nonviolence. An operation, then, and hardly an object of worship. Madhva otherwise spent his energies documenting, chapter and verse, the old charge that Advaita is Buddhism in disguise. The school’s later champion Vyāsatīrtha finished the job with what Doniger calls the final blow to any positive reading: “Buddha himself is an incarnation of God, and yet he deceived the people by false teaching.” Being God’s own descent, in this theology, tells you nothing about whether he was telling the truth.

So Rāmānuja’s school kept the Buddha out of the list and kept his name as an insult for its enemies. Caitanya’s school let him in and then split him in two so the adorable one could be quarantined from the historical one. Madhva’s school admitted that the descent had happened and denied that it deserved any devotion. A Vaiṣṇava’s position on the ninth avatāra was, in practice, a way of telling which lineage of initiation he had come up through.

A single text could not always hold to one position either. The Varāha Purāṇa lists the Buddha ninth in its fourth chapter, where the ten descents are called “steps for those who wish to attain the vision” of Viṣṇu. Eleven chapters later its Hymn of Ten Incarnations drops him and reaches ten by counting Rāma four separate times. Then chapter one hundred thirteen names him yet again. The book could not decide what to do with him any more than the schools could.

The annexation

Helmuth von Glasenapp, whom Gail quotes with approval, called the whole business an attempt “to annex Buddhism peacefully for Hinduism.” Annexation seems to me the exact word, because of what the arrangement did and did not include. The Buddha kept his name in the list. Everything else, the prestige, the antiquity, the claim on men’s reverence, went to Viṣṇu, and no cult of the adopted Buddha ever grew up inside the tradition that adopted him. Māyāmoha’s demons had at least been handed a doctrine to hold; the promoted Buddha was handed a rank. Holt, in the book he pointedly titled The Buddhist Viṣṇu, sums up the outcome without decoration: “cultic veneration of the Buddha within Hinduism is virtually absent.”

Paul Hacker, who spent years studying how Hindu traditions deal with rival claims, called this move inclusivism. You do not refute the rival system. You declare it a lower stage of your own, and near the end of his life Hacker put the point sourly: what looks to the European like tolerance, he wrote, is almost always inclusivism. A religion filing its rival’s founder as a junior form of its own god is about as pure a case of what he meant as there will ever be.

Not everyone inside the tradition was easy about the stratagem, and the unease got written into scripture. The Tamil Kāñcippurāṇam brings the Buddha to Kāñcī along with Nārada, the two of them worn down by guilt for the delusion they had spread, and it hands Śiva the verdict:

“The sin of teaching false doctrine can be expiated only after many millions of aeons. Even in order to help the gods it is a heinous sin to destroy someone’s faith in Siva.”

So the god on whose side the operation ran is made, in the tradition’s own book, to rule it a sin. That is a guilty conscience, recorded by the guilty party.

The Buddhists did not simply absorb all this either. Bhavya had catalogued the doctrine back in the sixth century, and a later commentator, Prajñāvarman, worked out the counter move: Viṣṇu became the Buddha, he suggested, to do penance for all the killing he had done as Rāma. The avatāra is real in this account, but he has come down to apologize, and the apology lands on the very mythology that had drafted him.

The doctrine kept moving after the Purāṇas were done with it. Radhakrishnan and the modern reformers turned its polarity clean around, recasting the Buddha as the avatāra who descended to purify Hinduism of its bloody rites, a reading that, as Holt points out, reverses more or less everything the Purāṇas actually say. Modern India completed the reclamation by making him a bhūmiputra, a son of the soil. And the discomfort never quite died. Gail noticed that the Poona critical edition of the Mahābhārata prints Kṛṣṇa where the Kumbhakonam edition’s ten avatāra verse reads Buddha, and permitted himself the observation that some Indian philologists too are subject to “dogmatic prejudgments that congeal into textual falsifications.” I checked what can be checked. In the critical edition’s supplementary passages the verse stands as star passage 835 to the Śānti Parvan, and its printed line runs “rāmo rāmaś ca rāmaś ca kṛṣṇaḥ kalkī ca te daśa”: Kṛṣṇa in the ninth seat. The Kumbhakonam edition and the seventh century Mahabalipuram inscription both read buddhaḥ in that seat. Whether the editors recorded the Buddha reading in their variant notes, or silently followed other manuscripts, only the printed page in Poona can settle, so Gail’s darkest interpretation stays unproven. The discrepancy itself is now verified. And as a parable of the whole subject it is almost too perfect: a tradition that put the Buddha into its god, and then found a way to print Kṛṣṇa over his name.

The last word belongs to a painting. At the Viṣṇu shrine of Devinuvara, on the southern tip of Sri Lanka, Buddhist temple artists in the 1950s painted the ten avatāras of Viṣṇu on the walls, with the Buddha standing ninth. A Buddhist temple, honouring the god who once descended to delude the Buddhists, by including the Buddha in his retinue. The annexation, in the end, ran both ways.

Sources:

• Adalbert Gail, Buddha als Avatāra Viṣṇus (1969).

• Wendy Doniger, The Origins of Evil in Hindu Mythology (1976).

• John Holt, The Buddhist Viṣṇu (2004).

• Paul Hacker, Kleine Schriften (1978).

• R. C. Hazra, Studies in the Purāṇic Records (1940).

• Johannes Schneider, his 2015 article on Bhavya.

• Edwin Bryant, on the date of the Bhāgavata and the Kāñcī temple.

• Simon Brodbeck, Divine Descent and the Four World-Ages in the Mahābhārata.

There is a moment, somewhere around three thousand years ago, when a man in eastern Iran asks a question into the dark:

“To what land to flee? Where shall I go to flee? They exclude (me) from my family and from my clan. The community with which I have associated has not satisfied me, nor those who are the deceitful rulers of the land. How, then, shall I satisfy Thee, Wise Lord?”

He goes on, in the next breath, to give his god an inventory of his poverty: “I know that (reason) because of which I am powerless, Wise One: by my condition of having few cattle, as well as (that) I am a person with few men. I lament to Thee. Take notice of it, Lord, offering the support which a friend should grant to a friend.”

A man with few cattle and fewer friends, thrown out by his own clan, complaining to his god the way you complain to someone who owes you money. That is Zarathustra in his own hymns, the seventeen songs called the Gathas, the oldest things in any Iranian language. Not a bearded lawgiver on a mountain. Not the founder of anything, yet. A working poet and priest in a cattle herding society, in trouble.

Almost everything commonly said about this man is said in vocabulary that did not exist when he lived. Prophet, religion, scripture, heaven, hell, faith: every one of those words carries a millennium of later luggage, most of it stamped in places Zarathustra never saw. He comes before. This essay reads him the way the evidence presents him: through his own verses and the world they assume, and through the scholars who spent their lives on the philology, Mary Boyce, Stanley Insler, Martin West, Helmut Humbach, Prods Oktor Skjærvø. Readers who know the Vedic world will find, again and again, that the fastest way to understand Zarathustra is to stop thinking of Moses and start thinking of a ṛṣi, a Vedic seer poet. That is not a devotional flourish. It is where the philology points.

The world he was born into

Start with the language. The Gathas are composed in a tongue so close to the Sanskrit of the Rigveda that scholars can sometimes convert one into the other with sound laws alone. Boyce puts it plainly: “The language of the Gathas is archaic, and close to that of the Rigveda (whose composition has been assigned to about 1700 B.C. onwards); and the picture of the world to be gained from them is correspondingly ancient.” Skjærvø agrees that the Gathas and the oldest Rigvedic hymns share one horizon. How deep that horizon lies is its own scholarly argument, Boyce opening the Rigveda’s composition around 1700 BCE where others begin its oldest hymns nearer the middle of the second millennium. Either way the two poetries are contemporaries, and on the older reckoning their conversation opens in the seventeenth century before the common era.

That kinship is the master key to everything. The Iranians and the Vedic Indians were, not long before Zarathustra, one people with one poetry and one ritual system. The priest who pours the offering is hotṛ in the Veda and zaotar in the Avesta: the same word. The sacred order that holds the world together is ṛta in the Veda and aša in the Avesta: the same word. The glowing plant pressed for the ritual drink is soma in the Veda and haoma in the Avesta: the same word. The old title of the great lords among the gods is asura in the Veda and ahura in the Avesta: the same word. When Zarathustra addresses his god as Ahura Mazdā, the Wise Lord, a Vedic ear hears the asura called “wise,” and the Rigveda itself addresses Varuṇa in exactly those terms. The two traditions are estranged brothers, and they even quarrel like brothers: gods called daēva were declared demons in Iran, while in India the asuras slid, over centuries, from lords to demons. Each side kept the whole vocabulary and reversed half the signs.

The society behind the hymns is a herdsman’s world. Wealth is counted in cattle, the poem’s imagery runs on pasture and raid, and the fee a poet expects for his praise songs is livestock: in Yasna 44 Zarathustra pointedly reminds his god of “the ten mares with stallion, and the camel” that were promised him. West notes drily that this “assimilates the situation to that of the priest-poet who, like the Vedic Rishi, expects and receives substantial rewards of livestock from his patron.” The power in this world belongs to petty princes the hymns call kavi. In India kaví means the inspired poet sage. In Iran it had become a title of rulers, and most of them, along with the sacrificing priests called karapan, were Zarathustra’s enemies.

When did he live? A late tradition, preserved in Zoroastrian books, put Zoroaster “258 years before Alexander,” which yields a sixth century BCE date, and older scholarship repeats it. The philologists broadly abandoned that: a sixth century man does not compose in a language contemporary with the oldest Rigveda. Boyce argued for something like 1700 to 1500 BCE; a common middle position is the late second millennium. This essay does not need to settle it. What matters is the world the hymns themselves show: Bronze Age, pastoral, pre imperial, and far closer to the Rigveda than to the Persia of Cyrus.

The man: a priest with a grievance

Zarathustra tells us his profession. “The priest who is just in harmony with truth is the offspring from the best spirit,” he sings in Yasna 33, speaking of himself; West’s comment on the verse is blunt: “he is an unimpeachable minister (zaotar-, the traditional word for the priest who poured out the libation).” Nigosian summarizes the situation exactly: “he calls himself a ‘zaotar,’ the Vedic ‘hotar’ meaning priest.” A zaotar was not a shepherd surprised by lightning. He was a trained professional, master of an intricate verse craft of fixed stanzas and counted syllables, kin to the meters of the Rigveda, formed by years of apprenticeship in a hereditary guild.

So the reformer came from inside the ritual system, and his instrument of reform was the system’s own high art: the praise hymn. This is why the ṛṣi comparison earns its keep. Like a Vedic seer, he composes metrical mantras to be sung in the presence of the sacred fire; like a Vedic seer, he expects his patron’s mares and camels; like a Vedic seer, he claims direct vision of the divine. And the vision passages are the most electric things in the Gathas. “I have already realized Thee to be virtuous, Wise Lord, when I saw Thee to be the First One at the creation of the world,” he sings in Yasna 43, and in Yasna 31: “I realized Thee to be (ever) young in mind, Wise One, when I grasped Thee in a vision to be the Father of good thinking.” A seer’s darśana, his direct sight, of god at the world’s beginning, put into first person verse.

The strangest and most intimate moment comes in Yasna 43, where the direction of speech reverses. The god’s spirit approaches and asks: “Who art thou? To which side dost thou belong?” And the man answers with his own name: “(I am) Zarathustra.” Nothing in later Zoroastrian tradition, and little in any scripture anywhere, sounds like that exchange: a god running an identity check, a poet answering like a man at a border crossing. The Gathas are full of this two way traffic. Yasna 44 is built entirely as a chain of questions, each stanza opening with the same formula, “This I ask Thee. Tell me truly, Lord”: who fixed the path of the sun and stars, who holds the earth below and the sky from falling, who yoked speed to horses. The form has a close Vedic cousin in the hymn where Vasiṣṭha questions Varuṇa about his guilt, and West traces the question formula itself to poetic habits the two traditions inherited together. India never let go of the form. Centuries later the Bhagavad Gītā was built on the same frame, one man and one god in conversation on the edge of a decision, though there it is the god who gives the answers.

The reform, point by point

What did this priest actually change? Six things, each visible in the hymns themselves. Take them one at a time.

1. One wise Lord, with a family of powers. The Gathas are addressed overwhelmingly to one god, Ahura Mazdā, the Wise Lord, maker and orderer of everything. Around him stand a small family of capitalized qualities: Good Thought, Best Order, Desirable Dominion, Devotion, Wholeness, Immortality. Later tradition would count them as the Amesha Spentas, the Life giving Immortals, and assign each a creation to guard. In the Gathas they are still fluid, aspects of the god, virtues of the worshipper, and divine persons all at once, and sometimes literally his household:

“I know the Wise One who created it to be the Father of effective good thinking. And His daughter is piety of good actions. The All-viewing Lord is not to be deceived.” (Yasna 45, Insler)

Zarathustra says he will worship these powers “with their own names” and serve them “with love.” One supreme lord, a family of powers at once distinct from him and identical with him: the architecture will feel familiar to anyone raised on the Indian side of the family.

2. A universe sorted by Order, entered by choice. One word rules the system: aša, cosmic Order, cognate with Vedic ṛta. Everything falls under aša or under its enemy, druj, the Lie. And the sorting is done by decision. The famous verses of Yasna 30 stage the drama: two primal spirits, one choosing rightly and one not, and then the instruction to the audience to reflect with a clear mind, “man by man for himself,” upon “the two choices of decision.” Humbach’s note on the passage insists the choice is existential rather than mythological. This is the piece of Zarathustra with no good Vedic parallel, and it may be his most consequential invention: a religion whose entry requirement is a decision. Every membership religion since has walked through the door he cut.

3. The cow gets legal standing in heaven. Yasna 29 may be the strangest opening scene in any sacred poetry. Before the reform is announced, before the poet even appears, the soul of the cow files a complaint in the divine court:

“To all of you the soul of the cow lamented: ‘For whom did ye shape me? Who fashioned me? (For) the cruelty of fury and violence, of bondage and might, holds me in captivity. I have no pastor other than you. Therefore appear to me with good pasturage.’” (Yasna 29, Insler)

Humbach’s note explains what “the soul of the cow” means: not one animal but the soul of the species itself, and in this scene specifically the cow that the deceitful have seized to kill for sacrifice. The court deliberates. The Wise Lord answers with a ruling about what the cow is actually for:

“Milk and butter, this is the prescript that the Lord, of one mind with Right, made for the cow.” (Yasna 29, West)

West’s comment is even blunter than his translation: “The cow is there to provide dairy products (not meat, as the sacrificers think).” And who will protect her? The court’s answer is not a warrior but a poet, Zarathustra, “the only one who has given ear to our commandments.” The cow is unimpressed:

“But thereupon the soul of the cow wept: ‘I who have recognized that my caretaker is powerless, (merely) the voice of a man without might, although I wish him to be one who possesses rule through power...’” (Yasna 29, Insler)

She wanted a king and got a poet, which is either the most human touch in ancient literature or its driest joke. Elsewhere the hymns indict the whole religious establishment by name for what it does to her: the karapan sacrificing priests and the kavi princes “have delivered the cow unto fury” and “have not been eager to prosper her and her pasturage with truth” (Yasna 44, Insler). For Indian readers this should land with a shock of recognition. The Rigveda already calls the cow aghnyā, “she who is not to be killed,” and the Purāṇas would later stage the same courtroom, the earth in the form of a weeping cow appealing to Viṣṇu for protection. The oldest layers of both civilizations put the mistreated cow at the moral center of the world, and each built a religion that answered her. Honesty note: this is protection from cruelty and wanton slaughter, not vegetarianism; Zoroastrian ritual went on offering animals for millennia, and scholars still argue over exactly what practice Zarathustra was condemning. But that the cow’s suffering is the opening grievance of the entire Gathic revelation is not an interpretation. It is the text.

4. The sacred drink called by its real name. The haoma is the pressed ritual intoxicant of the old Indo Iranian cult, the same word and the same rite as the Vedic soma, which the Rigveda celebrates with an entire book of hymns. Zarathustra’s hymns treat it as the fuel of everything he hates, the drink at whose ceremony “the cow is to be killed” (Yasna 32, Insler), and his question about it is one of the angriest in the Gathas:

“When, Wise One, shall men desist from murdering? When shall they fear the folly of that intoxicating drink, through the effects of which the Karpans as well as the evil rulers of the lands torture our (good) intentions in an evil way?” (Yasna 48, Insler)

West translates the drink’s epithet in that verse without ceremony: “this liquor piss.” Humbach reads “this urine-like intoxicant.” Three philologists, one verdict on the tone: contempt. There is a nuance worth keeping, and Humbach states it: the polemic may target the misuse of haoma, the drunkenness and the cow slaughter around its rite, rather than the plant itself. Either way, what happened next is the great irony of Zoroastrian history. The Younger Avesta rehabilitated haoma completely; in the later texts the same word names the death averting drink of immortality, with its own hymn of praise, and haoma pressing stands at the center of Zoroastrian ritual to this day. The reformer’s angriest verse lost to the liturgy he came from. Indian readers may permit themselves a smile here: the cousin religion also kept its soma, whatever the seers’ quarrels over it, long after anyone remembered which plant it was.

5. The offering turned inward. Zarathustra does not abolish sacrifice; he redirects it. What his god eats is the worshipper’s own integrity:

“Yes, both completeness and immortality are for Thy sustenance.” (Yasna 34, Insler)

Insler’s footnote spells out the theology: the completeness and immortality created by the faith and devotion of the truthful “is the only sustenance which Ahura Mazda requires.” In Yasna 33 the poet puts himself on the altar and offers “the breath of even his own person.” Readers of the Gītā will place this move instantly: the sacrifice internalized, the leaf and flower and water that God actually eats, patraṁ puṣpaṁ phalaṁ toyam, in second millennium Iranian dress.

6. The old war gods dethroned. The daēvas, the gods of raid and intoxicated frenzy, the Indra type, are cast out as false. The Rigveda itself preserves the fork in the road, in the extraordinary hymn where Varuṇa and Indra state their rival claims in turn. Boyce sets the two speeches side by side. Varuṇa:

“Lordship belongs indeed to me, the perpetual sovereign, as all the Immortals (acknowledge) to us... I let the dripping waters rise up, through rta I uphold the sky. Through rta I am the lord who rules through rta.” (Rigveda 4.42, in Boyce’s rendering)

And Indra:

“Men who drive swiftly, having good horses, call on me when surrounded in battle. I provoke strife, I the bountiful Indra. I whirl up the dust, my strength is overwhelming... When draughts of soma, when songs have made me drunk, then both the unbounded regions grow afraid.” (Rigveda 4.42, in Boyce’s rendering)

The sovereign of Order against the intoxicated storm of battle: the whole Indo Iranian religious crisis in one hymn’s debate. India held both claims in tension for centuries and worked out its own settlements. Iran, in one poet’s lifetime, ruled for the sovereign by decree, and sent Indra’s kind into the outer darkness under the name daēva.

Exile, the patron, and the price

The reform got him expelled. That is the meaning of the verse this essay opened with: excluded from family and clan, unsatisfied by the community, harassed by “the deceitful rulers of the land.” A herding society has no police and no priestly councils to appeal to; a priest who declares the princes’ gods false has exactly one option, which is to find a prince of his own.

He found one. The hymns name him with gratitude: “Which truthful person is thy committed friend for the great task? ... Yes, it is Kavi Vishtaspa, with whom thou art allied.” And again: “Kavi Vishtaspa reached this understanding of our task, which he respected in harmony with truth.” One kavi, one prince of the old poet king title, took the exile in, and the entire subsequent history of the religion hangs on that single act of Bronze Age patronage.

Vīštāspa deserves a longer look than history usually gives him, because he is the first royal convert of a world religion whose name we actually know, and the Gathas let us watch the conversion pay for itself. His name is a herdsman aristocrat’s name: it contains aspa, horse, the same word as Sanskrit aśva, and princes across both sister cultures wore horse names like regimental colors. That he kept the title kavi matters too. The other kavis are the villains of the Gathas; Vīštāspa is the one poet prince who crossed over, which is perhaps why the tradition never dropped his title even after the title’s whole class had been damned. Around him the hymns show a court taking shape. Frashaoshtra of the Hvōgva family is summoned by name, “come thou hither together with those sincere men whom we two wish to be in His favor”; Insler’s note identifies Frashaoshtra and his brother Jāmāspa as noblemen of Vīštāspa’s circle, and tradition makes Jāmāspa the prince’s adviser and, eventually, the husband of Zarathustra’s own daughter. The last of the seventeen hymns is that wedding’s song: in Yasna 53 the poet marries Pourucistā, “young one among Zarathustra’s daughters,” into the Hvōgva line, with Vīštāspa and Frashaoshtra named among the company. Reform, patron, marriage alliances: not an institution being founded but a kinship network being welded, which is exactly how anything durable was built in the Bronze Age. Later ages repaid Vīštāspa with an afterlife nearly as large as the poet’s. The Younger Avesta made him the last of the Kayanian line of kings; Ferdowsi’s Shahnameh remembers him as Goshtasp. And in one of history’s small ambushes, the same name returns a thousand years later as Hystaspes, the father of Darius the Great, a coincidence of aristocratic naming that may have helped ancient chronographers confuse the two men and drag the poet forward in time by a full millennium.

The community that formed around poet and patron appears in the Gathas under the word maga, a bond or exchange of gifts between god, singer, and sponsor, and some scholars have connected it to the same word family as the later magu, from which Greek gave us “magus” and every language got “magic.”

What did the poet promise in return? Healing, in the largest sense. His words for the coming helpers of the world are plural and modest: the saošyants, “strengtheners,” a title any righteous person could earn. “Those men shall be the saviors of the lands,” he sings of them, “fated to be the expellers of fury.” Once he begs his god for a single such figure, a “world-healer.” And he speaks of the great collective project in a verb that later doctrine would turn into a noun: “May we be the ones who will make this world splendid.” Frašō.kərəti, the Making Splendid, the renovation of the universe, one of the most influential ideas in the history of religions, sits in the Gathas as a first person plural wish. Not an apocalypse to wait for. A job to do.

He also told his followers what their choices would cost them. The dead, in his hymns, come to a crossing he calls the Bridge of the Judge: “with all these I shall cross over the Bridge of the Judge,” he sings of himself and his companions, while the souls of the deceitful, arriving at the same span, are turned away by their own consciences toward “the House of Deceit.” Requital comes “with the molten iron” at existence’s final turn. Judgment after death, weighed on conduct rather than on offerings, each soul alone at the crossing: the whole machinery that later traditions across half the world would inherit is sketched in these verses, in miniature, by a poet worried about his neighbors’ choices.

And the plural, human “strengtheners” of his hymns grew, in the centuries after him, into something more baroque. Later Avestan tradition expected a single wondrous Saošyant at the end of the age, born of the poet’s own preserved seed when a maiden bathes in a lake, arriving to drive the Lie out of the world once and for all. The legend is the tradition’s, and the tradition’s alone; but notice what it did. It took a job title that any righteous herdsman could hold and turned it into a miracle to wait for. Every reform, given enough centuries, is reabsorbed by the mythology it interrupted. Zarathustra’s was no exception, and the same centuries dressed the man himself in wonders: a mother who conceived him at fifteen in virginity, a laugh instead of a cry at birth. The working priest with few cattle disappeared behind the gold and legend of his own tradition, and it took modern philology to find him again.

What happened to his god

Zarathustra’s own story ends offstage; the Gathas do not narrate his death, and the legends (the virgin conception, the laugh, murder at the altar at seventy seven) are exactly that, legends, and late ones. What can be followed in evidence is the career of his god, and it is one of the great political stories of antiquity.

Roughly a millennium after the poet, a dynasty from southwestern Iran built the largest empire the world had yet seen, and its kings carved his god’s name at the top of their proclamations. Darius at Behistun credits every victory to Ahuramazdā. Boyce relayed the suggestion that the six noble Persians who stand with Darius on his tomb sculpture mirror the six Immortals around the Wise Lord: the king as the seventh figure of an earthly heptad, ruling as Ahuramazdā’s counterpart below. And Xerxes, his son, left the most Zarathustrian paragraph any king ever published, the so called daiva inscription found at Persepolis: “there was (a place) where previously the Daivas were worshipped. Afterwards, by the grace of Ahuramazda I destroyed that sanctuary of the Daivas, and I proclaimed: ‘The Daivas shall not be worshipped!’ Where previously the Daivas were worshipped, there I worshipped Ahuramazda with proper rites, in accordance with asha.” The poet’s war on the daēvas, fought with verses in the eastern pastures, reappears centuries later as imperial policy in the west, in stone. Whether Cyrus and Darius were Zoroastrians in any strict sense is a real scholarly fight, and this essay flags it as one; that their dynasty spoke the reformer’s language of Mazdā, aša, and the daivas is simply what the inscriptions say.

The dynasty’s later kings show the old inheritance still shifting underfoot. Artaxerxes II set a triad at the head of his inscriptions, Ahuramazdā, Anāhitā, Mithra, and Boyce traced what that promotion likely cost: the river goddess Anāhitā “came to be accorded, in Achaemenian worship, the place of the third Ahura, the ‘Son of the Waters,’ *Varuna Apąm Napat. Thereafter *Varuna faded into the background in popular devotion, although he retains his place to this day in Zoroastrian priestly rituals and prayers.” Readers on the Indian side should feel the rhyme in their bones: Varuṇa, the wise asura whom the Rigveda addresses in the very syllables of Ahura Mazdā’s name, faded in India too, shrinking from cosmic sovereign to god of the waters. The same great god lost the same throne twice, once in each brother civilization.

The story keeps moving east again after Alexander. On the coins of the Kushan emperors who ruled from Central Asia into India, the largest group of gods is Zoroastrian: Asha Vahishta and Khshathra Vairya, Mithra and Verethraghna, the divine Glory itself as Pharro, riding royal money through the Buddhist heartlands, side by side with the Buddha. The two estranged brothers of the Indo Iranian family, having reversed each other’s gods a thousand years before, ended up jingling together in the same purse.

The Rishi reading

Put the pieces in one frame. A hereditary priest of the old Indo Iranian ritual, master of the mantra craft, has a vision of the one wise Lord behind the world’s order. He recasts the inherited religion around that vision: one god, a family of luminous qualities, a universe sorted by Order against the Lie, entry by individual choice, sacrifice redirected from slaughter and intoxication toward devotion, the old war gods dethroned. He is exiled for it, finds a royal patron, founds a community bound by gift exchange, and promises that righteous men, plural, will heal and eventually make splendid the world. His hymns are memorized, syllable by syllable, for well over a thousand years before anyone writes them down, by priests who no longer fully understand the dialect, which is precisely why we can trust them: you do not smoothly forge in a language you cannot speak.

Every element of that portrait has its closest kin in India. The form of his hymns is Vedic. His job title is Vedic. His god’s title is Vedic. His central concept is Vedic ṛta under another pronunciation. His fee schedule is Vedic. His question dialogue with god has a Vasiṣṭha hymn for a twin and the Gītā for a distant descendant. Even his cow appealing to heaven has her sisters in the Purāṇas, where the earth in cow form comes weeping to Viṣṇu. The Western habit of reading him as a proto biblical prophet gets the man structurally wrong: prophets are seized by a word from outside; this man was a professional seer who worked in meter and billed in mares. Kellens and Skjærvø built the modern scholarly case for exactly this reading, and the Gathas reward it on every page.

He asked his god, in one hymn, for ten mares with a stallion and a camel. What his god eventually collected was the largest empire on earth, an ethics staked on each soul’s own choice, the idea of a world made splendid at history’s end, and a fire that his priests, from Yazd to Mumbai, have not let die. The exchange rate on Bronze Age poetry has never been matched.

Sources

• Stanley Insler, The Gathas of Zarathustra (Acta Iranica, 1975). All Gathic quotations follow Insler unless noted below. Passages quoted: Yasna 29, 30, 31, 32, 33, 34, 43, 44, 45, 46, 48, 51, 53.

• M. L. West, The Hymns of Zoroaster (2010). Three quotations follow West: the mares and camel of Yasna 44.18, the milk and butter prescript of Yasna 29.7, and “make this world splendid” in Yasna 30.9. His comments on Yasna 29.7, 33.6 and 44.18 are quoted as marked.

• Helmut Humbach and Pallan Ichaporia, The Heritage of Zarathushtra (1994). Renderings compared throughout; his notes on Yasna 29.1, 32.14 and 48.10 and his remark on the two spirits are cited as marked.

• Mary Boyce, Zoroastrians: Their Religious Beliefs and Practices and A History of Zoroastrianism Vol. II. Historical backbone of the Achaemenid section; the Varuṇa and Indra speeches of Rigveda 4.42 and the Xerxes daiva inscription (Persepolis H) are given in her renderings; the Kushan coin pantheon follows her account.

• Prods Oktor Skjærvø, The Spirit of Zoroastrianism. Dating of the Gathic horizon and comparative material.

• S. A. Nigosian, Indo Aryan Religion in Achaemenid Persia (1970). Used with caution on chronology.

• The rishi reading of Zarathustra follows the school of Jean Kellens and Skjærvø.

• On the disputed date of Zarathustra: Boyce argues for the early second millennium BCE against the traditional “258 years before Alexander” reckoning; the disagreement is flagged in the text.

PART ONE: THE CLAIM

An elephant is dying in a lake.

He is the head of his herd, immense and unbothered, and he has waded in to drink. Then a crocodile takes him by the foot and will not let go. The elephant is the stronger animal, but the water is the crocodile’s country, and strength counts for nothing there. The struggle lasts, in the telling, 1,000 years. His herd strains to save him, the cows crying out, the great bulls gripping him from behind to pull him back, and the Bhāgavata says plainly that they could not. At the end of everything, sinking, the elephant does the only thing left to him. He tears a lotus from the water with his trunk, lifts it toward a sky he can no longer clearly see, and cries out to the one he cannot see at all.

And the one he cannot see comes.

The Bhāgavata Purāṇa tells what happens next with a precision that is itself an argument.

so ‘ntaḥsarasy urubalena gṛhīta ārto, dṛṣṭvā garutmati hariṃ kha upāttacakram utkṣipya sāmbujakaraṃ giram āha kṛcchrāt, nārāyaṇākhilaguro bhagavan namas te

Seized within the lake by a thing of enormous strength, in agony, seeing Hari in the sky on the back of Garuḍa, his eagle, with his discus raised, he lifted his trunk with a lotus still in it and spoke this word in his pain: Nārāyaṇa, teacher of all, Bhagavān, homage to you. (Bhāgavata 8.3.32)

taṃ vīkṣya pīḍitam ajaḥ sahasāvatīrya, sagrāham āśu sarasaḥ kṛpayojjahāra grāhād vipāṭitamukhād ariṇā gajendraṃ, saṃpaśyatāṃ harir amūmucad ucchriyāṇām

Seeing him so afflicted, the birthless one alighted all at once and out of compassion swiftly pulled him from the lake, crocodile and all; and while the gods stood watching, Hari tore open the beast’s jaws with his discus and set Gajendra free. (Bhāgavata 8.3.33)

Note: the motive word in the verse is kṛpayā, out of compassion. It is the only reason the Bhāgavata gives him.

The word for the whole act is compassion, and the Lord in the story has no other reason for acting. The elephant has no lineage, no ritual, no merit to plead. He is an animal drowning. The one thing that reaches him is the cry.

This is the story of Gajendra, and it sits near the heart of the Bhāgavata Purāṇa, the book that calls itself the ripened fruit of the tree of the Vedas, because it contains, in miniature, the claim this essay exists to test. The claim is simple to state and enormous in what it asserts. The Vaiṣṇava tradition, the tradition of those who worship Viṣṇu as the highest, does not merely say that Viṣṇu is kind among his other qualities, the way a king might be generous between wars. It says that compassion is the innermost thing about him, the attribute beneath all the attributes, the reason he does anything at all. And it goes further. The Gītā and the Purāṇas argue, scene by scene and word by word, that no other conception of the highest makes mercy so central, so unconditional, so tender, and so willing to come down: not the old gods of the Veda, whose kindness moves inside the exchange of the sacrifice; not the serene compassion of the Buddha’s way, which shines on all alike but belongs to no one. The strongest form of the claim deserves to be stated plainly, because everything that follows is a test of it. Viṣṇu is the most compassionate Lord human beings have ever worshipped.

That is a theological claim, not a neutral fact, and the traditions it measures itself against have their own replies, which this essay will honour in their place. The way to find out is to open the three books and count. That is what I did.

A lexicon of divine mercy: from dayā to vātsalya

The three books are the Bhagavad Gītā, the Viṣṇu Purāṇa, and the Bhāgavata Purāṇa. Between them they hold most of a thousand years of Sanskrit devotion.

Compassion in Sanskrit is not one word but a family of them, and to weigh the theme honestly you have to hold the whole family in view. There is dayā, mercy, the plain word for a heart that melts at another’s pain. There is kṛpā, a tenderness that stoops toward the weak. There is karuṇā, compassion felt as a movement of the heart, the same word Buddhism made famous. There is anukampā, which means, quite literally, a trembling along with another, a shudder of fellow feeling. There is kāruṇya, the quality of being full of karuṇā. Around these sit the words for grace, anugraha and prasāda, favour freely given and never earned. And around those sit the words for a love closer than mercy, vātsalya, the fondness a cow feels for its calf, saulabhya, the astonishing quality of being easy to reach, and the family of refuge, śaraṇa and āśraya, the shelter the helpless run to. Beneath all of them lies an old and severe ethical word, ānṛśaṃsya, non cruelty, which the Mahābhārata calls the highest law of all.

These are not interchangeable. They form a rising scale, from pity for the suffering, through grace that rescues, to a love that simply wants to be near. Much of this primer is the story of Viṣṇu climbing that scale in the imagination of his devotees.

What the counting shows

Counting a theme is a crude instrument, but it is an honest one, and it settles the first question: is the impression real, or is it only the warm haze a believer carries away? The counting was done in two ways: across standard English editions, one consistent translator per book, and then again in the Sanskrit, where the mercy words can be traced at their roots.

The English counts show the direction. In the Viṣṇu Purāṇa the language of compassion, grace, and refuge appears at a rate of about one word in every 1,000; in the Bhāgavata, at roughly twice that rate. The Sanskrit gives the real scale. Take the six core words for mercy and tenderness, kṛpā, karuṇā, dayā, anukampā, anugraha, and vātsalya, and set them against the length of each book. The Gītā, in its 700 verses, sounds them about 10 times per 1,000 verses. The Viṣṇu Purāṇa, for all its rescues, sounds them at about a third of that rate, 3 or 4 times per 1,000. The Bhāgavata sounds them more than 40 times per 1,000 verses. In absolute terms, the six words occur roughly 570 times across the Bhāgavata’s 14,000 odd verses, against about 23 in the whole Viṣṇu Purāṇa and 7 in the Gītā. The younger Purāṇa is not twice as saturated in the vocabulary of divine tenderness as the older one. It is roughly 10 times as saturated.

The single most telling number is the smallest. Anukampā, the Gītā’s word for the Lord’s compassion at its most intimate, a trembling along with another, appears exactly once in the Gītā, in the verse the whole poem turns on. In the Viṣṇu Purāṇa it appears never, not once in six books. In the Bhāgavata it appears 45 times, in the mouths of Bhīṣma and Brahmā and Prahlāda, in the hymns of the demigods and the teaching of Kapila to his mother. The Bhāgavata found the Gītā’s rarest and most inward mercy word and made it a refrain.

One number cuts the other way, and ends up sharpening the point. The Viṣṇu Purāṇa actually uses the vocabulary of prasāda, favour, more often per verse than the Bhāgavata does. But its favour is overwhelmingly the ordinary favour of a hierarchical world: sages propitiated, kings gracious, gods pleased with an offering. What the Bhāgavata multiplies is not favour. It is tenderness. Where the older book has no fixed epithet for it at all, the Bhāgavata calls the Lord bhakta vatsala, tender to his devotees, 11 times; dīna vatsala, tender to the afflicted; and, in Prahlāda’s own hymn, kṛpaṇa vatsala, tender to the wretched. The words that grow are precisely the ones that mean a mercy with a face.

But a skeptic could stop me here. Compassion, the skeptic says, is just generic Indian ethics. Every tradition praises mercy. Sages are merciful, kings are merciful, the whole moral atmosphere is merciful. None of that makes compassion special to Viṣṇu. So I filtered again, keeping only the verses where the compassion is actually his: his mercy, his grace, a plea addressed to him, a rescue performed by him. The result answers the skeptic. In the Viṣṇu Purāṇa, under a fifth of the core mercy language attaches to Viṣṇu himself; most of it is the ordinary mercy of the world. In the Bhāgavata, that share rises to somewhere between a third and a half. The Bhāgavata does not merely use more compassion words. It redirects far more of them onto the Lord. Across the centuries, compassion migrates from the background ethic of a moral universe to the foreground trait of a particular person. It begins as something Viṣṇu has and ends as something Viṣṇu is.

A tradition reveals what it believes about its Lord by what it cannot stop saying about him. The rest of this essay asks why the tradition came to see mercy as the innermost thing about the Lord, and then walks the evidence, book by book, so you can watch that migration happen in the verses.

A word on method before we begin, since numbers deserve honesty. The Sanskrit counts are drawn from the complete electronic texts of the three books, searched at the root so that no form escapes, with the false catches weeded out by hand: the warrior named Kṛpa is not mercy, and gadayā, with the mace, which battle scenes use constantly, is not dayā. Words with double senses are counted honestly; prasāda, which can mean grace but also the settled clarity of a calm mind, is only counted as grace where it is grace. The English counts use one consistent translator per book. The Gītā verses are cited by chapter and verse; the Purāṇa verses by book or skandha and chapter and verse.

PART TWO: THE THEOLOGY OF DIVINE COMPASSION

Arjuna’s false compassion

The Bhagavad Gītā begins, famously, with a breakdown, and the first thing it does with compassion is refuse to trust it.

Arjuna, the finest archer of his age, looks across the battlefield at the cousins and teachers he must kill, and he cannot. His bow slips from his hand. The word the Gītā uses for what has overcome him is our word: he is filled with deep compassion, kṛpayā parayāviṣṭaḥ (1.27), and sinks down in his chariot (1.47). A little later he names his own condition, saying his very nature is stricken by the taint of pity, kārpaṇya doṣopahata svabhāvaḥ (2.7), the noun built from the same root, and for the first time he asks to be taught rather than obeyed.

Here is the surprise, and it is a large one. The Gītā opens by showing us compassion and then declining to praise it. Kṛṣṇa’s answer to the weeping man is not comfort but correction. Do not yield to this unmanliness, he says, klaibyaṃ mā sma gamaḥ (2.3); it does not become you; stand up. The most compassionate teacher in the tradition looks at a man dissolving in tenderness and tells him his tenderness is a weakness.

This is not cruelty. It is a definition arrived at by negation. What Arjuna feels is real feeling, but it is feeling curved back on itself, grief for his own loss wearing the costume of concern for others, a paralysis that would let a whole unjust order stand rather than pay the price of ending it. The Gītā calls this a counterfeit, and it spends the rest of its length minting the true coin. Real compassion, it will argue, is not the sentiment that cannot act. It is the love that rescues.

The Lord in the heart: anukampā

Before we watch him act, it is worth asking why he acts at all, and the deepest answer these books give is that he is not, in the end, outside us.

In the tenth chapter of the Gītā, Kṛṣṇa describes what he does within a person who loves him, and it is the verse this whole subject turns on.

teṣām evānukampārtham aham ajñānajaṃ tamaḥ nāśayāmy ātmabhāvastho jñānadīpena bhāsvatā

Out of compassion for them alone, dwelling within their very self, I destroy the darkness born of ignorance with the shining lamp of knowledge. (Gītā 10.11)

Note: the word is anukampā, which means, almost literally, a trembling along with another. And the compassion acts from inside: ātmabhāvasthaḥ, standing within their own being.

He is not a distant grantor of favours here. He is seated in the heart, and from that innermost seat he does the tenderest thing imaginable. He lights a lamp in the dark so the frightened one can see. The Gītā is explicit that this seat is real. I am seated in the hearts of all, he says, sarvasya cāhaṃ hṛdi sanniviṣṭaḥ (15.15). And it draws the consequence for us: the highest yogin is the one who sees the joy and pain of all beings as his own (6.32).

If the Lord dwells in every heart, then his compassion is not the pity of a stranger looking down. It is nearer than that, and more inevitable. To ask why he is compassionate becomes almost the same as asking why you flinch when your own hand is burned. He is closer to us than that. And once he is that close, the wonder is no longer that he feels for us. It is what he is willing to do about it.

Avatāra: mercy that comes down

The first thing he does is move. His compassion does not sit still; it takes a body and comes down.

The Gītā gives the reason in two of its most quoted lines.

yadā yadā hi dharmasya glānir bhavati bhārata abhyutthānam adharmasya tadātmānaṃ sṛjāmy aham paritrāṇāya sādhūnāṃ vināśāya ca duṣkṛtām dharma saṃsthāpanārthāya sambhavāmi yuge yuge

Whenever righteousness weakens and unrighteousness rises up, O Bhārata, then I bring myself into being; for the protection of the good, the destruction of evil, and the setting right of the law, I come into being age after age. (Gītā 4.7 to 4.8)

The word compassion is nowhere in the verse, and it does not need to be, because the verse is compassion with its sleeves rolled up. It is the moment a quality becomes an action.

This is what the tradition means by an avatāra, a word worth slowing down on, because it is almost always mistranslated as incarnation. It does not mean that. It means a descent, from a root that describes crossing down, the way one fords a river from the far bank to this one. An avatāra is the Lord coming down to where we are drowning. And once you hold that meaning, every descent story reads as the same story in a new key. It is never a display of power for its own sake. It is a rescue.

The Viṣṇu Purāṇa gives that rescue its most elemental form in the descent of the boar. The earth herself, in this telling, has slipped from her place and sunk down into the depths of the cosmic ocean, lost under the weight of the waters. And what the Lord does about it is not to command from above. He goes in after her.

He, the lord, concluding that within the waters lay the earth, and being desirous to raise it up... took the figure of a boar... and, the supporter of spiritual and material being, plunged into the ocean. (Viṣṇu Purāṇa I.4, Wilson’s translation)

Picture the direction of that sentence. The highest being in existence turns himself into an animal and dives, down through the whole depth of the sea, to find a drowning world and carry it back into the light on the point of his tusk. When he sets her down, the Earth lifts a hymn of pure relief. It is the same shape as the elephant and the lake, only now the drowning thing is the world, and the Lord still comes down into the water after it.

Once you see the shape, you see it everywhere in the descents. When a flood is about to drown the first man, he becomes a fish and tows his boat to safety. When the gods need the ocean churned for the nectar of life, he becomes a tortoise and takes the whole grinding weight of the churn on his own back. When the worlds have been seized, he comes as a dwarf and, in three quiet steps, wins them back. When a child who loves him is about to be killed, he tears his way out of a pillar in a form that is neither man nor animal, purely to be there in time. The costumes change. The direction never does. It is always downward, and always toward the one who cannot save itself. Transcendence, here, refuses to stay transcendent. The infinite keeps making itself small and specific and reachable, a boar in the water, a dwarf at a sacrifice, a charioteer beside a broken soldier, because a compassion that stayed in heaven would not be worth the name. The descents are the proof that his mercy is not a mood he watches us from within. It is a distance he is willing to cross.

So the boar in the deep and the promise in the Gītā are one teaching. I come into being age after age, he says, and he means it as a vow. Whatever it takes, whatever form, however far down, when the world is sinking he will not remain above it. He comes.

Śaraṇāgati: taking refuge

He comes down, then. But to whom, and on what terms? The stories are strangely unanimous. He comes for the helpless, and helplessness is the only ticket required.

There is a quality of the Lord that the tradition prizes above his power, and it is the strangest of all his qualities, because it is a kind of susceptibility. He can be reached. The word for it is saulabhya, the quality of being easy to attain, and its counterpart in us is the act the tradition calls śaraṇāgati, the taking of refuge. What is remarkable about refuge is that it is the one path to the Lord that asks for no strength. You do not climb to it. You fall into it, and you qualify by having nothing left.

The Gītā gathers this into the single verse it treasures above every other, the verse the commentators call the carama śloka, the final word.

sarva dharmān parityajya mām ekaṃ śaraṇaṃ vraja ahaṃ tvā sarva pāpebhyo mokṣayiṣyāmi mā śucaḥ

Abandon all the dharmas and take refuge in me alone; I will free you from all sins; do not grieve. (Gītā 18.66)

Note: the dharmas abandoned here are not morals but the whole apparatus of duties, laws, and careful ledgers of merit. What remains when the ledgers are dropped is him.

Read it slowly and notice what is missing. No condition, no examination, no list of what you must first become. Only the instruction to let go, the promise that answers it, and then, at the very end, two words spoken to a particular frightened man. Mā śucaḥ, do not grieve, is said to the same Arjuna who could not stop grieving in the first chapter. The book that opened with a man collapsing in sorrow closes with the Lord telling him the sorrow can be set down. Refuge is the bridge between the two.

The Gītā states the principle. The Viṣṇu Purāṇa shows it happening, and it chooses for its great scene of refuge not a saint but the gods themselves at the moment they stop being gods. They have been defeated. Stripped of their splendour, the whole shining hierarchy of heaven is reduced to fugitives, and they do the thing the powerful almost never do. They admit they cannot save themselves. They flee to Viṣṇu, and the hymn they raise is one long surrender.

Oh lord of all, great soul of all, asylum of all, undecayable, have pity upon thy servants; oh Viṣṇu, be manifest unto us... We, discomfited by the Daityas, have fled to thee, oh Viṣṇu, for refuge. Spirit of all, have compassion upon us; defend us with thy mighty power. (Viṣṇu Purāṇa I.9, Wilson’s translation)

Note: the Daityas are the demons who have driven the gods from their heaven.

Notice that these are gods. If anyone in the cosmos had standing to negotiate, to offer, to earn, it was they, and the Purāṇa strips it all away before it lets them approach. They come with empty hands. That is the point the scene is making. Refuge is not the last resort of the weak. It is the true condition of everyone, and the gods are simply the first to be honest about it.

And Viṣṇu answers, not with a lecture but with a rescue. He sets in motion the churning of the great ocean, and from that churning, among the treasures the sea gives up, rises Śrī, the goddess of fortune, seated on her lotus, whom the Purāṇa calls the mother of the world and the beloved of Viṣṇu. She is not a decoration on the story. In the tradition’s reading she is his compassion given a face and a name, his favour made into a person who can stand beside him. And what the Purāṇa gives her is the tenderest promise in the book, spoken in her own voice: I will never turn my face away from that mortal who praises me as Indra has praised me. The grace the gods fled toward turns out to have been a person who was already inclined to come. The door was open before they knocked.

If the gods make the principle grand, a small boy makes it unbearable. His name is Dhruva. He is a child, a prince, whose father’s second wife has humiliated him, telling him he has no right even to climb into his father’s lap. Wounded past what a child should have to bear, he resolves to leave. His mother, begging him to stay, tells him, thou art my only refuge, I have no hope but thou, and the five year old walks into the forest anyway, to find the refuge that stands behind every refuge. When he reaches the Lord in his meditation, his whole theology fits in a single sentence: I have taken refuge with that pure form of thine, oh supreme. And Viṣṇu comes to the child. Consider what he might have said, this infinite being, to a small boy who had wandered into the woods in a fit of hurt pride. He might have tested him. He might have corrected him. Instead the Purāṇa gives him words that read like a hand laid on a small head.

Son of Uttānapāda, be prosperous. Contented with thy devotions, I, the giver of boons, am present. Demand what boon thou desirest... I am well pleased with thee. (Viṣṇu Purāṇa I.12, Wilson’s translation)

The child who had been told he was worth nothing hears, from the highest being in existence, that he is seen, and cherished, and welcome. Dhruva had no merit. He had a wound and a longing and a single true sentence, and it was enough.

Put the three together and the shape is clear. The Gītā gives the doctrine, that you may abandon everything and simply take hold of him. The gods give the scale, that even the mighty come with empty hands. The child gives the intimacy, that the emptiest hands of all are the ones he fills first. The tradition asks only one qualification for his mercy, and it is a strange one to build a religion on. You qualify by having run out of everything else.

Anugraha: grace to the guilty

Helplessness is one thing. Guilt is another, and harder. It is one matter to rescue the innocent who cannot save themselves, and quite another to rescue the guilty who, by any fair accounting, should not be saved at all. Here the tradition makes its boldest claim, and it makes it on purpose, because a mercy that could be earned would not really be mercy.

The Gītā states it almost as a provocation.

api cet sudurācāro bhajate mām ananyabhāk sādhur eva sa mantavyaḥ samyag vyavasito hi saḥ kṣipraṃ bhavati dharmātmā śaśvacchāntiṃ nigacchati kaunteya pratijānīhi na me bhaktaḥ praṇaśyati

Even the very worst of evildoers, if he worships me with an undivided heart, is to be counted righteous, for he has resolved rightly; swiftly he becomes a soul of dharma and attains lasting peace. O son of Kuntī, declare it: my devotee never perishes. (Gītā 9.30 to 9.31)

And he closes the door against every exclusion the social order knew.

māṃ hi pārtha vyapāśritya ye ‘pi syuḥ pāpayonayaḥ striyo vaiśyās tathā śūdrās te ‘pi yānti parāṃ gatim

For those who take refuge in me, O Pārtha, even those of stained birth, women, merchants, and servants, they too reach the highest goal. (Gītā 9.32)

Note: the verse speaks in the categories of its age, the very categories its society used to exclude, and its point is to overrule them. Whoever the order pushed to the margins, the refuge reaches.

The turning of the heart comes first; the becoming good follows after, as its result. Grace, anugraha, means a favour freely given, and a favour you had to deserve would be a wage, not a gift.

The Bhāgavata pushes this as far as it can go, in the story of Ajāmila (Skandha 6). Born to learning and duty, Ajāmila abandoned all of it for a life of theft and lust, and wasted himself utterly. This is no sympathetic sinner with a hidden nobility; the Bhāgavata is blunt about how far he fell. And then he comes to die. Three terrible figures arrive with nooses, the servants of Yama, lord of the dead, and begin tearing the soul out of him, and in his terror he calls out, loudly and at length, to his youngest son, who is playing at a distance. The boy’s name is Nārāyaṇa, which is also a name of the Lord. Ajāmila is not praying. He is a frightened man shouting for his child. But the Name has been spoken. And hearing the Name of their master called aloud by the dying man, though he called it without meaning it, the messengers of Viṣṇu rush to the spot, four armed and lotus eyed, and stop the servants of Yama by force, wrenching the noose off a soul that by every entry in the ledger belonged to them.

The Bhāgavata refuses to let us soften this into a lesson about sincerity. It stresses that Ajāmila had no devotion and no proper intonation, that his mind was in no fit state, that he meant only to call his son. The whole point is that the mercy did not depend on him. It depended on the Name, and the messengers of the Lord state the doctrine over the dying man in a single verse.

sāṅketyaṃ pārihāsyaṃ vā stobhaṃ helanam eva vā vaikuṇṭhanāmagrahaṇam aśeṣāghaharaṃ viduḥ

Spoken to call someone, spoken in jest, spoken as a filler of song, or spoken even in contempt, the taking of the Name of Vaikuṇṭha destroys sin without remainder: so the wise know. (Bhāgavata 6.2.14)

Note: Vaikuṇṭha is the Lord’s own abode, and the Name of Vaikuṇṭha means his Name. The verse lists, one by one, every way a name can be said without meaning it, and closes every loophole in mercy’s favour. Aśeṣa means without remainder; nothing of the sin survives.

The Name belongs to a Lord whose compassion runs so far ahead of desert that it can be triggered by an accident. This is why the tradition chose, as its great parable of grace, not a good man who slipped but a wreck who had thrown his life away. Anyone can imagine mercy for the basically good. The Bhāgavata wants you to imagine it for the man you had already written off, because that is how you know the mercy is real. It reaches exactly the people who could never have reached it themselves.

There is a hard freedom in this that is easy to miss under the drama. If grace could be earned, then every soul would be forever calculating whether it had done enough, and the frightened would have most reason to despair. By placing the whole weight on the Name rather than on the merit of the one who speaks it, the Bhāgavata lifts that calculation off us entirely. The worst evildoer of the Gītā and the ruined brahmin of the Bhāgavata are not exceptions the tradition is embarrassed by. They are the proof it is proudest of.

Vātsalya: from mercy to love

Rescue the drowning, forgive the guilty, and you have described a very great mercy. But underneath even that, these books keep uncovering something that is not mercy at all. Mercy, for all its beauty, can still be the posture of a superior stooping to an inferior. The tradition digs under that posture and finds something warmer holding it up. It finds love, and the word it reaches for is not one of the grand ones. It is vātsalya, the ordinary, aching fondness a mother feels for her child.

The Gītā says it plainly, and saves the saying for the end, just before the final secret.

sarvaguhyatamaṃ bhūyaḥ śṛṇu me paramaṃ vacaḥ iṣṭo ‘si me dṛḍham iti tato vakṣyāmi te hitam

Hear again my highest word, the most secret of all: you are dearly loved by me, and so I will speak for your good. (Gītā 18.64)

manmanā bhava madbhakto madyājī māṃ namaskuru mām evaiṣyasi satyaṃ te pratijāne priyo ‘si me

Set your mind on me, be devoted to me, worship me, bow to me, and you will come to me. I promise you this truly, for you are dear to me. (Gītā 18.65)

This is no longer the language of a sovereign granting protection. It is a confession of affection, offered as the reason behind everything else.

The Bhāgavata turns that confession into a whole world, the boyhood of Kṛṣṇa in the cow country of Vraja, where Bhagavān is loved not as a king but as a child and a darling, and where the tenderness runs downward, into smallness, because that is what love does. One day, says the Bhāgavata, Yaśodā was fondling her little son in her lap when the babe suddenly grew too heavy for her arms.

ekadāroham ārūḍhaṃ lālayantī sutaṃ satī garimāṇaṃ śiśor voḍhuṃ na sehe girikūṭavat

Once, fondling her son whom she had lifted onto her lap, the good mother could no longer bear the child’s weight, as if it were the peak of a mountain. (Bhāgavata 10.7.18)

For an instant the weight of the whole universe presses through the body of a one year old, and then it is gone, and he is only her boy again. The tradition loves to tell how she once tried to tie that boy to a mortar to keep him from mischief, and the Bhāgavata gives the mother a word of her own as she reaches for the rope: arbhaka vatsalā, tender to her child (10.9.12). The Lord whom nothing in existence can bind stood before her with his face cast down and let the rope be knotted around him. Kuntī, remembering that scene years later in her great prayer, confesses that it bewilders her past all the cosmic visions.

gopy ādade tvayi kṛtāgasi dāma tāvad, yā te daśāśrukalilāñjanasambhramākṣam vaktraṃ ninīya bhayabhāvanayā sthitasya, sā māṃ vimohayati bhīr api yad bibheti

When you had misbehaved, the cowherd woman took up the rope to bind you, and you stood with frightened eyes brimming with tears, the black of them smearing, your face cast down in fear. That state of yours bewilders me, that Fear itself is afraid of you. (Bhāgavata 1.8.31)

The one who holds the cosmos consents to be held. That is what he is willing to become for the sake of being loved.

And when the cowherd women come to him in the night, the Bhāgavata says something that reads almost as a definition of grace as tenderness. Revealing himself in as many forms as there were women, he was with each of them, so that not one was left standing alone while another was embraced (10.33). No one is a spare. Each is danced with as though she were the only one. And then, in the middle of that overwhelming scene, the Bhāgavata pauses on a gesture so small it could be missed.

tāsāṃ rati vihāreṇa śrāntānāṃ vadanāni saḥ prāmṛjat karuṇaḥ premṇā śantamenāṅga pāṇinā

And he, the compassionate one, with love, wiped the faces of those girls, tired from the play of the dance, with the hand that soothes most. (Bhāgavata 10.33.21; some editions number it 10.33.20)

Note: every word is carrying weight. Karuṇaḥ, the compassionate one. Premṇā, with love. Śantamena pāṇinā, with the most soothing of hands.

The vocabulary we have followed through the drowning elephant and the descending boar is spent here on drying the sweat from a dancer’s cheek. The hand that broke the crocodile’s jaws is the hand that wipes a tired face.

This is why the whole subject finally resolves into love rather than mercy. Mercy answers need, and if need were the whole story, then once we were safe he would have no more reason to attend to us. But love is not answering a need. Love wants to be near. It is the difference between a rescuer, who is done when you are pulled from the water, and a mother, who is never done. The reason he comes when we call is not only that we are drowning. It is that he is fond of us, that he made himself small enough to be held, that he would multiply himself rather than leave one of us dancing alone. Take away the compassion for a moment and there is still this underneath it, warm and unhurried and a little astonishing. He likes us. He always did.

Brahmā’s verse: Bhāgavata 10.14.8

Before the comparison, one verse deserves to stand alone, because in it the Bhāgavata states its doctrine of mercy as doctrine, and lets the maker of the worlds be the one to say it. Brahmā, the creator, has just tested the cowherd boy of Vraja and been overwhelmed by what he found, and the hymn he pours out in his defeat contains the verse the later tradition would treasure above almost any other.

tat te ‘nukampāṃ susamīkṣamāṇo, bhuñjāna evātmakṛtaṃ vipākam hṛd vāg vapurbhir vidadhan namas te, jīveta yo muktipade sa dāyabhāk

The one who lives watching intently for your compassion, bearing whatever ripens from his own past deeds, offering homage with heart, speech, and body, is the rightful heir to the seat of liberation. (Bhāgavata 10.14.8)

Note: dāyabhāk is an inheritance word. An heir does not earn the estate. He is simply the son.

It is the whole theology of these books compressed into four lines. The heir of liberation is not the one who conquers his senses, though the tradition honours that, and not the one who perfects the rites or completes the knowledge. It is the one who watches for his compassion the way a farmer watches the sky for rain, bearing his own harvest in the meantime. And the word for the compassion he watches for is anukampā, the Gītā’s own trembling word from 10.11, which the Viṣṇu Purāṇa never uses at all and the Bhāgavata uses 45 times, here returned with interest at the very summit of the book. Watching his mercy is not preparation for the path. It is the path.

The rivals: the Vedic gods, Śiva, the Goddess, the Buddha

It is worth pausing to say what this compassion is not, because comparison sharpens it, and because the claim this essay opened with, that no Lord was ever more compassionate, cannot be made honestly without looking the rivals in the face.

Begin with the gods of the Veda, the world these three books grew out of. Their dealings with men move, above all, inside an exchange. Indra’s strength answers the pressed soma; Varuṇa’s mercy is the mercy of a judge, who may loosen the noose he himself has fitted; the sacrifice is a contract, and the kindness of the gods is real but purchasable. The Gītā itself draws this contrast from inside the tradition. The knowers of the three Vedas, it says, having worshipped with soma and won heaven, enjoy the celestial pleasures of the gods until the merit is spent, and then, when it is exhausted, they fall back into the mortal world (9.20 to 9.21). What the offering purchases, runs out. Against this it sets the devotee who thinks of nothing else, and gives him the sentence that broke the old economy in two.

ananyāś cintayanto māṃ ye janāḥ paryupāsate teṣāṃ nityābhiyuktānāṃ yogakṣemaṃ vahāmy aham

But those who worship me thinking of nothing else, who are constantly joined to me, I myself carry what they lack and preserve what they have. (Gītā 9.22)

The Vedic god receives the offering and pays. This Lord carries. The difference between a contract and a compassion is the whole distance between the two sentences.

Śiva must be spoken of with reverence here, and the Bhāgavata itself sets the tone: vaiṣṇavānāṃ yathā śambhuḥ, among the devotees of Viṣṇu, Śambhu is the greatest (12.13.16). The Śaiva traditions carry a rich theology of grace of their own, and any Śaiva will present it with texts in hand. Within these three books, Śiva is the swiftly pleased one, so generous that he gives without weighing the asker, and the Purāṇas tell with awe of boons granted even to those who would turn them against the worlds. The difference the counting points to is not one of generosity but of register. His is the glory of the mountain and the great austerity, the power at the edge of things; the mercy these books cannot stop describing in Viṣṇu is the tenderness of the lap and the rope, the parent’s register rather than the ascetic’s. Those are two different glories, and the tradition at its best honours both. The Goddess deserves the same honesty. Her mercy is real and vast: in the Śākta vision she is the Mother absolutely, the one who takes up arms the instant her children are threatened, and millions have fled to her as the most immediate refuge they know. The distinction, once more, is not the amount of the love but its direction. She is the mother who fights for the child. The Bhāgavata’s boldest image runs the other way: the Lord becomes the child, puts himself below the mother, and lets her bind him to a mortar. One vision arms the mother; this one lets a mother tie up the Lord. The claim of this essay is only about where unearned tenderness toward the helpless is the defining note, and the counting answers that it is in Viṣṇu.

Then there is the strongest rival of all. The word karuṇā, which the Gītā uses for the mark of the devotee the Lord loves (12.13), is also the great word of Buddhism, the compassion of the bodhisattva who vows to save all beings. But the two compassions have a different shape. Buddhist karuṇā flows from seeing that there is, finally, no separate self anywhere, and at its height it is serene and impersonal, the sun shining equally on all. The compassion of Viṣṇu is personal to its core. It has a face and a name. It plays favourites, bending toward the one who turns to it. It weeps, it descends, it wipes a tired face. It is not the calm of one who has seen through the self, but the ardour of one who loves particular creatures and will cross any distance to reach them. Both are called karuṇā. Only one comes leaping down from an eagle.

Further west, the comparison grows friendlier. A personal Lord who loves, grieves, and gives himself for his people is the closest analogue the world’s faiths offer, and a Christian reader of the Bhāgavata will recognize much. The distinctive Vaiṣṇava notes are two: the descent is not a single event in history but a standing vow renewed age after age, and the tenderness reaches a register the theologies of majesty rarely allow, the Lord as an infant in a mother’s lap, a boy caught stealing butter, a dancer wiping sweat from a girl’s face. Where other traditions guard the dignity of the highest, this one spends it, and counts the spending as glory.

These comparisons are the tradition’s own case, stated as fairly as possible, and each rival tradition has its replies, which deserve their own essays. The claim is not that mercy is absent elsewhere. It is that nowhere else is mercy the first fact, the unearned fact, the descending fact, the tender fact, the universal fact, and the self lowering fact all at once, and about one person. That conjunction is what the verses themselves now have to show.

The compassion he leaves behind

There is a last turn in the theology, and it is the one that keeps this whole subject from being merely consoling. In every one of these books, the compassion of the Lord is contagious. It does not stay in heaven. It reproduces itself in the people it touches.

The Gītā said so when it listed compassion toward all beings, dayā bhūteṣu (16.2), among the marks of a divine nature, with harshness and wrath heading the demonic list a few verses later. The Viṣṇu Purāṇa made it a scene almost too hard to bear. The child Prahlāda, tortured by his father for loving Viṣṇu, is commanded to renounce the Lord he praises, and his answer to the tyrant deserves to be set apart.

Viṣṇu, who is Brahma, is the creator and protector, not of me alone, but of all human beings, and even, father, of you: he is the supreme lord of all. (Viṣṇu Purāṇa I.17, Wilson’s translation)

The child extends the divine protection even to the man tormenting him, and through the chapters that follow he blesses the servants sent to kill him, because he has caught the Lord’s own compassion from the Lord who dwells in every heart and he cannot unlearn it. Behind all of this stands an old and severe conviction, older than any of these books, that non cruelty, ānṛśaṃsya, is the highest law there is, higher than caste, higher than ritual, higher than duty. When your particular duty and simple mercy come into conflict, that conviction says, choose mercy.

This is why the compassion of Viṣṇu is not sentimentality. It makes a demand. If the deepest thing about the Lord is that he trembles with the suffering of a drowning animal, then the deepest thing we can become is someone who trembles too.

With that theology in hand, we can now walk the evidence itself, and watch these five movements surface, book by book, in the verses.

PART THREE: THE TEXTUAL SURVEY, BOOK BY BOOK

This is the survey a primer owes you. We take each text in turn, walk its compassion verses, quote the ones that carry weight, and account for the rest by reference so nothing is hidden. The five movements of Part two are the lens: watch for the Lord who indwells, descends, shelters, forgives, and loves.

The Bhagavad Gītā

The Gītā is short, so we can name every one of its compassion verses. They fall into five natural groups, and together they are the seed of everything the Purāṇas will later grow.

The pity that is corrected. The Gītā’s first three uses of the mercy vocabulary are all negative, and deliberately so. Arjuna is overcome by compassion, kṛpayā parayāviṣṭaḥ, at the sight of his kinsmen (1.27), his eyes filling with tears, kṛpayāviṣṭam aśru pūrṇa (2.1), until he confesses that his nature is stricken by the taint of pity, kārpaṇya (2.7). Kṛṣṇa’s rebuke, do not yield to unmanliness (2.3), is the hinge of the whole poem: the feeling Arjuna calls compassion is the feeling the Gītā means to replace. These are the only verses where the mercy word names a human failing rather than a divine glory, and the contrast is the point.

His compassion. Against that counterfeit stands the true coin, and it is named exactly once, at the center of the poem, in the verse this entire tradition turns on. Out of compassion for them, anukampā, dwelling in their hearts, I destroy the darkness of ignorance with the lamp of knowledge (10.11). Here compassion is the Lord’s own inner act. Its precondition is stated at 15.15, I am seated in the hearts of all, and its human echo at 12.13, the devotee dear to the Lord is friendly and compassionate, maitraḥ karuṇa eva ca, and at 16.2, where compassion toward beings, dayā bhūteṣu, stands among the qualities of the divine nature, with harshness, pāruṣya, listed among the demonic marks two verses later (16.4).

His grace. The Gītā’s language of grace is the word prasāda and the word anugraha. Arjuna receives the supreme teaching, he says, by the Lord’s grace, mad anugrahāya (11.1). And the promise of grace runs through the closing chapter like a refrain: by my grace you will pass beyond all difficulties (18.58), by his grace you attain supreme peace and the eternal abode (18.62), and at the very end Arjuna’s own confession, by your grace, tvat prasādāt, my delusion is destroyed and my memory regained (18.73), with the earlier promise that by my grace, mat prasādāt, the devotee attains the imperishable abode (18.56). Grace here is not earned; it is received.

His refuge. The Gītā’s supreme verse is a verse of refuge. Abandon all dharmas and take refuge in me alone; I will free you from all sins; do not grieve (18.66), which it prepares four verses earlier.

tam eva śaraṇaṃ gaccha sarvabhāvena bhārata tatprasādāt parāṃ śāntiṃ sthānaṃ prāpsyasi śāśvatam

Go to him alone for refuge with your whole being, O Bhārata; by his grace you will attain the supreme peace and the eternal abode. (Gītā 18.62)

And it echoes with those who take refuge in me alone cross beyond māyā, his own bewildering power that no effort crosses (7.14). The refuge is open to all without exception, even, the poem insists, the worst evildoer (9.30 to 9.31) and those of the lowliest birth (9.32).

His love. Beneath refuge lies affection. The friend of all beings, suhṛt (5.29), becomes, by the last chapter, a lover: you are dearly loved by me (18.64), you are dear to me (18.65), and the whole litany of the devotee who is dear to the Lord (12.13 to 12.20). Even the terror of the cosmic vision ends in a plea to the Lord’s tenderness.

tasmāt praṇamya praṇidhāya kāyaṃ, prasādaye tvām aham īśam īḍyam piteva putrasya sakheva sakhyuḥ, priyaḥ priyāyārhasi deva soḍhum

Therefore, bowing down and prostrating my body, I beg your grace, O Lord worthy of worship: as a father bears with a son, as a friend with a friend, as a lover with the beloved, so should you bear with me, O Deva. (Gītā 11.44)

Note: Arjuna has just seen the form that swallows worlds, and the analogies he reaches for are the three most intimate relationships a human being has.

That is the entire compassion vocabulary of the Gītā: 30 loci, 30 distinct places where the mercy vocabulary occurs, by exact count in the Sanskrit, and of them all but three, the cluster of Arjuna’s corrected pity, belong to the Lord’s own mercy, grace, refuge, or love, 9 in every 10. For so short a text to place its single use of anukampā at its exact center, and to close on I will free you from all sins, do not grieve, tells you where its heart lies.

The Viṣṇu Purāṇa

The Viṣṇu Purāṇa carries 111 verses where the mercy is his, and they cluster exactly where you would expect: heaviest in Book One, which tells the great devotional stories, and in Book Five, the life of Kṛṣṇa, and thinnest in the books of cosmography and genealogy between them. That distribution is itself a lesson. Compassion gathers around narrative, around a person in trouble and a Lord who answers.

Book One holds the richest seam, 31 loci spread across chapters 1, 2, 4, 7, 8, 9, and 11 through 20. It opens in a key of grace, the sage receiving his knowledge through the favour of the Lord (I.1), and rises through the cosmogony to the descent of the boar, who plunges into the ocean to raise the drowning Earth (I.4), the first great picture of compassion as rescue. Its devotional heart is a run of four episodes. The gods, defeated, flee to Viṣṇu as the asylum of all, crying have pity upon thy servants, and from the churning that follows rises Śrī, his grace made into a goddess (I.9). The slighted child Dhruva calls him his only refuge and hears in return, I am well pleased with you (I.11 to I.12). The ideal king Pṛthu, a portion of Viṣṇu, is praised as compassionate and kind (I.13). And the tortured child Prahlāda declares that Viṣṇu is the creator and protector not of himself alone but of all beings, even of the father tormenting him (I.17 to I.20). When the Lord at last appears before him, the boy’s greeting compresses the whole book into an epithet.

deva prapannārtihara prasādaṃ kuru keśava avalokanadānena bhūyo māṃ pāvayācyuta

O Deva, remover of the pain of those who come to you, be gracious, Keśava; purify me again, Acyuta, by the gift of your glance. (Viṣṇu Purāṇa I.20.16)

Note: prapannārtihara, remover of the pain of the refuged, is the Purāṇa’s own compound for the whole doctrine, and it returns on other lips later in the book (V.23.31).

Chapters 14 and 15 add the propitiation hymns of Dhruva’s line, where Viṣṇu is the refuge and the giver of boons.

Book Two, the book of cosmography and the earth’s geography, is nearly silent on mercy, with only 3 loci (chapters 2, 6, 12). The one that matters comes in the tale of the royal ascetic Bharata, where the language of taking refuge in the supreme surfaces amid an otherwise diagrammatic book, a reminder that even the map of the worlds is drawn around a Lord one can flee to.

Book Three, on the Vedas, the sages, and dharma, gives 6 loci (chapters 2, 7, 8). Here compassion appears less as rescue than as law.

dayā samastabhūteṣu titikṣā nātimānitā satyaṃ śaucam anāyāso maṅgalaṃ priyavāditā

Compassion toward all beings, forbearance, freedom from arrogance, truthfulness, purity, contentment, auspicious conduct, and gentle speech. (Viṣṇu Purāṇa III.8.36, from the duties by which Viṣṇu is pleased)

Dayā heads the list of the duties that please him, so that mercy toward creatures becomes a form of worship. The book’s teaching on right conduct quietly makes kindness the substance of religion.

Book Four, the long dynastic chronicle, carries 13 loci (chapters 2, 3, 4, 6, 7, 8, 13, 15, 18). Amid the king lists, devotion keeps breaking through: the refrain to that Viṣṇu I fly for refuge recurs, and the story of the pious king Ambarīṣa, the great devotee of Hari, shows the Lord defending the one who takes shelter in him. Even a genealogy, in this Purāṇa, cannot go many chapters without someone fleeing to the Lord.

Book Five, the life of Kṛṣṇa, is the second great reservoir, 52 loci across more than 20 chapters (1, 3, 6, 7, 10 through 13, 18 through 20, 25 through 30, 33 through 38). Here compassion becomes biography. Kṛṣṇa grants liberation to the demoness Pūtanā, who came to kill him and received the reward of a mother instead (V.5 in the wider tradition, echoed here); he subdues the serpent Kāliya but spares his life at the pleading of the serpent’s wives; he lifts Mount Govardhana as a literal shelter, holding the mountain over the cowherds and their cattle to save them from the storm (V.10 to V.13); and through the Vraja chapters he is the protector, the refuge, the beloved of the cowherd people. The concentration of mercy language here, second only to Book One, is the Viṣṇu Purāṇa already leaning toward the Bhāgavata’s Kṛṣṇa.

Book Six, on the dissolution of the world and the horrors of the Kali age, the last and darkest age of the cosmic cycle, closes the Purāṇa with 6 loci (chapters 7, 8) and one consoling doctrine: at the end of time, when every other refuge fails, the mere remembrance of Viṣṇu saves. The book that describes the darkest age makes the darkest age the easiest to be rescued from, because the Lord’s compassion meets the Kali yuga devotee with the least required of any. It is a fitting last word, mercy scaled to human weakness.

Read across all six books, the Viṣṇu Purāṇa shows compassion as something Viṣṇu reliably does, a settled attribute expressed in rescue after rescue. What it has not yet done, and what the Bhāgavata will, is make that mercy the very air the book breathes.

The Bhāgavata Purāṇa

Here the theme floods the book. There are 626 loci where the mercy is his, spread across all twelve skandhas, the trunks, as the Bhāgavata calls its books, and the introductory Māhātmya, with the greatest concentration by far in the tenth skandha, the life of Kṛṣṇa. If the Viṣṇu Purāṇa shows compassion as something Viṣṇu does, the Bhāgavata shows it as the weather of the whole world he moves through. We take it skandha by skandha.

The Māhātmya, the preamble in praise of the Bhāgavata itself, already sets the key with 17 loci (discourses 1, 2, 4, 5, 6). It tells of Devotion personified as a grieving mother restored to youth by the recitation of the Lord’s stories, and of sinners lifted to liberation simply by hearing them. Before the book proper begins, its subject is announced: the mercy of Hari, and the ease with which it can be received.

Skandha One, 44 loci across nearly every discourse (1 through 19), is the frame, and it is thick with compassion because it is a book of deaths and refuges. Kṛṣṇa is remembered as the friend whom, out of affection, Yudhiṣṭhira made his counsellor, envoy, and charioteer. Kuntī prays to him recalling how the mother Yaśodā once bound him with a cord while he stood with his face cast down. And the dying Bhīṣma, the grand old warrior of the Mahābhārata, on his bed of arrows, gives the skandha its great testimony.

tathāpy ekāntabhakteṣu paśya bhūpānukampitam yan me ‘sūṃs tyajataḥ sākṣāt kṛṣṇo darśanam āgataḥ

Yet see, O king, his compassion for those who are devoted to him alone: Kṛṣṇa himself has come into my sight at this hour, as I give up my life. (Bhāgavata 1.9.22)

The word Bhīṣma chooses is anukampitam, the trembling word, and the book opens with the Lord attending personally at a deathbed, which is the whole argument in a single image.

Skandha Two, 19 loci (discourses 2, 4, 5, 7 through 10, 19), is the cosmology taught by Śuka, the young sage who narrates the whole Purāṇa, and its compassion is metaphysical: the Lord assumes his forms and unfolds the worlds not from need but as an act of grace toward the beings who will inhabit them. Even creation, here, is a kindness. And before Śuka will teach a syllable of that cosmology, he bows, and his invocation names exactly whom this Lord purifies.

kirātahūṇāndhrapulindapulkaśā, ābhīraśumbhā yavanāḥ khasādayaḥ ye ‘nye ca pāpā yadapāśrayāśrayāḥ, śudhyanti tasmai prabhaviṣṇave namaḥ

Kirātas, Hūṇas, Āndhras, Pulindas, Pulkaśas, Ābhīras, Śumbhas, Yavanas, Khasas, and whatever other sinners there be: taking shelter of those who have taken shelter of him, they are purified. To that all powerful Viṣṇu I bow. (Bhāgavata 2.4.18)

Note: these are the frontier peoples the old order held furthest from the sacred, hunters and mountain tribes and foreigners, Greeks and Huns among them. The description of the universe begins by bowing to the Lord who purifies precisely them.

Skandha Three, 64 loci and the densest of the early books (discourses 1 through 33, especially 13 through 33), gives two great pictures. It tells again the descent of the boar, who lifts the Earth from the waters, and it culminates in the teaching of Kapila, the sage the book counts among the Lord’s own descents, to his mother Devahūti, where the doctrine is stated outright.

bhṛtyānukampitadhiyeha gṛhītamūrteḥ sañcintayed bhagavato vadanāravindam

Let the devotee meditate on the lotus face of the Lord, who has taken form in this world with a mind moved by compassion for his servants. (Bhāgavata 3.28.29)

Divine embodiment is defined, in so many words, as an act of compassion.

Skandha Four, 59 loci (discourses 1, 6 through 25, 29 through 31), is the book of Dhruva and Pṛthu. Here the Bhāgavata tells its own fuller version of the slighted child Dhruva, whose longing draws the Lord to him and who is met not with a test but with grace, the same episode the Viṣṇu Purāṇa told, now dwelt on at length. Around it stand the tales of Dakṣa and Pṛthu, kingship measured again by mercy.

Skandha Five, 44 loci (discourses 1 through 20, 24, 25, 31), turns to Ṛṣabha, the renunciate avatāra, and to the strange, moving career of Bharata, a king who gave up a throne for the forest and was undone there by a single act of tenderness. A pregnant doe, startled at the river by a lion’s roar, miscarried into the current and died, and the royal sage saw the newborn fawn being swept downstream, wretched, kṛpaṇam, and took it up, says the Bhāgavata, out of compassion, anukampayā, like a kinsman (5.8.7). It is the Lord’s own vocabulary spent on a human act, the kṛpaṇa and the anukampā of the great rescues given to a fawn. Bharata raised it, loved it past all his vows, died thinking of it, was reborn as a deer for that attachment, and was then given a human birth in which he finally walked free. The skandha reads his whole arc the way it reads everything else it contains: as the patience of a Lord who follows a soul through many births and many errors and does not abandon it. Even the man undone by compassion is not, in the end, undone. And even in this book’s descriptions of the hells, the Lord’s mercy is the thread by which the fallen are eventually raised.

Skandha Six, 27 loci (discourses 1 through 19), is the book of Ajāmila, and so it is one of the tradition’s central testimonies to grace beyond desert. The fallen brahmin cries his son’s name Nārāyaṇa at the hour of death, and the messengers of Viṣṇu wrench Yama’s noose from him, on the strength of a Name spoken without meaning. Around this stand the hymns of Dakṣa and the doctrine, stated plainly here, that to develop love for the Lord through his Name is the paramount duty of all, because the Name reaches even those who cannot reach it themselves.

Skandha Seven, 32 loci (discourses 1, 4 through 15), is the book of Prahlāda, and therefore the book where divine compassion and human compassion meet. The Lord bursts from the pillar as the man lion to save the child devotee, and the child, standing before that terrifying rescue, addresses him with an epithet that compresses the whole theology.

trasto ‘smy ahaṃ kṛpaṇavatsala duḥsahogra, saṃsāracakrakadanād grasatāṃ praṇītaḥ baddhaḥ svakarmabhir uśattama te ‘ṅghrimūlaṃ, prīto ‘pavargaśaraṇaṃ hvayase kadā nu

I am terrified, O you who are tender to the wretched, driven among devourers by the unbearable, grinding wheel of births, and bound by my own deeds. When, O most radiant one, will you be pleased, and call me to the base of your feet, the refuge that is liberation? (Bhāgavata 7.9.16)

Note: kṛpaṇa vatsala, tender to the wretched. The child saved from the pillar does not ask for safety. He asks for the feet.

And he in turn refuses to hate his persecutors and pleads even for his demon father, having caught his mercy from the Lord he loves. Prahlāda’s discourse on devotion makes compassion toward all beings the very substance of loving the Lord.

Skandha Eight, 47 loci (discourses 4 through 24), is a book of rescues, and it holds the elephant. Gajendra, seized in the lake, lifts his lotus and cries out, and the Lord alights from his eagle out of extreme compassion and draws him from the water (8.3 to 8.4). Around this stand the churning of the ocean, the saving of Manu from the flood by the fish, and the descent of the dwarf Vāmana, who grants even the defeated demon king Bali not destruction but a lasting place, mercy extended to an enemy who kept his word. If you wanted a single skandha to teach the theme, this is the one.

Skandha Nine, 26 loci (discourses 1 through 24), the dynastic book, keeps the theme alive through the royal lines, above all in Ambarīṣa, the devotee king whom the Lord’s discus defends against the wrath of a great sage, so that the one who takes refuge in Viṣṇu is shielded even from a brahmin’s curse. And it is in this story, when the terrified sage reaches Vaikuṇṭha itself, that the Lord makes the most astonishing confession in the three books.

ahaṃ bhaktaparādhīno hy asvatantra iva dvija sādhubhir grastahṛdayo bhaktair bhaktajanapriyaḥ

I am wholly under the power of my devotees, O brāhmaṇa, as if I had no independence at all; my heart is seized by the saintly, for I am beloved of my devotees and they of me. (Bhāgavata 9.4.63)

ye dārāgāraputrāpta prāṇān vittam imaṃ param hitvā māṃ śaraṇaṃ yātāḥ kathaṃ tāṃs tyaktum utsahe

They have left wife, home, children, kinsmen, wealth, and life itself, and come to me for refuge. How could I bear to abandon them? (Bhāgavata 9.4.65)

sādhavo hṛdayaṃ mahyaṃ sādhūnāṃ hṛdayaṃ tv aham

The saintly are my heart, and I am the heart of the saintly. (Bhāgavata 9.4.68)

Note: asvatantra, without independence. The sovereign of everything describes himself as the one being in existence who is not free, because love has already committed him.

Nowhere does the supremacy claim of this essay stand on firmer text. Other scriptures show the highest being merciful. This one shows him owned. The genealogies, as in the Viṣṇu Purāṇa, keep pausing to show the Lord protecting those who are his.

Skandha Ten, 210 loci, more than a third of the whole book’s mercy language, is the life of Kṛṣṇa and the summit of the entire subject. Here compassion becomes intimacy. It is the book of vātsalya, of Yaśodā and the babe grown suddenly heavy as a mountain in her lap, of the Lord whom nothing can bind consenting to a mother’s rope. Shelter runs all through it too, most literally when Kṛṣṇa lifts Mount Govardhana over the whole of Vraja to save the people and their cattle from the storm.

atyāsārātivātena paśavo jātavepanāḥ gopā gopyaś ca śītārtā govindaṃ śaraṇaṃ yayuḥ

The animals trembling under the driving rain and wind, the cowherd men and women stricken with cold, they went to Govinda for refuge. (Bhāgavata 10.25.11)

He answers by holding up a mountain for seven days, and the chapter names him bhakta vatsala, tender to his devotees (10.25), one of the 11 times the Bhāgavata reaches for that epithet. Even before any of this, the skandha’s opening hymn has the gods tell the unborn child that earth and heaven themselves will become objects of his compassion.

diṣṭyāṅkitāṃ tvatpadakaiḥ suśobhanair, drakṣyāma gāṃ dyāṃ ca tavānukampitām

By good fortune we shall see this earth marked with the beauty of your footprints, and heaven too, made an object of your compassion. (Bhāgavata 10.2.38)

Grace here reaches even those who came to kill him, and the protection over the cowherds never lapses. And at the center of it all stands the night dance, where he reveals himself in as many forms as there are women so that none is left alone, and where, in the gesture the whole primer has been circling, the compassionate Lord with his own blissful hand wipes the faces of the girls made tired by dancing (10.33.21). Across its discourses, from the childhood to the Vraja years to the deeds at Mathurā and Dvārakā, the tenth skandha is one sustained demonstration that the Lord of the cosmos is also the most tender presence in a human life.

Skandha Eleven, 23 loci (discourses 2 through 9, 23, 26, 29 through 31), is the Lord’s final teaching before his departure, the Uddhava discourse, and its compassion turns inward and ethical. Among the marks of the true saint, it lists mercy toward all creatures, friendliness, and the absence of cruelty, so that the compassion the Lord has shown becomes the compassion the devotee is asked to become. The book that has spent ten skandhas showing divine mercy now hands it to us as a practice.

Skandha Twelve, 14 loci (discourses 6, 8, 10 through 13), closes the Purāṇa in the Kali age, and its last word on compassion is the gentlest and most democratic.

kaler doṣanidhe rājann asti hy eko mahān guṇaḥ kīrtanād eva kṛṣṇasya muktasaṅgaḥ paraṃ vrajet

The age of Kali is an ocean of faults, O king, but it holds one great virtue: by the chanting of Kṛṣṇa’s name alone, a person is freed from every bond and reaches the highest. (Bhāgavata 12.3.51)

In the dark age, when every discipline is beyond most people, the sole sufficient grace is the chanting of the Lord’s name, freely available to anyone, requiring neither purity nor learning nor strength. The Bhāgavata ends where Ajāmila began, on the mercy carried in a Name, now offered to everyone alive in the hardest of times.

Read whole, the Bhāgavata does not argue that Viṣṇu is compassionate. It assumes it on every page and then shows it, from the deathbed of Bhīṣma to the lake of the elephant to the dust of Vraja, until mercy is no longer one of the things the book says about the Lord but the thing the book is.

Conclusion: The Lord who cannot look away

Set the three books side by side and the movement is unmistakable. The Gītā defines compassion by first refusing a counterfeit, then placing the Lord’s own anukampā at its exact center. The Viṣṇu Purāṇa turns that definition into a series of rescues, compassion as a reliable divine act. And the Bhāgavata lets mercy saturate everything, until it is the innermost truth of the Lord and the whole reason he takes a body at all. Something Viṣṇu has becomes something Viṣṇu is.

This essay opened with a claim that must have sounded like devotional excess: that Viṣṇu is the most compassionate Lord human beings have ever worshipped. It is time to gather what the three books have shown and let the claim stand on its evidence. Six things, taken together, are the case.

First, mercy is his defining attribute, not one quality on a list. The counting showed it: the vocabulary of tenderness does not merely appear in these books, it multiplies across them, migrates onto his person, and hardens into his very epithets, tender to his devotees, tender to the afflicted, tender to the wretched. Other conceptions of the highest are defined first by something else, by power, by law, by austerity, by stillness. This one is defined by the trembling.

Second, his grace reaches the guilty. Boons in the wider tradition are wages; even demons earn them by austerity. His anugraha breaks the contract. Ajāmila is saved by a Name spoken without meaning it, and the worst of evildoers, the Gītā insists, is to be counted righteous the moment his heart turns. A mercy that must be earned is a payment. A mercy that outruns desert is the only kind that deserves the name.

Third, his compassion descends. Not once, as a concession, but age after age, as a standing vow. He has been a fish for a drowning man, a tortoise under a churning mountain, a boar in the depths after a drowning world, a dwarf before a demon king, a man lion for a tortured child, a cowherd boy in the dust of Vraja. A compassion that stayed in heaven would be a sentiment. His takes a body, every time, and the direction of the body is always downward.

Fourth, it is personal and tender. The great rival compassion, the karuṇā of the Buddha’s way, is serene, impersonal, the sun on all alike. His has a face and a fondness. It plays favourites with whoever turns to it. It weeps at a deathbed, lets a mother tie it to a mortar, and spends the language of world rescue on wiping the sweat from a dancer’s cheek.

Fifth, it is universal in reach. It has descended for an animal, a fallen priest, a slighted child, defeated gods, women the order slighted and the low born the order despised, and, in the end, even for the demons who came to kill him, one of whom received the reward of a mother. No one is structurally outside it. The breadth of a mercy is a measure of the mercy.

And sixth, it lowers itself. Saulabhya, the tradition calls it, the quality of being easy to reach. The one the Upaniṣads cannot find words for consents to be small, holdable, bindable, biddable, a messenger, a charioteer, a boy with his face cast down before his mother’s rope. Other conceptions guard the transcendence of the highest. This one spends its transcendence, gladly, on being reachable. Condescension, in its oldest and cleanest sense, a descending to be with, is the final proof.

Each of these, taken alone, can be found somewhere else. There is grace in the Śaiva traditions, real and richly theorized. There is compassion at the very center of the Buddhist path. There is a personal, loving, self giving Lord in the faiths of the West. The Vaiṣṇava case does not rest on any single pillar. It rests on the conjunction, mercy as the defining attribute, unearned, descending in the flesh again and again, tender rather than serene, excluding no one, and stooping all the way down, and on the claim, borne out by the counting, that no other vision of the highest holds all six at once. That is the tradition’s case, made from inside it, and the other traditions have their own replies. But it is not excess. It is an argument, and its evidence is a thousand years of text that cannot stop saying one thing.

And the last turn is the one that returns the whole subject to us. This compassion is contagious. The child Prahlāda forgives his torturers because he has caught it. The saint of the eleventh skandha is known by his mercy toward all beings. The oldest law the tradition knows, older than caste or ritual, is non cruelty, ānṛśaṃsya, the highest dharma. If the deepest thing about the Lord is that he trembles at the cry of a drowning animal, then the deepest thing we can become is someone who trembles too.

He is the Lord who comes when you call. The texts spend a thousand years saying so, in a hundred voices, through an elephant and a boar, a fugitive host of gods and a slighted child, a fallen priest and a mother with a rope and a weeping archer. And the quiet claim beneath all of it is that we were made in the image of exactly that, and are only truly ourselves when we cannot look away either.

• Bhagavad Gītā: Winthrop Sargeant’s translation.

• Viṣṇu Purāṇa: H. H. Wilson’s 1840 translation, for the English passages marked as his.

• Bhāgavata Purāṇa: Gītā Press edition, with the Motilal edition of G. V. Tagare for cross reference.

• The translations set beneath the Sanskrit in this essay are my own, except where marked as Wilson’s.

• Sanskrit counts: the complete GRETIL electronic texts of both Purāṇas, and the sanskritdocuments.org text of the Gītā.

• Jan Gonda, Aspects of Early Viṣṇuism, on Śrī and the preserver identity.

• S. M. Srinivasa Chari, Vaiṣṇavism, where compassion is counted among the essential auspicious attributes of the Lord.

Most of the famous Upaniṣads open in the middle of a conversation. In the Kaṭha Upaniṣad, a boy named Naciketas badgers Death himself about what survives the body. In the Bṛhadāraṇyaka, the sage Yājñavalkya spars with a king in open court, and, elsewhere in that same Upaniṣad, tries to explain the self to his wife Maitreyī before he leaves home for good. The Śvetāśvatara opens differently. It opens with a seminar.

A circle of men described as Brahmavādins, people who discuss Brahman, the ultimate reality behind the world, are sitting together and they put a question on the table:

kiṃ kāraṇaṃ brahma kutaḥ sma jātā
jīvāma kena kva ca saṃpratiṣṭhāḥ
adhiṣṭhitāḥ kena sukhetareṣu
vartāmahe brahmavido vyavasthām

“What is the cause? Is it brahman? From where were we born? By what do we live, and where do we come to rest? Governed by whom do we live out our lives in pleasure and in pain, you who know brahman?” (1.1)

Then they run through the possible answers one by one:

kālaḥ svabhāvo niyatir yadṛcchā
bhūtāni yoniḥ puruṣa iti cintyā

“Time, inherent nature, necessity, chance, the elements, the womb of things, the person: so runs the inquiry.” (1.2)

Is the cause of everything time? Is it Svabhāva, the inherent nature of things? Is it necessity, or chance, or the basic physical elements, or Puruṣa, the conscious person? The Upaniṣad runs through the theories the early Indian thinkers actually argued over, and then it gives an answer that almost no other Upaniṣad gives in quite this voice:

te dhyānayogānugatā apaśyan
devātmaśaktiṃ svaguṇair nigūḍhām
yaḥ kāraṇāni nikhilāni tāni
kālātmayuktāny adhitiṣṭhaty ekaḥ

“The sages, absorbed in meditation, saw the self power of God, hidden by its own qualities. He, the One, alone presides over all those causes, from time to the self.” (1.3)

That answer tells you what kind of Upaniṣad you are holding. The sages who went deep in meditation found, behind time and nature and chance and the self, a single creative power that belongs to a Deva, a god. The word the verse gives that power, Devātmaśakti, the divine’s own power, is the earliest appearance of a term that the later traditions of the Goddess would treasure. And the cause turns out to be a person. One God, who stands behind all the impersonal candidates the seminar had listed, and rules them.

The Śvetāśvatara is a short Upaniṣad, six chapters of verse, and it is the one where the impersonal Absolute of the older Upaniṣads becomes a personal God, an Īśvara, a Lord with a will and a face, who creates, who rules, who hides inside every being, and who can be loved. It is the Upaniṣad that the later devotional traditions, both the worshippers of Śiva and the worshippers of Viṣṇu, would reach for again and again, because it is the place in the oldest layer of Indian scripture where philosophy and devotion are already fused. Almost everything that Hinduism becomes over the next two thousand years, the personal God or Īśvara, the practice of meditative yoga, the devotee’s surrender to the Lord, the doctrine of grace, the union of reason and love, is already here in seed. This is a guide to that Upaniṣad: to what its verses actually say, to the people who have spent their lives translating it, and to why the three great schools of Vedānta could read the very same lines and walk away with three different Gods.

A note on how to read along. I give the famous verses in the original, in transliteration, with a plain translation beneath, because the Śvetāśvatara is a text where the choice of an English word is often a choice of theology, and a reader who can see the Sanskrit sitting under the English is in a far better position to judge. The Sanskrit terms get explained the first time they appear. And the scholars and teachers I lean on, the philologists Max Müller, Patrick Olivelle, and Signe Cohen, the philosopher Sarvepalli Radhakrishnan, and the three founding masters of Vedānta, Śaṅkara, Rāmānuja, and Madhva, get introduced before I put words in their mouths.

The Upaniṣad signs its own name

Start with something the Śvetāśvatara does that no other classical Upaniṣad does. It tells you who wrote it.

The Upaniṣads are, as a rule, anonymous. They are named for the schools of recitation that preserved them, or for a teacher inside the story, and the actual composers stay invisible, which suited a tradition that liked to present its wisdom as timeless and authorless, breathed out with the Veda itself. The Śvetāśvatara breaks the pattern. Near the very end, in the verse numbered 6.21, it names its author:

tapaḥprabhāvād devaprasādāc ca
brahma ha śvetāśvataro ‘tha vidvān
atyāśramibhyaḥ paramaṃ pavitraṃ
provāca samyag ṛṣisaṃghajuṣṭam

“Through the power of his austerity and through the grace of God, the wise Śvetāśvatara rightly proclaimed brahman, the supreme and holy, to the most advanced renouncers, and it was welcomed by the company of seers.” (6.21)

The Upaniṣad is named after a man, and the man is named inside it. Signe Cohen, an American scholar of Sanskrit whose 2008 book Text and Authority in the Older Upaniṣads is one of the more careful modern studies of how these texts were built, makes this her starting point. Cohen's point is that the Śvetāśvatara is the only one of the classical Upaniṣads ascribed to an individual, historical author, a named person rather than a school of recitation. The name itself is debated, and its two natural readings pull in different directions. Split one way, as śveta plus aśvatara, the ordinary Sanskrit word for a mule, it means "the one with white mules," which is how the standard lexicon sees it. Split the other way, as śvetāśva, white horse, plus a suffix, it points to white horses instead. Some scholars even take the name as that of a Vedic school rather than a man. Tradition reads it as the personal name of the sage, and some later interpreters heard in it a symbol of purified senses, though that gloss is devotional embroidery rather than settled philology. Whoever he was, he wanted his name on the work.

By Cohen’s analysis this signature belongs to the oldest layer of the text. The impulse to claim personal authorship was there from the start. This is a composer who wanted you to know that a particular person, by a particular discipline and a particular gift of grace, had come to see the Deva, the God, and was now telling you about it. The Upaniṣad is, in that sense, a testimony, and notice that the very signature verse already contains the two ideas the whole work will live by: austerity, the seeker’s effort, and Devaprasāda, the grace of God. Effort and grace, named together, in the line where the author signs his name.

How old is it?

Before going further into what the Upaniṣad teaches, it helps to know roughly when it was speaking, because the Śvetāśvatara is usually filed among the later principal Upaniṣads, and it is worth seeing exactly what that does and does not mean.

Patrick Olivelle, whose 1998 critical edition and translation, The Early Upaniṣads, is the standard scholarly text in English, places it after the Kaṭha and the Īśā and before the Muṇḍaka, and dates it, with appropriate caution, to “the last few centuries BCE.” He quotes the German scholar Thomas Oberlies to the effect that it can hardly have been composed long before the first century before the common era.

Cohen complicates that judgment in a way worth slowing down for, because it changes how you read the whole thing. Rather than date the text as a single block, she dates its layers. Old Indian verse comes in different metres, and the metres carry chronological fingerprints. Most of the Śvetāśvatara is composed in the older, looser metres called triṣṭubh and jagatī, the same metrical world as the oldest Upaniṣads. Running through it is a second layer in the tighter, later śloka metre, the classical verse form, and that layer often reads like a teacher’s explanatory gloss that has been absorbed into the text and now sits right beside the verse it was explaining. Her conclusion is careful: the core was “composed around the same time as some of the oldest Upaniṣads, but likely later than the very oldest,” and what we now hold is a work that grew in layers over a long stretch of time. She is insistent that we are reading the “extant version,” the text as it ended up, not some pristine original.

So the answer to “how old is the Śvetāśvatara” is that it has an old heart inside a body that kept growing. The questions about the cause of the world, the vision of the one God, the signature of the sage, these belong to the deep layer. Some of the more systematic philosophical verses came later. Putting the evidence together, a defensible working date for the Śvetāśvatara as we have it is roughly 400 to 200 BCE, with a core that may reach back further still. Olivelle himself warns that any attempt to date these documents more precisely than a few centuries is, in his phrase, “as stable as a house of cards,” so a range, and not a single year, is the honest answer. Read this way, the occasional unevenness of the Upaniṣad’s surface stops being a puzzle and becomes a record of how a living tradition kept thinking with the same short scripture for centuries.

The one God

At the center of the Śvetāśvatara stands a God, and everything distinctive about the Upaniṣad flows from him.

The older Upaniṣads are famous for an impersonal vision of the ultimate. Brahman, in the Bṛhadāraṇyaka and the Chāndogya, is reality itself, consciousness without a second, approached by stripping away every limited description until only the bare “not this, not this” remains. It is the ground of being, and it is identical with your own deepest self, the Ātman, but it is not a person you pray to. The Śvetāśvatara keeps all of that metaphysical depth and gives the Absolute a face.

The face it gives is Rudra, the fierce Vedic god of storm and wild places who would later be worshipped as Śiva. The Upaniṣad is emphatic about his oneness:

eko hi rudro na dvitīyāya tasthur
ya imāṃl lokān īśata īśanībhiḥ
pratyaṅ janāṃs tiṣṭhati sañcukocānta-
kāle saṃsṛjya viśvā bhuvanāni gopāḥ

“Rudra is truly one; they did not allow a second. He rules these worlds by his ruling powers. He stands facing all beings; having created all the worlds, he, the protector, draws them back at the end of time.” (3.2)

There is a deliberate insistence here, a hammering on the word one, that Olivelle notices and names. A major aim of the author, he writes, is “to establish that the God who creates and from whom one expects salvation is one,” and the Upaniṣad drives this home by repeating the little word Eka, one, again and again. This is one supreme God who has gathered up all the functions the older texts had spread across an impersonal Absolute and a crowd of deities.

And this God is described in language of startling reach. He is the great Puruṣa, the cosmic Person, sunlike, and to know him is to pass beyond death. This is one of the most chanted verses in the whole tradition:

vedāham etaṃ puruṣaṃ mahāntam
ādityavarṇaṃ tamasaḥ parastāt
tam eva viditvāti mṛtyum eti
nānyaḥ panthā vidyate ‘yanāya

“I know that great Person, the color of the sun, beyond the darkness. Only by knowing him does one go beyond death. There is no other path to go.” (3.8)

That last line, nānyaḥ panthā vidyate ‘yanāya, “there is no other path,” became a stock formula of the whole liberation tradition. We will come back to this verse, because it carries a secret about where the Śvetāśvatara found its material.

The God of this Upaniṣad is also described through paradox, in a verse that became the model for a whole style of negative theology:

apāṇipādo javano grahītā
paśyaty acakṣuḥ sa śṛṇoty akarṇaḥ
sa vetti vedyaṃ na ca tasyāsti vettā
tam āhur agryaṃ puruṣaṃ mahāntam

“Without hands and feet, he moves swiftly and grasps; he sees without eyes, he hears without ears. He knows whatever can be known, but of him there is no knower. They call him the foremost, the great Person.” (3.19)

And, most expansively of all, this God is every gender and every age, present in all things:

tvaṃ strī tvaṃ pumān asi
tvaṃ kumāra uta vā kumārī
tvaṃ jīrṇo daṇḍena vañcasi
tvaṃ jāto bhavasi viśvatomukhaḥ

“You are woman, you are man; you are the youth and also the maiden. You, grown old, totter along on a staff; you, once born, come to be facing in every direction.” (4.3)

Most beautiful of all, in the line that gives this essay its title, he is hidden:

eko devaḥ sarvabhūteṣu gūḍhaḥ
sarvavyāpī sarvabhūtāntarātmā
karmādhyakṣaḥ sarvabhūtādhivāsaḥ
sākṣī cetā kevalo nirguṇaś ca

“The one God, hidden in all beings, all pervading, the inner self of all beings, overseer of all action, dwelling in all beings, the witness, the conscious one, alone, and beyond all qualities.” (6.11)

Eko devaḥ sarvabhūteṣu gūḍhaḥ. The one God, concealed in everything that exists, present in you as your own innermost witness. The Absolute has not vanished. It has been given a personal face, and the ground of being is now someone who sees you. Notice the last line of that verse, where a single God is called both Sākṣī, the witness, and Nirguṇa, beyond qualities. In one breath the Upaniṣad holds together the warm personal God of devotion and the attributeless Absolute of the philosophers, and that double grip is exactly why every later school could find itself here.

The borrowed jewels

Here is one of the most interesting things about the Śvetāśvatara, and one that popular accounts almost never mention. Some of its most famous verses are not original to it. The Upaniṣad is quoting older scripture, and once you know where to look, the seams are visible, and they tell a story.

Take that great verse 3.8, “I know that great Person, the color of the sun.” It feels like the heart of the Upaniṣad, and it is, but it did not begin there. It is a verse from the later layer of the Puruṣa Sūkta, the ancient hymn of the cosmic Person, transmitted in the Yajurveda, in the Taittirīya Āraṇyaka, and in the Mahānārāyaṇa Upaniṣad. A small correction worth making, because popular sources blur it: this verse is not from the original Ṛgvedic Puruṣa Sūkta of book ten. It belongs to the appended verses preserved in the Yajurveda recensions. The Śvetāśvatara took an old liturgical jewel and set it into its own crown.

The same is true, even more strikingly, of the most famous image in the entire Upaniṣad, the two birds, which we will read in full in a moment. That verse is word for word identical to a verse of the Muṇḍaka Upaniṣad, and both of them reproduce, syllable for syllable, a verse of the Ṛgveda itself, the riddle hymn of book one. A single image traveled from the oldest Indian poetry, through the Muṇḍaka, into the Śvetāśvatara, unchanged.

And then there is a quieter borrowing that shows the Śvetāśvatara’s hand more clearly than any other. One of the most quoted lines in all of Hindu scripture is the paradox of the divine size, smaller than the smallest and larger than the largest. The Śvetāśvatara has it:

aṇor aṇīyān mahato mahīyān
ātmā guhāyāṃ nihito ‘sya jantoḥ
tam akratuḥ paśyati vītaśoko
dhātuḥ prasādān mahimānam īśam

“Smaller than the small, greater than the great, the self is set in the secret cave of the heart of this creature. The one without craving, freed from sorrow, beholds by the grace of the Ordainer the majesty, the Lord.” (3.20)

This verse also appears, almost identically, in the older Kaṭha Upaniṣad. The two are the same for three of their four lines. But the Śvetāśvatara changes the final word. Where the Kaṭha ends by saying that the seeker beholds “the greatness of the self,” mahimānam ātmanaḥ, the Śvetāśvatara ends by saying he beholds “the majesty, the Lord,” mahimānam īśam. One word is swapped, and an impersonal verse about the self becomes a theistic verse about a personal Lord. That single substitution is the whole project of the Śvetāśvatara in miniature. It inherits the meditative depth of the older Upaniṣads and turns it, gently and deliberately, toward a God.

So the Śvetāśvatara is, among other things, a great anthologist. It gathers the inherited jewels of the tradition, the cosmic Person from the Yajurveda, the two birds from the Ṛgveda and the Muṇḍaka, the size paradox from the Kaṭha, and it resets each one in a frame where the ultimate reality has a face and a name. The borrowing is not a weakness. It is the method. The Upaniṣad is showing you that the God it proclaims was there in the oldest scripture all along, waiting to be recognized.

Two birds on one tree

Now to that most beloved image, and to the place where the great schools of Vedānta divide.

dvā suparṇā sayujā sakhāyā
samānaṃ vṛkṣaṃ pariṣasvajāte
tayor anyaḥ pippalaṃ svādv atty
anaśnann anyo abhicākaśīti

“Two birds, companions ever joined, cling to the same tree. One of the two eats the sweet fruit; the other, not eating, looks on.” (4.6)

samāne vṛkṣe puruṣo nimagno
‘nīśayā śocati muhyamānaḥ
juṣṭaṃ yadā paśyaty anyam īśam asya
mahimānam iti vītaśokaḥ

“On the same tree the person, sunk in matter, grieves, deluded by his own powerlessness. But when he sees the other, the Lord, worshipped, and his greatness, then his grief passes away.” (4.7)

Two birds in one tree. One eats the fruit of experience, its pleasures and pains. The other simply watches, serene, tasting nothing. And the liberation of the grieving soul comes when it lifts its eyes to the watching bird and recognizes its glory. The image is so vivid, and so old, that every school of Vedānta claimed it and built on it, and the question they fought over is simple to state: who exactly are the two birds, and how are they related?

The eighth century master Śaṅkara, founder of Advaita, the school of pure nonduality, reads the two birds as the individual soul and God, and then dissolves the distance between them. Commenting on the same verse where it appears in the Muṇḍaka, he says the first bird, the individual soul, eats the fruits of pleasure and pain “owing to non discrimination,” while the second, God, “merely by his presence as the eternal witness,” tastes nothing and directs everything, the way a king presides without acting. And liberation, for Śaṅkara, is the moment the soul realizes that the two birds were never truly two. The soul sees the witness and understands, “I am this One who is the self of all,” and that the eating, suffering self was a misperception. For Advaita, the duality is a teaching device that points beyond itself to a final identity. In the end there is one bird.

The eleventh century teacher Rāmānuja, founder of Viśiṣṭādvaita, the school of qualified nonduality, reads the same image and stops short of that last step. For him the two birds are the soul and God, and they remain two, but they are inseparable, joined as a body is joined to the self that animates it. The soul is real and forever distinct, yet it has no existence apart from the God who is its inner ruler. Liberation is the soul, still itself, resting at last in loving nearness to a God who was always its support, rather than any discovery that the soul was God all along. There are two birds, bound together, forever.

The thirteenth century teacher Madhva, founder of Dvaita, the school of pure dualism, reads the two birds and keeps them fully and eternally apart. For Madhva the soul and God are different in the way light is different from the source of light, dependent but never identical, and the dependence itself, and the love that flows along it, is eternal. The soul never becomes God, in bondage or in liberation. There are two birds, and they are two forever.

One image, and three destinations for the soul. The Śvetāśvatara gives you the tree and the birds and the moment of looking up, and leaves the tradition to argue, for more than a thousand years, about what the soul sees when it looks. This is the Upaniṣad doing what it does best. It hands you a picture vivid enough to carry the weight of an entire philosophy, and old enough that all three schools can honestly claim it.

There is a detail here that almost no one outside the specialist literature mentions, and it is worth pausing on. None of these three great teachers actually wrote a surviving, accepted commentary on the Śvetāśvatara itself. There is a commentary on it that circulates under Śaṅkara’s name, but modern scholarship, most thoroughly in a 2018 study by Ivan Andrijanić, judges it the work of a later hand, because its vocabulary reflects developments in Advaita thought that came after Śaṅkara and it quotes texts younger than he was. Rāmānuja wrote no verse by verse commentary on any Upaniṣad. Madhva commented on ten Upaniṣads, and the Śvetāśvatara is not among them. What we have instead is something more revealing: all three traditions reached into this one short Upaniṣad and pulled out the verses they needed, and read the two birds through the lens of their own systems. The Śvetāśvatara was not so much commented upon as quarried.

The meeting of the schools

The Śvetāśvatara is also a crossroads of philosophies, and learning to see them is part of the pleasure of reading it.

That opening list of candidate causes, time and nature and necessity and chance, is a snapshot of the live philosophical debates of the day, and two of the systems it engages leave deep marks on its language. The first is Sāṃkhya, the analytical philosophy that explains the world as the interplay of two ultimate principles, puruṣa, pure consciousness, and prakṛti, primal matter, which unfolds into everything through the mixing of its three strands or guṇas. The second is Yoga, the discipline of stilling body and mind to reach liberating insight.

The Sāṃkhya vocabulary surfaces most famously in a verse that puzzled commentators for centuries:

ajām ekāṃ lohitaśuklakṛṣṇāṃ
bahvīḥ prajāḥ sṛjamānāṃ sarūpāḥ
ajo hy eko juṣamāṇo ‘nuśete
jahāty enāṃ bhuktabhogām ajo ‘nyaḥ

“There is one unborn female, red and white and black, giving birth to many creatures like herself. One unborn male lies by her and enjoys her; another unborn male leaves her once his enjoyment is done.” (4.5)

The red, white, and black female who gives birth to the world is read as prakṛti, nature, with her three guṇas, and the two unborn males as two souls, one bound to her and one free. The whole verse is a riddle in the Sāṃkhya key. The Upaniṣad even names the legendary founder of that system:

yo yoniṃ yonim adhitiṣṭhaty eko
viśvāni rūpāṇi yonīś ca sarvāḥ
ṛṣiṃ prasūtaṃ kapilaṃ yas tam agre
jñānair bibharti jāyamānaṃ ca paśyet

“The One who presides over each and every womb, over all forms and all sources, who bore the seer Kapila, born at the beginning, and watched over him as he was being born.” (5.2)

Whether Kapila here means the philosopher Kapila, the founder of Sāṃkhya, or simply means “tawny” and points to a golden cosmic being, is one of those genuinely open questions the readers handle differently, and we will see the divide in a moment. What is not in doubt is that the Upaniṣad is in living conversation with named schools of thought.

It also contains some of the oldest practical instructions for yoga meditation anywhere in the Upaniṣads:

trir unnataṃ sthāpya samaṃ śarīraṃ
hṛdīndriyāṇi manasā sanniveśya
brahmoḍupena pratareta vidvān
srotāṃsi sarvāṇi bhayānakāni

“Holding the body steady, with the three upper parts erect and even, drawing the senses and the mind into the heart, the wise man should cross all the fearful currents by the raft of brahman.” (2.8)

This is the classic posture verse, the seated, upright spine of meditation, and the yoga it teaches is Dhyāna, absorption, supported by restraint of the breath and the senses, the meditative kind rather than the postural gymnastics the word suggests today. It comes centuries before Patañjali gathered such practices into the systematic eight limbed yoga of the Yoga Sūtras, a work current scholarship places around the fourth or fifth century of the common era, well after this Upaniṣad. And the thread the Śvetāśvatara strikes here, meditation turned toward a personal Lord, is one Patañjali would keep: among his disciplines is Īśvarapraṇidhāna, surrender to the Lord. The same posture verse is echoed almost exactly in the Bhagavad Gītā’s instruction to hold the body, head, and neck straight and still. The Śvetāśvatara sits at the headwaters of the yoga tradition as surely as it does the devotional one.

But notice what the Śvetāśvatara does with the philosophy it borrows. Sāṃkhya, in its classic form, has no God. It explains the world through two impersonal principles and leaves it there. The Śvetāśvatara takes the whole Sāṃkhya apparatus and sets a personal Lord above it, a God who wields prakṛti rather than merely standing beside it. The dualism of consciousness and matter is kept as a description of the world, and then both are gathered under one God who is their source and ruler. The Upaniṣad is doing theology with the tools of philosophy, and it is among the earliest texts to do it so fully.

That theology reaches its high point in two verses that the later schools would never stop quoting. The first describes a God who acts without effort or instrument:

na tasya kāryaṃ karaṇaṃ ca vidyate
na tatsamaś cābhyadhikaś ca dṛśyate
parāsya śaktir vividhaiva śrūyate
svābhāvikī jñānabalakriyā ca

“Of him there is no work to be done and no instrument of action. No one is seen equal to him or greater. His supreme power is heard to be manifold, and natural, expressing itself as knowledge, strength, and action.” (6.8)

That last phrase, the threefold natural power of knowledge and strength and action, jñāna bala kriyā, became a technical foundation of later theology, the way the worshippers of Viṣṇu and the worshippers of Śiva alike would describe the inner life of God. The second is the verse that the dualist schools would make their banner, and we turn to it next.

The God of the philosophers and the difference of souls

There is a verse near the end of the Śvetāśvatara that became one of the single most important proof texts in all of Vedānta, especially for the schools that insist the soul and God are permanently distinct.

nityo nityānāṃ cetanaś cetanānām
eko bahūnāṃ yo vidadhāti kāmān
taṃ kāraṇaṃ sāṃkhyayogādhigamyaṃ
jñātvā devaṃ mucyate sarvapāśaiḥ

“The one eternal among the many eternals, the one conscious among the many conscious, who, being one, grants the desires of the many: knowing him, the God who is the cause, to be grasped through Sāṃkhya and Yoga, one is freed from all bonds.” (6.13)

Read that first line slowly, because the grammar is doing theology. “The eternal among eternals, the conscious among the conscious.” The God is in the singular, and the beings he stands among are in the plural. The verse asserts, in plain grammatical form, that there are many eternal conscious selves, and one supreme God among them who fulfils their desires. For the philosopher who believes the individual soul is finally identical with the Absolute, this is an awkward sentence. For the philosopher who believes the soul and God are forever distinct, it is a gift.

This is why the verse became a cornerstone for both of the great theistic, difference affirming schools. Rāmānuja, in his Śrī Bhāṣya, leans on it to establish that conscious souls are real, many, and eternal, each a distinct mode of the one God. Madhva and the Dvaita tradition go further and read it as scripture’s flat declaration of Bheda, eternal difference. Madhva built his whole system on what his followers call the five differences, the eternal distinctions between God and soul, between God and matter, between soul and soul, between soul and matter, and between one piece of matter and another, and the Śvetāśvatara is one of his richest quarries for them. The plurality verse 6.13, the framing of the world as enjoyer and enjoyed and ruler in the first chapter, the two unborn knowing and unknowing selves, all of these the dualist tradition reads as the Upaniṣad saying, in so many words, that you and God are not the same, and never will be.

So here is the irony at the heart of the Śvetāśvatara’s reception, the one specialists know and almost no one else does. This is, on its surface, a Śaiva Upaniṣad. Its supreme God is named Rudra and Hara, the names of Śiva. One scholar has remarked that the Śvetāśvatara served the worship of Śiva much as the Bhagavad Gītā served the worship of Viṣṇu, as its great scriptural charter. And yet the two teachers who quarried it most eagerly for the difference between the soul and God, Rāmānuja and Madhva, were Vaiṣṇavas, worshippers of Viṣṇu, building theologies in which the supreme God is not Śiva at all. An Upaniṣad that names Śiva as the one God became a favorite source of proof texts for the schools devoted to Viṣṇu. The reason is the one we keep returning to. The Śvetāśvatara pitched its deepest claims about God at a level above any single name, and so more than one devotion could come home in it.

One verse, three Gods

If the two birds show the philosophers dividing, there is a single verse that shows the translators dividing, and watching them do so is one of the best ways to understand how alive this Upaniṣad still is.

māyāṃ tu prakṛtiṃ vidyān
māyinaṃ tu maheśvaram
tasyāvayavabhūtais tu
vyāptaṃ sarvam idaṃ jagat

“Know that māyā is prakṛti, nature, and that the wielder of māyā is the Great Lord. This whole world is pervaded by beings that are parts of him.” (4.10)

The word at the center is Māyā. In later Hindu philosophy māyā becomes one of the most loaded terms in the language, usually translated as illusion, the cosmic appearance that makes the one reality look like a world of many separate things. But the word is older than that settled meaning, and in the Śvetāśvatara it is doing something the great translators read in three different ways, each reading shaped by the translator’s own sense of God.

Max Müller, the German born Oxford philologist who edited the Sacred Books of the East in the nineteenth century, renders māyā here as “art” and the Lord as the “maker,” the artist, while admitting in a footnote that the word also carries the sense of illusion, because for the nondual philosophy he favored all the Lord’s making shades into mere appearance. Sarvepalli Radhakrishnan, the philosopher who became independent India’s second president and whose 1953 The Principal Upaniṣads reads the texts as a believing Vedāntin, renders prakṛti as māyā and calls the Lord “the wielder of māyā,” and then insists in his commentary that māyā here is not the illusoriness of the world at all but the Lord’s positive creative power, his śakti, the real energy by which a real God makes a real world. Patrick Olivelle, in the cooler register of modern critical scholarship, translates it “one should recognize the illusory power as primal matter, and the illusionist as the great Lord,” and then notes carefully that māyā here “probably means more a trick or magic than cosmic illusion as in later philosophies that denied the existence of the world.” The scholar Dominic Goodall has made the same point from the philological side, that the māyā of the Śvetāśvatara is closer to a creative, supernatural power than to the later doctrine of an unreal universe.

Sit with that for a moment. Three superb readers, one verse, and three different Gods come out the other side. For Müller, schooled in the nonduality of Śaṅkara, the Lord is an artist whose art shades into illusion. For Radhakrishnan, the believing theist, the Lord is a real creator whose māyā is his genuine power to bring forth a genuine world. For Olivelle and Goodall, the careful philologists, the Lord is something closer to a magician working a marvel, and the heavy later doctrine of an illusory universe is read back into the verse only by those who came afterward. The Sanskrit sits there on the page, the same four lines, while the God who emerges from them changes with the reader. This is the single best illustration of why the Śvetāśvatara repays slow reading. It is an early text, written before the great systems hardened, and its words still have the openness of that early moment.

The verse that opened the door to devotion

For all its philosophy, the Śvetāśvatara is remembered most for one short verse near its close, because that verse is where Indian scripture speaks, with full clarity, of love for God.

yasya deve parā bhaktir
yathā deve tathā gurau
tasyaite kathitā hy arthāḥ
prakāśante mahātmanaḥ

“These truths, declared here, shine forth only in the great soul who has the highest devotion for God, and for the teacher as for God.” (6.23)

The word for that highest devotion is Bhakti, and this is the only occurrence of the word in the principal Upaniṣads. It is widely regarded as the earliest clear Upaniṣadic statement of loving devotion to a personal God, and it ties that devotion, from the very start, to two things the whole later tradition would keep together: love of God, and reverence for the teacher who reveals him, yathā deve tathā gurau, “as toward God, so toward the guru.”

It is hard to overstate how much grows from this small seed. Bhakti, loving devotion to a personal God, becomes the central current of Hindu religious life over the following centuries, the force behind the Tamil saints, the great devotional movements that sweep across the subcontinent, the poetry of surrender in a dozen languages. And here it is, already explicit, tied to the love of God and the reverence for the guru, in the closing verse of the Śvetāśvatara. The Upaniṣad even doubles its final words for emphasis, a small liturgical flourish that marks its end, as if the composer knew he had arrived somewhere important. One honest note belongs here, because the careful reader deserves it: Müller observed long ago that this is the single appearance of the word in the text and may belong to its latest layer, an epilogue. Whether or not that is so, the verse sits in the received scripture and became one of the most quoted lines in all of devotional Hinduism.

There is grace here too. The Śvetāśvatara had already said, in the signature verse, that the sage came to know brahman “through the grace of God,” Devaprasāda, and earlier still, in that recast verse about the divine size, that the seeker beholds the Lord “by the grace of the Ordainer,” Dhātuḥ Prasāda. This is among the earliest clear language of divine grace in the Upaniṣads, the sense that God is not only a truth to be worked out but a person who gives himself to be known. A small scholarly honesty: the line about the grace of the Ordainer is shared with the Kaṭha Upaniṣad, so the Śvetāśvatara is not strictly the first to use it, but the frankly theistic version, the grace of God, the Devaprasāda of the signature verse, is its own. Knowledge and love and grace are braided together in this Upaniṣad, and that braid is the devotional religion of India in miniature.

The Gītā in the distance

One last thread ties the Śvetāśvatara to the most beloved scripture of all, and it is the kind of connection a careful reader loves to find.

Several verses of the Śvetāśvatara stand very close to verses of the Bhagavad Gītā. The clearest case is the great description of the all pervading being whose limbs are everywhere. The Śvetāśvatara has it at 3.16, “with hands and feet on every side, with eyes and heads and faces on every side, with ears on every side, he stands encompassing all,” and the Bhagavad Gītā, in its thirteenth chapter, has the very same verse, nearly word for word. The seated posture of meditation we read above is shared with the Gītā too, and the whole framework of the field and the knower of the field, of the three guṇas, of a personal supreme God who is yet the inner self of all, runs through both.

The relationship is worth stating carefully, because it is easy to get wrong. Both the Śvetāśvatara and the Gītā draw, in this shared verse, on an even older source, a hymn of the Ṛgveda about the god whose eyes and faces are on every side. So the safest way to put it is that both belong to a single emerging theistic stream, the Śvetāśvatara standing at or near its source and the Gītā among its later and fuller flowerings, rather than that one simply copied the other. Olivelle places the Śvetāśvatara just before the formative stage of the Gītā, which fits the picture exactly. When you read the Śvetāśvatara, you are reading the older relative of the Gītā, the place where the personal God of the Gītā is already speaking, in a quieter and more compressed voice.

Why this small Upaniṣad matters so much

Step back from the verses and look at what the Śvetāśvatara holds together in six short chapters.

It has the old Upaniṣadic depth, the question of the cause of the world, the identity of the self and the Absolute, the longing to pass beyond death. It has a personal God, one Lord, creator and ruler, hidden in every heart, who can be known and loved. It has the philosophy of its age, the Sāṃkhya analysis of nature and the discipline of Yoga, taken up and set under that God. It has the oldest seated meditation in the Upaniṣads. It has the word māyā poised at the exact moment before it hardened into a single meaning, still open enough that three later theologies could each claim it. It has the two birds, the image on which Śaṅkara, Rāmānuja, and Madhva would build three different worlds. It has the plurality verse that the dualists made their banner, and the all genders verse that the devotional heart never tires of, and the paradox of a God who grasps without hands. And it has Bhakti, the love of God and the reverence for the teacher, named in scripture for what may be the first time. Almost the entire future of Hindu religious thought is present here, alive and whole.

That is why an Upaniṣad this short has cast a shadow this long. The later traditions found their God already speaking in the Śvetāśvatara, and spent the next two thousand years drawing out what he had said. The personal God of the bhakti saints, the theistic Vedānta of Rāmānuja and Madhva, the meditative practice of the yogins, the Śaiva and Vaiṣṇava theologies that shaped the devotional life of a civilization, all of them keep their roots in this one slim, signed, much quoted Upaniṣad. It is the common ancestor, the place where the personal God of Indian devotion stands before the great traditions divided the territory among themselves.

It opened, you remember, with a seminar, a circle of seekers asking where the world comes from and at whose command they live in joy and in pain. By the end it has answered its own question, and the answer is a person. One God, hidden in all beings, the witness in the heart, to be reached by knowledge and by discipline and, in the end, by love. The men in that opening circle wanted to know the cause of things. The Śvetāśvatara told them the cause has a face, and that the face is turned toward them.

Sources and further reading

• Patrick Olivelle, The Early Upaniṣads: Annotated Text and Translation (Oxford, 1998).

• S. Radhakrishnan, The Principal Upaniṣads (1953).

• F. Max Müller, The Upanishads, Part II, Sacred Books of the East volume 15 (1884).

• Signe Cohen, Text and Authority in the Older Upaniṣads (Brill, 2008).

• Swami Gambhirananda, Eight Upaniṣads with the Commentary of Śaṅkarācārya (Advaita Ashrama).

• Ivan Andrijanić, “Śaṅkara and the Authorship of Śvetāśvataropaniṣad Bhāṣya,” The Journal of Hindu Studies 12.3 (2018).

Picture the scene. It is around 150 BCE, on the hot northwestern edge of the Indian world, and a Greek king is looking for a fight. Not with armies, he has plenty of those, but with words. His name is Menander, he rules Bactria and the Punjab, and the Buddhist texts that remember him call him Milinda. They do not remember him as a pushover. He is, they say, hard to equal in argument and harder still to beat, the acknowledged better of every school’s founder he has ever debated. He collects sages the way other kings collect elephants, and he has worn most of them out.

Then he meets a monk named Nāgasena, and he asks the question he has presumably been asking everyone. Buddhism teaches that there is no soul. Fine. So when you die and are reborn, what exactly is reborn? If nothing permanent crosses over, in what sense is the next life yours? And here is the sting in the tail: why should anyone carry the karmic bill for what some vanished stranger did?

It is a genuinely good question, the kind that does not go away when you wave your hand at it. And the Buddhist tradition, to its enormous credit, did not wave its hand. It sat with the problem for centuries and came back with two answers. The first is older and more orderly. The second is from Nāgārjuna, and it is the stranger of the two by a wide margin. I want to give both of them their fair hearing, and, in passing, say that I side with neither.

Why it is a hard question

Start with what is actually being denied, because the whole drama depends on it.

The picture Buddhism throws out is the one the Bhagavad Gītā had painted so beautifully just up the road. There, Krishna comforts a panicking Arjuna by telling him the self cannot really be killed. It simply changes bodies the way you change clothes. In Gita, the self, “after casting away worn out garments,” picks up fresh ones, and so it slips from a worn body into a new one and carries on. The bodies are disposable. The self is the thing that lasts. That self is the soul doing precisely what a soul is for: stringing your lives together like beads on a single thread, so that the one who acts and the one who pays are obviously the same.

Buddhism pulls out the thread. There is no bead holder. No ātman, no little permanent you sitting behind your eyes and owning the show.

And the moment you remove it, Menander’s question turns sharp. With a soul, rebirth was easy and so was fairness: the same self that acted is the one who later reaps. Take the self away and you have a real mess on your hands. If nobody owns my karma, what stops me from waking up next time stuck with a stranger’s debts, or a stranger from waking up with mine?

You might think the secular modern move is to keep Buddhism’s kindness and quietly drop the rebirth part, treating it as a poetic metaphor. German philosopher and Madhyamaka scholar Jan Westerhoff points out a snag here. The whole goal of the Buddhist path is nirvāṇa, the final end of craving and suffering, the candle of the restless self finally blown out. Now suppose there is no rebirth, and death is simply the lights going out for good. Then death already delivers that final end, automatically, to absolutely everyone. The saint who meditated for forty years and the man who never gave it a thought get the exact same total extinction the moment they die. If that were true, the entire long effort of the path would buy you nothing you were not going to get for free anyway. That is why the tradition never took the cheap exit. It held on to both rebirth and no soul at once, and rolled up its sleeves to make sense of the pair.

The first answer: the Abhidharma

The first answer comes from the Abhidharma. A quick word on what that is, because the name will keep coming up. The early Buddhist scriptures were gathered into three great collections, traditionally called the “three baskets.” One basket holds the rules for monks. One holds the Buddha’s discourses, his teachings more or less as he spoke them, full of stories and parables. The third basket is the Abhidharma, the analytical one. It is the tradition’s attempt to take those loose, story shaped teachings and reorganize them into a rigorous system, breaking the whole blur of experience down into its smallest moving parts and charting how each part hands off to the next. If the discourses are the Buddha talking, the Abhidharma is the tradition doing the homework.

So what are you, on this view? Not a soul, but a pudgala. The word just means “person,” but the Abhidharma loads it with a particular claim: it names you as a process rather than a thing. What truly exists, the Abhidharma says, are countless fleeting events called dharmas. A dharma is the smallest real flicker of experience: a single moment of seeing, a flash of warmth, one passing impulse of the mind. We bundle these flickers, out of habit and convenience, into five “heaps” called skandhas, one heap of physical stuff and four of mind. Nobody owns the bundle. Nobody stands on top of it. The pieces arise, do their bit, and pass, each one nudging the next into being.

Holding it all up is a single stubborn principle: the parts are real, the whole is just a handy name. This is the chariot Nāgasena famously parks in front of King Milinda. Where, he asks, is the “chariot”? Is it the pole? The axle? The wheels, the frame, the ropes, the yoke, the spokes, the goad? None of those is the chariot, and there is no extra “chariot” floating above them. “Chariot” is simply the word we reach for when those parts are arranged chariot wise. And you, says Nāgasena, are exactly the same. Real parts, doing their thing, and over them no owner, only a name.

It is worth slowing down on what these parts actually are, because the cartoon version gets it wrong. It is tempting to imagine the Abhidharma swapping out the soul for a swarm of tiny mind atoms, little solid beads rattling along in a line. The scholar Noa Ronkin makes the case that the early Pali material is better read as a world of processes and events than of little substances. The person, she writes, is “but the sum total of the processes” that make her up, a string of momentary happenings with nothing solid underneath. The parts, in other words, are more like verbs than nouns. Dig down as far as you like and you never reach solid ground, only more happening.

Which lands us right back on the hard question. If you are just a stream of ownerless flickers, what makes the stream a single stream, and what keeps your karma stuck to you? Since Menander is a king and not a graduate student, Nāgasena answers in pictures.

First, a flame. A lamp burns all night. Is the flame at dawn the same flame that was lit at dusk? No. A totally different flame? Also no. It is one continuous burning, “neither the same nor another,” in the lovely Pali phrase na ca so na ca añño. That, says Nāgasena, is you, down the length of a life and across the gap of death. Not one frozen thing hanging on, and not a parade of unrelated strangers, but a single fire passing itself forward.

Second, and this one answers the fairness worry head on, a man lights a fire in the cold season to warm his hands, then strolls off and leaves it crackling. It jumps to a neighbour’s field and burns it down. Dragged in front of the magistrate, he pleads that the fire that ate the field was not his fire, the cozy little one he lit. Nāgasena is not having it. The chain of cause runs unbroken from his match to the ruined crop, and that chain is enough to pin it on him. He runs the same play with a mango thief who cannot disown the ripe fruit grown from the pit he stole, and with milk thickening into curds and then butter, continuous through every change without being the same at any two moments.

There is the first answer, whole. No soul travels. What ties your lives together is a rope of causes, fire lighting fire, milk turning to curd, each instant handing itself to the next. And here is the part to hold on to: for the Abhidharma that rope is really there. It is the bedrock under everything. In principle you could follow it knot by knot.

The later technical literature spells the machinery out with almost alarming precision. In the Paṭṭhāna, the Abhidharma’s great manual of conditions, continuity runs on what are called the proximity conditions: each moment of consciousness, as it dies, cues up the very next “without any pause.” Ronkin notes these relations “were probably necessary in order to account for the continuity of phenomena without relying on any metaphysical substance.” The same seamless handoff connects the last flicker of a dying mind to the first flicker of the next life. Rebirth is not a soul catching the next bus. It is simply the final moment of one life cueing the first moment of the next, the same way one thought cues the next over breakfast. And karma is an act of will that has long since ceased but left traces in the stream, traces that ripen when the conditions are right, sometimes lifetimes later. No self is keeping the account. The unbroken chain of causes does everything a soul used to do.

It is a powerful solution. It dissolves the self and still explains why you wake up tomorrow as the same person, and why your actions keep following you from one life to the next. Just keep one thing in view, because the second answer goes straight for it: the rope is real.

Nāgārjuna

Before the second answer, a word about the man, because he is no minor figure. Nāgārjuna lived around the second century CE, likely in South India(modern day Andhra Pradesh or Telangana), and he is a strong candidate for the most important Buddhist thinker after the Buddha himself. He founded the Madhyamaka, the “middle way” school, and whole traditions of later Buddhism across Tibet, China, and beyond trace their philosophy back through him. His central idea, the one everything else hangs from, is śūnyatā, usually translated “emptiness.” It is the claim that nothing whatsoever exists on its own, in and of itself. Everything is what it is only by leaning on other things, all the way down. He argues this with a patient, almost dizzying logic that pulls the rug out from under one comfortable assumption after another, and now it is rebirth’s turn.

The second answer: emptiness

In a short verse text traditionally pinned to him, the Pratītyasamutpādahṛdaya, the “Heart of Dependent Origination,” Nāgārjuna reaches for a handful of everyday images, and you have to watch them closely because each is a small trap. A teacher recites a verse and the student learns it. A seal presses into soft wax and lifts away, leaving its crest behind. You hear the word for a sour fruit and your mouth floods.

Look hard at any one of them. Something real plainly happens, an effect genuinely shows up, and yet nothing actually makes the trip across. The teacher does not lose the verse when the student gains it; no little parcel of poetry leaps from skull to skull. The verse just arises again, fresh, in the student. No metal crosses from the seal into the wax; only the shape turns up there. No sourness travels from the word onto your tongue; there is only a sound in the air, and still the saliva comes. Cause, then effect, reliably, with nobody and nothing in the seat between them.

Nāgārjuna takes this all the way to the edge. The verses say it without blinking: the five skandhas, the very stuff the person was made of, do not travel onward. Not so much as an atom passes from this life to the next. The next life is genuinely other than this one. What carries over, to steal Westerhoff’s neat modern phrase, is closer to information than to any thing or substance, a pattern that repeats rather than a passenger that rides.

On the surface this can read like the Abhidharma with the dial cranked up. Both deny the soul. Both lean on lamps and seeds and continuity without identity. But here is the one spot where they split, and it is a chasm.

For the Abhidharma, the rope of causes between your lives is really there, a true thread you could trace knot by knot. Nāgārjuna looks at that rope and says it has no svabhāva. The word is the hinge of his whole philosophy, so it is worth getting right. Svabhāva means “own being” or “self nature”: the property of existing in your own right, from your own side, without leaning on anything else to be what you are. A thing with svabhāva would be self contained, standing on its own feet, needing nobody’s permission to exist. Nāgārjuna’s signature move is to say that nothing at all has it. The rope of causes is real only the way “a river” is real: a useful name for water that is never the same twice, with no “river itself” tucked underneath. The Abhidharma had already dissolved the self into a stream of causes. Nāgārjuna turns around and dissolves the stream.

Sit with how far that reaches. What is left is rebirth as a pattern that keeps reassembling, with no soul to carry it and no solid rope beneath it either. Just one moment leaning on the next, dependence all the way down, and no floor at the bottom of the staircase.

Where they actually disagree

It would be easy to leave it there, as a tidy face off: Abhidharma keeps the rope, Nāgārjuna burns it. That is roughly right, and it is the version I have told before. But there is a sharper truth underneath, and it actually makes the quarrel more interesting rather than less.

The trick is that svabhāva, “own nature,” did not mean the same thing to everybody. Nāgārjuna’s real target is narrower than “the Abhidharma” in general.

One northern school, the Sarvāstivāda, used the word at full strength. To have a svabhāva, for them, was to be a dravya, a genuine substance, a real little thing with its own intrinsic nature, so real that the Sarvāstivādins were willing to say such things exist in the past and future as well as the present. That is a properly solid, substance heavy picture, and it is exactly the thing Nāgārjuna spends his life taking apart. When he says nothing has svabhāva, this beefed up, stands on its own kind of dharma is who he is talking to.

The southern Theravāda, whose canon we have been quoting all along, meant something much humbler by the word. There, sabhāva just tells you what a given dharma is, what makes it different from the others, without claiming it exists independently from its own side. In Ronkin’s crisp formulation, it “determines what a dhamma is and not that a dhamma is.” They allowed only fleeting present dharmas and politely declined to turn a dharma into a substance.

Why should you care? Because it tells you what the fight is really over. It is not quite “the Abhidharma believes in a real thread and Nāgārjuna says it is empty,” full stop. It is that the substance loving wing of the Abhidharma had handed the building blocks of the causal stream their own private reality, and Nāgārjuna’s emptiness is the claim that nothing gets that, not the self, not the parts, not even the causal link tying them together. No exemptions. You borrow your being from whatever holds you up, and that thing borrows from whatever holds it up, and so on forever, with no one in the chain ever actually owning the stuff. Dependence is the whole of the story.

What neither answer settles

The honest thing to admit is that neither answer ties off cleanly, and the tradition knew it perfectly well.

Lean on the Abhidharma and you can ask whether its real, karma toting causal stream is just the banished soul wearing a fake moustache. If something hangs together firmly enough to keep my karma reliably mine and to convict the field burner, how far is that, really, from a thin little self slipped back in through the side door under a more respectable name?

Lean on Nāgārjuna and the worry flips. If the link between my lives is only a convention with no nature of its own, what keeps karma lawful instead of random? Why does the pattern keep snapping back into focus as me, with my debts attached, instead of fraying into static or reassembling as somebody else? “Dependence all the way down” is dizzying, and the dizziness is precisely the objection.

Eighteen centuries later, neither side has blinked. That is half the reason the debate is worth your evening. This is not a museum piece behind glass. It is two of the most serious attempts anyone has ever made to explain how a person can keep going with nobody home at the center, and the two are still arguing.

Where it leaves us

Two answers, one question, and the same refusal at the root of both. Each denies the soul. The entire distance between them comes down to a single thing: how real is the thread between one life and the next. The Abhidharma keeps the thread, and builds a gorgeous clockwork of momentary causes to carry you over the gap. Nāgārjuna says the thread itself is empty, that there is leaning and nothing underneath the leaning, and that somehow, even so, you still open your eyes on the far side.

For what it is worth, my own sympathies lie with neither, but with the Viśiṣṭādvaita of Rāmānuja, which keeps the self real and eternal. That is an argument for another day, and these two have earned a hearing on their own terms.

A Greek king asked the question on a dusty frontier two thousand years ago and could not get a straight answer he was happy with. The fact that we are still chewing on it is the surest sign he asked a very good one.

Sources and further reading

The dialogue itself. The Milindapañha, translated by T. W. Rhys Davids as The Questions of King Milinda (Sacred Books of the East, 1890). This is where the flame, the burning field, the mango, the milk, and the chariot all come from. Free to read at the Internet Sacred Text Archive.

Nāgārjuna’s answer. The Pratītyasamutpādahṛdaya, the “Heart of Dependent Origination,” with the seal, teacher, and sour word similes. Studied by Carmen Dragonetti (1978), who notes the attribution to Nāgārjuna is traditional rather than certain, though Candrakīrti quotes the key verse as his.

The Abhidharma, explained well. Noa Ronkin, Early Buddhist Metaphysics (Routledge, 2005). The source for the “process not substance” reading and the crucial Sarvāstivāda versus Theravāda split over svabhāva.

The two answers, framed. Jan Westerhoff, Nāgārjuna’s Madhyamaka (Oxford, 2009), and his lecture “The Place of Rebirth in Buddhist Philosophy,” which is also where the argument that you cannot simply drop rebirth comes from.

The Gītā line. The “worn out garments” verse is Bhagavad Gītā 2.22, here in Winthrop Sargeant’s The Bhagavad Gītā (SUNY Press).

The Scythians should not have a religion we can read.

They built no temples. Had no scripture. They raised one kind of idol only, and it was a Sword. They had no priestly class that wrote anything down, no creation hymn we can recite, no name for their own high god in their own preserved words. By every rule that lets us read a dead faith, theirs should be gone, scattered with the ash of their funeral fires across four thousand miles of Eurasian grass.

And yet we can name their gods. We can sort them. We can stand each one in a lineup and say, with reasonable confidence, this one is a cousin of the gods in the Veda, this one is a borrowed foreigner in a fresh coat of paint, this one is older than the Iranian language itself, and this one is just a Greek name thrown like a blanket over something we cannot quite see underneath.

I want to walk through the Scythian pantheon god by god, and for each one ask the only question that really matters once you have met them: where did you come from? Are you Indian, in the sense of sharing a grandparent with Agni and the waters of the Ṛgveda? Are you Iranian, a sibling of the gods Darius prayed to? Are you a deep Indo-European inheritance, as old as the word for “sky”? Are you a guest from Babylon and the Levant wearing steppe clothes? Or are you something older than all of them, something that was already haunting Siberia before a single Iranian word was spoken there?

I covered the Scythians as a people in an earlier piece, Children of the Golden Sword: who they were, how they fought, how they buried their kings, how they ended up in India and dissolved into it. That one had a chapter on the religion. This one is the chapter that grew into a substack piece.

How can we know any of this?

Three witnesses let us back into a religion with no books.

The first is a Greek tourist. Around 450 BCE a man from Halicarnassus named Herodotus traveled to the Greek colonies on the north shore of the Black Sea, talked to people who dealt with Scythians every day, and wrote it all down in Book IV of his Histories. He is an outsider who never names a Scythian god in its own language without immediately translating it into a Greek one, which is a little like trying to learn Hindu theology from a visitor who insists on calling Viṣṇu “the Greek Apollo” and Śiva “their version of Dionysus.” Useful, maddening, and all we have.

The second witness is the ground itself. For two centuries archaeologists have been opening the kurgans, the great burial mounds of the steppe, and in a few miraculous cases the Siberian permafrost handed back tombs frozen whole, with the bodies, the felt, the horses, and the drug kits still inside. Again and again the excavations have confirmed the exact details that made scholars dismiss Herodotus as a fabulist. I told that story in the last piece and will not repeat it. For our purposes here the ground does one crucial thing: it proves the religion was real and practiced, not a Greek fantasy.

The third witness is the strangest and the most beautiful. High in the Caucasus mountains lives a people called the Ossetians, around two and a half million of them, who still speak an Iranian language that descends in a direct line from Scythian. Their national epic, the Nart sagas, preserves under a thin Christian glaze a pre-Christian world of gods and heroes that keeps lining up, point for point, with what Herodotus saw twenty-five centuries ago. They are as close to living Scythians as the earth still holds, and at the end of this piece they will hand us our last god, still warm.

One honest warning. Almost everything that follows is reconstruction. We are triangulating a faith from a foreign tourist, a pile of grave goods, and a poem recited three thousand miles and two and a half millennia downstream. I will flag what is solid, what is a strong scholarly bet, and what is somebody’s beautiful guess.

The creation, in brief

Before the pantheon there is a family tree, and it is worth one fast pass because three of our gods are born inside it.

The Scythians told it like this. The first man was Targitaos, child of the sky god and a daughter of the river. He had three sons. While they ruled, four objects of gold fell burning from the sky: a plough with its yoke, a battle-axe, and a cup. When the two elder brothers reached for the gold it blazed up and drove them back. Only when the youngest came near did the fire lie down and let itself be lifted, and so the youngest became king, and the royal line of the Scythians descended from him.

I told this story in full last time, including the Greek rewrite in which Herakles fathers the line on a snake legged woman, and the long scholarly fight over whether the three sons stand for the three classes of society. If you set the two family trees side by side, do not read the six names as a single lineage; they are rival casts for one plot. The Scythians’ own version makes the three brothers sons of Targitaos (Arpoxais, Lipoxais, and Kolaxais), while the Greek version gives Herakles and the snake legged queen a wholly different three (Agathyrsos, Gelonos, and Skythes). Only the shape carries across, the youngest son always taking the crown, never the names. Rather than talk about it again, here is only the part this piece needs: the receipts in the names.

The three brothers are not random. Read through the Iranian, each is a “Lord of X,” built on the same royal word, -xšaya, “king” or “ruler,” the very word that sits inside Persian shah. Each son rules one zone of the world:

• Kolaxais, the youngest, the Sun-King

• Lipoxais, the eldest, the Lord of the mountain and the earth

• Arpoxais, the middle son, the Lord of the watery depths

Sky fire, mountain, and water: a small cosmos divided among three brothers.

And the burning gold that only the rightful king can hold has a precise Iranian name too. It is xvarǝnah, the divine radiance of legitimate kingship, the same glowing fortune that haloes kings in the Avesta, and the same word that survives in the Caucasus today as Ossetic farn, “grace” or “blessing.” A Scythian king was the man the fire did not burn. That idea is pure inherited Indo-Iranian royal ideology, and we will watch it survive to the end of the piece.

None of which means the Scythians swallowed it as innocent folklore. Bruce Lincoln, who reads origin myths as weapons rather than fairy tales, calls this one “the story Scythian kings and their apologists liked to tell,” a charter for the right to rule dressed up as a tale of falling gold.

There is one more receipt, and it points straight at India. The gifts that fell burning from the sky, the plough with its yoke, the axe, and the cup, are not random treasure. The scholar Mikhail Bukharin has argued that they are the regalia of a coronation, the same kit an Indian king handled at the rājasūya, the Vedic rite that turned a man into a king. The plough and yoke stood for the land he would break and bound; the cup held the anointing waters poured over him; the weapon marked his martial power. By taking up all of them at once, the king carried every function of society in his own body, which is exactly the claim the Scythian myth makes when only the youngest son can lift the gold. Bukharin even reaches for a particular Indian king, Pṛthu, the first sovereign ever consecrated, who took a wild and barren land, taught his people the plough, and made the earth bear fruit, the same role the Scythian heir plays when he alone can hold the fire. This is one scholar’s reconstruction and it runs out ahead of the evidence, so treat it as a strong parallel rather than a proven link. But it is the kind of parallel that does not happen twice by accident. The steppe and the Ganges were doing the same thing with the same objects, because they had inherited the same idea of what makes a king.

Now the gods proper. Herodotus gives us the whole pantheon in a single passage, Book IV chapter 59, naming seven gods worshipped by every Scythian and an eighth reserved for the royal clan. The great Cambridge scholar Ellis Minns, whose 1913 Scythians and Greeks is still a foundation stone of the field, laid it out in one long breath:

The following deities were common to all, Tabiti, Hestia, who was the principal object of their veneration, next to her Papaeus, Zeus, with Apia, Ge, husband and wife, after them Goetosyrus, Apollo, Argimpasa, Aphrodite Urania, and Ares. Thamimasadas, Poseidon, was peculiar to the Royal Scyths. They raised no statues, altars or temples to their gods, save to Ares alone.

Eight gods, one altar, no idols but a sword. We will take them in Herodotus’s order, because the order itself tells a story: he starts at the top, and the top is not where you would expect.

1. Tabiti, the fire-queen

Here is the first surprise, and it is a big one. The chief deity of the Scythians, the one Herodotus says they honored above all others, the one whose name a Scythian king swore by when he wanted an oath that could not be broken, was not the thundering sky father and not the blood fed war-god. It was a goddess, and her domain was fire on the hearth.

Her name is Tabiti, and the name is a gift, because it is transparent. Strip the Greek spelling and you get something close to Tapatī, which Martin West, the Oxford philologist who mapped this entire family of gods, glosses as “the Burning one,” from an ancient root for heat and flame that runs straight through the family. The same root surfaces all over the Indian and Iranian world:

• Sanskrit tapati, “it burns” or “it glows”

• tapas, the inner heat of ascetic and ritual power that any reader of Indian religion knows

• the Vedic fire-god Agni and the Iranian sacred fire Ātar, her brothers by another name

She is family. When a Scythian banked the coals of the royal hearth and called them holy, he was tending the same divine fire that an Indian priest fed with ghee a thousand miles to the southeast, the two of them performing a rite they had both inherited from a common ancestor neither remembered.

A small caution, because I promised to keep the seams visible: the words being cousins does not prove the Scythians thought about fire exactly the way the Vedic tradition thought about tapas. Shared root, not shared theology. But the shared root is real, and it is the firmest etymology in the whole pantheon.

Then there is the puzzle that makes Tabiti genuinely strange. Across the Indo-European world the fire-god is usually male. Agni is male. Ātar is male. But the Scythian fire is a queen, and so is the Roman hearth-fire, Vesta, and so is her Greek mask here, Hestia. Something in the deep tradition let the hearth-fire be imagined as a woman, the still center of the home and the realm, while the sacrificial fire stayed a man. The Scythians, a people we picture as nothing but galloping men with bows, put a goddess of the hearth at the very top of heaven. Hold onto her. Of all eight gods in this lineup, she is the one who will still be alive at the end.

2. Papaios, literally “Papa”

After the fire-queen comes the sky-father, and his name is almost too charming to be true. The Scythians called their Zeus Papaios, and Herodotus, who usually just translates a name and moves on, stops to give this one his personal endorsement. “Zeus,” he writes, “in my judgment most rightly so called, Papaeus.” The reason he approves is that Papaios is, near enough, the word “Papa.” It is the baby’s word for father, the reduplicated pa-pa that toddlers babble in language after language, promoted all the way up to the name of the highest god of the sky.

And that babble is ancient. The Indo-European sky-god was conceived as a father from the beginning, and his oldest title is exactly “Sky Father.” You can watch it surface across the whole family like the same melody played on different instruments:

• Greek Zeus patēr

• Latin Jupiter, from Dyeu-piter

• Vedic Dyaus Pitar, the Sky Father of the Ṛgveda

Papaios is that god. He is the Scythian Dyaus, the Scythian Jupiter, the bright father of the daylight sky, and his folksy nursery name is not a downgrade but a survival, the oldest and warmest thing you can call a god. The supreme being of the steppe was, to the people who feared him, just Dad.

In the creation story he is the sky half of the first divine couple, the husband of the earth. Which brings us to his wife.

3. Api, the earth and the waters

Papaios needs a partner, and the partnership itself is the most predictable thing in any Indo-European pantheon: Sky marries Earth, the bright father above lies with the broad mother below, and from their union the world is furnished. The Vedic poets sing of Dyaus and Pṛthivī, Heaven and Earth, as a married pair so constantly that they often fuse into a single dual word. The Scythians have the same marriage. Papaios the sky takes Api the earth, whom Herodotus matched to the Greek earth-goddess Gē, and the royal line grows from the seam where they meet.

Api’s job is clear and inherited. Her name is the knot, and the scholars split two ways on it:

• The old Soviet reading, still the popular one, takes Api from the Iranian āp-, “water,” making her the Earth-and-Waters and tying her straight to the Vedic Āpaḥ, “the Waters,” the divine female waters of the Ṛgveda. If that is right, she is a sister of the Indian sacred rivers.

• Martin West, who wrote the standard book on this material, doubts the water etymology and reads her name as something like “the Kindly One,” a respectful nickname for the earth rather than a label meaning Waters. He even resists calling her “Mother.”

So the verdict on Api comes in two layers. The structure, Sky-father married to Earth-mother, is rock-solid inherited Indo-European and Indo-Iranian, no argument. The name, and whether she is specifically a water-goddess and therefore a near-twin of the Vedic Āpaḥ, is genuinely unsettled. I lean toward the watery reading because it fits her partner and her people, but I am telling you it is a lean, not a proof.

4. Goitosyros, the cattle-lord sun

Herodotus calls him Apollo. His name says he is Mithra in disguise.

Goitosyros, sometimes written Oitosyros, is the god you would expect to know least, because he left us no myth, no recorded rite, no slab of art we can point to and say there he is. He is a name and a Greek equation, and for a long time that made him the dull cousin of the pantheon. Then you crack the name, and the dull cousin turns out to be one of the most important gods in the Iranian world wearing a false beard.

Read through the Iranian, Goitosyros resolves to something very close to Gaiϑā-sūra-, “the Mighty Lord of the Pastures,” or more bluntly, “Strong-in-Cattle.” Break the name in two:

• gaiϑā means herds, grazing land, and possessions, the wealth of a people who counted their fortune in livestock

• sūra- means “mighty,” the same word that turns up as the Avestan epithet Sūrā on the great goddess and as Vedic śūra-, “hero”

And here is the catch that wakes the whole section up. "Mighty Lord of the Pastures" is not a generic phrase. It is a known title of Mithra, the towering Iranian god of the sun, the sworn oath, and the covenant, who carries in the Avesta the standing epithet Vouru-gaoyaoiti-, "of the Wide Pastures." The Scythian sun-god's name is, letter for letter, a Mithra epithet.

So the obscure one is not obscure at all in origin. He is the steppe face of Mithra, the cattle-herder's sun, the god the common Scythian who lived off his animals would have prayed to most, dressed up by Herodotus as Greek Apollo because Apollo was the Greek sun. We genuinely do not know his myths, and I will not invent them. But we know exactly where he comes from, and it is a long way from "we barely know him."

5. Artimpasa, the goddess in borrowed clothes

If Tabiti is the surprise at the top, Artimpasa, also written Argimpasa, whom Herodotus matched to Aphrodite Ourania, the Heavenly Aphrodite, is the puzzle at the heart. She is where this religion stops looking tidy and starts looking like a real place where caravans and conquests and old local spirits all left fingerprints on the same statue. She is the great goddess, and she holds an improbable spread of powers at once:

• love and fertility

• war and sovereignty

• the wild animals, as Mistress of Beasts

Three or four goddesses in one, which is exactly the clue to her history, because she did not start that way. She accumulated. And you can scratch down through her like coats of paint.

Start at the core. Underneath everything, Artimpasa is an Iranian goddess of the same family as Arti or Aši, the Iranian personification of fortune and reward, and she shades into the great Iranian water and fertility goddess Anāhitā, whom the Persian kings worshipped. That is her bone structure, and it is solidly Indo-Iranian. So far she is a respectable steppe relative of the Persian goddesses.

Now the borrowed coat. The Scythians of the western steppe lived next to the Greek colonies of the Black Sea, and through those colonies ran the long tendrils of the Near Eastern goddess: the armed, lion riding, sex and war goddess of the whole ancient world, called Ishtar in Babylon, Astarte by the Phoenicians, Cybele in Anatolia. Herodotus even hands us the moment of contact, and it reads like a curse out of a folktale. Generations before his own day, he says, a band of Scythians riding home from a raid into the Levant stopped at Ascalon and stripped the temple of Heavenly Aphrodite, “the most ancient,” he notes, “of all the temples of this goddess.” The goddess did not forget. She struck the robbers, and their sons, and their sons’ sons, with what Herodotus calls the “female sickness,” and the affliction ran in the blood for centuries. That temple at Ascalon was a temple of Astarte.

So in Herodotus’s own telling, the Scythian great goddess and the Levantine Ishtar-Astarte have already met, already fused, already exchanged a curse. The scholar Ustinova, who traced this overlay in detail, is admirably honest about how tangled it gets. By the time we see Artimpasa clearly, she says, it is “impossible to determine” whether a given feature of hers is Iranian, Levantine, or Greek. She is all three at once, which is what happens to a goddess who lives on a trade route.

Two borrowings shaped her most, and they are worth naming, because they explain how one goddess could hold so many jobs. From the Iranian side came Anāhitā, whose full Avestan title, Arəduuī Sūrā Anāhitā, names her three times over as the Humid, the Strong, and the Immaculate. Those three epithets are not decoration; read with Ustinova they line up with the three powers Artimpasa already carried, fertility, sovereignty, and priestly force, the whole of society folded into one figure. Anāhitā was also a patron of kings, and as she flowed into Artimpasa the Scythian goddess set hard into exactly that, the divine sponsor of the royal house who grants a man his right to rule.

From the Levantine side came something stranger, and it solves a puzzle the piece has been circling. At Ascalon the great Astarte did not stand alone; beside her was a lesser, half-bestial goddess, Atargatis, a Great Goddess shadowed by a smaller, semi-human one. Ustinova’s argument is that this pairing is the template the Scythians carried home, and that their own snake-legged foremother, human above and serpent below, slid neatly into the role of the lesser goddess attached to the greater. It lets you hold two facts at once that otherwise fight: Artimpasa and the snake-woman are almost the same figure, and yet not. Artimpasa grants the king his power and shares his bed, but she is never called the mother of the Scythians. The serpent-woman is. The borrowing shows in the seam between them.

And under even that, there is a floor. We will come back to it in the origins section, but flag it now: beneath the Iranian goddess and her Near Eastern coat lies something older still, a Mistress of the Animals descended from the deer mother of Inner Asia, the spirit who ruled the beasts long before any Iranian rode west.

She also kept the most unsettling priesthood in the ancient world, and they deserve their own section.

The Enarei, the un-men

Artimpasa was served by a class of diviners called the Enarei, and the Greeks could not stop staring at them. They were born male but lived as women, dressed as women, spoke and worked as women, and were said to have received both their changed nature and their gift of prophecy from the goddess herself. Herodotus, ever the one to reach for an embarrassing physical cause, says the goddess afflicted them with that same “female sickness” from the Ascalon curse, passing it down the generations as a kind of divine hereditary condition.

They told the future by their own peculiar method, cutting strips of lime-tree bark and braiding and unbraiding them through their fingers while they spoke the answer. The ordinary Scythian diviners read willow wands; the Enarei alone worked the linden bark, a craft the goddess was said to have handed them along with their changed bodies.

The name itself is the giveaway. Enarei comes from an Iranian phrase, a-nar-ya-, “un-manly” or “not-men.” And whatever Herodotus’s prudish medical theory, what the Enarei actually look like to a historian of religion is a recognizable type: the gender-crossing shaman-priest, the sacred third figure who is neither and both, found from the eunuch galli of Cybele in Anatolia to shamans across Siberia. They are one of the very few clearly documented cases anywhere in the ancient world of a third-gender priesthood holding real religious authority, and they belonged to the goddess of the borrowed coats, which fits, because she herself was a being who would not stay inside one category. Their very name, the un-men, will surface one last time at the end of this piece, fossilized inside a language still spoken in the mountains.

6. The Sword, the god you could hold

The Scythians raised no statues and built no temples, with exactly one exception, and the exception is the most Scythian thing imaginable. Their war-god, whom Herodotus calls Ares, had no human image at all. He was a weapon. In each district the Scythians piled up an enormous mound of brushwood and drove into its summit a single ancient iron sword. Herodotus saw what that blade was for:

On this pile there is set for each people an ancient scimitar of iron, which is their image of Ares; to this scimitar they bring yearly sacrifice of sheep and goats and horses.

He goes on, and it gets much darker, the wine poured over the heads of human captives and their throats cut over the iron. I staged that full rite in Children of the Golden Sword and will not run the blood twice here. Remember only Minns’s dry verdict on the whole faith: they built nothing for their gods, “save to Ares alone.” Every other Scythian god was an idea. This one was an object you could pick up in your hand, and feed. What matters for the sorting is where this strange god comes from, and the answer has two parts welded together:

• The deified weapon is an inherited Indo-European pattern, not a Scythian eccentricity. The idea that a divine power lives inside a particular blade, that a sword can be a god or hold a god, runs through the family wherever we can see it, and at the very end of this piece it will give us the most famous sword in European legend.

• The war-god inside it is the specific Indo-Iranian smasher of resistance, the one the Avesta calls Vǝrǝθraγna (Verethragna), whose Ossetian descendant we are about to meet swinging a blade of tempered steel.

Two ancient inheritances, fused into one blade, and the only god on this whole list that could draw blood by itself.

7. Herakles, who is really Targitaos

Herodotus lists a Scythian Herakles among the gods, and this one we can unmask quickly, because we have already met the face under the mask. The Greeks loved to find Herakles everywhere, and on the Black Sea they cheerfully made him the father of the Scythian royal line through the snake-legged woman of their preferred origin story. But the Scythians’ own first ancestor, the one who actually sat at the root of their family tree, was Targitaos, whose name reads in Iranian as something like darga-tavah-, “long-lasting strength” or “enduring power.” A good name for a progenitor.

Targitaos belongs to a recognizable Indo-Iranian type: the first man and culture-founder, the monster slaying ancestor from whom the people descend, cousin to figures like the Iranian Hušang and the dragon slayer Θraētaona of the Avesta. The Greeks looked at this powerful primordial strongman and saw their own club-swinging hero, so they called him Herakles and moved on. Peel the mask and he is the steppe’s own founding father, the human root of the genealogical myth we ran through earlier.

8. Thagimasadas, the royal outlier

The first seven gods belonged to all Scythians. The eighth belonged only to the Royal Scythians, the dominant tribe, and to no one else, which already makes him interesting: a god you worshipped to show which clan you were. Herodotus calls him Thagimasadas and equates him with Poseidon, god of the sea and, crucially for a horse people, god of horses. A landlocked nation of riders had no use for a sea-god, but Poseidon was also Hippios, Poseidon of the Horses, lord of the freshwater springs and the galloping herds, and that is the side of him the steppe wanted.

Thagimasadas is the hardest of the eight to pin down etymologically, and he does something structurally odd: he breaks the clean set. A pantheon of seven all-Scythian gods is a tidy, even an Iranian-feeling number, echoing the seven great holy beings of Zoroastrian Persia. Then this eighth god sticks out past the edge, belonging to one royal clan as a kind of dynastic patron of horses and waters. The best reading makes him a specifically Iranian, specifically tribal god, the divine ancestor and horse-lord of the ruling house, worshipped precisely because he was theirs and not everyone’s. Every aristocracy likes a god the commoners cannot claim.

The faces around the edges

Those eight are the canonical pantheon, but a few more divine figures crowd in at the margins of the sources, and they are worth a quick roll-call because each one teaches something the main eight cannot.

The snake legged goddess. A woman who is human above the waist and twin serpents below, the cave-dwelling foremother of the royal line in the Greek-flavored origin story. She may be a face of Api, or a separate chthonic ancestress, and she is everywhere in Scythian metalwork, coiled at the center of gold plaques and horse-gear. She is the one Scythian deity we can actually see, again and again, even though we are least sure of her name. Indo-Iranian ancestress under a Greek story.

The sun-god of the Massagetae. The eastern Saka people of Tomyris, the warrior queen who killed Cyrus the Great and, the story goes, dropped his severed head into a skin full of blood so the conqueror could drink his fill at last, kept the barest and cleanest religion on the steppe. Herodotus sums up their whole faith in one line: “The sun is the only god whom they worship; to him they sacrifice horses; the reason of it is that he is the swiftest of the gods, and therefore they give him the swiftest of mortal things.” That image, the sun honored with a galloping horse, is deeply Indo-Iranian: it is the same instinct behind the great Vedic horse sacrifice, the aśvamedha, where the horse is bound to the solar and royal order. Out on the far eastern edge, Tomyris and her people may have kept the oldest and barest steppe layer of all. Indo-Iranian, the eastern edition.

Ditagoia. She matters precisely because she does not fit. Centuries after Herodotus, the late Scythians who settled in the Crimea and built a small capital called Scythian Neapolis worshipped a goddess of this name, whom they equated with the Greek Artemis or the underworld Hecate. She is a different goddess wearing a different Greek mask in a different century, and she proves something important: there was never one fixed “Scythian goddess” frozen for all time. The divine feminine on the steppe kept shifting shape, absorbing whatever powerful goddess the Scythians ran into next, from Levantine Astarte in Herodotus’s day to Greek Hecate in the Crimea five hundred years later. The goddess was a sponge, and that is a fact about Scythian religion, not a flaw in our knowledge of it. A late Greek mask over a local goddess, and a lesson in how the steppe did theology.

So where did the gods come from?

We have met them all. Now we can answer the question the whole piece has been circling, the one a curious reader actually wants settled: taken together, is this an Indian religion, an Iranian one, a European one, a Babylonian one? The honest, satisfying answer is that it is a building with four floors, and you can name what is on each.

The first floor is inherited Indo-European, the deepest and oldest stone. The fire that is holy. The Sky-Father married to the Earth. The Father-god whose name is “Papa.” The holy weapon that holds a god. The whole habit of dividing the cosmos and society into the king, the warrior, and the producer. None of this is borrowed from anyone, because it is older than the borrowing nations existed. It is the shared inheritance the Scythians carried out of the same ancestral world that produced Greek and Latin and Sanskrit. When you feel the Ṛgveda flicker behind Tabiti and Papaios and Api, that is this floor showing through, and it is the closest the Scythians come to being “Indian,” not by descent from India but by sharing its grandparents.

The second floor is the Iranian house they actually lived in. Most of the god-names, once you crack them, are good Iranian words: the “Lord of X” princes, xvarǝnah the royal glory, a-nar-ya- the un-men, sūra- the mighty. The same gods stand in the Avesta under Persian names:

• Anāhitā behind the great goddess

• Mithra likely behind the sun

• Verethragna behind the war-god

• Ātar behind the hearth-fire

The Scythians were not generic Indo-Europeans; they were specifically Iranians, the northern cousins of the Persians, and their pantheon is an Iranian pantheon first and foremost. This is the main structure, the load-bearing one.

The third floor is a single borrowed coat, and it hangs on one figure only: the goddess. This is the cleanest and most containable part of the whole story, which is why I trust it. The Near Eastern influence, the Ishtar-Astarte-Cybele material, did not flood the pantheon. It concentrated almost entirely on Artimpasa, the Heavenly Aphrodite, picked up along the Black Sea trade routes and at temples like Ascalon. The sky-father stayed Indo-European. The fire-queen stayed Indo-Iranian. The war-sword stayed inherited. Only the great goddess put on foreign jewelry, which is exactly what great goddesses did all across the ancient world, because she was the figure most easily identified with whatever powerful goddess the next culture offered. One borrowed coat, on one shoulder.

The fourth floor is the basement, and it is older than all of them: the Siberian substrate. Before any Iranian word was spoken on this grass, Inner Asia already had its animal spirits and its deer-mother, the Mistress of the Animals who ruled the beasts. You can see her oldest monuments in the deer-stones, standing stones carved with leaping deer that radiocarbon dating places three to five centuries before the earliest Scythian art, in the Late Bronze Age cultures of Mongolia and southern Siberia that came before them.

When the Iranians rode in, they did not erase this older spirit-world; they built on top of it. The deer-mother sank down into the foundations of Artimpasa as her Mistress-of-Animals aspect, while the whole dazzling Scythian “animal style” of curled predators and folded stags kept her visual language alive on every belt and scabbard. I should flag, because honest scholars do, that whether there was one unified “Deer Goddess” is contested: some specialists see a single great animal-mother, others see only a scatter of animal spirits. But that something pre-Iranian and Siberian underlies the steppe’s sacred art is not seriously in doubt. It is the floor you stand on without noticing.

Put the four floors together and you get the real answer. Scythian religion is an Indo-Iranian system, inherited and structured, the genuine steppe sibling of the religion in the Veda and the Avesta. It wears a Near Eastern coat on the shoulders of its goddess, and it stands on a Siberian floor it never tore out. It is not Indian, exactly, and not Persian, exactly, and certainly not Greek, despite all of Herodotus’s masks. It is the third sibling, raised on the open grass between them, who kept the family fire burning in its own way.

These gods got two afterlives, and they ran in opposite directions. The eastern Saka carried the pantheon south into India, where you can watch it change religions on their own coins, a steppe war-god shading toward Śiva, a Saka dynasty endowing Buddhist monasteries within a few generations of crossing the mountains. That second life is a whole essay of its own, and it lives in Children of the Golden Sword rather than here. The other afterlife, the one I want to end on, ran the opposite way, west into the mountains, and it is the stranger of the two.

The god who is still alive

Religions do not usually get an epilogue. This one does.

Remember the Ossetians, our third witness, the Caucasus people who still speak a daughter of the Scythian tongue. Their Nart epic is a working museum of steppe religion. One thing to say plainly: the Nart sagas are not the Ossetians’ alone. They are the common heritage of the Caucasus, carried by the Adyghe and Kabardians, the Abkhaz, the Abazin, and the Ubykh as much as by the Ossetes, and the same heroes live on in all of them: the matriarch I call Satana is Satanaya to the Circassians and Abkhaz, and the stone-born Soslan is their Sosruko, born from the same rock in every telling.
Vasily Abaev, who introduces the very volume I draw on here, makes the point himself: he lists these peoples one by one and insists the Narts share not a few stray motifs but one foundation, the same chief heroes under the same name, even tracing how Soslan’s name shuttled from Ossetian into Circassian, was preserved in Abkhaz, and passed back again. I dwell on the Ossetian telling for one narrow reason: Ossetic is an Iranian language descended in a straight line from Scythian, so its version preserves the linguistic thread, farn from xvarǝnah and the rest, that ties the whole epic back to the steppe. That is a reason to foreground one telling, not to erase the others. Abaev spent a lifetime laying the Scythians and the Ossetians side by side, and he caught their relationship in a single sentence: “that which with the Scythians was a living custom becomes with the Ossetians a theme for their Nart epic.” A thing that was once done in blood on the steppe survives in the mountains as a song. And the proof that this is descent and not coincidence runs through the language itself. The sound changes that turn old Scythian words into modern Ossetic are regular and traceable, so the sacred vocabulary comes through intact: xvarǝnah, the burning royal glory of the founding myth, surfaces as Ossetic farn, “divine grace,” and the a-nar-ya- “un-men” leave their mark too. The words carry the line.

And so do the gods, who walk through the Nart sagas in light disguise:

• The great goddess is still there, as Satana, the immortal mother and sorceress at the center of the Nart world, the wise woman without whom the heroes are lost. Scholars line her up with the powerful, governing Sarmatian and Scythian women the ancients kept reporting, the Tomyris type, the queen who rules.

• The sun-hero is still there, as Soslan, born miraculously from a stone, married to the daughter of the Sun, and killed in the end by a rolling solar wheel, his story still carrying buried Herodotean details like the warrior’s grim cloak and the straw-stuffed funeral horse.

• The war-god and his holy sword are there too, in the steel bodied hero Batraz, whose death by his own sword cast into the sea I told at the close of the last piece and will let that ending keep. If you have not read it, that is the door back to it.

But I promised you the fire-queen would outlast them all, and she does. At the top of Herodotus’s list, before the sky-father, before the sword, stood Tabiti, the hearth-fire goddess, the chief deity, the oath that could not be broken. Walk into a traditional Ossetian house twenty-five centuries later and the holiest object in it is the hearth, and above the hearth hangs an iron chain over the central fire, the rækhys. It is the most sacred thing in the home, and it governs the most charged moments of a life:

• a guest becomes sacred and untouchable the moment his hand closes on it

• a new bride circles it three times to enter her husband’s family

• a mother settling her child for the night keeps one hand on it and commends him to its keeping

To swear by it is the most binding oath a person can take; to defile it is the gravest insult a family can suffer. The Ossetians credit the chain to a hearth-god they call Safa, who gave it to humankind, but strip the later name and what you are looking at is unmistakable: the sacred hearth-fire, the unbreakable oath, the still center of the house and the realm, the very things that made Tabiti the chief god of the Scythians.

Twenty-five hundred years after a Greek tourist wrote down that the Scythians swore their deepest oaths by the royal hearth, a shepherd in the Caucasus still swears his by the chain above the fire. The sky-father faded into a folk memory. The sword-god rode off into the legend of a king and a lake. The great goddess scattered into a hundred borrowed shapes. But the fire-queen, the first name on the list, the burning one whose name is the sister of the Vedic word for sacred heat, never went out. She is the oldest thing the Scythians believed, and she is the last thing still believed, banked in the coals of a mountain house, holy, warm, and patiently waiting for the next oath.

One last honest word, because I have tried to keep the seams showing the whole way down. The match between the Narts and the Scythians is built on comparison and inference, not on a signed confession; some of the heroes’ names are borrowed from neighbors even where their stories are not, and a careful scholar holds the line that the cycles and the words carry the continuity while staying skeptical about any single tidy equation. But the fire on the hearth needs no etymology to defend it. It is just still burning, where they left it.

Sources and further reading

The ancient texts. Herodotus, Histories, is the spine of everything here: Book IV for the pantheon list (4.59), the war-god’s sword (4.62), the founding myth (4.5 to 4.10), and the Enarei (4.67); Book I for the Ascalon curse (1.105) and the Massagetae’s sun-and-horse worship (1.216). Hippokrates, Airs, Waters, Places, adds the Enarei from the medical side. For the Iranian cousins of these gods I lean on the Avesta, above all the Ābān Yašt, which gives Anāhitā her threefold name, and the Mihr Yašt, which gives Mithra his title “lord of the wide pastures.” Quotations follow the public-domain translations of A. D. Godley and George Rawlinson.

Reconstructing the myth and the names. D. S. Raevsky, Scythian Mythology, is the fullest attempt to read the pantheon as one structured system. Bruce Lincoln’s essay “Once Again ‘the Scythian’ Myth of Origins” gives the skeptical, propaganda-minded reading, and supplies the two family-tree diagrams. The hard Iranian etymologies, the three “Lord of X” sons, xvarǝnah and its Ossetic survival farn, Targitaos as “enduring strength,” and Goitosyros as a Mithra epithet, come from the recent philological work of Dzitstsoyty on the Scythian language, building on Abaev and Grantovsky.

The Indic parallels. The bolder comparisons with the Vedic rājasūya consecration and the myth of King Pṛthu are argued by Mikhail Bukharin, “Colaxais and His Brothers.” They are a striking parallel rather than a settled fact, and I have flagged them that way in the text.

The Indo-European frame. M. L. West, Indo-European Poetry and Myth, is the authority for the deep-family verdicts: Tabiti as “the Burning one,” Papaios as Father Dyeus, and the cautions about Api and about over-reading trifunctional structure into the myth.

The goddess and her borrowed coat. Yulia Ustinova, The Supreme Gods of the Bosporan Kingdom, is the close study behind the whole Artimpasa section: the Iranian Anāhitā with her threefold name, the Levantine Astarte of Ascalon and her lesser, half-bestial twin Atargatis, and the snake-legged foremother who slid into Atargatis’s role. Ustinova is refreshingly honest that the strands are impossible to fully separate. Yuri Zaytsev, The Scythian Neapolis, covers the late Crimean goddess Ditagoia.

The Siberian floor. Esther Jacobson-Tepfer’s work on the deer-mother and the Mistress of the Animals, and the radiocarbon dating of the deer-stones, with the honest counter-voices who doubt a single unified “Deer Goddess.”

The religion as lived, and the people behind it. For the archaeology that confirms the rites, the genetics of who the Scythians were, the death-cult, the cannabis, and the Saka road into India and toward Buddhism, see my earlier piece, Children of the Golden Sword, and the sources gathered there (Rolle, Rudenko, Artamonov, and the recent ancient-DNA studies).

The god still alive. John Colarusso and Tamirlan Salbiev, Tales of the Narts, with V. I. Abaev’s introduction, is the gateway to the Ossetian survivals: Satana, Soslan, Batraz, and the sacred hearth-chain. Georges Dumézil’s lifetime of work on Scythian-to-Ossetian continuity stands behind all of it.

Open the Ṛgveda, the oldest large body of poetry in any Indo-European language, and you meet a god who measured out the universe in three strides:

“Now shall I proclaim the heroic deeds of Viṣṇu, who measured out the earthly realms, who propped up the higher seat, having stridden out three times, the wide ranging one.”

The Ṛgveda took shape in the river country of the Punjab across the late second millennium BCE, while pharaohs still ruled Egypt and centuries before Homer, a thousand years before the Buddha and before the oldest verse of the Hebrew Bible. Its hymns are not all of one age. The fullest telling of the three strides, the one above, belongs to the first of the ten books, which is among the later strata. But the deed itself is older than the hymn that celebrates it: the strides are already there in the family books, Maṇḍalas six and seven, the bedrock core of the collection. This is not a late embellishment on Viṣṇu. It is the oldest thing we know about him. And the god who paces them out is worshipped this morning, under the same name, by something close to a billion people. Few gods anywhere have been called by name, without a break, for so long.

And yet, by the count, Viṣṇu is almost no one in that oldest book. Five or six hymns out of more than a thousand, and a scatter of passing mentions. The handbooks write him down as “minor” and move on. But open a Vedic ritual text only a little younger and you find the same god set at the very summit of the pantheon: agnir vai devānām avamo viṣṇuḥ paramas, tadantareṇa sarvā anyā devatāḥ, “Agni is the lowest of the gods, Viṣṇu the highest, and all the rest stand in between.” Barely counted in one breath, and crowned in the next. A minor deity is not spoken of that way. Something has been mismeasured.

That mismeasurement is what this primer is about. The usual story explains it by chronology:

• Viṣṇu rose late, a small solar deity of the early Veda who climbed, somehow and against the odds, to the head of the gods.

• Vaiṣṇavism, the religion built on him, was a late assembly, brahmanised and systematised in the last few centuries BCE.

A god older than the alphabet, filed under “rose to prominence late.”

That story, I think, mistakes one question for three. “When did Vaiṣṇavism begin?” has three different answers, depending on what you mean by it.

There is the deity. Viṣṇu is a late Bronze Age god, and he has what look like cousins across the Indo-European world, a silent striding god among the Norse who takes one enormous step at the end of time, a far-seeing judge among the Iranians, kin that would make him an inheritance rather than an invention. By the close of the Vedic period, in any case, he is no minor anything; he has become the sacrifice itself, the act on which the standing of the cosmos depends.

There is the worship of him as the supreme God, cut into stone, by an Indian king and by a foreign convert alike, two thousand years ago and more.

And there is the system: the metaphysics of emanation, the fixed roll of ten incarnations, the wedding of the goddess to the god, the warring schools of philosophy. That part really is late, the work of roughly 200 BCE to 500 CE and after.

The standard story takes the youngest of the three and lets its lateness stand for the whole. It dates the religion by the date of its system, and a god first hymned a thousand years before the Buddha ends up written off as a child of late antiquity. One line carries the whole correction: systematization is not origin. The system is late. The god and the worship are not.

Two things keep this from being merely my assertion. The first is that you need not take it from any single school, and the schools here agree on almost nothing. The Victorian philologists called him a faded sun; the Dutch structuralists, the cosmic centre; the Marxist historians, a tribal cult dressed up by priests; the traditional theologians, the eternal absolute. They cannot all be right, and they spend their footnotes saying so. But on the few points this argument rests on, they converge, and a claim that colonial philology, Hindu theology, and Indian Marxism all underwrite is not the artefact of any one of their biases. The second is a mechanism hiding in plain sight. Viṣṇu did not rise from nothing in the middle Vedic period. He rose with the priesthood whose god he had always been, and the rise left a fingerprint you can count.

A word on method. The genuinely late things, I grant up front. The cowherd Kṛṣṇa of the butter and the flute, the goddess Rādhā, the consort at Viṣṇu’s side, the full theology of the descents: these are late, and I will say so each time they appear.

So, the three ages in turn, beginning with the god.

Part One. The Strider

Start with the hymn count

Open the Ṛgveda, the oldest layer of Indian religious poetry, composed in the late Bronze Age somewhere in the centuries around 1500 to 1200 BCE. Of its 1,028 hymns, only five or six are addressed to Viṣṇu alone. By raw frequency he is a modest presence. Indra, the warrior king who breaks the dragon and releases the waters, holds something like 250 hymns. Agni, the sacred fire, around 200. Counting the name across the whole collection gives about 113 occurrences, most of them passing references inside hymns to other gods.

The Victorian philologists who first catalogued the Vedic pantheon read those numbers in the obvious way. Arthur Macdonell, whose Vedic Mythology of 1898 is still a standard reference, opened his section on the god with the verdict that has stuck: “the main characteristic of the nature of Viṣṇu, a minor deity in the Ṛgveda, is his three steps.” His three strides, Macdonell decided, were the daily or yearly course of the sun, and “Viṣṇu probably is a personification of the activity of the sun.” Indra, he noted, “is the only deity most closely associated with him.”

That is the baseline, and it is fair. If you count hymns, Viṣṇu is not in the front rank of the Ṛgveda, and his most visible Vedic trait can be read, plausibly, as solar.

But the date matters, and it is the first thing the standard story skips past. A god who already has his own hymns in the late Bronze Age, a signature cosmic deed, and a named paradise is not a beginner. He is an inherited figure of a tradition older than nearly every religion still practised on earth. Three thousand years ago, someone was already singing to him by name.

What the hymns actually say

Now look at the content, which carries more weight than the count.

The defining deed of the Ṛgvedic Viṣṇu is the three strides. In three vast steps he paces out the cosmos and makes the room in which living things can exist. Its fullest telling is the hundred and fifty-fourth hymn of the first book, six verses by the poet Dīrghatamas, though the same act is sung in the older family books as well. It opens:

“Now shall I proclaim the heroic deeds of Viṣṇu, who measured out the earthly realms, who propped up the higher seat, having stridden out three times, the wide ranging one.”

And it tells you what the strides are for:

“In this way Viṣṇu will be praised for his heroic deed, he who is like a fearsome wild beast, living in the mountains and roaming wherever it wants, in whose three wide strides dwell all living beings.”

That is world-building work. Before the strides there is no measured space; after them there are three worlds, earth, midspace, and heaven, each wide enough for its inhabitants. The sun crosses the sky every day, but the sun does not build the room the sky encloses. In this poetry, Viṣṇu does. His standing epithets say it: urugāya and urukrama, the wide going, the wide striding one, and the verb for striding across the worlds is reserved almost entirely for him.

Then comes the third step, and here the later religion is already visible in outline. Two of Viṣṇu’s strides are within mortal reach; the third, his highest, lies beyond it. The hymns call it the paramaṃ padam, the highest footstep, and they describe it as a place of honey and light, the goal the singer longs to reach. The 154th hymn again:

“Might I reach that dear cattle pen of his, where men seeking the gods find elation, for exactly that is the bond to the wide striding one, the wellspring of honey in the highest step of Viṣṇu.”

And in the 22nd hymn, the line that the whole later tradition would hang upon:

“That is the highest footstep of Viṣṇu. The patrons always behold it, stretched out like the eye in heaven.”

Tad viṣṇoḥ paramaṃ padam. The American Vedicist E. W. Hopkins, a careful and unsentimental reader, called this the earliest Indian image of a heaven of the blessed, the home of departed souls, and the Indian scholar R. N. Dandekar followed him. In other words, the destination that classical Vaiṣṇavism names Vaikuṇṭha, the paradise of the liberated, is already present in the Ṛgveda in rudimentary form: honey, light, the goal of the soul’s journey. Centuries later the Kaṭha Upaniṣad would quote that exact phrase, the highest step of Viṣṇu, as the end of the spiritual road. The seed of the Vaiṣṇava heaven is Bronze Age.

I want to be fair about the relationships too, including where they cut against me. In the Ṛgveda Viṣṇu is most often Indra’s companion, and the junior of the two. When Indra braces to kill Vṛtra and free the imprisoned waters, he turns to his ally:

“Comrade Viṣṇu, stride out widely. Heaven, grant a place for the mace to prop apart. We two will smite Vṛtra, we two will give leave to the rivers.”

Viṣṇu makes the space, Indra strikes the blow. The later tradition will turn this completely around, subordinating Indra to a cosmic Viṣṇu. That inversion is real growth, and I will not pretend it was finished in 1200 BCE. It was not.

But notice what already clings to him besides the strides. His name most likely comes from the root meaning to pervade, to penetrate, to be active throughout, which is why the Purāṇas would later gloss it as “the one who enters or pervades everything.” He is invoked to protect the embryo and to grant offspring. He is called ghṛtayoni, ghee wombed, lodged in the clarified butter that the Vedic texts treat as the life sap of the universe. The great Dutch Indologist Jan Gonda devoted a whole monograph, Aspects of Early Viṣṇuism, to drawing this out: alongside the solar strider there is, from the very start, a Viṣṇu bound up with fertility, with the sustaining of life, with food itself. That is the seed of his classical office as the preserver, the maintainer of the worlds, and it is Vedic, not a Puranic afterthought.

Where the “minor god” verdict came from

The modern study of this god began in the nineteenth century, and so did most of the confusion about his age.

The first European Vedicists worked under the spell of solar mythology and what Max Müller called henotheism, the habit of the Vedic poets of exalting whatever god they happened to be addressing as if he were supreme. Müller treated this as “an unconscious urge towards monotheism,” a stage on the way to higher religion. Within that scheme Viṣṇu was slotted as one solar deity among several, and the three strides were read as sunrise, zenith, and sunset, the reading Macdonell inherited and codified. Hermann Oldenberg dissented even then, holding that “every definite trace of solar character is lacking” and that Viṣṇu was from the beginning simply a traverser of wide space. So the solar reading was never unanimous, and much of the “minor god” verdict rests on it.

A second strand of nineteenth century scholarship reached for a non Aryan or specifically Dravidian origin for the deities that did not fit the Indra centred warrior pantheon. The name Nārāyaṇa, which becomes central to our story, was a favourite target. One school derived it from Dravidian roots, nār meaning water and ayana a resting place, so that Nārāyaṇa would be “the one who lies on the waters,” the cosmic sleeper.

I take the case against that reading from Suvira Jaiswal, and since she will be the most quoted scholar in these pages, let me say who she is. Her 1967 book, The Origin and Development of Vaiṣṇavism, is the standard academic history of how the cult formed, written from the materialist school of D. D. Kosambi and R. S. Sharma, which treats religion as a by-product of social and economic life and dates that formation to roughly 200 BCE to 500 CE. She is, for most of this essay, the scholar with the most professional reason to call Viṣṇu young. Which is why it matters that, again and again, she ends up calling him old.

On the Dravidian etymology she is firmly sceptical. She notes that the theory “does not explain how nir was changed into nar, involving the substitution of the vowel i by a,” and that the watery reading was itself the epic poet’s invention, since “the poet of the Mahābhārata seems to have taken particular delight in inventing fanciful derivations of the names of the deities.” The Dravidian origin of Viṣṇu and Nārāyaṇa is part of the history of the question rather than a settled result, and I flag it because it still circulates.

The most consequential framing, though, came from the missionary scholars, and it is the root of the “late and borrowed” picture. Horace Hayman Wilson, lecturing at Oxford in 1840, was candid that one studied Hindu religion in order, in his words, “to confute the falsities of Hinduism” and win Hindus to Christianity. In that climate the German Indologist Albrecht Weber and others advanced, in the 1870s, the theory that the bhakti of Kṛṣṇa, the devotion of loving surrender, had been borrowed from Christianity, carried east by Nestorian missionaries. Whatever the intent, the effect was to make the whole devotional tradition look late and derivative by construction.

It was Indian scholars who took this apart, and they did it with the very evidence we have been handling. R. G. Bhandarkar in 1913, R. P. Chanda working from the archaeology in 1920, and H. C. Raychaudhuri in 1921 used the inscriptions, Pāṇini, and Megasthenes to push the cult back into the centuries BCE and to sever it from any Christian source. Jaiswal, writing decades later from the opposite end of the ideological spectrum, records the outcome flatly: the Christian origin theory “is now abandoned altogether on chronological grounds.” The “recent and borrowed” picture was refuted long ago, inside the scholarship. It survives mostly in the popular afterglow, and in the looser kind of textbook.

What the hymn count hides

If the solar minimizers were one camp, there has always been another, and it includes some of the most serious Vedicists of the last century.

The Dutch scholar F. B. J. Kuiper argued, in his work on Vedic cosmogony, that the hymn count badly understates the god. In the Veda, he wrote, Viṣṇu “must have been a central figure, of greater importance than Indra himself,” because his third stride represents the transcendent totality that reconciles being and non being, the whole that stands above the cosmos rather than inside it. Gonda, citing the earlier judgment of Sten Konow, put it more cautiously but to the same end: in his Vedic form “the seeds were already present from which a great god could develop.” And A. B. Keith, no sentimentalist, warned that the relative prominence of the gods “is not necessarily fully brought out” by the Ṛgveda, “the range of interests of which was not very wide.” The hymnal was the work of a warrior and priestly elite. It records what they cared about. A deity beloved of women, of cultivators, of the unlettered, would be under reported by the very nature of the text. Silence in the Ṛgveda is not absence in the world.

The most striking modern contribution to this side of the argument is also the most concrete, and it comes from the contemporary scholar who writes under the name Manasataramgini. He simply counted. He tallied the occurrences of the name Viṣṇu across the entire span of Vedic literature, text by text, and the pattern is hard to unsee. In the Ṛgveda the name appears about 113 times. Then look at the Yajurveda, the Veda of the priest who performs the physical acts of the sacrifice, the adhvaryu: in its core collections the count leaps to 175 in the Taittirīya Saṃhitā, 197 in the Kāṭha, 233 in the Maitrāyaṇī, and 223 in the Śatapatha Brāhmaṇa. In the Sāmaveda, the Veda of chant, he stays scarce, around 22. These are his own tallies rather than figures from a published concordance, and the exact numbers would move a little with the method of counting.

That pattern is not noise, and it is not a quirk of text size. It tells you something the hymn count hides. The Vedic priesthood was not one body but several lineages, each with its own collection and its own ritual role, and Viṣṇu was always the special deity of one of them, the ādhvaryava tradition of the Yajurveda, the tradition of physical ritual action. His apparent rise across the middle Vedic period, on this reading, is really the rise of that priestly tradition to dominance, carrying its god up with it. Manasataramgini puts the whole thing under a single Vedic sentence that we will meet again in a moment: yajño vai viṣṇuḥ, the sacrifice is Viṣṇu. The deity who owns the sacrifice rises with the priesthood of the sacrifice.

He adds another piece, easy to check. There is the deity-pair, the devatā dvandva, called Agnāviṣṇū, Agni-and-Viṣṇu, absent from the Ṛgveda but routine in the Yajurveda and the Brāhmaṇas, and it always sets Viṣṇu last, in the slot of supremacy. That ordering is the structure behind the line this essay opened with, the Aitareya Brāhmaṇa’s “Agni the lowest, Viṣṇu the highest, all the rest in between.” It is no flourish. It is the ritual hierarchy, written down, in the very period when Viṣṇu is supposed to have been a nobody.

And there is the deep background, the part that reaches back toward the steppe, where I want to be candid that the ground gets softer. The contemporary scholar Manasataramgini reconstructs Viṣṇu as an inherited Indo-European deity, older than the Indo-Aryans, by setting him beside possible cousins in the other branches. One Vedic detail is solid and suggestive: in a hymn Indra urges Viṣṇu to stride vitaram, “wider,” and that same root for wideness has been read into the name of the Norse Víðarr, the silent god, strongest after Thor, who plants his great shoe to tear open the jaws of the Fenris wolf at Ragnarök and survives into the world reborn. Further out he points to the Iranian Rashnu, the far seeing judge who rides at Mithra’s right hand as the Vedic Viṣṇu stands close to Mitra, and, more speculatively still, the Celtic thunderer Taranis. This is a comparison, not a proof, and not every link will survive scrutiny; the etymology of Víðarr is itself disputed, and the Celtic parallel is the thinnest of the three. But the argument runs the same way as everything else around it. A god who can even plausibly be set beside Germanic and Iranian counterparts looks less like an Indian invention of the iron age than like an inheritance carried in from much deeper time.

There is independent confirmation that the Vedic gods are this old, from an unexpected quarter. In the fourteenth century BCE, far to the west of India, a treaty was sworn between the Hittites and a kingdom called Mitanni in northern Syria, and the gods called to witness it, inscribed at Boğazköy, are unmistakably the gods of the Veda: Indara, Mitra, the two Nāsatya, and Uruwana, that is Indra, Mitra, the Aśvins, and Varuṇa. Paul Thieme argued that these names are specifically Indo-Aryan, not generic Indo-Iranian, which means a recognizably Vedic pantheon was being invoked, in the Near East, before or around the time the Ṛgveda itself was being composed in India. The Indologist Hermann Jacobi went so far as to say the Mitannians were already worshipping the Vedic gods in full perfection in the fourteenth century. Edwin Bryant, whose survey of the whole Indo-Aryan question is the most balanced we have, treats this as among the firmest pegs for dating the Ṛgveda to “around the middle of the second millennium.” I should be honest about one thing: Viṣṇu is not among the four Mitanni names, which are the sovereignty and warrior gods, Mitra, Varuṇa, Indra, the Aśvins. The Mitanni tablets do not prove the antiquity of Viṣṇu specifically. What they prove is that the Vedic pantheon he belongs to was already old and already abroad in the Bronze Age, which is the frame within which his own Ṛgvedic hymns, his Indo-European cousins, and his cosmic strides have to be read. He is an old god among old gods, not a latecomer to a young religion.

Viṣṇu becomes the sacrifice

Here is the hinge the whole development turns on, and the evidence is unusually clean.

In the ritual prose of the Yajurveda and the Brāhmaṇas, Viṣṇu stops being one god among many and simply becomes the sacrifice. Yajño vai viṣṇuḥ, say the Śatapatha Brāhmaṇa and, earlier, the Taittirīya Saṃhitā: the sacrifice is Viṣṇu. The equation is so routine that Griffith, Englishing the White Yajurveda, gives up explaining it and just glosses the pronoun at the altar verses: “thee,” he notes, means “Viṣṇu, that is, the sacrifice.” To a modern reader that lands as dry ritual bookkeeping. To the men who wrote the Brāhmaṇas it was the largest claim a sentence could carry. For them the sacrifice was the axis of the cosmos, the act that kept the worlds from sliding back into nothing, and to call a god the sacrifice was to set him at the centre of everything without troubling to write him a single new hymn. A marginal god does not get fused with the one act that holds the universe up.

This is why the count and the formula say the same thing from two directions. By the late Vedic period Viṣṇu already holds the function that classical Vaiṣṇavism will later universalize. He is the cosmic maintainer, identified with the act that sustains the world. The Śatapatha Brāhmaṇa even makes him, in one passage, equal in size to the sacrifice, and identifies him with food, the all supporting. Macdonell, no friend of the maximal reading, conceded that “a special feature of the Brāhmaṇas is the constant identification of Viṣṇu with the sacrifice.”

The strides follow the same path from poetry into rite. In the Yajurveda the worshipper himself paces out three steps, the viṣṇu krama, identifying his own body with Viṣṇu measuring the worlds. The text he speaks as he steps is plain about it: “O Viṣṇu, stride thou widely forth, give ample room for our abode. Drink butter, homed in butter. Still speed on the sacrifice’s lord.” The Bronze Age hymn has become a thing the priest does with his feet. The Atharvaveda, meanwhile, keeps the god in the household register, invoked over the ghee and, above all, for the making of a child: “Let Viṣṇu form and mould the womb, let Tvaṣṭar duly shape the forms, Prajāpati infuse the stream, and Dhātar lay for thee the germ.” This is the homely, life giving Viṣṇu, the shaper of the womb, the god of the embryo, exactly the side that the warrior poets of the Ṛgveda had little reason to record and that the household priests of the Atharvan kept alive. The god runs continuously through all four Vedas, and across the three priestly traditions, gaining weight as he goes. By the close of the Vedic period he is already the god who is the sacrifice, and the god whose highest step is where the soul wants to end up. No later age had to discover him.

The navel of the world

One more Vedic image bridges the strider of the hymns and the cosmic God of the temples, and it shows how little of the later theology was invented from nothing. It is the image of the navel.

In the ritual thought of the Brāhmaṇas the sacrifice is not only the axis of the cosmos but its centre, its navel. The Ṛgveda already says it, in the great riddle hymn: “the altar is the extreme limit of the earth, the sacrifice the navel of all existence.” And the navel, the Śatapatha Brāhmaṇa adds, “means the centre, and the centre is safe from danger.” Now recall that in these same texts Viṣṇu is the sacrifice. The god who is the sacrifice is, by the same logic, the navel of the world.

Gonda traced this idea across the whole Vedic and post Vedic literature, and the trail leads straight to one of the most familiar images in all of Hinduism. The navel is where life begins. A mother goddess of fertility broods over the navel of the earth, as the Greeks set their omphalos stone in the temple at Delphi. The cosmic pillar, the skambha of the Atharvaveda, rises from the centre and links earth to heaven, and the sacrificial post, the yūpa, which the texts say belongs to Viṣṇu “because Viṣṇu is the sacrifice,” is that pillar in wood, the axis of the world and the ladder by which the sacrificer climbs to the sky. The hymn of the All Maker, Viśvakarman, already speaks of the One “fixed upon the navel of the Unborn, on which all existence is established.” And out of all this comes the picture every Hindu child knows: Viṣṇu asleep on the cosmic serpent in the ocean of dissolution, and from his navel a lotus rising, and seated on the lotus the four faced Brahmā, beginning the world again. The god’s title for this is Padmanābha, the lotus naveled one. Gonda’s conclusion: the lotus from the navel “clearly forebodes the well known conception of later times,” for the navel of Viṣṇu is “the centre of the universe, the source of existence,” and it is no accident that the deity who came to occupy that centre was Viṣṇu, “who in Vedic times was the sacrifice.”

This is the same cosmic centrality that Kuiper reached from a different direction when he called Viṣṇu more important than Indra, the totality that stands above the dual cosmos. Two of the most careful readers of the Veda, one working from the strides and the other from the navel, arrive at the same place. And Gonda’s whole book closes on the word that names it. The essence of the god, he decided after weighing every aspect, is pervasiveness: “the actual penetration of the provinces of the universe must indeed be considered a central feature,” the name itself coming from the root to pervade. The strider strides because he pervades, and a pervader is, of necessity, everywhere at the centre.

Part Two. The Cosmic Person

To follow the story forward we have to bring in a second name, because the supreme God of classical Vaiṣṇavism is carried, in the earliest texts, as much by the name Nārāyaṇa as by the name Viṣṇu. This is one of the most important and least appreciated facts in the whole subject, and it sharpens the antiquity argument rather than weakening it.

Nārāyaṇa enters

Nārāyaṇa is, on the face of it, a non Vedic deity. He does not appear in the Ṛgveda at all. He enters the record in the Śatapatha Brāhmaṇa, the great ritual prose of the White Yajurveda, and he enters already cosmic. In one passage he sends forth the classes of gods, the Vasus, Rudras, and Ādityas, through the three daily pressings of the sacrifice, and then, by performing a five day rite called the pañcarātra sattra, “gained superiority over all beings and became identical with all beings.” His litany, the text says, is the Puruṣa Sūkta, the hymn of the cosmic Person whose body becomes the universe. From his first appearance, then, Nārāyaṇa is identified with Puruṣa, the thousand headed primeval Man of Ṛgveda 10.90. Jaiswal draws the conclusion: “already at the time of the composition of the Śatapatha Brāhmaṇa Nārāyaṇa was a deity of considerable eminence.”

The standard etymology she settles on, after rejecting the Dravidian one, derives the name from nara plus ayana, the goal or resting place where men go, so that Nārāyaṇa is the collective entity of Man, with Nara as the individual. It is this collective Man sense that generates the famous epic pairing of Nara and Nārāyaṇa, the twin sages, later read as Arjuna and Kṛṣṇa.

A memory of human sacrifice

Here Jaiswal makes her boldest and strangest claim, the kind of thing popular accounts skip. The five day rite that Nārāyaṇa performs in the Śatapatha, the pañcarātra sattra, was, in the older stratum of the text, the Puruṣamedha, the human sacrifice. And since medha means food, Puruṣamedha means, literally, human food. Reading Nārāyaṇa as the collectivity of Man, Jaiswal traces the deity to a tribal institution of human sacrifice, a man killed for the good and the food of the whole community, only later turned symbolic. In her words, the Śatapatha passage “undoubtedly discloses Nārāyaṇa’s connection with the institution of human sacrifice,” and “if Nārāyaṇa signified the collectivity of man, it was for the good of the whole tribe that a human being was killed, originally to provide food for the entire community of men, and later for ritual purposes.”

She finds the survival of this in a ritual that the manuals still describe, the Nārāyaṇa bali, an offering to Nārāyaṇa performed for those who die an unnatural death, by suicide or violence, for the uncremated, and for those guilty of the five great sins. She reads it as “a remnant of some earlier ritual in which human beings were unnaturally put to death in honour of Nārāyaṇa.” And she sets it against its twin, the Viṣṇu bali, a rite performed in the eighth month of pregnancy to secure an easy delivery, a fertility rite. The two never blur in the texts, and the contrast tells her something about origins: Nārāyaṇa carries the memory of human sacrifice, Viṣṇu the memory of fertility, two originally distinct figures who would later fuse.

This theme has a striking afterlife in the Vaiṣṇava scriptures themselves, where the tradition turns against the very practice its oldest stratum may remember. In the Bhāgavata Purāṇa, in the chapter that describes the hells, the human sacrificers are consigned to torment, and the punishment is a perfect reversal:

“There are men and women in this world who sacrifice human beings, and then eat their victims’ flesh. Those who perform such sacrifices are taken after death to the abode of Yamarāja, where their victims, having taken the form of Rākṣasas, cut them to pieces with sharpened swords. Just as in this world the man eaters drank their victims’ blood, dancing and singing in jubilation, their victims now enjoy drinking the blood of the sacrificers and celebrating in the same way.”

The Sanskrit names the act puruṣa medhena yajante, “they worship by human sacrifice.” Whatever the deep origin of Nārāyaṇa, the developed Vaiṣṇava tradition damns the literal rite to hell. The seed and the flower can be very far apart.

How Nārāyaṇa became Viṣṇu

How did this non Vedic cosmic Person become Viṣṇu? Jaiswal and the historians of the Bharatiya Vidya Bhavan school reconstruct a two step absorption, and the steps matter for the dating.

Nārāyaṇa was identified first with Prajāpati Brahmā, the creator, which makes sense, since both are cosmic Persons associated with the Puruṣa and with creation. Crucially, the animal forms that would later become Viṣṇu’s avatāras, the fish, the tortoise, the boar, are in the Brāhmaṇas the forms of Prajāpati, not of Viṣṇu. Only afterward was Nārāyaṇa identified with Viṣṇu, by which time Brahmā is demoted to the subordinate who sprouts from Viṣṇu’s navel. The Taittirīya Āraṇyaka seals the threefold identity in a single formula of meditation that any Vaiṣṇava can still recite:

nārāyaṇāya vidmahe vāsudevāya dhīmahi tan no viṣṇuḥ prachodayāt

“We contemplate Nārāyaṇa, we meditate on Vāsudeva, may Viṣṇu impel us.” Three names, one deity. R. C. Majumdar, the nationalist historian who edited the vast History and Culture of the Indian People, notes that this formula “regards Nārāyaṇa, Vāsudeva and Viṣṇu as one and the same deity,” and that the Mahābhārata, telling the same story, calls the supreme god Nārāyaṇa far more often than it calls him Viṣṇu. The name that would win in the end, Vaiṣṇava, the Viṣṇu name, was at this stage the rarer of the two. The dominance of the Viṣṇu element, as Jaiswal puts it, “is a later development, and in the beginning Nārāyaṇa was the deity par excellence.”

This is the honest complication I promised, and it makes the case stronger, not weaker. The supreme God theology of Vaiṣṇavism, the cosmic Person, the all pervader, the object of surrender, attaches in the earliest texts to Nārāyaṇa, who is non Vedic but Brāhmaṇa old and identified from the start with the Puruṣa. The Vedic Viṣṇu, the strider with the honey heaven, is the partner he fuses with. What we call Vaiṣṇavism is the confluence of more than one ancient river, the Vedic Viṣṇu, the cosmic Nārāyaṇa, and, as we will see next, the heroic Vāsudeva. All three are old. You can point to each one separately and date it deep.

Part Three. The Hero and the Bhāgavata Cults

That completes the first age, the deity, with its three converging streams of Vedic Viṣṇu, cosmic Nārāyaṇa, and the third we turn to now. The second age is the worship, and it gathers around that third stream: Vāsudeva, the hero god of the Vṛṣṇi clan, the figure who would become Kṛṣṇa, and the one a Greek ambassador would soon put on a pillar as “the god of gods.” His cult is the best attested of the three in the last centuries BCE, and it is the matrix in which the words Bhāgavata and bhakti are first carved into stone.

The man who became a god

The starting point for modern work on this is H. C. Raychaudhuri’s thesis, which Jaiswal summarizes and endorses: “Vāsudeva Kṛṣṇa, a hero of the Sāttvata sect of the Yādavas, was apotheosised first by his clansmen who imparted his worship to others.” A real man, or at least a legendary clan hero, deified by his own people, the Sātvatas, and then exported. Bhandarkar, two generations earlier, had already concluded from the evidence that this worship “must have originated long before the establishment of the Maurya dynasty,” that is, well before 320 BCE.

The witnesses line up across the fourth and third centuries. Pāṇini, the grammarian, perhaps in the fifth or fourth century BCE, frames a rule about a word for “one whose object of devotion is Vāsudeva,” and Patañjali, commenting on it around 150 BCE, glosses Vāsudeva as the name not of a warrior but of “the honourable one,” the god. Megasthenes, the Greek envoy to the Mauryan court around 300 BCE, reported that the people he called the Sourasenoi, the Śūrasena Yādavas of the Mathurā region, “held Herakles in especial honour,” and there is wide, though not universal, agreement that this Herakles is Vāsudeva Kṛṣṇa. Jaiswal, characteristically, reads the phrase against the maximalists: the words “especial honour,” she says, “clearly indicate that Kṛṣṇa was still a minor divinity, far from being the supreme god,” in 300 BCE. Even read at its most cautious, the witness puts an organized cult of the hero god at Mathurā around 300 BCE.

There is also an older, more uncertain thread. A Kṛṣṇa appears in the Ṛgveda itself, a dark skinned figure defeated by Indra on the river Aṃśumatī, the Jamna, and Jaiswal takes the name to “indicate a non Aryan genesis.” And the Chāndogya Upaniṣad, perhaps in the seventh or sixth century BCE, names a “Kṛṣṇa son of Devakī,” pupil of the sage Ghora Āṅgirasa. Edwin Bryant, who has surveyed this evidence as carefully as anyone, calls the Chāndogya reference “plausible but still questionable,” plausible because it gives Kṛṣṇa his mother’s name, Devakī, questionable because nothing later connects him with the sage Ghora. He leaves it open. But he then lays out a chain of references that is much harder to wave away, and the cumulative weight of it is the real argument. Follow the links. Yāska’s Nirukta, around the fifth century BCE, alludes to the Syamantaka jewel, a motif from a well known Kṛṣṇa story. Pāṇini, in the same period, frames his rule about the devotee of Vāsudeva and names the very clans of the Kṛṣṇa legend, the Andhakas and Vṛṣṇis. The Baudhāyana Dharma Sūtra invokes Viṣṇu under a string of twelve names, Keśava, Govinda, Dāmodara, names that “reveal an awareness of several stories that are fully developed in later texts.” The Buddhist Niddesa, in the Pali canon, speaks “somewhat derogatorily of those devoted to Vāsudeva and Baladeva,” and a hostile witness who bothers to sneer is a witness to a cult worth sneering at. And the gāyatrī of the Mahānārāyaṇa Upaniṣad fuses the three names, Nārāyaṇa, Vāsudeva, Viṣṇu, in one breath, while, Bryant notes, “the same mantra appears in the much earlier Yajur Veda without mentioning Vāsudeva,” letting us watch the hero’s name being threaded into an older Vedic formula. From Pāṇini to the Heliodoros pillar the testimony is continuous, and Bryant’s conclusion is the sober one: the worship of the divine Vāsudeva “had already developed deep roots” by the time a Greek bowed to it.

The receipts in stone

Four inscriptions, cut between roughly the second century BCE and the first CE, settle the matter. Together they establish, beyond reasonable doubt, an organized, named, image worshipping cult of Vāsudeva in the centuries BCE.

The most striking of them was raised by a man who was not even an Indian. Around 113 BCE, on the road outside Vidiśā, a Greek named Heliodoros, son of Dion, an ambassador of the Indo-Greek king Antialkidas, set up a tall stone pillar crowned with Garuḍa, the eagle that carries Viṣṇu, and on it he carved this:

“This Garuḍa standard of Vāsudeva, the god of gods, was set up here by Heliodora the Bhāgavata, son of Dion, a man of Takṣaśilā, the Greek ambassador who came from the great king Antialkidas to king Kāśīputra Bhāgabhadra the saviour, prospering in the fourteenth year of his reign.”

A second inscription on the shaft adds three steps that lead to heaven, self restraint, charity, and vigilance, so he knew the ethics as well as the name. A foreign diplomat, two thousand years ago, calling himself a Bhāgavata and raising a standard to Vāsudeva as “the god of gods.” It was on the strength of this and the inscriptions beside it that Bhandarkar called the early evidence “irrefragable” and judged the worship of Vāsudeva “as old as Pāṇini.”

The Ghosundi inscription of the first century BCE, near Nagari in Rajasthan, records a Bhāgavata king named Sarvatāta, who had performed a horse sacrifice, building a stone enclosure of worship, a Nārāyaṇa vāṭikā, for the gods Saṃkarṣaṇa and Vāsudeva, whom it calls sarveśvara, the lords of all, the undefeated. Two things matter here. The pairing of Saṃkarṣaṇa with Vāsudeva, and the placing of their shrine inside a Nārāyaṇa enclosure, which already hints at the fusion of the hero cult with the cosmic Nārāyaṇa, in stone, in the first century BCE.

The Nānāghāṭ cave inscription of about the same date, set up by the Sātavāhana queen Nāganikā, opens its invocation with a row of gods, “Dharma, Indra, Saṃkarṣaṇa, Vāsudeva, the Moon, the Sun, and the four guardians of the quarters.” The hero pair has entered the formal, royal, almost Vedic invocation.

And the Mora well inscription near Mathurā, of the first century CE, records a woman named Toṣā setting up, in a stone shrine, the images of “the holy five heroes of the Vṛṣṇis,” the bhāgavataṃ pañcavīrāṇām, meant for worship. Stone images, in a temple, of the deified Vṛṣṇi heroes, by the first century CE. The image cult is not a medieval development.

The five heroes, the lost goddess, and the elder brother

That phrase, the five heroes of the Vṛṣṇis, opens a window onto something older and stranger than the tidy theology that came later, and it is one of the richest parts of Jaiswal’s book.

Before Vāsudeva became the supreme God, he was one of a group of deified clansmen, and Jaiswal argues that the oldest layer of the Vṛṣṇi cult was centred not on a male hero at all but on a goddess. Her name was Ekānaṃśā, the sister of Vāsudeva and Baladeva, “the black goddess,” whom Jaiswal identifies as “the tribal goddess of the Vṛṣṇis.” Her case is detailed: a Kuṣāṇa relief from Mathurā shows the goddess standing between Balarāma and Keśava, and as late as the sixth century the astronomer Varāhamihira still prescribes flanking her image with the two brothers. From this she draws a striking conclusion: “the cult of Ekānaṃśā is much older than that of Baladeva and Vāsudeva, who appeared in the beginning simply as associate male gods of the more important female divinity; but with the changeover of social conditions, from a matriarchal to a patriarchal state, the goddess lost her importance and the male gods received greater prominence.” The scaffolding she hangs this on, a primeval matriarchy giving way to patriarchy, is a nineteenth century evolutionary scheme that anthropology has long since set aside, and it is the weakest timber in her case. But the observation underneath it does not depend on the scheme: the goddess is demonstrably there early, flanked by the two brothers, and she demonstrably fades. The brothers, in this reading, began as the goddess’s attendants and ended as gods in their own right, while she faded into Subhadrā, the sister at Jagannātha Purī, or was absorbed into Durgā and Śrī.

The elder brother, Saṃkarṣaṇa Baladeva, is himself a study in absorption. Jaiswal traces him to a serpent cult and an agrarian one at once. He is mentioned first in the Arthaśāstra, where spies disguised as his ascetic devotees drug an enemy’s cowherds, and “his predilection towards strong drinks was known.” His weapons are the plough and the pestle, the tools of farming. His very name means the act of ploughing or furrowing. He is “an incarnation of Śeṣanāga,” the cosmic serpent, and a snake comes out of his mouth at his death. “If we dissociate the nāga element from the character of Saṃkarṣaṇa Baladeva,” Jaiswal writes, “hardly anything distinctive remains.” Megasthenes, she suggests, glimpsed him as a kind of Dionysos, the god who “first yoked oxen to the plough,” and Arrian’s note that this Indian Dionysos was older than Herakles by fifteen generations she reads as a memory that Saṃkarṣaṇa, the elder brother, was felt to be the older god. The later theology would invert this too, demoting “the elder brother to the subordinate, anomalous position of a mere emanatory form of the supreme god Vāsudeva.”

The system is younger than the cult

Out of this cluster of clan heroes the developed Bhāgavata theology built its most distinctive doctrine, the vyūhas, the four emanations. Jaiswal states the full cosmogony:

“the supreme god bhagavat in his transcendental highest form Vāsudeva creates from himself the vyūha Saṃkarṣaṇa and prakṛti, the primeval matter. From the association of Saṃkarṣaṇa and prakṛti arise the vyūha Pradyumna and manas, the cosmic mind; and from the association of these two emanate the vyūha Aniruddha and ahaṃkāra, cosmic self consciousness.”

Vāsudeva, Saṃkarṣaṇa, Pradyumna, Aniruddha, four deified Vṛṣṇi kinsmen turned into a metaphysics of the unfolding of God into the world. And here the dating is precise and agreed, because it is the very thing that tells us the system is later than the cult. Bhandarkar pointed out long ago that the Bhagavad Gītā “contains no allusion to the vyūhas,” so the Gītā predates the doctrine, and that the first century BCE inscriptions know only Vāsudeva and Saṃkarṣaṇa, “so that the system of four vyūhas was not fully developed up to that time.” Jaiswal puts the developed vyūha theology in the third or fourth century CE. The twenty four forms of Viṣṇu are later still; Jaiswal goes so far as to derive them from the twenty four Jain Tīrthaṅkaras, though that particular link is her own conjecture, and most would read the shared number as coincidence rather than borrowing.

It is the same split as before. The worship of Vāsudeva is pre Mauryan; the four fold metaphysics that explains him is of the Gupta age. Hold those two apart and most of the paradoxes in the subject go away.

The white island

The literary home of this theology is a long and strange section of the Mahābhārata called the Nārāyaṇīya, set into the great didactic book, the Śānti Parva, and there the epic openly theorizes a one God religion. Bhīṣma tells of a far northern land beyond the ocean of milk, Śvetadvīpa, the White Island, whose inhabitants are “radiant as the moon and devoted to Nārāyaṇa.” The decisive phrase, in Bibek Debroy’s translation, is exact and astonishing for a text of two thousand years ago: “those men, who live in Śvetadvīpa, worship one person alone.” They are ekāntins, men of single minded, exclusive devotion; “they are devoid of senses, they do not eat, their eyes do not blink”; and “they knew about the beneficial rituals of pañcarātra.” This is exclusive, single-God devotion, named and located and equipped with a ritual system, inside the epic, a long way from the vague henotheism the Veda is usually reduced to.

The God they worship is defined in the same section in terms that gather every thread of this essay. He is the cosmic Person beyond the principles of the philosophers: “he is beyond the twenty four principles and is known as the twenty fifth. He is known as Vāsudeva, the eternal supreme soul.” He is the source and end of all: “everything in the universe, mobile and immobile, is generated from me. Those who are devoted to me enter me and are emancipated.” And he unfolds himself in the four forms the Bhāgavatas worshipped, the vyūhas: from him comes Saṃkarṣaṇa, from Saṃkarṣaṇa Pradyumna, from Pradyumna Aniruddha, and from Aniruddha the lotus born Brahmā who makes the world. The cosmic Nārāyaṇa, the Vṛṣṇi Vāsudeva, the four fold God of the Bhāgavata cult, and the exclusive devotion of the white islanders are all one doctrine here, the doctrine the epic calls the Sātvata way, to be told, it warns, only to a devotee.

The sceptic in the hall

The epic also knows that this is a contested claim, and it stages the contest. At the great royal sacrifice in the Sabhā Parva, when Bhīṣma directs that the first honour of the assembly be given to Kṛṣṇa, a king named Śiśupāla rises to object, and his objection is precisely the rationalist, status conscious one a modern sceptic might make. Kṛṣṇa, he says, “is not even a king,” “not a sacrificial priest, or a preceptor”; to honour him above crowned kings is mere favouritism, “like a dog that has found an offering of sacrificial ghee and consumes it in private.” We have seen, he sneers, “what Vāsudeva is. We have seen everything exactly as it is.” It is the voice of the man who counts only what can be counted. And the answer Bhīṣma gives is the answer of this whole tradition: “the entire universe and everything are established in Vāsudeva. The worlds owe their origin to Kṛṣṇa, and in him are they dissolved. Since he does not know that Kṛṣṇa is always everywhere, this Śiśupāla is only a child.” The argument over whether the cowherd’s kinsman is also the ground of the world is not a modern invention. The epic has it out, in the hall, before the war.

The romance really is late

Now the concession I promised, made in full, because the argument is strong enough to carry it and because hiding it would be dishonest.

The Kṛṣṇa most people picture first, the butter thief of Vṛndāvana, the flute player, the lover of the cowherd girls, the gopīs, is not the oldest Kṛṣṇa. He is, in Jaiswal’s careful reconstruction, a later graft, “largely the result of his identification and amalgamation with the worship of some youthful god of the Ābhīra tribe,” a pastoral herding people of the early centuries CE. The Jain accounts of Kṛṣṇa, she notes, “show no knowledge of a pastoral and infant god Kṛṣṇa,” and the Padma Purāṇa has Viṣṇu announce that he will be born among the Ābhīras. The erotic element, the dance with the gopīs, “appears to have been largely responsible” for the introduction of the amorous material, reflecting “the exigencies of a nomadic life” that “required a greater freedom of sexes.” R. C. Majumdar, no minimizer of Hindu antiquity, reached the same verdict independently: this stratum “did not form part of the original creed.”

And Rādhā, the beloved of the later devotional imagination, is later still, essentially absent from the early sources altogether, a figure of the medieval bhakti efflorescence that runs through the Tamil Āḻvārs and, later, Bengal.

One honest qualification runs the other way, and it belongs here because it fits the same pattern as everything else. Already in the Ṛgveda, in the very hymn of the three strides, Viṣṇu is called gopā, the herdsman: trīṇi padā vi cakrame viṣṇur gopā adābhyaḥ, which Griffith renders “Viṣṇu, the Guardian, he whom none deceiveth, made three steps.” The word means, literally, a keeper of cattle, and Sāyaṇa glosses it here as the protector of all the worlds, the guardian who cannot be deceived. This is not yet the boy with the butter and the flute. But the cattle keeping coloring is on Viṣṇu from the beginning, of a piece with the god of the embryo and the ghee we met in Part One, and when a cowherd god finally rose to the centre of the religion he turned out to be a form of Viṣṇu and no one else. Even the cowherd has a Vedic seed. What is genuinely late is the romance grown around him, the gopīs and the butter and Rādhā, not the herdsman himself.

So the romantic Kṛṣṇa really is young. Saying so costs the case nothing. The Vāsudeva whom Heliodoros worshipped around 113 BCE was the warrior god, the kinsman hero, the Bhāgavata’s “god of gods,” not the cowherd of Vṛndāvana. The antiquity I am defending is the antiquity of that god, and of the Vedic Viṣṇu and the cosmic Nārāyaṇa he merged with, not of every later flower on the tree.

The goddess at his side

One more stream completes the classical figure, the goddess Śrī Lakṣmī, and her history is a small model of the whole.

She begins as two goddesses, Śrī and Lakṣmī, fertility and fortune, merged by the time of the Brāhmaṇas. The earliest hymn to her, the Śrī Sūkta, is a supplement attached to the Ṛgveda. Jaiswal, and a line of German scholars before her, read her original character as a pre Aryan goddess of fertility and vegetation: she is “moist,” “ever nourished,” “abounding in cow dung,” “born through the mire,” and her oldest images, the Gaja Lakṣmī type with the goddess on a lotus flanked by elephants, go back to the second and first centuries BCE at Bharhut and Sanchi. For most of her early life she is independent, “worshipped in her own right quite unallied with any sectarian cult,” and where she does attach to a male god it is first to Indra, or to Kubera the lord of wealth, or to Kārttikeya, not to Viṣṇu.

Her marriage to Viṣṇu is a Gupta age event, and Jaiswal dates it with unusual sharpness: “the goddess was united with the god towards the beginning of the Gupta period,” in “the third or fourth centuries A.D.” The verses of the Śrī Sūkta that link her to Viṣṇu, she notes, “are very late interpolations, certainly post Gupta.” The Gupta inscriptions are the first to call Viṣṇu the permanent abode of Lakṣmī. The lotus, her emblem, is the last of the four attributes to be added to Viṣṇu’s hands, and Jaiswal reads its arrival as the sign of her absorption. By the sixth century the Ahirbudhnya Saṃhitā has made her his śakti, his power.

The divine couple at the centre of later Vaiṣṇavism is an ancient, independent goddess married to the god in the Gupta age: old goddess, old god, late marriage.

Part Four. The Teacher

No account of Viṣṇu can skip the text in which the whole theology speaks in the first person, the Bhagavad Gītā, the seven hundred verses set into the sixth book of the Mahābhārata in which Kṛṣṇa, on the field of Kurukṣetra, reveals himself to Arjuna as God. It is, in Jaiswal’s words, “a Bhāgavata document,” and it is where the doctrine of devotion “finds its first clear exposition.”

Whose god speaks in the Gītā

There is a subtlety here that most readers miss, and it ties the Gītā straight back into the argument of Part Two. The God of the Gītā names himself in several registers, and he treats both Viṣṇu and Vāsudeva as instances rather than as his own name. In the tenth chapter, listing the forms in which he is supreme, he says ādityānām ahaṃ viṣṇuḥ, “among the Ādityas I am Viṣṇu,” and vṛṣṇīnāṃ vāsudevo ‘smi, “of the Vṛṣṇis I am Vāsudeva.” He stands behind both names. Jaiswal draws the natural inference, that the supreme speaker of the Gītā is the cosmic Nārāyaṇa, the bhagavat, into whom both the Vedic Viṣṇu and the Vṛṣṇi Vāsudeva are being gathered. The text is even called, in tradition, the Nārāyaṇīya Gītā, and the Buddhist writer Āryadeva, around the end of the second century CE, attributes its verses to Nārāyaṇa. The Gītā is the literary act of fusion itself, the moment the three rivers are declared to be one water.

The descent declared

The Gītā is also where the doctrine of the avatāra, the descent of God into the world, receives its classic statement, in the fourth chapter:

yadā yadā hi dharmasya glānir bhavati bhārata abhyutthānam adharmasya tadātmānaṃ sṛjāmy aham paritrāṇāya sādhūnāṃ vināśāya ca duṣkṛtām dharma saṃsthāpanārthāya sambhavāmi yuge yuge

“Whenever there is a withering of dharma, Bhārata, and a rising up of adharma, then I send forth myself. For the protection of the good and the destruction of evildoers, to set dharma firmly in place, I come into being age after age.”

Two things here matter. First, the verb. As the philologists point out, the word avatāra itself, the descent, is not yet used here. The Gītā says sambhavāmi, I come into being, sṛjāmi, I send forth. The noun “avatāra” is a later coinage; the Gītā and the Nārāyaṇīya speak of births and appearances, prādurbhāva. The technical doctrine is later than the idea. Second, the priority of purpose. The line puts the protection of the good first and the destruction of the wicked second, an ordering the later theologians, as we will see, made into a principle.

God mistaken for a man

The Gītā knows, and says, that a God who walks the earth as a man will be taken for one. “The deluded despise Me,” Kṛṣṇa says, “clad in human form, not knowing My higher being as the great Lord of beings” (9.11). It is a remarkable verse, because it is the theological mirror image of the modern historian’s question. The scholar asks whether Kṛṣṇa was a man who was later raised to God; the Gītā answers that he is God who came down as a man and was mistaken, by the dull, for less. Whichever way you read the history, the text itself is built around exactly that paradox, the supreme being in a human body, “than whom nothing higher exists,” yet driving another man’s chariot. “On Me all this universe is strung,” he says in the same breath, “like pearls on a thread” (7.7).

And the Gītā ends on the note that the whole later tradition would build into its central doctrine. After all the paths, the knowledge and the works and the meditation, the last word is surrender: “Abandoning all duties, take refuge in Me alone. I shall liberate you from all evils; do not grieve” (18.66). This single verse, the caramaśloka or final verse, becomes the seed of prapatti, the doctrine of self surrender that Rāmānuja, a thousand years later, would make the path open to everyone, of every caste and either sex. The road from the Bronze Age strider runs through this line.

The cosmic form

And the Gītā is where the god shows what he is. In the eleventh chapter Arjuna asks to see the divine form behind the friend and charioteer, and Kṛṣṇa grants it, first giving him the divine eye to see with, then unveiling the viśvarūpa, the universal form, the whole cosmos blazing in a single body. “If a thousand suns should rise all at once in the sky,” the text says, “such splendour would resemble the brightness of that Great Self.” It is the Puruṣa of the Ṛgveda, the thousand headed Person whose body is the universe, made visible to one terrified man, and at its heart is the most chilling self definition in the poem: kālo ‘smi, “I am Time, the mighty cause of world destruction, who has come forth to annihilate the worlds.”

Two details of this vision matter for the argument of this whole essay. First, when Arjuna, overwhelmed, cries out, he does not call the form Kṛṣṇa. He calls it Viṣṇu. “I find neither courage nor tranquillity, O Viṣṇu,” he says (11.24), and again, “your terrible rays blaze forth, O Viṣṇu” (11.30). At the climax of the revelation the friend and the cosmic God are named as one, and the name is the Vedic one. Second, when Arjuna has had enough of the terror and asks for the familiar shape back, he asks for an icon: “I desire to see you wearing a crown, armed with a club, discus in hand. Become that four armed form” (11.46). The crown, the mace, the discus: this is the temple Viṣṇu, the four armed image of the Gupta sculptors, requested by name inside a text of the last centuries BCE. Jaiswal notes that the universal form “appears to have been originally associated with Nārāyaṇa,” its prototype the vision of the whole world inside the deity’s body in an older episode of the epic, when the sage Mārkaṇḍeya is swallowed by the cosmic child on the waters. The Gītā gathers the Vedic Puruṣa, the cosmic Nārāyaṇa, the temple icon, and the Vṛṣṇi hero into one revelation, and lets a trembling man name them all Viṣṇu.

Where did devotion come from?

Finally, the Gītā is the charter of bhakti, devotion, the path it commends above sacrifice and above austere knowledge: give me your mind, your offering, your reverence, and you will come to me. The later tradition would build everything on this. And it is over the origin of this one word that the modern schools divide most sharply, in a way that shows cleanly how method drives conclusion.

The dominant Indian historiography of the twentieth century, the materialist tradition of D. D. Kosambi, R. S. Sharma, and Jaiswal, derives bhakti not from feeling but from economics. Jaiswal’s etymology is the load bearing move of her whole reconstruction, and here it is in full:

“The terms bhagavat, bhakta, bhakti and Bhāgavata are derived from the root bhaj, which in its earliest use in the Vedic literature means to divide, distribute, allot, share with, or partake of. It is never used in the sense of to adore, or to serve, which is the usual meaning of the verb in later works. Similarly bhaga in the Ṛgveda meant wealth, share, a lucky share.”

On this reading, bhagavat, the one later worshipped as the Blessed Lord, originally meant simply the possessor of wealth, the tribal collective that owned the common goods; bhakti meant a share of that wealth; and bhakta, the devotee, originally meant the member who received his portion, which is why, she notes, the word bhakta still means a meal, and in agricultural society came to mean boiled rice. Devotion, in this account, begins as the distribution of food. The earliest religious sense, she adds, is “material and concrete,” and the early Vaiṣṇava inscriptions confirm the flavour: the Traikūṭaka kings describe themselves as “the hired labourers of the bhagavat,” and the relationship of god and devotee is “that of the master and the slave.” Loving, ecstatic devotion, she insists, comes later, with the Tamil Āḻvārs and later still with Bengal.

Against this stands the traditional theological reading, given its fullest modern statement by S. M. Srinivasa Chari, writing from inside the Śrī Vaiṣṇava school. Chari derives bhakti from the same root bhaj but in its other sense, to serve, to be devoted, and glosses it as snehapūrvam anudhyānam, “unceasing meditation suffused with love.” For him devotion is not a social residue but the central spiritual discipline, and its roots are Upaniṣadic. The two readings of one word, the economic and the devotional, mark the deepest fault line in the whole field, and we will come back to what it tells us. For now it is enough to note that both schools agree the Gītā is where the doctrine of bhakti receives its first clear exposition, in the last centuries BCE.

Part Five. The Descents

With the avatāras we reach the third age, the system, the part of all this that really is late. The daśāvatāra, the ten descents of Viṣṇu, is its showcase: fish, tortoise, boar, man lion, dwarf, axe wielding Rāma, Rāma of Ayodhyā, Kṛṣṇa, the Buddha, and the rider yet to come. And yet even in this last layer the same trick is at work. The fixed list is late; several of the figures on it are not. Some are Vedic, older than almost anything else in the religion. Some are cosmogonic acts annexed from another god. Some are deified men. One is a hostile takeover of a rival faith. Go through them one at a time and you watch Vaiṣṇavism build itself.

How the list grew, and kept growing

First, the doctrine. Jaiswal locates the engine of the whole theory in the fusion we have already traced: “the theory of incarnation seems to have evolved with the identification of bhagavat Nārāyaṇa with the hero god Vāsudeva Kṛṣṇa, who thus came to be looked upon as the human incarnation of the former.” The model, in other words, was Kṛṣṇa. Once one human form of God was admitted, the others followed.

She and Bhandarkar agree, against the popular impression, that the famous ten are a late and unstable aggregation. The earliest list, in the Nārāyaṇīya section of the Mahābhārata, gives only four to six: in Bhandarkar’s reading of the text, “Varāha, Narasiṃha, the oppressor of Bali, Rāma of the Bhṛgu race, Rāma Dāśarathi, and he who will come for the destruction of Kaṃsa.” There is no fish, no tortoise, no Buddha, no Kalki in that oldest nucleus. The verse that lists the ten, Jaiswal notes, “is now proved to be a late interpolation.” The number grew over time, four to six to ten to, in the Bhāgavata Purāṇa, twenty two and even twenty four, and in the Pāñcarātra Sāttvata Saṃhitā, thirty nine. The standard ten, she observes, “did not find general acceptance much before the eighth century A.D.” The avatāra list is less an ancient creed than a roster that kept growing as the religion absorbed new gods.

A philological detail underlines the point. The very word avatāra, the descent, “is not used in early works.” The Gītā, the Nārāyaṇīya, and the Harivaṃśa speak of births, appearances, manifestations, prādurbhāva. The noun came after the idea. Jaiswal also stresses, and this matters for a reader of Indian history, that the doctrine was shaped in dialogue with Buddhism: “in this form the doctrine appears to have been considerably influenced by the Buddhist concept of the former Buddhas, whose prime attribute is compassion,” a concept with “chronological priority,” since “some of the former Buddhas are known to have been worshipped as early as the third century B.C.” The endless incarnations of Viṣṇu are modelled, in part, on the endless prior Buddhas.

Against all this stands the theologian’s view, and both belong in the picture. Chari, from inside the tradition, insists the descents are “not illusory manifestations but real,” that God takes them “out of His own free will,” not under the compulsion of karma, and that “His body in the avatāra stage is constituted of śuddha sattva, the pure spiritual substance,” so that the deity is never defiled by the world he enters. The purpose, he says, drawing the order straight from the Gītā, is twofold but unequal: “the protection of the ardent devotees and the destruction of the evil forces,” with protection primary and destruction “incidental,” since evil “can be carried out by God through His will without an incarnation.” Where the historian sees a roster of absorbed cults, the theologian sees the free self manifestation of the one God. Both readings will help us understand the individual descents.

Matsya, the fish

A fish warns Manu of a coming flood, is reared until it is huge, and tows his boat to safety; in the Puranic retelling it also recovers the stolen Vedas. The flood story is among the oldest in the world, with its Mesopotamian cousins in Utnapishtim and Ziusudra, and the Indian version is textually very old: it is told in the Śatapatha Brāhmaṇa, around 800 to 600 BCE. But, and this is the recurring surprise, the fish there is not Viṣṇu. As Jaiswal notes, in the Mahābhārata the fish that saves Manu and the tortoise that supports the earth are mentioned “but it does not connect them with Viṣṇu. In the Brāhmaṇas, both figure as forms of Prajāpati, the creator god; and it appears that they were originally totems of some tribes.” Matsya becomes a descent of Viṣṇu only in the Puranic layer. Chari, tellingly, leaves the fish out of his Pāñcarātra list of thirty nine; it reaches the canonical ten from the Vedic and Puranic stream, not from the theological one.

Kūrma, the tortoise

Viṣṇu becomes a tortoise to serve as the pivot under Mount Mandara when the gods and demons churn the ocean for the nectar of immortality. Again the figure is old and again it begins elsewhere. In the Śatapatha Brāhmaṇa and the Taittirīya Saṃhitā the cosmogonic tortoise is Prajāpati, the creator, identified with Kaśyapa; the older Vedic image of the earth resting on a tortoise’s back is simply repurposed. Chari cites the Taittirīya Āraṇyaka for “the avatāra of Viṣṇu as tortoise,” but, as with the fish, Kūrma is absent from his thirty nine. The churning of the ocean, with the tortoise at its base, becomes a Viṣṇu story only in the epics and Purāṇas.

Varāha, the boar

A boar dives into the cosmic waters and lifts the drowned earth on his tusks. This is the clearest case of all, the textbook example of a cosmogonic act annexed from Prajāpati. In the Taittirīya Saṃhitā, “Prajāpati, who was roaming as the wind, acquires the form of a cosmogonic boar lifting the earth goddess from the primeval waters.” The Śatapatha names the boar Emūṣa, echoing a Ṛgvedic epithet. The earth raising boar is Prajāpati’s deed, and only later, in the Purāṇas, is it transferred wholesale to Viṣṇu, with the demon Hiraṇyākṣa added as an antagonist. Here the historians and the theologians nearly touch: Varāha is one of the genuinely old figures, and Chari keeps him, at number sixteen, in his Pāñcarātra thirty nine. Jaiswal adds the sociological reading that runs through her whole treatment: “It is probable that the boar was a totem of some powerful non Aryan tribe undergoing brahmanisation in the age represented by the Brāhmaṇas.” The boar’s stronghold was the Vindhya country, and it was the most popular of all the descents in the Gupta age; the Hūṇa king Toramāṇa endowed a temple to “Nārāyaṇa who has the form of a boar” around the year 500.

Narasiṃha, the man lion

To kill the demon Hiraṇyakaśipu, whom a boon protects from death by man or beast, by day or night, indoors or out, and to save the demon’s devout son Prahlāda, Viṣṇu bursts from a pillar at twilight as a being neither man nor lion and tears him open on a threshold, in neither category of the boon. This one is genuinely late. “With the exception of a late passage of the Taittirīya Āraṇyaka,” Jaiswal writes, “there is no reference to the god Narasiṃha in the Vedic literature.” She reads him as “originally a dreadful god,” a local terror of the Punjab to be propitiated, absorbed into Viṣṇu only in the early centuries CE, and scholars have long noticed that the plot of the boon outwitted is the old Vedic story of Indra killing Namuci, transferred to Viṣṇu. Chari ranks Narasiṃha seventeenth in his thirty nine and places him in the first age of the world; for the theologian he is the protector of the devotee Prahlāda, the descent that proves God comes for the sake of love before the sake of destruction.

Vāmana, the dwarf

A small Brahmin approaches the generous demon king Bali, begs three paces of land, and then swells to cosmic size, striding over earth and heaven and, with his third step on Bali’s head, pressing the demon down to the underworld. This is the oldest descent of all, and the reason is now familiar: it is the Ṛgvedic Viṣṇu himself, the strider of the three steps, turned into a story. Chari is explicit and correct: “the repeated statements in the Ṛgveda about the three strides refer to the Trivikrama avatāra of the Vedic period.” The Aitareya Brāhmaṇa has the three strides; the Śatapatha already calls Viṣṇu a dwarf, vāmana, in the passage Eggeling and Muir flagged as “the germ of the Dwarf Incarnation.” Everything that is distinctively Vedic about Viṣṇu, the measuring of the worlds, the making of room, the third step beyond mortal sight, is here. The dwarf disguise and the demon Bali are the later narrative scaffolding around an act that is three thousand years old. Jaiswal notes that the Bali story may even preserve “the suppression of a cult of Bali,” since the demon king himself received a kind of worship, remembered to this day in the Kerala festival of Onam, where Bali is the beloved king and Viṣṇu the trickster who displaced him.

Paraśurāma, the axe Rāma

A Brahmin warrior, son of Jamadagni, avenges his father by exterminating the warrior caste “twenty one times.” He is not Vedic, and he is not, at first, divine. He appears in the Mahābhārata as a warrior Brahmin “with no divine or avatāra status,” and his identification with Viṣṇu “occurs only once” in the epic. Jaiswal reads him as a sectarian fabrication: “the apotheosis of Bhārgava Rāma, better known as Paraśurāma, was the work of the Bhṛgu Brahmins, the redactors of the Mahābhārata,” who “evolved between the second and the fourth century A.D.” his legends of glory and humiliation. Chari, revealingly, classes him not as a full descent but as an āveśāvatāra, a possession incarnation, “in whom God pervaded his body,” a partial and lesser mode. A descent that even the tradition ranks as second tier, and that the historians read as a caste charter dressed up as theology.

Rāma of Ayodhyā

The prince of Ayodhyā, exiled, whose wife Sītā is carried off by Rāvaṇa of Laṅkā, whom he then destroys, the very pattern of dharma. His epic, the Vālmīki Rāmāyaṇa, has a long compositional history, and the deifying material is concentrated, the philologists agree, in the first and last books, the Bāla and Uttara Kāṇḍas, “generally regarded as later additions made presumably in the second century A.D.” In the older core Rāma is, as the consensus puts it, “more human than divine,” and in the Pāli Dasaratha Jātaka he is a Bodhisattva, with Sītā as his sister become wife and no abduction at all, “the most primitive feature of the Rāma legends available to us.” Jaiswal’s verdict is exact: there is nothing to show that Rāma Dāśarathi was adored as an incarnation of Viṣṇu in the centuries BCE, “but he was certainly worshipped as such in the Gupta period.” She even recovers an older Sītā beneath the queen, the furrow goddess of the agricultural rites, “the goddess of the farmers,” once flanked, like Ekānaṃśā, by two brother gods. Chari, by contrast, ranks Rāma among the two pūrṇa or perfect descents, “perfect ideal human beings symbolising the very dharma,” whose special gift was “to reveal the secret of śaraṇāgati,” self surrender, as the path to God. The historian dates the deification to the Gupta age; the theologian reads the perfected man as the eternal God. Both are looking at the same prince.

Kṛṣṇa

We have already met him as the Vṛṣṇi hero, the Bhāgavata’s god, the speaker of the Gītā, and the figure onto whom the late Ābhīra cowherd was grafted. In the logic of the avatāra doctrine he is not one descent among ten but the first and the model, the human form of God whose success generated the whole theory. Bhandarkar’s sharpest observation belongs here: even in the Gītā, “the identification of Vāsudeva with Nārāyaṇa had not taken place, nor had his being an incarnation of Viṣṇu come to be acknowledged.” Kṛṣṇa’s own status as an avatāra of Viṣṇu is itself a development; he begins as a god in his own right, the hero of the Sātvatas, and is folded into Viṣṇu rather than budding from him. For Chari he is the supreme pūrṇa descent, in the deepest Śrī Vaiṣṇava reading the very Para Vāsudeva come down in full.

There is a final twist here, and it shows how alive this material stayed. In the Mahābhārata, the Harivaṃśa, and the Viṣṇu Purāṇa, Kṛṣṇa is without question an incarnation of the transcendent Viṣṇu. But the Bhāgavata Purāṇa quietly turns it around. Where the other descents are called partial, a portion or a ray of God, of Kṛṣṇa alone it says kṛṣṇas tu bhagavān svayam, “but Kṛṣṇa is God himself.” Bryant notes that this half line “becomes something of a mahāvākya,” a great utterance, for the Kṛṣṇa sects, and that the sixteenth century theologians, Vallabha and Caitanya, took it to its logical end and held that it is “not Kṛṣṇa who is an incarnation of Viṣṇu but Viṣṇu who is a partial manifestation of Kṛṣṇa.” The cowherd who began as the junior partner ends up swallowing the senior. A tradition that can flip its own hierarchy and keep going for centuries is not something brittle or freshly bolted together.

The Buddha, and the brother he replaced

The ninth descent is the showpiece of the whole syncretic machinery, and also the place where two of the traditions in our corpus openly disagree. In the standard Puranic list Viṣṇu is born as the Buddha, and in the orthodox framing he does it in order to delude the impious into abandoning the Vedas, so that they can be destroyed, a backhanded canonization of the rival founder as a deliberate divine deception. The identification is firmly datable: Hindus came to regard the Buddha as a descent of Viṣṇu only between about 450 and the sixth century CE, first in the Viṣṇu Purāṇa. Jaiswal, against the cynical reading, argues that the move was not a brahmin trick but a popular rapprochement, that “the identification of the Buddha with Nārāyaṇa was not the result of a clever machination on the part of the brāhmaṇas,” noting that the Buddhist Lalitavistara itself already calls the Buddha Nārāyaṇa and Mahāpuruṣa.

And here is a detail that says everything about how constructed the famous ten really are. In the Śrī Vaiṣṇava tradition that Chari represents, the ninth descent is not the Buddha at all. Chari’s list of ten names, in the ninth slot, Balarāma, the elder brother, and omits the Buddha entirely. The Buddha had to be inserted into the popular list around the sixth century, displacing Balarāma, and a major theological school never accepted the substitution. The “canonical ten” are canonical only by sect and by century. There is no fixed, eternal, agreed list, and the fact that one tradition worships Balarāma where another worships the Buddha in the same numbered slot is the proof.

Kalki, the rider to come

The tenth descent has not happened. At the end of the present dark age, the Kali Yuga, Viṣṇu will be born as Kalki, a Brahmin of the village of Śambhala, and ride out on a white horse with a blazing sword to destroy the wicked and restore the age of truth. The figure is in the Mahābhārata’s forest book and elaborated in the Purāṇas. Jaiswal suggests his conception “may have been inspired by the idea of Maitreya, the Buddha of the future,” and that in the Purāṇas he becomes “the symbol of the ultimate triumph of brahmanism.” Chari lists him, at number thirty eight of the thirty nine, as “the form of Viṣṇu riding over a horse, yet to take place at the end of the Kali yuga,” whose purpose is “re establishing the lost dharma.” A future descent is a useful thing for a religion to keep in reserve, and it has been claimed by more than one ambitious man across the centuries.

What the ten really are

Step back from the individual stories and a clear shape appears, and it is the shape of the whole essay in miniature. The fish, the tortoise, and the boar are cosmogonic acts of Prajāpati, the creator, annexed to Viṣṇu in the Puranic period; the man lion and the axe Rāma are late absorptions of fierce local or sectarian gods, the one carrying the old Indra and Namuci plot, the other a caste charter; the dwarf alone was always Viṣṇu’s, the Ṛgvedic strider in narrative dress; Rāma and Kṛṣṇa are deified heroes, the human descents that drove the doctrine; the Buddha is a sixth century polemical annexation that one major school never accepted; and Kalki is the apocalypse held in reserve. The ten are not a single ancient revelation. They are the sediment of a religion absorbing the gods around it across two thousand years, with the oldest layer, the dwarf, reaching back to the Bronze Age strider with whom we began.

Part Six. The Long Afterlife

The story does not stop at the Gupta temples. If the first thousand years of Viṣṇu’s history is the slow fusion of the Vedic strider, the cosmic Nārāyaṇa, and the Vṛṣṇi hero, the next thousand is the long flowering of devotion to the god they became. It is here that the bhakti the colonial scholars wanted to credit to Christianity actually unfolds, out of entirely Indian roots, into the religion a third of India still practises. Bhandarkar’s 1913 book remains the clearest map of it, and most of what follows is his.

The temple of love, and the god who is feared

Bhandarkar begins with a psychological observation that organizes everything. Of the two Vedic gods who survived into supreme status, Viṣṇu and Rudra-Śiva, the one is approached through love and the other through fear. “What contributed to the formation of Vaiṣṇavism,” he writes, “were the appearances and occurrences which excited love, admiration and a spirit of worship; while in Rudra-Śaivism the sentiment of fear is at the bottom.” And he puts it in a single sentence that is hard to improve on: “in India the god that is loved is Viṣṇu-Nārāyaṇa-Vāsudeva-Kṛṣṇa, while the god that is feared is Rudra-Śiva.” Vaiṣṇavism is, at its core, the religion of the God who came down to be loved.

The word for that love, bhakti, is older than its later passion. Bhandarkar’s philology is careful: the term “in the sense of love is not to be found except in a verse” of the Śvetāśvatara Upaniṣad, and even in Rāmānuja’s hands “Bhakti means constant meditation and corresponds to the Upāsana of the Upaniṣads.” The word “etymologically signifies resorting to and then loving the thing resorted to.” The thing came before the word: the love of the soul for the Supreme “was an idea implied and often expressed by the word priya, dear, in the Upaniṣad period.” Devotion grew out of the Upaniṣads, not out of any foreign import.

The Āḻvārs and the Tamil Veda

The emotional flood begins in the Tamil south, with the poet saints called the Āḻvārs, “those immersed in God,” twelve of them, whom Bhandarkar redates from the fantastic traditional figures to “about the fifth or sixth century” and after, though current scholarship tends to place the group a little later, across roughly the sixth to ninth centuries. Their Tamil hymns, gathered into the four thousand verses of the Nālāyira Divya Prabandham, are “considered as very sacred and spoken of as the Vaiṣṇava Veda.” Here, for the first time, God is sung to in the language of the people and in the accents of helpless love, and the songs drove Buddhism and Jainism out of the Tamil country. Vasudha Narayanan, in Bryant’s sourcebook, makes the point that for this tradition the Tamil hymns stand beside the Sanskrit Veda as a second revelation, the ubhaya Vedānta, the dual scripture.

The Ācāryas, and the war over Śaṅkara

The Āḻvārs supplied the feeling; the Ācāryas supplied the philosophy, and they did it under pressure. By the eighth century the great Advaitin Śaṅkara had taught that the world is illusion and the personal God a lower truth dissolved, at the summit, into an attributeless absolute. To a religion built on love between a real soul and a real God, this was, in Bhandarkar’s words, “considered as destructive of the Bhakti.” The whole later history of Vaiṣṇava thought is the building of theologies that could hold devotion and metaphysics together against Śaṅkara’s world-illusion.

The line begins with Nāthamuni, who gathered the Āḻvār hymns, and his successor Yāmuna, who, unable “to maintain the doctrine of Bhakti in the face of the theory of Advaita,” directed his own successor to answer Śaṅkara on his own ground, in a commentary on the Brahma Sūtras. That successor was Rāmānuja, born by the traditional reckoning around 1017, who lived and taught at Śrīraṅgam, was persecuted by a Śaiva Chola king, and built the system called Viśiṣṭādvaita, qualified non dualism. Its elegance is that it makes the world and the souls real without making them separate from God: “the individual soul and the insensate world are the attributes of the Supreme Soul. They constitute his body, just as the body and the indwelling soul constitute the human being.” God is one, but a one with a real body of real souls, loved and loving. Rāmānuja’s bhakti, Bhandarkar notes, “is reduced to the form of a continuous meditation,” the disciplined upāsanā of the Upaniṣads, and to it he joined prapatti, self surrender, the throwing of oneself on God’s grace, which “can be practised by all orders, including Śūdras.” His followers later split over how that grace works, the famous quarrel of the monkey and the cat: does the soul cling to God as a baby monkey grips its mother, with some effort of its own, or does God seize the soul as a cat carries its kitten, by grace alone. It is a real theological fault line, and it is the kind a religion only develops when it has been thinking hard about love for a very long time.

After Rāmānuja the schools multiply, each giving a different answer to the same question of how the soul and God are one and not one. Madhva, in the thirteenth century, found even Rāmānuja too monist and taught outright dualism, Dvaita, with five eternal and unbridgeable distinctions between God, souls, and matter; his supreme name is mostly Viṣṇu, and Bhandarkar notes that in his hands “the Gopāla-Kṛṣṇa element seems to be entirely absent,” the religion swung back from the cowherd toward the cosmic God. Nimbārka taught identity in difference and gave Kṛṣṇa and his beloved Rādhā an “almost exclusive prominence.” Rāmānanda carried the movement north into the vernacular, worshipping Rāma and Sītā and making “no distinction between Brāhmaṇas and members of the degraded castes,” and from his circle came Kabīr, the Muslim weaver who preached a stark imageless monotheism with the name of Rāma, and, later, Tulsīdās, whose Hindi Rāmcaritmānas became the Bible of the north. Vallabha, around 1500, built the path of the divine child, Puṣṭimārga, the way of grace, around the boy Kṛṣṇa of Braj, with daily temple worship that wakes, bathes, dresses, feeds, and cradles the god as an infant. And in Bengal, his contemporary Caitanya took the emotional road to its end, teaching that the highest love, when it reaches its peak, “constitutes itself into Rādhā,” and losing himself in the ecstatic chanting of the divine names.

The late making of Rādhā

Here the honest historian has to mark a late arrival, because it bears directly on the antiquity argument. Rādhā, the beloved who dominates the later imagination, is a genuinely late creation. She “is not mentioned by name in the Harivaṃśa, Viṣṇu Purāṇa and the Bhāgavata,” the very texts that tell the cowherd Kṛṣṇa’s story. She emerges fully only around the sixteenth century, in the Nārada Pāñcarātra and the theologies of Nimbārka, Vallabha, and Caitanya, spun out of a hint in the Bhāgavata’s account of the one favoured gopī. Bhandarkar, a Victorian reformer, disapproved, calling her elevation “a degrading element.” You need not share his disapproval to accept his chronology. The romantic theology of Rādhā and Kṛṣṇa is medieval. It is the furthest flower on the tree, and nobody should mistake it for the root.

The centre never moves

Two thousand years of this, and the centre still does not move. The religion never stops elaborating: the Tamil hymns, the six rival Vedāntas, the boy of Braj, the saint of Bengal, the Maratha poets of the sober god Viṭṭhala who “discarded Rādhā-Kṛṣṇaism” for a purer devotion. And yet, as Bhandarkar says, through all of it “Vāsudeva as the name of the Supreme Being stands in the background of all,” and the spiritual core is “essentially derived from the Bhagavadgītā.” The bhakti that a missionary scholar in 1840 wanted to credit to Christianity is, when you actually trace it, the flowering of a devotion whose first clear statement is the Gītā, whose vocabulary is Upaniṣadic, and whose object is the Vedic Viṣṇu. It grew from inside, over three thousand years, from a Bronze Age root.

Part Seven. The Case, and the Quarrel

We have followed the god through his three ages: the Bronze Age deity, the worship of the last centuries BCE, the late system. The evidence for the first two has been laid out along the way, in the hymns and the inscriptions, and I will not relist it here. Two things remain, and they are the two I promised at the start: to show that the argument does not depend on me, and to ask how the opposite impression ever became the default.

Where even the opponents agree

The strongest form of the argument does not lean on any single school. It rests on the points where schools that agree on nothing else converge, because those are the claims that no ideological lens can be accused of manufacturing.

There are at least four such points. First, that Viṣṇu’s identification with the sacrifice, yajño vai viṣṇuḥ, is the hinge of his rise: the colonial Macdonell notes it, the structuralist Gonda builds on it, the materialist Jaiswal cites it, the traditionalist Chari claims it as Vedic, and the comparativist Manasataramgini turns it into a measurable curve. Second, that the Heliodoros pillar proves a foreigner-attested cult of Vāsudeva in the second century BCE: Bhandarkar, Jaiswal, Majumdar, and Chari all build on it. Third, that the developed system, the vyūha metaphysics and the fixed list of descents, is a later crystallization; even Chari, the eternalist, concedes that the vyūha doctrine is a Pāñcarātra development absent from the Gītā. Fourth, that the oldest descents, the dwarf and the boar, sit in the late Vedic Brāhmaṇa layer, which Macdonell, Keith, the epigraphist D. C. Sircar, and Manasataramgini all grant.

And the device that makes these land hardest is to let each concession come from the school least inclined to give it. The Marxist Jaiswal, whose whole book places the formation of Vaiṣṇavism in the last centuries BCE and the first CE, nonetheless writes of Ṛgvedic Viṣṇu that “even in the Ṛgvedic age he was considered to be a great god,” and “he may have been more popular among the masses whose ideas and inclinations are not recorded in the Vedas.” She even sides with Gonda’s pervasion reading over the old solar one. The author of the standard “late” book concedes the seed and the argument from silence in a single breath. The traditional theologian Chari, who would prefer everything eternal, concedes that the vyūha is later. The nationalist Majumdar, who defends Hindu antiquity at every turn, concedes that the cowherd Kṛṣṇa is late. Each admission is worth more for coming from the witness with the most reason to withhold it.

Why we were taught otherwise

If the evidence runs this way, why is the reverse the default? The answer is partly about the lenses through which the question has been asked, and naming them is the point.

The simplest distortion is the hymn count itself. To call Viṣṇu “minor” because he has few Ṛgvedic hymns quietly carries a second claim, that he was therefore unimportant. The first does not entail the second, and the scholars who know the text best say so.

The deeper distortion is older and uglier, and we met it earlier: the missionary scholarship of the nineteenth century, Wilson studying Hinduism to refute it, Weber deriving Kṛṣṇa devotion from Christianity, a framing built to make the tradition look late and borrowed. That theory was dismantled a century ago by Bhandarkar, Chanda, and Raychaudhuri, using the very inscriptions we have been reading, and even Jaiswal records that it is now abandoned. It survives in the popular afterglow, not in the scholarship.

The most respectable and most influential distortion is the modern one, and it is subtle because it is not a mistake so much as a choice. The dominant school of twentieth century Indian history, the materialist tradition of Kosambi, Sharma, and Jaiswal, reads religion as a reflex of social and economic life. Its instinct is to foreground what can be dated and attached to a social formation, so it trains its attention on the period when the cult was systematized, brahmanized, given temples and texts and a settled theology, the last centuries BCE and the first CE, and it leaves the undatable Vedic root in shadow. We saw the method at its purest in the etymology of bhakti, where devotion is traced to the sharing out of tribal food, and in the reading of every Vaiṣṇava element as an originally separate tribal or totemic cult, the boar a pig totem, Narasiṃha a dreadful Punjab god, Saṃkarṣaṇa a serpent of the furrows, Śrī a fertility goddess, the cowherd Kṛṣṇa an Ābhīra deity, all absorbed into a brahmanical synthesis. This is a powerful method, and it has produced real knowledge. But a method is a lens, and no lens is neutral. One built to illuminate the datable synthesis will show you the datable synthesis, and then the synthesis starts to look like an origin.

And underneath all three distortions lies the single confusion this essay began with. Systematization is not origin. The lateness of the system, the vyūha metaphysics, the fixed ten, the divine marriage, the schools, is not in dispute; even Chari, from the opposite extreme, grants it. What is in dispute is only whether the system is the same thing as the beginning. The frequency reading, the colonial frame, and the materialist lens all, in their different ways, answer yes. The evidence answers no.

What has to change

Let me be precise about the claim, because precision is the whole point, and because the careless version of this argument does real damage to the careful one.

Keep the scholarly date for the system. The four fold emanation, the canonical ten, the divine couple, the great temples of the Gupta age and after, all of that really is the work of roughly 200 BCE to 500 CE and later. No honest case requires moving it, and the careful gopī romance, the consort goddess, the Buddha in the ninth slot, the systematic theology of the descents, all of it is genuinely young.

What has to change is the slide from there to the impression that Viṣṇu and the devotion to him are correspondingly recent. They are not. The deity is Bronze Age, the organized worship is more than two thousand years old, and only the system is late. To date the religion from the age of its theology is to date the Nile from the year of its first dam. The river was already running, for a very long time, before anyone built the thing we can put a date on.

Return, at the end, to Heliodoros and his pillar. The Greek who knelt to Vāsudeva around 113 BCE was not present at a birth. He was a latecomer to something already ancient, systematic enough to have an ethics he could quote, confident enough to draw a convert from across the Hellenistic world. He could become a Bhāgavata because the tradition he joined had not begun in his century, or anywhere near it.

And the pillar is still there. It stands on the riverbank near Vidiśā, and the fishing communities along the water still smear it with vermilion and call it Khamba Baba, the pillar-saint, and pray to it for the birth of a son. A Greek ambassador’s column to “the god of gods,” two thousand years on, is still a place where Indians come to worship. That is the real span of the thing. Follow it back as far as the evidence runs and you do not reach a beginning two thousand years ago. You reach a Bronze Age poet singing of the god who paced out the world in three strides and kept a wellspring of honey in his highest heaven, the god whose footstep the wise have always watched, stretched out like the eye in the sky.

The tradition kept the threads, even when it forgot it was keeping them. When the Mahābhārata wanted to name the supreme God, it gave him a thousand names, and the litany of the thousand, still chanted every day in Vaiṣṇava homes, opens with two of them: viśvaṃ viṣṇur vaṣaṭkāraḥ. The universe. Viṣṇu. The ritual call. In the first breath of his own thousand named hymn the god is the all, the pervader, and the cry of the sacrifice. Those are the two things this essay has followed from the Bronze Age forward: the one who strides through everything, and the one who is the rite that holds everything together. Both names are older than the religion is usually said to be.

This is not a recent religion. It is one of the oldest objects of continuous human devotion that we can actually name and date, and it is long past time our standard account of it said so.

Sources and further reading

Primary texts

• Ṛgveda, trans. Stephanie Jamison and Joel Brereton. Especially 1.154, 1.155, and 1.22 (the three strides and the highest step), 7.99 and 7.100, the Vṛtra alliance verse 8.100.12, the Puruṣa Sūkta 10.90, and the riddle hymn 1.164 (the navel of existence).

• White Yajurveda (Vājasaneyi Saṃhitā), Sāmaveda, and Atharvaveda, trans. R. T. H. Griffith. For “Viṣṇu, that is, the sacrifice,” the Viṣṇu stride rite, the “loftiest place of Viṣṇu,” and the womb shaping charm (Atharvaveda 5.25).

• Śatapatha Brāhmaṇa, trans. Julius Eggeling. For “the sacrifice is Viṣṇu,” the dwarf, Nārāyaṇa, and the Puruṣamedha.

• Taittirīya Saṃhitā, trans. A. B. Keith. For the dwarf, the boar, and Prajāpati.

• Aitareya Brāhmaṇa. For “Agni is the lowest, Viṣṇu the highest.”

• Bhagavad Gītā, Winthrop Sargeant’s word by word edition. Chapters 4, 7, 9, 10, 11, and 18.

• Mahābhārata, trans. Bibek Debroy (unabridged). The Nārāyaṇīya and the Śvetadvīpa episode (Śānti Parva), the Śiśupāla scene (Sabhā Parva), the Nara-Nārāyaṇa formula, and the Viṣṇu-sahasranāma (Anuśāsana Parva).

• Bhāgavata Purāṇa. 5.26.31 (the hell of the human sacrificers) and 1.3.28 (kṛṣṇas tu bhagavān svayam).

• Inscriptions in D. C. Sircar, Select Inscriptions Bearing on Indian History and Civilization. The Besnagar (Heliodoros), Ghosundi, Nānāghāṭ, and Mora records.

Scholarship

• A. A. Macdonell, Vedic Mythology.

• Jan Gonda, Aspects of Early Viṣṇuism — the strides, the sacrifice, the navel and Padmanābha, the avatāras, kingship, Śrī, and the pervasiveness thesis.

• F. B. J. Kuiper, Ancient Indian Cosmogony.

• R. N. Dandekar, Vedic Mythological Tracts.

• R. G. Bhandarkar, Vaiṣṇavism, Śaivism and Minor Religious Systems — the three stream model, the bhakti philology, the Āḻvārs and Ācāryas from Rāmānuja to Caitanya, the loved Viṣṇu and feared Śiva, and the late making of Rādhā.

• Suvira Jaiswal, The Origin and Development of Vaiṣṇavism.

• Edwin Bryant (ed.), Krishna: A Sourcebook; and Bryant, The Quest for the Origins of Vedic Culture — the early Vāsudeva chain, the Bhāgavata reversal, and the Mitanni evidence.

• R. C. Majumdar and others, The History and Culture of the Indian People (the Vedic Age, the Age of Imperial Unity, the Classical Age, the Age of Imperial Kanauj); and Majumdar’s Ancient India.

• S. M. Srinivasa Chari, Vaiṣṇavism: Its Philosophy, Theology and Religious Discipline.

• Doris Srinivasan, Many Heads, Arms and Eyes.

• J. N. Banerjea, The Development of Hindu Iconography.

• Manasataramgini, “The roots of Vaiṣṇavam: a view from the numerology of Vedic texts.”

• Colonial framing (Part One): H. H. Wilson and A. Weber, as reported and refuted by Bhandarkar, R. P. Chanda, and H. C. Raychaudhuri, and summarized by Jaiswal in her opening chapter.

Sometime around 530 BCE, a Persian army crossed a river in Central Asia, the Araxes (probably the Syr Darya or the Amu Darya), to conquer a horse nomad people called the Massagetae. The Persians were led by Cyrus the Great, founder of the largest empire the world had yet seen, a man whose rule extended over Babylon, Lydia, and Media. The nomads were led by a widow named Tomyris.

Cyrus tricked the first wave of Massagetae nomads with a banquet of wine, a drink these nomads barely knew, and slaughtered them while they lay drunk. Among the captured was the queen’s son, who killed himself in shame. Tomyris sent Cyrus a message: “Your wine is a poison that makes men mad. You thirst for blood, Cyrus, and I swear by the Sun, the great god of my people, I will give you your fill of it.” Then she met him on the battlefield and destroyed his army. She found Cyrus’s corpse, cut off his head, and shoved it into a skin filled with human blood. She then told the Shahanshah, “Now, drink your fill.”

We’re entering a world of horse archers, golden art, and human sacrifice. Women ruled, men resisted; the Greeks called them Scythians, the Persians Saka, and the Indians Shaka, eventually absorbing them. They left no written records, so our knowledge comes from enemies, contemporaries, graves, and, more recently, genetics.

I. Origins: where the Scythians came from

The Scythians come from the Eurasian steppe, which stretches from the plains of Hungary across the Black Sea, the Volga, and the Caspian, through modern-day Kazakhstan up into the Altai, the largest temperate grassland on Earth, an ocean of grass with no trees to stop the wind or the horse.

In the classical imagination, they appear suddenly in the eighth and seventh centuries BCE, pouring out of the east, pushing aside an earlier people called the Cimmerians, raiding down into Assyria and Anatolia and the fringes of the Greek world. For a generation in the seventh century BCE they were a terror across the Near East, and they were considered to be among the wolves that helped take down the dying Assyrian Empire. Much of what we know about their origin story comes from Herodotus, the famous Greek historian who wrote about them around 430 BCE.

The Scythians’ own legend, as he recorded it, runs like this. The first man was called Targitaus, son of Zeus and a daughter of the river Borysthenes (the Dnipro). Targitaus had three sons: Lipoxais, Arpoxais, and Kolaxais. While they ruled, four golden objects fell from the sky:

• a plough and a yoke(together)

• a battle-axe

• a cup

When the two elder brothers approached the gold, it burst into flame. Only when the youngest, Kolaxais, came near did the fire cool, and so he gathered up the gold and became king. The scholar Bruce Lincoln (University of Chicago) reads the origin story like this:

• the plough and yoke are the tools of the farmer

• the axe is the weapon of the warrior

• the cup is the vessel of the priest

It is the same three-way division of society, into producer, fighter, and priest, that ran through every early Indo-European society from Rome to India.

There is a second origin tale, the one the Greeks who lived around the Black Sea preferred, likely because it put their own hero at the base of the myth. Heracles, wandering the empty north, fell asleep and woke to find his horses stolen. He found them in the keeping of a creature that was a beautiful woman above the waist and a serpent below. She would return the horses only if he lay with her. He complied, and she bore three sons. The youngest, Scythes, was the only one strong enough to string his father’s great bow and buckle his belt, and so the land and the line were his.

The Scythian genetics tell a very interesting story. For now, hold this fact. The Scythians were not a single tribe with a single homeland. “Scythian” is a label, like “Viking” or “Celt,” for a whole family of related horse peoples spread across thousands of miles, speaking Iranian dialects, sharing a way of life, a style of art, and sometimes a similar way of death. The Greeks met the western ones on the Black Sea. The Persians met the eastern ones, whom they called Saka.

The Persians knew them well, because they kept trying to conquer the Scythians and kept failing. Darius the Great carved a great inscription into a cliff at Behistun around 519 BCE listing the peoples he ruled, and among them are the Saka (likely the first appearance of the name in the historical record). Persian sources sort them into three kinds:

• the Saka Haumavarga, the “haoma-drinking” Saka, named for the sacred plant of the Iranian priests

• the Saka Tigraxauda, the “pointed-cap” Saka, instantly known by their tall conical hats

• the Saka Paradraya, the Saka “beyond the sea,” the western Scythians of the Black Sea steppe

Even this seemingly neat threefold division isn’t really doing justice to the reality, because the modern interpretation keeps getting it wrong (you can see this all over X when Scythians are discussed). “Saka” is not a nation name. It was the Persians’ catch-all term for the horse nomads beyond their frontier, the way every settled empire slaps a single word on the blur of tribes at its edge. There was no united Saka confederation and no single Saka people in the genetic sense. The Scythian/Saka label covered dozens of tribes and clans, who were sometimes at war with one another, spread over thousands of miles, related in language and way of life but ruled from no central city, and they most definitely did not spring from a single stock. When the genomes finally came in and got analysed, as we will see, they confirmed exactly this.

Darius learned the same lesson Cyrus had. Around 513 BCE he led a vast army across the Danube into Scythian territory to crush them once and for all. The Scythians simply refused to fight a pitched battle. They retreated, destroyed their own water supply, burned the grass behind them, and let the Persian army wander and starve across an empty land until Darius gave up and went home. It is one of the oldest recorded examples of a strategy that would later humble Napoleon and Hitler in the same part of the world.

You cannot conquer people who have nothing to defend and nowhere you can pin them down.

So where did they actually come from? For a long time, scholars argued between a western origin, on the Black Sea steppe, and an eastern one, deep in Siberia. The newest evidence, including a single richly buried Iron Age warrior from Chinge-Tey in the Tuva region of southern Siberia, leans east. That warrior is older than the famous frozen tombs of the Altai and far older than Herodotus’s Black Sea Scythians, and he is genetically more East Eurasian than any of them. The whole Scythian package as we now know it, the animal art, the kurgan burials, the mounted-archer way of life, seems to have taken shape in the eastern steppe and rolled west, not the other way round as some scholars suggested earlier. The riders Herodotus met on the edge of the Greek world were the far western tip of something that had begun thousands of miles away.

And they came to India, eventually, not as conquerors but as refugees. Far to the east, on the frontier of China, a confederation called the Xiongnu pressed on a people called the Yuezhi. The Yuezhi, pushed off their land, crashed into the Saka of the Ili valley and drove them south. The Saka, in turn, spilled through the mountain passes into Bactria and down into the Indus. It was a chain reaction where the last domino fell in India. The Greeks and Romans even renamed the lower Indus after them: Indo-Scythia, the land of the Saka. They did not choose India. They were pushed into it. And then, as we will see, India did to them what India does.

II. Hearth, sword, and smoke: Scythian religion

From what we know so far about the Scythians, it’s fair to assume that they built no temples or major religious monuments. To understand them, you have to understand a religion with no temples and few idols; it was a faith carried on horseback and written in fire, blood, and hemp smoke.

Herodotus depicts the whole Scythian pantheon in a single paragraph, naming each Scythian god by its nearest Greek equivalent. Edward Minns, whose great 1913 book Scythians and Greeks is still a foundation of the field, lays it out:

The following deities were common to all, Tabiti, Hestia, who was the principal object of their veneration, next to her Papaeus, Zeus, with Apia, Ge, husband and wife, after them Goetosyrus, Apollo, Argimpasa, Aphrodite Urania, and Ares. Thamimasadas, Poseidon, was peculiar to the Royal Scyths. They raised no statues, altars or temples to their gods, save to Ares alone.

When you look at who sits at the top, it’s not the typical sky father we see in other Indo-European societies, not the war-god, but Tabiti, the goddess of the hearth and the sacred fire. The Russian scholar Bessonova built her whole reconstruction of Scythian religion around this hierarchy. According to her, the pantheon is hearth-centered and deeply Iranian, a seven-god system arranged in tiers that mirrors the religion of the Persians across the mountains, with their supreme god and his six holy immortals. To swear by the royal hearth of the Scythians was the most binding oath a man could take, exactly as the Persians swore by the royal fire. These were Iranian people, cousins of the Medes and Persians, and their gods were Iranian gods wearing Greek names that Herodotus pinned on them so his readers could follow along.

Note: “Tabiti” is the Greek spelling of a Scythian word that linguists reconstruct as Tapatī, “the Burning One,” from an old Indo-Iranian root for heat and flame. It is the same root that gives Sanskrit tapati, “it burns,” and tapas, the Vedic word for the inner heat of ritual and ascetic power, and even the name of the Vedic figure Tapati. The fact that the words are cousins does not prove the Scythians thought about her exactly as the Vedic tradition thought about tapas.

Then there is the exception, the one god who got an altar. The Scythian war god, called Ares, was worshipped not as a statue but as a sword. In every district, the Scythians piled up an enormous mound of brushwood, three furlongs broad and long, and on top of it they planted an ancient iron blade. Herodotus describes the rite as follows (translation by A. D. Godley):

On this pile there is set for each people an ancient scimitar of iron, which is their image of Ares; to this scimitar they bring yearly sacrifice of sheep and goats and horses, offering to these symbols even more than they do to the other gods. Of all their enemies that they take alive, they sacrifice one man in every hundred, not according to their fashion of sacrificing sheep and goats, but differently. They pour wine on the men’s heads and cut their throats over a vessel; then they carry the blood up on to the pile of sticks and pour it on the scimitar.

A god you could hold in your hand, and feed with the blood of your enemies. The sword in the sacred mound is one of the most haunting images the ancient Scythian world has left us, and, as we will see at the very end of this piece, it did not die with the Scythians. It rode west into a legend we already know.

The Scythians also had a class of diviners that fascinated and unsettled the Greeks: the Enarei (priestesses and soothsayers), who were born male but lived as women, dressed as women, and were said to have received their second sight, and their changed nature, as a gift or a curse from the goddess of love. Herodotus put it down to a “feminine sickness” sent by Aphrodite in punishment for some ancestor’s looting of her temple. The Enarei look like one of the few documented institutions of a recognized third gender holding religious authority anywhere in the ancient world. They told the future by unwinding bundles of willow rods through their fingers, a method of divination that, remarkably, still survived among their descendants in the Caucasus thousands of years later.

And then there was death, which the Scythians did with grandeur. When a king died, his body was embalmed, drawn through his territory on a wagon so every subject and tribe could mourn him, and finally laid in a great chamber under the earth. Herodotus again tells what went into the grave with the king:

They strangle and lay in the vacant room within the tomb one of the dead man’s concubines, and his cupbearer, his cook, his groom, and his messenger and horses, and cups of gold. When they have done this they make a great mound, vying with each other to make it as great as possible. After the lapse of a year they take fifty of the king’s best attendants, and fifty of the finest horses, slay them, and stuff them with chaff. One of the fifty strangled youths is then put astride of each horse. So these horses are set in a circle about the tomb.

Note: the frozen Pazyryk tombs found in the Altai vindicated Herodotus’s account centuries later.

Let’s talk about one more Scythian rite, which is talked about quite often in popular discourse around Scythians and… cannabis. Herodotus reports that after a funeral, the Scythians purified themselves not with water but with smoke. They set up a small tent, heated stones inside it, threw hemp seed on the stones, and sat in the rising vapor. Godley’s translation says this:

The Scythians then take the seed of this hemp and, creeping under the rugs, they throw it on the red-hot stones; and, being so thrown, it smoulders and sends forth so much steam that no Greek vapour-bath could surpass it. The Scythians howl in joy for the vapour-bath.

For many centuries this was just a traveller’s tale, and academia did not believe most of what Herodotus wrote about them (however, he was vindicated again). In 1948, the Soviet archaeologist Sergei Rudenko opened the frozen tombs of Pazyryk in the Altai and found, beside the bodies, exactly what Herodotus had described: little six-legged pole frames for a tent, bronze censers, and inside them the charred seeds of Cannabis sativa. Rudenko quotes an ancient lexicon that the Scythian “smoking” was so strong it brought a sweat out of anyone who tried it, and he notes dryly that he found the apparatus buried with both a man and a woman, “since both men and women probably smoked.” Twenty-four centuries after Herodotus, the seeds were still in the burner.

III. The golden beasts: Scythian art

If the Scythians had given the world nothing but their art, they would still be unforgettable. Academia refers to their art style as the Animal Style.

Picture a stag with its legs folded beneath it and its antlers sweeping back along its spine in a single rhythmic line. A panther curled into a perfect circle, biting its own tail. A griffin sinking its beak into the haunch of a horse, an eagle tearing a hare, a tangle of predators and prey worked in gold and bronze on every object a nomad owned: belt plaques, sword scabbards, horse harness, drinking cups, the gorgeous fittings of a saddle. The Scythians decorated their world with animals locked in the moment before the kill. Rudenko, who knew this art better than almost anyone, was careful about what it meant. The obvious guess is that they were sacred tribal animals. He rejected it:

Neither the Greek historians nor the Chinese chroniclers make any reference to an animal cult among the Scyths. In sets of decoration from one outfit, that is, all worn by one horse, several different animals may be represented. This lends weight to our inference about the impossibility of the animals having a totemic significance.

It was not religion in the narrow sense. It was a way of seeing life, and the art represented that. The art and artifacts found so far show, in their way, a life lived among herds and hunters on an open plain, where everything was either predator or prey.

If the beasts were not totems (spirit beings or sacred objects), some scholars suspect they were a map of the universe. When you closely study their art, you notice the same arrangement again and again: birds of prey above, hooved grazers in the middle, fanged predators sinking into them from below. Some authors and Scythian experts have suggested that the art encodes their world into three tiers:

• a sky of raptors

• a middle earth of stags and horses

• an underworld of monsters, with the predator forever falling on the prey at the seam between the worlds

Please do note that this is just an interpretation, and a very speculative one, since the Scythians left no literature for us to reference.

But I must say it does rhyme with something we have already discussed above. The four golden objects that fell from the sky in the Targitaus myth, the plough and yoke, the axe, and the cup, split society into three. The animal art may split the cosmos into three the same way. The habit of seeing the world in threes runs deep in Iranian-speaking peoples, as it does across the wider Indo-European family.

The frozen tombs of Pazyryk were a significant discovery. The Siberian ice had preserved them like no other site could. Among the things the Russians uncovered were felt wall-hangings, one of them four and a half meters by six and a half, showing a goddess (likely Tabiti) enthroned with a rider before her, one of the few images of a Scythian deity we possess. There was the oldest knotted-pile carpet on Earth, worked at three thousand six hundred knots to the square decimeter, well over a million knots in all, bordered with rows of grazing stags and mounted men. Then there were the bodies themselves, embalmed with a surgical care that Rudenko describes: the organs removed, the cavity filled and sewn shut with horsehair thread, the skull packed with herbs.

And there were the tattoos. On the body of a chieftain, Rudenko found skin covered from shoulder to wrist in the Animal Style: a horned beast over the heart, a donkey, a winged creature, a mountain ram, a griffin, all inked into him. He thought they “most probably signified noble birth or was a mark of manhood, or perhaps both.” In the Pazyryk burials we find the Animal Style not on a belt buckle but on a human arm. This clearly indicates that the Scythians considered animal art part of their expression, not mere body decoration.

Then there is the gold. The Scythian kings were buried with a great deal of gold and wealth, and much of the finest work was made for them by Greek craftsmen along the Black Sea coast, who, as the scholar Artamonov put it, “seem to have been very willing to work for Scythian patrons,” being “perhaps too nervous of their behaviour, when angered, not to do so.” As you’ll see later, this Scytho-Greek connection lasts for a long time, and we see expressions of it in ancient India and Gandhara. My research on this topic indicates that, of all their contemporaries, the Greeks always had a very deep and intricate understanding of Scythian art and subjects. One such example is the great silver-gilt vase found at Chertomlyk. Here we see a very clear example of Greek depictions of Scythians breaking wild horses, dressed in their traditional clothing, typically long-haired, the older men bearded. On a gold comb from Solokha, three Scythian warriors fight to the death in exquisite miniature. Across the steppe, from the Kelermes panther curled on a shield to the later golden stags and pectorals dug from the kurgans, the same vision repeats.

Much of what we know about the Scythians came to light by accident and greed. At the end of the seventeenth century, Russian treasure hunters pushing east into Siberia, the bugrovshchiki, or mound-robbers, began breaking into the frozen kurgans of the Altai and pulling out the gold. So much of it flowed back west that in 1715 the governor of Siberia, Prince Gagarin, started shipping it to the Tsar. Peter the Great understood what it was, and in 1718 decreed that every ancient and unusual object dug from the ground be bought for the treasury and sent to St Petersburg, rather than melted down. The roughly 250 gold pieces that survived became the Siberian Collection of Peter the Great, the first archaeological collection in Russian history.

IV. The genetics

For most of history the Scythians were a riddle made of secondhand stories. In the last decade, ancient DNA has begun to answer this riddle.

In the Bronze Age, long before the Scythians, the eastern steppe was held by herding peoples we call Sintashta and Andronovo, descendants of the great westward and eastward expansion of the Yamnaya. They were overwhelmingly carriers of one male lineage, the Y-chromosome haplogroup R1a, and specifically its Asian branch, R1a-Z93. That lineage is the genetic thread that ties the Bronze Age steppe to the later Scythians.

Here comes the surprise. When the geneticist Gnecchi-Ruscone and colleagues sequenced one hundred and eleven individuals from the Kazakh steppe spanning fifteen hundred years, they found a turnover, not a continuity. The classic Scythians were a new blend, roughly 60% descended from that Bronze Age steppe stock and roughly 40% from an eastern Eurasian, Siberian-related source. The steppe had been remixed. And that eastern fraction was not even across the map. It rose as you moved east and it rose as time went on, so that the Scythians of the Altai carried far more East Asian ancestry than their cousins on the Black Sea, who looked more European. A study of Iron Age nomads by Unterländer and colleagues concluded that the eastern and western Scythians had arisen largely independently, in place, linked by a steady drift of genes from east to west, rather than spreading from a single homeland. “Scythian” really was a culture worn by a genetic mosaic.

The western half of that mosaic has now been read too. Andreeva and colleagues sequenced one hundred and thirty-one individuals from the classic Scythian heartland north of the Black Sea. The royal Scythians of the fourth and third centuries BCE turned out to be genetically tight and homogeneous, carrying close to half Yamnaya-derived ancestry, and overwhelmingly of the R1a male line, in some elite clans every single buried man sharing one paternal lineage. Their reconstructed faces, from the pigmentation genes, match Herodotus and the Greek vase painters: mostly brown or fair hair, a fair number with blue eyes. The same study even caught a medical fingerprint of their dominance, a gene variant for fructose intolerance that spread across western Eurasia in the centuries the Scythians ruled the steppe and now sits in the DNA of millions of Europeans.

At a Saka cemetery at Boz-Barmak in Kyrgyzstan, all the men sequenced shared a single Y-lineage, while every individual had a different maternal line, the signature of a society where men stayed put and wives came in from outside. The Pazyryk anthropology had hinted at exactly this decades earlier, when Rudenko found enormous physical variety inside single tombs and guessed at “obligatory exogamy.”

Two caveats before we move on, because this field is still moving fast. The very newest studies, the single Tuva warrior and the Boz-Barmak cemetery, are based on small samples, and one is still a preprint that has not finished peer review. And the headline lineage, R1a-Z93, is reported at coarse resolution in some of the Scythian papers, so treat “the Scythians were Z93” as a strong generalization rather than a claim about every individual.

V. The peoples they touched: Tocharians, Yuezhi, Greeks, and the road west

The Scythians who ranged from Hungary to Mongolia did not live alone. Over the centuries the Scythians and Sakas traded with, fought, and married into a dozen other peoples, and some of those entanglements get so badly confused in popular accounts (especially on X) that it’s worth taking them one at a time.

The Tocharians first, because the confusion there is nearly total. In the oasis towns of the Tarim Basin, in what is now western China, manuscripts survive in two extinct languages that scholars labeled Tocharian A and Tocharian B. Tocharian is Indo-European, but it is not Iranian, not related to the speech of the Sakas at all. It belongs to a separate and very early branch of the family, and strangely enough it patterns with the western Indo-European tongues like Latin and Celtic, even though it was spoken at the far eastern edge of the Indo-European world. Its ancestors apparently split off near the very beginning and were carried east thousands of years before any Scythian existed, probably with the Afanasievo migration around 3700 BCE.

So the Tocharians and the Sakas were different peoples speaking unrelated languages, separated by millennia. They just ended up as neighbors around the Tarim and Bactria, where they traded and intermarried like neighbors usually do. The famous Tarim Basin mummies make the muddle worse. Those strikingly preserved, sometimes red-haired bodies are forever being paraded as Indo-European, but their genomes say otherwise. A 2021 study found that the earliest Tarim people were a genetically isolated local population, roughly 70% descended from an ancient Ice Age Siberian lineage, with no detectable steppe ancestry whatsoever. They had picked up herding and farming from their neighbors without being replaced by them (yes, this happens, and there are many such examples). So the mummies, the Tocharian language, and the Iranian Sakas are three different things, however stubbornly the popular imagination keeps welding them into one.

The Yuezhi are the ones who matter most here. This was the confederation the Xiongnu drove west, which in turn drove the Sakas into India. But the Yuezhi kept going. One of their five tribes, the Guishuang, rose above the rest and built an empire of its own across Bactria and northern India. These were the Kushans, who would become one of the greatest royal patrons Buddhism ever had. We will come back to them in a bit.

To the southwest, the Sakas met the Greeks of Bactria, the far-flung heirs of Alexander (yawn), who had built a brilliant Hellenistic kingdom on the Oxus and then pushed into India. The Sakas did not destroy this world so much as marry into it. They kept its bilingual coinage and its Greek divine images, kept the court styles, and their kings sometimes carried Greek names. The numismatist Robert Senior showed that you often cannot tell a Saka king from a Greek one by his coins or his name. Further west sat the Parthians, the Iranian dynasty ruling Persia and Mesopotamia, and the Saka and Parthian worlds blurred together in the borderland of Sakastan, or Sistan.

The Scythians’ eastern cousins were eventually absorbed by the Turkic peoples who came to dominate the steppe. Their western relatives had a stranger fate. The Sarmatians, and after them the Alans, rode the grass highway straight into the heart of the late Roman world. Alans fought in Gaul, Iberia, and North Africa, and some scholars trace a contingent of steppe cavalry stationed in Roman Britain (still contested). They carried the culture of the steppe into Europe at the very moment Rome was coming apart. One fragment of them never left the mountains where they took shelter. High in the Caucasus there is a people who still speak an Iranian language descended from that of the Alans, who tell stories reaching back to the Scythians Herodotus described, and who in their own way have kept the old gods. We will meet them at the end. They are the closest thing to living Scythians left on earth.

VI. How Sakas became Hindus and Buddhists

Here the story turns. A people whose god was a sword planted in blood, whose kings rode into the earth with strangled retainers, came into India and within a few generations were endowing Buddhist monasteries and turning Greek gods into Hindu ones on their coins. How did that happen?

Partly because the Sakas were never zealots. They were pragmatists ruling over conquered peoples, and Indian religion offered them something their sword cult never could: a way to be legitimate kings of settled subjects.

The Sakas were not the only steppe people to come down the Bactrian road into India. Close behind them came their cousins the Kushans. The Sakas began the turn toward Indian religion; the Kushans, a little later, made it imperial policy.

The clearest place to watch the change is on the coins. Follow one figure across two hundred years of minting, as the scholar Frohlich did, and you can see a god change religions in three stages:

• Under the Greeks: he is Poseidon, a purely Greek sea-god holding a trident, an odd choice for a landlocked kingdom, but Greek through and through.

• Under the Sakas: the same trident-bearer is quietly localized. On Maues’s coins a small river-god appears at his feet, a detail with no Greek parallel; later issues add Iranian flame-markings on his shoulders. He is, in Frohlich’s words, “no more Poseidon as the Greeks conceived him, but not yet Shiva as the Kushans used to worship him.”

• Under the Kushans: he arrives. He stands beside a bull, the trident is now a trishula, and the coin names him Oesho. He is unmistakably Shiva.

No single king redrew him from scratch. He was reinterpreted coin after coin and reign after reign, a Greek form emptied of its Greek meaning and refilled, first with Iranian content, then with Hindu, until Poseidon’s trident had become Shiva’s trishula.

The Greek strongman Heracles went the same way, sliding toward two Indian fates at once. In Gandhara sculpture he survives as Vajrapani, the Buddha’s guardian, still carrying Heracles’ club, now turned into a thunderbolt. Elsewhere he shades into Balarama, elder brother of Lord Kṛṣṇa.

Rosenfield’s study of their art shows the Kushans kept no state religion at all, a radical thing in their world. Their greatest king, Kanishka, struck gold coins showing himself at a fire altar on one face and, on the other, a rotating cast of gods drawn from every people he ruled, each labeled in Greek letters: the Iranian sun Miiro and moon Mao, the goddess Nana, the fortune-goddess Ardoxsho, the Greek Herakles, the Hindu Oesho who is Shiva, and, most astonishing of all, the Buddha himself, labeled in Greek script BODDO. The earliest datable image of the Buddha on any coin was likely struck by a king descended from steppe nomads. Kanishka went further still. He convened a great Buddhist council, raised a stupa at Peshawar (in present-day Pakistan) that Chinese pilgrims called the tallest building in the world, and funded the missions that carried Buddhism up the Silk Road into China. A faith that began under a tree in Bihar became a world religion in the hands of the children of horse archers.

The royal titles tell the same story of a crown draped in every legitimacy at once. One inscription styles the king:

maharajasa rajatirajasa devaputrasa kaisarasa

Great King, in the Indian style. King of Kings, in the Iranian. Son of God, or Son of Heaven, calqued from the Chinese. And Caesar, from Rome. Four civilizations’ worth of kingship.

And it was not only the emperors. The Western Kshatrapas, the Saka satraps who ruled western India for nearly four centuries, were open-handed Buddhist donors from their first generation. The richest record survives at the Buddhist caves of Nashik, where around 105 CE a Saka nobleman named Ushavadata, son-in-law of the Saka king Nahapana, recorded gift after gift: thousands of gold coins to feed the monks, a cave cut and handed to the order, houses and gardens and water-tanks endowed in town after town along the trade roads, whole herds of cows given away. His wife, Nahapana’s own daughter, endowed a cave of her own. A Saka prince with a Persian name, two generations off the steppe, pouring his fortune into the Buddhist cause.

The precedent for all of it had been set a century earlier, by a Greek. At Besnagar in central India stands a stone pillar raised around 113 BCE by Heliodorus, envoy of the Indo-Greek king Antialcidas, who had become a devotee of Vasudeva, that is, Lord Kṛṣṇa. He called himself a Bhagavata, a follower of the Lord, and crowned his pillar with Viṣṇu’s eagle, Garuda. This pattern outlasted the Sakas. A small engraved seal from Gandhara, now in the British Museum, shows a royal worshipper of steppe stock kneeling before a four-armed Viṣṇu, mace and discus and lotus in his hands, the prayer cut in Bactrian script. Scholars argue over the king, the Kushan Huvishka, or a later Huna or Kidarite prince, but the image is the same either way: a descendant of horse nomads, bowing to Viṣṇu.

VII. What the Sakas left in India

The Sakas ruled parts of India for the better part of five centuries, longer than the Mughals, and the marks they left are deeper and stranger than most Indians realize, because some of them are things Indians now think of as simply, eternally Indian.

Their rule came in two main waves.

The northern Indo-Scythians, in Gandhara and the Punjab:

• Maues (around 120 BCE): the first Saka king in India, who took the grand title King of Kings.

• Azes (around 58 to 57 BCE): swept away the last Indo-Greek kingdoms and unified Gandhara. He even founded a dynastic era of his own, the Azes era, long used to date inscriptions across the northwest. Some earlier scholars linked the start of the Vikrama Samvat era to Azes; this was disproven later.

The western Sakas, the long-lived Western Kshatrapas, who ruled Gujarat, Malwa, and the Konkan from their capital at Ujjain for nearly 380 years:

• Bhumaka and Nahapana, of the Kshaharata line.

• then Chastana, and, greatest of all, Rudradaman I (around 150 CE).

That western Saka kingdom did not go unchallenged. Early in the second century CE the great Deccan power, the Satavahana emperor Gautamiputra Satakarni, fell on Nahapana and destroyed him. The proof survives in metal: a hoard of Nahapana’s silver coins, dug up at Jogalthembi, every one re-struck with Gautamiputra’s own stamp. The long inscription his mother raised at Nashik calls him the destroyer of the Shakas, Yavanas and Pahlavas, the king who uprooted the Kshaharata line and stopped the mixing of the four varnas. For a moment the foreign kings had been driven out.

It did not hold. Nahapana’s house was finished, but a second Saka line, the Kardamakas of Chastana, rose in its place, and under Chastana’s grandson Rudradaman the Western Kshatrapas came roaring back, this time beating the Satavahanas in turn. And it is Rudradaman who left India one of its most consequential gifts, though it is the last gift you would expect from a man the orthodox texts called a fallen barbarian.

On a great rock at Junagadh in Gujarat, the same rock that already bore an edict of the emperor Ashoka, Rudradaman had carved a long inscription recording how he repaired a broken irrigation dam, the Sudarshana lake, at his own expense, without burdening his subjects with new taxes or forced labor. The dam is not the point. The language is. Every major Indian royal inscription before Rudradaman, the edicts of Ashoka, the records of the Satavahanas, the dedications of the Indo-Greeks, had been written in Prakrit, the everyday vernacular. Rudradaman’s Junagadh inscription is the first long royal inscription in Sanskrit.

And then there is the calendar that did not stop. The Shaka era, beginning in 78 CE, now most often linked to the accession of the Saka ruler Chastana. The Shaka era spread across India and Southeast Asia, was used to date temples and dynasties for nearly two millennia, and in 1957 the Republic of India adopted it, as the Shaka calendar, for its official national reckoning. Open a government of India gazette and the date is given in the era of the Sakas. A people who came to India as refugees off the steppe gave the modern Indian state the very name of its year.

The Sakas also left art. Under their patronage and that of the Kushans, the Gandhara school fused Greek technique with Indian devotion to produce, for the first time, the Buddha in human form, the serene face that would be copied from Afghanistan to Japan. They left an administrative model, the satrap-as-viceroy system, that later dynasties borrowed. They left their coinage standard, so prized that when the Gupta emperor Chandragupta II finally conquered the last Western Kshatrapa around 400 CE, he kept minting coins in the loser’s design.

VIII. How the Mahabharata judged them

Indian literature did not ignore the Sakas. It wrestled with them, and the way it did so reveals something subtle about how Indian civilization handled outsiders, very differently from how the Greeks and other Indo-Europeans handled theirs.

In the Mahabharata, the Sakas (Shakas) appear again and again, almost always in the same company: the Yavanas (Greeks), the Kambojas, the Pahlavas (Parthians), the Tusharas, the Pulindas, the Kiratas. They are foreign warrior peoples on the edges of the known world. They turn up in the great catalogues of the Kaurava army, fighting in their thousands. They appear at Yudhishthira’s imperial sacrifice, bringing tribute of horses “that could travel great and long distances.” They are, in the epic’s geography, real and powerful peoples who must be reckoned with.

But the epic also explains what is wrong with them. The key passage comes in the Shanti Parva:

It is because there are no brahmanas among them that Shakas, Yavanas, Kambojas and other similar kshatriya tribes are regarded as the equals of vrishalas. It is because there are no brahmanas among them that Dramilas, Kalingas, Pulindas, Ushinaras, Kolas, Sarpas, Mahishakas and other similar kshatriya tribes are regarded as the equals of vrishalas.

The Shakas are called kshatriya tribes. Their fallen status, equal to a vrishala or shudra, is explicitly conditional: it holds “because there are no brahmanas among them,” because they have lost the priestly scaffolding of ritual, not because of anything innate in their blood. And, most tellingly of all, the Shakas and Yavanas are named in the very same breath as the Dramilas, the Tamils, and the Kalingas and Pulindas, peoples of India itself. This is not a wall built to keep foreigners out.

The epic even imagines these peoples being born inside the Brahmanical cosmos. In the story of the sage Vasishtha and his miraculous cow, when the cow is attacked she generates whole armies from her own body to defend her master, and among the warriors she produces are the Pahlavas, the Shakas, and the Yavanas, brought forth from her dung and her urine and her foam. It is hardly flattering, but notice what it does: it places the steppe and Mediterranean peoples inside the Indian sacred order, as warriors summoned to defend a brahmana.

There is a more famous degradation story, the one where King Sagara, advised by his guru, spares the Shakas, Yavanas, Kambojas, Paradas, and Pahlavas but punishes them by shaving their heads and stripping them of the Vedas. It is worth a careful note, because it is often misattributed: that tale lives in the Harivamsha and the Puranas, not in the main body of the Mahabharata.

Set this against how the Greeks saw the Scythians. To the Greek imagination, as the scholar Edith Hall has shown, the Scythian was the archetypal barbarian, the savage opposite against whom Greekness defined itself, milk-drinking, lawless, alien by essence. The Indian view was harsher in its language, but softer in its logic. The Shaka was a kshatriya who had lost his way and could, in principle, be brought back into the fold. And that, in the end, is precisely what happened: the Sakas married into Indian dynasties, endowed Indian temples and monasteries, wrote inscriptions in Sanskrit, and dissolved into the Indian population. As the historian Chattopadhyaya put it in a line that sums up the whole drama:

Ancient India has always captured her captors. The Sakas were also captured and completely Indianised.

IX. Where the children of the Scythians are now

So where did they all go? The answer comes in three parts: one solid, one contested, and one that is a calendar.

1. The solid part: the Ossetians. High in the Caucasus live the Ossetians, a people of around two and a half million who still speak an Iranian language. They descend from the Alans, who descended from the Sarmatians, who were the eastern cousins and successors of the Scythians. They are the closest thing to living Scythians on earth, and the proof is not only in their genes but in their stories. Their great oral epic, the Nart sagas, preserves across some twenty-five centuries the recognizable bones of what Herodotus reported. Georges Dumézil spent decades matching the two. Among the survivals:

• divination by bundles of rods, just as Herodotus describes the Scythian diviners

• the sacred blood-brotherhood oath, sworn over a shared cup of wine and blood

• the bow as the symbol of kingship

• a magical self-filling cup that rises to the lips of whoever tells the truth

• a forty-day mourning, and taboos that echo the Iranian Avesta

A reviewer of Dumézil called the Nart sagas “a treasure of epic narratives that confirms, illuminates, enriches the meagre ancient materials” we have from antiquity.

And the Narts give us the most beautiful ending this story could ask for. Their greatest hero is Batradz, a warrior with a body of tempered steel, forged and hardened in the sea by the smith of the gods. Batradz cannot die while his great sword remains in the world. At the end, dying, he commands his comrades to throw his sword into the sea. Twice they lie and say they have done it; twice he knows from the calm of the waters that they have not. Only when the sword finally goes in, and the sea turns to blood and storm, does Batradz die. If that rings a bell, it should. It is the death of King Arthur, Excalibur returned to the lake, a thousand years and three thousand miles from the Caucasus. And it is, scholars argue, the same image Herodotus saw on the steppe: the sword that is a god, the blade that holds a hero’s life. The Scythian Ares, the iron sword in the brushwood mound, rode west into the Caucasus as Batradz’s blade, and rode further west still into the legend of Camelot. The most famous sword in the European imagination may be a Scythian one.

2. The contested part: India. This is where it gets sensitive, so it is worth stating precisely what the evidence supports. Several martial communities of the northwest carry long traditions, many of them hardened in the colonial period, of Scythian or Saka descent:

• the Jats

• some Rajput clans

• the Ahirs

• the Gujjars

• and, even more loosely, the Pashtuns

The romance is understandable, since these are warrior communities in exactly the regions the Sakas ruled. But the genetics does not support a clean “we are the Scythians” story for any of them. The most careful study of Jat Y-chromosomes found at least nine distinct ancestral sources, with no single origin, Scythian or otherwise, accounting for them. The careful way to put it: some northwestern lineages may well carry a Saka thread, real but partial, woven into a population that is, like every Indian population, a deep and ancient mixture.

But the Sakas have far more Indian descendants than any caste genealogy could capture, and it has nothing to do with blood. They were absorbed. That is the whole point, the thing that makes the Indian story the opposite of the Scythian story in Europe: the Sakas did not stay a separate people who could be tracked. They married in, worshipped in, gave in, ruled in, and dissolved. Their genes scattered through the northwest in fragments too mixed to label.

3. The part that needs no DNA: the calendar. Every time the government of India prints an official date in the Shaka era, the era that still bears the Sakas' name, the Scythians are still here. They have no monuments that crowds visit, no nation that bears their name, no language anyone in India still speaks. But they left their fingerprint on the most ordinary thing of all, the count of the years, and a billion people carry it without knowing whose it is.

Sources and further reading

Classical sources

• Herodotus, Histories, Book IV (the Scythian account: origins, the seven gods, the sword of Ares IV.62, the Enarei, royal burials IV.71-72, the hemp purification IV.73-75) and Book I (Cyrus and Tomyris of the Massagetae, I.201-214).

• Hippocrates, Airs, Waters, Places (on the Enarei).

Scythian history, religion, and art

• Ellis H. Minns, Scythians and Greeks (Cambridge, 1913), the foundational survey; source of the pantheon and burial translations quoted here.

• Sergei I. Rudenko, Frozen Tombs of Siberia: The Pazyryk Burials of Iron Age Horsemen (trans. 1970), the Pazyryk excavations: tattoos, embalming, the hemp apparatus, the oldest pile carpet, the felt goddess.

• M. I. Artamonov, Treasures from Scythian Tombs (1969), the Hermitage gold and Greco-Scythian art.

• S. S. Bessonova, on the reconstruction of the Scythian pantheon (Iranian, hearth-centered).

• Edith Hall, Inventing the Barbarian (Oxford, 1989), the Greek construction of the Scythian as savage other.

Genetics

• Allentoft et al. (2015, Nature): Bronze Age Eurasia, Andronovo, R1a.

• Unterländer et al. (2017, Nature Communications): Iron Age nomads, east-west independence and asymmetry.

• Damgaard et al. (2018, Nature): 137 steppe genomes, Saka admixture.

• Krzewińska et al. (2018, Science Advances): Cimmerians, Scythians, Sarmatians; the Srubnaya source.

• Gnecchi-Ruscone et al. (2021, Science Advances): 111 genomes, the Bronze-to-Iron-Age turnover (about 60 percent steppe, 40 percent eastern Eurasian).

• Andreeva et al. (2025, Science Advances): 131 western Scythians, R1a dominance, phenotype.

• Rymbekova et al. (2025, preprint): Saka of Boz-Barmak, female exogamy (unreviewed, treat as provisional).

• Nedoluzhko et al. (2025, BMC Genomics): the Chinge-Tey warrior and the eastern-origin case.

• Zhang et al. (2021, Nature): the Tarim Basin mummies as a local Ancient North Eurasian population, not steppe migrants.

• Mahal & Matsoukas (2017, Frontiers in Genetics): Jat Y-chromosomes, nine ancestral origins, against any single-origin Scythian claim.

Sakas in India

• B. N. Shrava, The Sakas in India (1947); D. C. Sircar and S. Chattopadhyaya, The Sakas in India (1955): the migration, Indianization, and the “captured her captors” thesis.

• Robert Senior, Indo-Scythian Coins and History: Maues, Azes, the single-Azes revision, the Vikrama era.

• John Rosenfield, The Dynastic Arts of the Kushans (1967): Kanishka, the coin pantheon, religious pluralism.

• Frohlich, on the deities of Indo-Scythian coins (Poseidon to Shiva, Heracles to Vajrapani and Balarama).

• The Junagadh rock inscription of Rudradaman I (around 150 CE), ed. Kielhorn, Epigraphia Indica VIII; the Nashik cave inscriptions of Ushavadata; the Heliodorus pillar at Besnagar.

• The Mahabharata (Bibek Debroy translation): the Shanti Parva passage on the Shakas as fallen kshatriyas; the Vasishtha cow myth; the army and tribute catalogues.

Descendants and survivals

• Georges Dumézil, work on the Ossetian Nart sagas and Scythian continuity; V. I. Abaev on the Narts; on Batradz and the sword-in-the-sea, see the Encyclopaedia Iranica entry “Batradz.”

Around 8,000 BCE, two groups of farmers lived at opposite ends of the same mountain crescent and had practically nothing to do with each other. One sat in the hills of central Anatolia. The other lived over 1,000 km east, in the foothills of the Zagros. They grew wheat and barley, and they also herded goats and sheep, but genetically they were about as far apart as a Norwegian is from a Han Chinese.

When David Reich’s lab sequenced the first farmers of the Near East in 2016, the data backed that up. The early farmers of Anatolia and Zagros were, in the team’s words, strongly differentiated, each descended from its own local hunter-gatherers. Reich’s lab summarizes it by saying something like this: these populations were as different from one another as present-day Europeans are from East Asians.

So it’s fair to say the crescent held two separate beginnings, and they had their own paths. The western one ran into Europe. The eastern one ran into India. That crescent, Anatolia through Armenia into the Zagros, ended up behind both Europe and India. Same source, two roads, and they would not meet again for thousands of years.

Strangers at either end

The northern rim of the Fertile Crescent runs in a long arc of hill country, a few hundred kilometers wide, curving from western Anatolia around to the Zagros. Farming got invented along it more or less all at once. The people doing the inventing were not one population spreading one idea. They were locals, valley by valley, taming whatever grew and grazed in front of them. And some of those valleys were a very long way apart.

At the western tip of the arc sat Anatolian farmers, the people of Çatalhöyük with its bull horns and plastered skulls. At the eastern tip, in the Zagros, sat a separate farming people around sites like Ganj Dareh, where the goats were penned between the houses and the dead were buried under the floors. Same tools, same animals, two different worlds, and a thousand years of living a few hundred kilometers apart without blending in any meaningful way.

The difference went deeper than farming and crops. Both poles carried a deep streak of the same ancient lineage, one that geneticists and researchers call Basal Eurasian. In the Zagros it made up close to half their ancestry; in Anatolia it was less, but still a large share. It branched off near the very start of the human journey outside Africa, roughly 50,000 years ago, before the other non-African populations had even separated from one another. It also stayed clear of most of the Neanderthal interbreeding that swept the rest of that founding population around the same time, which is why it carried almost no Neanderthal DNA. So the oldest layer of ancestry in this story is a southern one, and it sits right here in the arc. Tens of thousands of years later, when the steppe finally had its turn around 4,000 BCE, it was building on a base that was already there.

Two peoples, then, sharing a crescent, a crop list, and one deep root far in the past, but already pulled apart on everything since. From here their paths fork, and stay forked for thousands of years.

The western road

Around 6,700 BCE the Anatolian farmers ran out of room, and they started moving.

They went two ways at once. One stream took to the water, moving across the Aegean island by island, Crete to the Cyclades to the Greek mainland, carrying seed grain and live animals in open boats across the short blue gaps. The other stream went overland, up out of Greece and into the Balkans along the river valleys, following the Vardar and the Struma into the heart of the continent. By 6,000 BCE the wheat had reached the Carpathian basin on the middle Danube.

For decades nobody could say whether these farmers were newcomers or just local foragers who had learned to plant. Ancient DNA settled it much later. The farmers who turned up in Europe were a different people, freshly arrived, whose closest genetic match was Neolithic Anatolia. They came from the arc.

Europe’s hunter-gatherers had held the continent for fourteen thousand years, and they had dark skin and pale blue eyes, a combination that has nearly vanished today. The men from La Braña in Spain and Loschbour in Luxembourg carried exactly that look. The incoming farmers were close to the reverse: they carried the strongest light-skin variant in West Eurasia at about 99 percent, and tended toward darker eyes. The modern European face did not exist yet. It had to be mixed, slowly, out of these two strangers meeting along the frontier.

They had one thing in common with the people they replaced. Neither could drink milk. The gene that lets adults digest fresh milk was rare in farmers and foragers alike, and it stayed rare for thousands of years after the first cows were milked, rising to high frequency only in the Bronze and Iron Ages. The herders came first; the milk drinkers came much later.

The frontier did not move smoothly. North of the Danube the soil turns heavy and cold and wet, nothing like the light Mediterranean ground the Anatolian toolkit was built for, and for a thousand years the farming wave simply stopped, pinned against the edge of the Hungarian plain. Then around 5,500 BCE a new culture cracked the heavy northern soil, and the frontier surged again, all the way to the Atlantic and the North Sea.

When the dust settled, Europe had been rebuilt from the southeast. The western pole of the arc was now the bedrock of the continent.

That bedrock did not hold. Over the millennia that followed, the first-farmer world of Europe came apart with real violence. Mass graves from the period hold dozens of people struck from behind and buried without their heads. At a cave in northern Spain, one farming group butchered, cooked, and ate at least eleven of their neighbors in a single episode around 3700 BCE, and the data shows the victims were locals, not strangers. Plague was moving through the same villages at the same time. The people who rebuilt Europe were themselves nearly erased before a new population arrived from the east. How that collapse happened, and how much of it was warfare, hunger, and disease, is a story for its own piece, and I will come back to it in the future.

The eastern road

The Zagros pole never went to Europe. It turned the other way.

From the foothills of western Iran, this lineage spread out across the Iranian plateau and pushed east, into the oasis country of Central Asia and eventually to the doorstep of the Indus. The great Bronze Age civilization of Central Asia, BMAC, sits on this road like a hinge, the turning point where the eastern arm of the arc swings down toward the subcontinent.

By the time it reached the edge of the Indus Valley, this Iranian-related ancestry had mixed with a deep, indigenous South Asian lineage that we call AASI, the oldest human stratum in India. The blend shows up across the borderlands of the Indus world, in the genomes from sites like Shahr-i-Sokhta and in the individual recovered from the Harappan cemetery at Rakhigarhi. That mixture, Iranian-related plus indigenous AASI, became the western pole of South Asian ancestry, and it is still there. The two biggest modern surveys, a study of more than 2,700 whole genomes in 2024 and the national GenomeIndia atlas in 2026, both find Iranian-farmer ancestry as the largest single western component across the subcontinent, from Punjab to Tamil Nadu.

The eastern road carries a twist the western one does not. The Iranian-related ancestry in South Asia broke away from the Iranian plateau lineage very early, before anyone in the region had planted a seed, before 8,000 BCE. I must admit that timing is awkward, and the field has not fully closed the case on it. If the ancestry runs deeper than farming, did it really spread with farmers at all, or with some earlier movement of hunter-gatherers we are only starting to see? In recent years, Narasimhan and the Reich group lean toward a farming model; some others read it as older. The one thing nobody disputes is that this western pole of India is not steppe, and it is not Anatolian. It was moving east, in whatever form, while the Anatolian farmers were still loading their boats.

The eastern pole turns up almost everywhere researchers look. Sequence a West Eurasian population from 9,000 BCE onward and a Zagros related component is usually somewhere in the mix. The Anatolian farmers have some. So do the Caucasus foragers, the steppe pastoralists, the people of the Indus, and living Iranians, Kurds, Baloch, Gujaratis, and Tamils. For a lineage that started in one stretch of foothills, it got around quite a bit.

Where the two roads meet

For thousands of years the two poles ran on separate tracks. The Anatolian side built Europe. The Iranian side built the foundation of India. They did not meet in the arc, and they did not meet in Europe or in India. They met on the grassland in between.

North of the Black and Caspian Seas, the Caucasus hunter-gatherers, who are near-twins of the Iranian pole, drifted up onto the Pontic-Caspian steppe and mixed with the Eastern European foragers already living there. The result of that mixture is the population everyone has heard of (especially if you are chronically online on X): the Yamnaya, the wagon-driving kurgan-building pastoralists of the steppe. By the Reich lab’s own decomposition the Yamnaya are roughly 40% Caucasus ancestry and 60% Eastern European forager. The internet famous steppe people are already close to half southern by the time they exist, around 4,000 BCE. Nearly half their ancestry came up from the Caucasus, the same highland south that produced the Iranian pole.

So the grassland is not a third source standing apart from the arc. It is the place the two roads crossed AGAIN, after a long detour, and produced something new. The steppe sits downstream of the crescent, not upstream of it.

Then the grassland boomed. Around 3,500 BCE, the Yamnaya expanded off the steppe in every direction, one of the largest and fastest population movements in European prehistory, pushing a fresh layer of steppe ancestry west into Europe and, more than a thousand years later, south through BMAC toward the Indus. Both halves of the arc’s scattered family got a second visit, this time from their reunited cousins riding wagons. So in both cases, Europe and India, the steppe was delivering a recombined version of ancestry the arc had supplied in the first place.

Recap

So let’s summarize what we learnt so far: A single bow of highland country invented farming. Its western tip moved into Europe and became the continent’s first farmers, the ancestry almost every modern European still carries. Its eastern tip flowed into Iran and on to the Indus and became the western half of nearly every South Asian. The two lines did not touch again until their descendants met on the steppe and braided together into the peoples of the Bronze Age.

So in many ways Europe and India are cousins, in the literal genetic sense, raised at opposite ends of the same crescent and sent off in opposite directions. Most of the deep map of who we are keeps running back to that one stretch of mountain country between Anatolia and the Zagros.

It comes down to two groups of farmers, nine thousand years ago, living at opposite ends of the arc. They never met. They spoke different languages and they didn’t look alike. But between them they carried the ancestry that would one day reach from Europe all the way to South Asia.

The genetics keeps pointing to the same thing. There was no single homeland that fanned out in every direction. There was an arc, and it ran two ways: west into Europe and east into India.

Call it ZAA. It is where both roads began.

Sources

• Lazaridis et al. (2016), “Genomic insights into the origin of farming in the ancient Near East,” Nature 536:419 to 424. The Anatolian and Zagros farmer poles strongly differentiated, each from local hunter-gatherers, as different from one another as present-day Europeans are from East Asians; the Zagros pole close to half Basal Eurasian, the Anatolian pole somewhat less.

• Haak et al. (2015) and Mathieson et al. (2015, Nature): early European farmers as a distinct Anatolian-derived population; pigmentation (SLC24A5) and the dark-skinned, pale-eyed hunter-gatherers.

• Gronenborn, Horejs, Börner and Ober (2023): mapped chronology of the farming expansion across Europe, the Aegean and Balkan routes, the Danube stall.

• Narasimhan et al. (2019, Science) and Shinde et al. (2019, Cell): the Indus Periphery cline, Iranian-related plus AASI ancestry, with the Iranian component diverging before 8,000 BCE; steppe ancestry reaching South Asia only after about 2000 BCE.

• Kerdoncuff et al. (2024) and GenomeIndia (2026): Iranian-farmer ancestry as the largest western component across modern South Asia.

• Lazaridis et al. (2022), “The genetic history of the Southern Arc,” Science: Yamnaya formed roughly forty percent Caucasus hunter-gatherer and sixty percent Eastern European hunter-gatherer.

• Curry (2025), Science: organized violence among Europe’s early farmers, including headless mass graves, and the late-Neolithic population collapse that preceded the steppe arrival.

• Saladié et al. (2025), “Evidence of neolithic cannibalism among farming communities at El Mirador cave, Sierra de Atapuerca, Spain,” Scientific Reports: at least eleven local individuals butchered and consumed in a single episode around 3700 BCE.

The standard story is that the Indo-European languages, the family that now includes English, Hindi, Persian, Greek, Russian, and Latin, came galloping out of the Pontic-Caspian steppe around 3,500 BCE on horseback. That story is half right, and the half it gets wrong is the most important half. The horsemen were real. The expansion was real. But they were the delivery truck, not the factory. The factory sat fifteen hundred kilometers to the south, in a belt of highland country that I will call the arc: Zagros, Armenia, Anatolia, the northern rim of the Fertile Crescent.

This essay makes the case for that arc from the ground up. I am going to lean hard on one rhetorical advantage along the way. The strongest single body of evidence for a southern homeland was produced by the people most committed to the steppe, the labs of David Reich, Iosif Lazaridis, and David Anthony. I do not share their conclusions. I do think their data convicts them. When a hostile witness keeps handing you receipts, you take the receipts.

Start where everything starts. With farming.

I. The first revolution happened in the arc

Around 9,500 BCE, in the hills of the northern Fertile Crescent, people stopped chasing food and started growing it. Wheat and barley were pulled out of wild grasses on those slopes. Lentils, peas, chickpeas, flax. Goats and sheep were folded into human life within a few centuries and a few hundred kilometers of each other, all of it inside the arc. Not one discovery in one place but a slow regional convergence across the whole crescent.

Before the grain was even fully tame, the people of the arc were building on a scale no one had attempted before. At Göbekli Tepe, in southeastern Anatolia, between roughly 9,600 and 8,000 BCE, hunter-gatherers who had barely begun to farm quarried limestone pillars weighing up to ten tons, carved them into great T-shapes, decorated them with foxes and snakes and vultures and scorpions, and stood them in rings. Then they buried them and built more. This is the oldest monumental sacred architecture on Earth, seven thousand years older than Stonehenge, older than pottery, older than the wheel, older than writing by six millennia, and older than agriculture itself. The temple came before the farm.

A few hundred kilometers and a thousand years away, at Çatalhöyük on the Konya plain, several thousand people packed into a honeycomb of mud-brick houses with no streets, entering through holes in the roofs and walking home across their neighbors’ rooftops. They set the horn cores of wild bulls into their walls. They modeled plaster breasts over the skulls of vultures and the jaws of foxes. They buried their dead under the sleeping platforms, and sometimes dug the skulls back up, packed the faces with plaster, and kept them in circulation among the living. Ian Hodder’s decades at the site have documented all of it.

By 7,000 BCE the Anatolian farmers were a dense, surplus-producing civilization of mud-brick towns, thousands of people each, trading obsidian and weaving cloth. A growing farming population does one thing reliably. It runs out of land.

II. The farmers moved, and we can trace the routes

Around 6,700 BCE they began leaving, and they went two ways.

One stream took the water. Aegean farmers moved west island by island, Crete and the Cyclades and the Greek mainland, carrying wheat and goats and pottery in small boats across the short blue gaps of the Aegean. The other stream walked, pushing up out of Greece into the Balkans along the river valleys, following the Vardar and the Struma into the heart of the continent. By 6,000 BCE farming had reached the Carpathian basin on the middle Danube.

There the wave stalled for a thousand years. North of the Danube the soil turns heavy, cold, and wet, nothing like the light dry Mediterranean hillsides the Aegean toolkit was built for. Southern farming did not work on it. So the frontier sat pinned against the edge of the Hungarian plain for ten centuries, until a culture called the Linearbandkeramik finally cracked the loess around 5,500 BCE and the frontier surged again, this time all the way to the Atlantic.

That pause matters because of who was waiting on the far side of it, and because of what ancient DNA later revealed about the meeting.

III. The first farmers were not the people already there

For decades, the spread of farming into Europe was an unsettled question. Did people move and carry farming with them, or did Europe’s old hunter-gatherers simply pick up the new techniques and stay put?

Ancient DNA ended the argument, and the answer was migration. When the first genomes came out of early European farmer skeletons, beginning with Haak and colleagues around 2010 and then Mathieson and colleagues with hundreds of ancient individuals in 2015, Europe’s first farmers turned out to be a distinct people whose closest match was the Neolithic population of Anatolia. They came from the arc, and they came in person.

The two populations did not even look alike. The hunter-gatherers had held Europe for fourteen thousand continuous years, and their genomes describe a face that has almost vanished today: dark skin with blue or pale eyes. The seven-thousand-year-old man from Loschbour and the pair from La Braña in Spain carried exactly that combination. The incoming farmers were close to the reverse, carrying the derived variant of the pigmentation gene SLC24A5, the single strongest light-skin allele in West Eurasia, at around ninety-nine percent, with a tendency toward lighter skin and darker eyes. The modern European face did not exist yet. It had to be mixed, over thousands of years, out of these two strangers meeting at the edge of the loess.

One detail is worth holding onto for later, because it warns against a common error. Neither population could digest milk as adults. The lactase-persistence variant was rare in farmers and foragers alike, and stayed rare for another four to five thousand years, rising only in the Bronze and Iron Ages. Even the Yamnaya pastoralists, the supposed milk-drinking supermen of the steppe, carried it at a few percent at most. Genes and culture do not move in lockstep, and the people we are about to follow onto the grassland were not yet the people their reputation describes.

So by 5,000 BCE the arc had already remade Europe once, in the bodies of real migrants, before any Indo-European word was spoken on any steppe.

IV. One arc, two farmers, as different as Europeans and East Asians

Here is the structural fact that the whole homeland debate turns on, and it comes straight from the Reich lab.

The arc was not one population. At the moment farming began it held two genetically distinct West Eurasian lineages, each of which made its own independent leap from foraging to farming at opposite ends of the same highland belt. Lazaridis and colleagues established this in 2016, in a Nature paper on the origin of farming in the Near East, and they did not hedge the language. The first farmers of the Levant and the first farmers of the Zagros were, in their words, “strongly genetically differentiated,” and each “descended from local hunter-gatherers.” How differentiated? The press summary spelled it out: these early Near Eastern populations were “as different from one another as present-day Europeans are from East Asians.”

Sit with that comparison. Inside one arc, before anyone moved anywhere, sat farming lineages as far apart as a Norwegian and a Han Chinese. The arc was not a single seed. It was a basin of deep diversity, and its two poles went on to seed two different halves of the world.

The western pole, the Anatolian Neolithic farmers, flooded into Europe as the Early European Farmers. When a geneticist says a modern Italian carries Anatolian farmer ancestry, this is the pole they mean.

The eastern pole, the Iran Neolithic farmers of the Zagros, around sites like Ganj Dareh, never went to Europe. They went east and south, into the Iranian plateau, into the great Central Asian Bronze Age civilization called the BMAC, onto the edges of the Indus Valley, and through those into the deep ancestry of nearly every modern South Asian.

Lazaridis 2016 even drew the arrows itself: farmers related to Anatolia “spread westward into Europe,” farmers related to Iran “spread northward into the Eurasian steppe,” and people related to both the Iranian farmers and the steppe pastoralists “spread eastward into South Asia.” That single sentence, written by the steppe camp’s own flagship lab, is the two-pole thesis in full. Europe’s western ancestry is the Anatolian pole. South Asia’s western ancestry is the Iranian pole. The two only fuse much later, and much further north, on the grassland. Which tells you what the grassland actually was.

There is a deeper layer under both poles. Lazaridis 2016 found that each of them drew roughly half its ancestry from a ghost lineage called Basal Eurasian, a population that split off before all other non-Africans diverged from one another and that carried little or no Neanderthal ancestry. Basal Eurasian sits in the south, in West Asia, not on any steppe. The deepest root of the West Eurasian family tree points down into the arc.

V. The language is too old for the steppe

The genes are half the case. Indo-European is fundamentally a language problem: one mother tongue, spoken somewhere at some time, whose daughters are now spoken by nearly half the human race. The reigning answer locates her on the Pontic-Caspian grassland around 4,000 to 3,500 BCE among horse-riding, wagon-driving pastoralists. The linguistics has never sat comfortably in that chair, and the person who has done the most to say so is Paul Heggarty.

In 2023, Heggarty and a large international team published in Science the most rigorous Bayesian dating of the Indo-European family yet attempted: 161 languages, 109 living and 52 ancient, scored on a carefully vetted database of core vocabulary, run through a statistical engine that estimates a family’s age from the rate its words diverge. The root came out at roughly 8,120 years before present, around 6,100 BCE. That is two thousand years too old for a homeland that only got moving on the steppe around 4,000 BCE. A root that deep lands squarely in the Neolithic of the arc, at the time and place farming and its vocabulary were spreading outward.

Heggarty’s model is a hybrid. It grants that the steppe was a major secondary staging ground for the European branches. It locates the primary homeland, the place of the deepest splits, in the highlands south of the Caucasus. This is not a fringe position dreamed up to annoy Harvard. It revives a homeland that the linguists Gamkrelidze and Ivanov argued for on purely linguistic grounds back in the 1980s, well before anyone could read ancient DNA.

The method had a sharp side benefit worth naming, because it reframes how you read the oldest texts. Heggarty’s “sampled ancestors” approach shows that Vedic Sanskrit is a sibling of the modern Indic languages, not their ancestor, just as Classical Latin is a sibling of Romance rather than its parent. The diversification of these families is older than their oldest surviving scriptures. The Rigveda is not the beginning. It is one preserved branch of something already old.

And the model gets one thing right that the pure steppe story cannot. Anatolian.

VI. Anatolian breaks the steppe-only model, and Lazaridis admits it

Anatolian, the branch that includes Hittite, is the oldest and most divergent limb of Indo-European, the first to split from everything else. If the family was born on the steppe, then Hittite, spoken in the heart of Anatolia, should trace back to steppe migrants. It does not, and the people who proved it did not want to.

The flagship of the steppe hypothesis right now is Lazaridis and colleagues, 2025, in Nature, “The genetic origin of the Indo-Europeans.” It is a serious and impressive paper: 428 ancient individuals, the densest genomic transect of steppe prehistory ever built, with David Anthony among the authors. It dates the formation of the core Yamnaya to 4,038 BCE, traces an explosive Yamnaya expansion between roughly 3,750 and 3,350 BCE that fanned five thousand kilometers from Hungary to Kazakhstan with almost no mixing into the locals it rolled over, and proposes that the deepest unity, “Proto-Indo-Anatolian,” was spoken between 4,400 and 4,000 BCE by people on a steppe-edge gradient it names the Caucasus Lower Volga, or CLV, cline.

Read at speed, that looks like the steppe homeland confirmed. Read closely, it is a retreat, and Heggarty said so in a Nature commentary in February 2025 with a title that did not pull punches: a beating of the retreat from the steppe hypothesis. Three things give the game away, and all three are in Lazaridis’s own data.

First, the direction of flow. The paper presents the CLV gradient as a neutral cline, as if ancestry sloshed evenly along it. But the southern, Caucasus-and-Zagros ancestry is densest at the southern end, around Aknashen near complete and Maikop close behind, and thins as you move north onto the steppe. That is the fingerprint of a population spreading from the south into the grassland, a founder effect radiating northward. The deep ancestry that built the steppe Indo-Europeans was itself moving up out of the arc. As Heggarty put it, the key movement heads north from the Caucasus, not south from the Volga.

Second, Anatolian. Lazaridis concedes in print that Bronze Age central Anatolians, the actual speakers of Hittite, carry only “at least a tenth,” about ten percent, of CLV steppe-related ancestry. He writes that “Anatolian is remarkable for its lack of steppe ancestry down to the Bronze Age.” The original steppe hypothesis claimed all branches, Anatolian included, came from the steppe. If the Hittites are ninety percent non-steppe, the oldest branch of the family did not come from the steppe, and once one branch is conceded, the burden on every other branch shifts. The 2022 Southern Arc papers from the same lab say it even more plainly: Anatolia before 3,000 BCE “derived almost entirely from local West Asian ancestry, with no significant steppe contribution.”

Third, the relabeling. Faced with an Anatolian branch that will not cooperate, the paper redraws the tree so that “Indo-European” now technically excludes Anatolian, which is folded into a deeper node called “Indo-Anatolian.” This lets you keep saying the Indo-Europeans came from the steppe while the inconvenient branch quietly exits through a door marked with a different word. The shot missed, so the goalposts moved.

There is fresh corroboration from this year, and I want to state it precisely because its headline is about something else. In April 2026, Akbari and the Reich lab published in Nature the largest ancient-DNA study ever assembled, 15,836 West Eurasian genomes, ten thousand of them new, a single paper that doubled the entire published record. Its subject is natural selection, not language homelands. But when the team traced a specific immune variant that rose sharply in Yamnaya-related populations, they found the selection had not begun on the steppe. It began earlier, south of the Caucasus, in people who carried no steppe ancestry at all. Even the biology we file under “steppe” keeps resolving, on inspection, to a source further south.

VII. The steppe was a mixing floor, and it was half-southern

Push one level deeper and the cradle-versus-engine line gets impossible for the steppe partisans to hold, because the Yamnaya themselves were, by ancestry, substantially southern.

Take the lab’s own decomposition. At formation, Yamnaya is roughly forty percent Caucasus hunter-gatherer and sixty percent Eastern European hunter-gatherer (Lazaridis and colleagues, the 2022 Southern Arc study, 727 individuals). The Caucasus component, CHG, is a near-twin of the Iran Neolithic, the eastern pole of our arc. So before Yamnaya even exists around 4,000 BCE, something close to forty percent of it is already southern, already drawn up out of the highlands. Layer in the CLV model from the 2025 paper, where Yamnaya is downstream of that Caucasus-Lower-Volga population, and the steppe looks less like a wellspring and more like a basin where two arc-derived streams and a northern forager stream met, blended, and invented a mobile pastoral economy.

The grassland supplied real things, and I am not waving them away: the horse, the wagon, the open road, and a demographic detonation that reshaped Europe and reached India. What it did not supply was the deep ancestry or, on Heggarty’s dating, the parent language. Both of those came from the arc. The steppe is where the family caught fire. It is not where the spark was struck.

VIII. The receipt from home: zero steppe in the Indus Valley

Turn to the eastern arm of the arc and the case stops being an argument about ancient Russians. It becomes a direct genetic receipt out of the Indian subcontinent.

In 2019, Shinde, Narasimhan, and colleagues published in Cell the first genome ever recovered from the Indus Valley Civilization. The individual, a woman from the cemetery at Rakhigarhi in Haryana, lived at the civilization’s mature peak, around 2,800 to 2,300 BCE. The paper’s title states the result without ornament: an ancient Harappan genome lacking ancestry from steppe pastoralists or Iranian farmers. She carried about half Iran-related Neolithic ancestry and about a tenth AASI, the deep indigenous South Asian hunter-gatherer lineage. Of steppe ancestry she carried, at the height of the Indus cities, statistically zero, ruled out at better than two standard deviations. The Yamnaya were galloping across Eurasia while she lived, and not one detectable thread of them had reached her.

The Iranian ancestry she did carry points the same direction the rest of this essay points. It is not the kind of Iranian ancestry that would have spread alongside Anatolian farmers. It is an early-diverging Iranian-plateau lineage that split off before farming was even invented, before 8,000 BCE. The western pole of South Asian ancestry does not trace to Anatolia and does not trace to the steppe. It traces to the Zagros Neolithic, the eastern arm of the arc, and to a branch of it so old it predates the farm.

The steppe ancestry did eventually arrive in South Asia. The same Narasimhan study, 523 ancient genomes, shows it: Central Asian steppe ancestry is absent before 2000 BCE, reaches the region around 2000 to 1500 BCE, after the Indus cities had already declined, and enters with a strong male bias, an estimated two to three steppe-carrying men for every woman, riding the R1a lineages that the later record tracks. It came late, it came partial, and it came as men marrying into a resident population whose foundation was already Iranian and AASI.

This is where the modern data closes the loop, and the closing is the part I find hardest to argue with. Three independent mega-studies, built on completely different samples, all reach the same verdict. Narasimhan 2019 with 523 ancient genomes. Kerdoncuff and colleagues in 2024 with 2,762 modern whole genomes, who put Iranian-farmer ancestry across the subcontinent at roughly thirty-five to seventy-five percent and steppe at a smaller, male-biased ten to forty percent. And the GenomeIndia atlas of 2026, nearly ten thousand whole genomes across more than eighty populations, where the R1a frequency tracks the western, steppe-derived component through the male line while the maternal lineages stay overwhelmingly local. Every dataset, ancient and modern, agrees: the largest western component in Indian ancestry is the Iranian pole, and the steppe is a late, male-mediated minority laid on top of it.

Then the line I keep returning to, because for me it settles the question of who is central. Open any West Eurasian genome from 9,000 BCE forward and Iran Neolithic ancestry is inside it. It is in the Anatolian farmers. It is in the Caucasus hunter-gatherers. It is in the Yamnaya, through the CHG. It is in the Indus periphery, in every modern Iranian and Kurd and Baloch and Gujarati and Brahmin and Tamil. The one population on Earth that carried Iran_N ancestry before anyone else carried any was the Zagros Neolithic population itself, because it was Iran_N. You cannot stand further upstream than being the source everyone else drank from.

IX. The cradle of the gods

The genetic and linguistic case stands on its own. But it would miss the part that makes the arc feel like a place rather than a smear of ancestry on a map. The arc was also where the religious imagination of half the world took its familiar shapes, and many of those shapes can be traced moving outward from it.

The deep-time evidence is the architecture itself. Göbekli Tepe’s rings of carved pillars, raised before the harvest. Çatalhöyük’s houses full of bull horns and plastered ancestor skulls and a goddess seated between two leopards. A jar from Hajji Firuz Tepe in the northern Zagros, dated to about 5,400 to 5,000 BCE, that Patrick McGovern’s chemical analysis showed had held resinated grape wine, the oldest such evidence in the Near East. The record has since been pushed back a few centuries by jars from Georgia in the South Caucasus, which is to say still inside the arc. Wine, the bull, the goddess, the ancestor cult, the sacred enclosure, all of it assembled in the arc thousands of years before any Indo-European word reached the steppe.

The traveling part needs care, because some of it is rock-solid and some of it is reaching, and I will tell you which is which.

The solid core is the comparative philology. Calvert Watkins spent a career showing that one mytheme, a hero or storm god slaying a serpent, recurs across the Indo-European world in phrasing so precisely parallel it must descend from a single Proto-Indo-European original. Indra kills Vritra in the Rigveda. The Hittite storm god Tarhunt kills the serpent Illuyanka, inside the arc. Apollo kills Python, Thor fights the world-serpent. The single sharpest receipt is that the slayer’s very name reconstructs across the Indo-Iranian split: Vedic Vritrahan, “Vritra-slayer,” is the exact cognate of Avestan Verethragna. The same god, killing the same dragon, in two languages that had only just parted. And the structure surfaces just outside the family, in the Bronze Age West Asian world the arc was embedded in: Baal kills Yam the sea in the Ugaritic texts, Marduk splits Tiamat in the Babylonian creation epic, and the Hebrew Bible keeps Yahweh defeating Leviathan and Rahab in Isaiah, Job, and the Psalms.

A second structure runs even deeper. Bruce Lincoln reconstructed a Proto-Indo-European creation myth in which the first man, Manu, sacrifices his twin, Yemo, and a primal ox, and the cosmos is built from the pieces. The names survive across the family: Yemo gives Vedic Yama, Iranian Yima, Norse Ymir, and Roman Remus, while Manu gives Vedic Manu, Germanic Mannus, and the Tuisto of Tacitus. The homology survives too, the dismembered body mapped onto the world, flesh to earth, bone to stone, blood to water, eye to sun. You can read it in the Rigveda’s Purusha hymn, where the cosmic man is divided to make the social order, and in the Norse Grimnismal, where Ymir’s body becomes the land. One creation myth, told in a dozen daughter tongues.

A third pattern is the vocabulary of the holy itself. The luminous radiance that clothes gods and kings has a recognizable cognate-cluster across the whole zone: Hebrew kavod, Akkadian melammu, Vedic tejas, Iranian khvarenah, and Greek doxa, which is the word the Septuagint chose to render kavod, carrying the idea straight into Christianity. And the chaos-water at the opening of Genesis, tehom, the deep over which the spirit hovers, is the same word as Babylonian Tiamat, the sea-serpent that Marduk slays. The deep that the Hebrew creator tames is, etymologically, the dragon his neighbors killed.

The most consequential transmission runs through Iran. Bodily resurrection, heaven and hell as moral destinations, a named adversary called Satan, ranked armies of angels, the apocalypse, a coming savior: this furniture is thin in the oldest layers of the Hebrew Bible and dense in Iranian religion that predates the long Persian contact with the Jewish exiles. The mainstream reading, argued by Bruce Lincoln among others, is that this eschatology entered Western religion during the Persian period, out of the Iranian plateau, the eastern arm of the arc. The geography is exact. The cradle of farming and language was also the source of the Western afterlife.

There is even a clean fossil of the Indo-Iranian split inside this religious record. The Sanskrit asura and the Iranian ahura are the same word, and so are Vedic deva and Iranian daeva, but the two traditions reversed their values: where India’s devas are gods and some asuras turn demonic, Iran’s ahura is the supreme good and the daevas become demons. We can even watch the reform happen in stone. Xerxes left an inscription boasting that “where previously the daivas were worshipped, there I worshipped Ahuramazda.” One people, one vocabulary, split near the arc, and one side flipped the polarity of heaven.

Now the honest accounting, because a strong case names its own seams. The dragon-slayer formula, the Vritrahan-Verethragna cognate, the Manu-Yemo cosmogony, the kavod-melammu-tejas cluster, and the tehom-Tiamat identity are established comparative scholarship. The Iranian origin of Western eschatology is mainstream, debated at the edges. The larger claim, that all of this is one “religious kit” exported from a single cradle, is my synthesis, not a settled finding, and a careful scholar could pull on any individual thread. I offer the synthesis as a pattern worth taking seriously. The individual threads I offer as receipts. What I will stand on flatly is the geography. The first temple, the first ancestor cult, the first wine, the dragon-slaying storm god, the radiance-word, and the inherited afterlife all sit in the same highland crescent, in the same span of millennia, upstream of everyone.

X. The honest objections, and why they do not sink the case

A case that only lists its own evidence is advertising, not argument. Here are the strongest objections, stated at full strength.

The steppe authors themselves read their data the other way. This is the most important one to be straight about. Shinde and Narasimhan do not think Rakhigarhi supports a southern homeland. They read the zero-steppe result as proof that steppe ancestry, and with it Indo-Aryan language, arrived later, after 2000 BCE, exactly as a steppe migration into India would predict. Their data is genuinely useful to the arc thesis on the question of the deep eastern pole. Their conclusions are not on my side, and pretending otherwise would be dishonest. What I am claiming is narrower and, I think, sturdier: the deep foundational ancestry of both Europe and South Asia is arc-derived, and the language root, on the best current linguistics, is older and more southern than the steppe model allows.

The wheel and the wagon. The single best card the steppe model holds is linguistic. The reconstructed Proto-Indo-European vocabulary includes solid words for wheel, axle, and wagon, and wheeled vehicles do not appear archaeologically until around 3,500 BCE. If the proto-language already had those words, it cannot be much older than the wheel, which collides head-on with Heggarty’s 8,120-year date. There are answers, that the wheel terms belong to a later, post-Anatolian layer spreading through an already-dispersing dialect network, and that linguistic paleontology is shakier than its fans admit, but I will not pretend the tension is resolved. It is the real soft spot in the early-and-southern dating, and an honest reader should hold it open.

No steppe artifacts in India. Critics note that the material culture of the steppe, Andronovo metalwork and pottery, does not show up in South Asia, which sits awkwardly beside a claimed genetic migration. The standard reply is that language and genes can spread through elite dominance without a full artifact package, and the male-biased ancestry pattern fits that. Both halves of this are worth keeping in view.

The symmetry problem cuts at everyone. Heggarty’s sharpest jab is that Indic speakers carry meaningful steppe ancestry while Iranic speakers carry very little, despite sharing a recent common Indo-Iranian ancestor, which is hard to square with a single steppe expansion driving both languages. Fair. But the mirror objection is also fair: if steppe migration fails to explain South Asia, why does it work so cleanly for Europe, where Corded Ware populations are something like three-quarters steppe? The honest position is that the steppe expansion was demographically real and enormous, and that being real and enormous is not the same as being the cradle. An engine can be the largest thing in the story and still not be the source.

None of these objections touches the spine of the case. Two divergent farmer poles in the arc, as different as Europeans from East Asians. Europe built from the western pole, South Asia from the eastern pole, the two fusing only later on the steppe. A language root too old for the grassland. An Anatolian branch the steppe model cannot keep. Iran Neolithic ancestry upstream of every West Eurasian on the planet. The objections nibble the edges. The center holds.

XI. The arc

Picture the whole thing at once. A crescent of highland country curving from the Aegean coast of Anatolia, through the Armenian heights, down the spine of the Zagros to the edge of the Iranian plateau. Eleven thousand years ago the people of that crescent invented farming at both of its ends, in two lineages so distinct they were as different from each other as a European is from an East Asian. They raised the first temples while they were still half wild, packed themselves into towns of bull horns and plastered skulls, pressed the first wine. The western pole walked and sailed into Europe and rebuilt it from the soil up. The eastern pole poured down into Iran and across to the Indus and laid the foundation of India. The deepest layers of their language went with them, the first branch never even leaving home, the rest drifting north onto the grass basin between the seas, where they met the horse and the wagon, fused into the Yamnaya, and detonated outward carrying daughter tongues that had already begun to part in the south.

The steppe was the engine. It was loud and fast and it changed the map of the world, and the scholars who study it are studying something real. But the cradle was the arc. The wheat came from there, and the word for father, and the storm god who kills the serpent, and the radiance on the face of the holy, and a very large share of the genome you are reading this with. The most striking part is that you do not have to take a heterodox scholar’s word for any of the genetics. The Reich lab measured the two poles and called them as far apart as Europeans and East Asians. Lazaridis drew the arrows from the arc into Europe and India. Anthony put his name on the paper that makes Yamnaya a downstream population of a Caucasus source. They were trying to map the steppe homeland. They mapped the road out of the arc instead.

The homeland was never the steppe. It was always, quietly, upstream.

Sources and further reading

Farming origins and the Neolithic arc

• Excavation and chronology of Göbekli Tepe (Klaus Schmidt and successors), Çatalhöyük (Ian Hodder and the Çatalhöyük Research Project), Jericho, and Ganj Dareh.

• Patrick McGovern, chemical residue analysis of the Hajji Firuz Tepe wine jars, University of Pennsylvania Museum: earliest Near Eastern grape wine, about 5,400 to 5,000 BCE, later superseded by South Caucasus finds near 6,000 BCE.

The two poles and the genetics of the first farmers

• Lazaridis et al. (2016), “Genomic insights into the origin of farming in the ancient Near East,” Nature 536:419 to 424. The Levant and Iran farmer poles “strongly genetically differentiated,” each from local hunter-gatherers, and “as different from one another as present-day Europeans are from East Asians.” Both poles roughly half Basal Eurasian. The directional sentence: Anatolian farmers west into Europe, Iranian farmers north into the steppe, both plus steppe east into South Asia.

• Haak et al. (2010, 2015) and Mathieson et al. (2015, Nature): early European farmer genomes, the migration into Europe, SLC24A5 and pigmentation, and the late rise of lactase persistence.

Indo-European homeland, the two models

• Lazaridis et al. (2025), “The genetic origin of the Indo-Europeans,” Nature 639:132 to 142. 428 ancient genomes (David Anthony among the authors); Yamnaya formation about 4,038 BCE; the CLV cline; Proto-Indo-Anatolian 4,400 to 4,000 BCE; Anatolian only about ten percent steppe ancestry.

• Lazaridis et al. (2022), “The genetic history of the Southern Arc,” Science, 727 individuals: Yamnaya formed about forty percent CHG and sixty percent EHG; Anatolia before 3,000 BCE “almost entirely from local West Asian ancestry, with no significant steppe contribution.”

• Heggarty et al. (2023), “Language trees with sampled ancestors support a hybrid model for the origin of Indo-European languages,” Science 381. 161 languages; root about 8,120 years before present; primary homeland south of the Caucasus; Vedic and Latin as siblings, not ancestors, of their living families.

• Heggarty (2025), “Beating the retreat from the Steppe hypothesis,” Nature commentary on Lazaridis 2025.

• Akbari et al. (2026), “Ancient DNA reveals pervasive directional selection across West Eurasia,” Nature. 15,836 genomes; directional selection on a Yamnaya-associated immune variant beginning south of the Caucasus, before steppe ancestry.

• Background, the steppe case at full strength: David W. Anthony, The Horse, the Wheel, and Language (2007), the standard statement of the wheel-and-wagon argument that any southern-and-early dating must answer.

South Asia

• Shinde et al. (2019, Cell 179:729): the Rakhigarhi Harappan genome, about half Iran Neolithic and a tenth AASI, statistically zero steppe at the IVC peak; the Iranian component diverging before 8,000 BCE. The authors read the result as consistent with a later steppe arrival.

• Narasimhan et al. (2019, Science 365): the formation of populations in South and Central Asia; steppe ancestry absent before 2000 BCE, arriving roughly 2000 to 1500 BCE with a two-to-three male bias.

• Kerdoncuff et al. (2024): 2,762 whole genomes; Iranian-farmer ancestry roughly thirty-five to seventy-five percent across the cline, steppe a smaller male-biased minority.

• GenomeIndia consortium (2026): nearly 10,000 whole genomes, more than 80 populations; the western steppe-derived component male-biased via R1a while maternal lineages stay local.

Religion and comparative myth

• Calvert Watkins, How to Kill a Dragon (Oxford, 1995): the Proto-Indo-European dragon-slayer formula; the Vritrahan and Verethragna cognate.

• Bruce Lincoln, Myth, Cosmos, and Society and Death, War, and Sacrifice: the Manu and Yemo creation myth and its cognates; later work on Iranian religion and the transmission of eschatology.

• Benjamin Sommer, The Bodies of God and the World of Ancient Israel (Cambridge, 2009): divine fluidity and embodiment in the Bronze Age West Asian world.

• Mark S. Smith, The Early History of God and The Origins of Biblical Monotheism: the Northwest Semitic and Israelite background, including the kavod and melammu radiance vocabulary.

A note on epistemic standing. The genetic and linguistic claims in sections I through VIII rest on peer-reviewed primary literature, much of it produced by scholars who favor the steppe hypothesis and whose data I am reading against their own conclusions. The religious-transmission argument in section IX rests on solid individual receipts (the dragon-slayer formula, the Vritrahan-Verethragna cognate, the Manu-Yemo cosmogony, the kavod cluster, the tehom-Tiamat identity, the Iranian sources of Western eschatology) but the claim that they form a single exported “religious kit” is an interpretive synthesis, offered as a pattern and not a proof.