Devas and Asuras

In the third book of the Rigveda there is a hymn of twenty-two verses, and every one of them ends on the same line. The hymn itself wanders. It speaks of the dawns that came before, of a calf with two mothers, of Vishnu guarding the highest pasture, of a king who is one and the same yet borne in many places at once. It is among the most obscure hymns in the collection, and scholars have argued for a century about what it is describing. But at the close of each verse, twenty-two times, the refrain comes back like a bell struck at the end of a corridor.

उषसः पूर्वा अध यद्व्यूषुर्महद्वि जज्ञे अक्षरं पदे गोः । व्रता देवानामुप नु प्रभूषन्महद्देवानामसुरत्वमेकम् ॥

uṣasaḥ pūrvā adha yad vyūṣur mahad vi jajñe akṣaram pade goḥ vratā devānām upa nu prabhūṣan mahad devānām asuratvam ekam

“When the earlier dawns broke, the great imperishable was born in the track of the cow. The observances of the gods are upheld. Great and one is the asura-hood of the gods.”

Ṛgveda 3.55.1 #

Read the last clause again. Asuratvam, asura-ness, belongs to the gods. It is what the hymn praises them for. Not a quality of their enemies, and no kind of slur: the very substance of divine power, named twenty-two times as though it were the one thing about the gods worth repeating.

Three thousand years later, in ordinary usage across every Indian language, asura means demon.

Something happened to that word. Following it is one of the more revealing exercises the tradition offers, because it does not resolve into a story about a good side and a bad one. It ends somewhere stranger, and then it walks straight into a live political argument that is going on in India this year.

The word before the war

The tradition’s own great reader of the Rigveda does not flinch at that refrain. Sayana, working at Vijayanagara in the fourteenth century, produced the commentary that remains the orthodox gloss on the hymns, and on this line, as Wilson reports him, he explains asuratvam as the abstract noun formed from asura, and asura itself as prabala: strong, powerful. The abstract, he writes, is prabalyam or aishvaryam, might or lordship. And ekam, one, he takes as mukhyam, chief.

Sayana lived at least two thousand years after the hymn was composed, in a world where asura had long since come to mean demon. He read it as sovereignty anyway.

The usage is not confined to one obscure hymn. In the fifth book, a verse addressing Rudra names the god twice over.

तमु ष्टुहि यः स्विषुः सुधन्वा यो विश्वस्य क्षयति भेषजस्य । यक्ष्वा महे सौमनसाय रुद्रं नमोभिर्देवमसुरं दुवस्य ॥

tam u ṣṭuhi yaḥ sviṣuḥ sudhanvā yo viśvasya kṣayati bheṣajasya yakṣvā mahe saumanasāya rudraṃ namobhir devam asuraṃ duvasya

“Praise him of the sure arrow and the strong bow, who rules over every healing herb. Sacrifice to Rudra for great favour. Honour with obeisance the deva, the asura.”

Ṛgveda 5.42.11 #

Devam asuram. Two words in the accusative, side by side, describing one being. Horace Hayman Wilson, whose translation of the fifth mandala appeared in 1857 and who was working through Sayana rather than around him, rendered it “adore the powerful divinity with prostrations.” That is not independent testimony, since Wilson is transmitting Sayana. It does show that the reading survived intact into the first English Rigveda, before any of the arguments that would later attach themselves to the word.

Varuna in particular is asura again and again, and the Indian Vedicist R. N. Dandekar built an entire study around that fact, “Asura Varuna,” published in the Annals of the Bhandarkar Oriental Research Institute in 1939. The god of cosmic order, the keeper of rita, carries the title that would later name his enemies.

Where the word actually turned

The clean version of this story has the Rigveda honorific and everything after it hostile. The clean version is wrong, and the correction matters, because it removes the temptation to look for a single dramatic event.

Wash Edward Hale’s Asura in Early Vedic Religion, published by Motilal Banarsidass in 1986 and still the standing monograph, works through the Vedic occurrences and finds the change already beginning inside the Rigveda itself. The singular asura stays honorific to the end. The plural, asurah, drifts toward naming a hostile class, and the drift is weighted toward the later books. There is no clean line to draw.

The layer where the opposition actually hardens is the one most popular accounts skip. The Atharvaveda, the fourth of the Vedas, already uses the word pejoratively. Then the Brahmanas, the great ritual prose texts of roughly the early first millennium BCE, become saturated with devasura combat myths: the gods and the asuras contend for this world and that, and the gods win by knowing some particular ritual technique the asuras did not know. The Shatapatha Brahmana returns to the pattern dozens of times. It is in that literature, over centuries, that two words become two armies.

Sukumari Bhattacharji’s The Indian Theogony, published by Cambridge in 1970 and subtitled a comparative study of Indian mythology from the Vedas to the Puranas, follows exactly this kind of movement across the whole span, which is the scale at which the change becomes visible at all. A single text will not show it. Only the sequence will.

So when the popular explanation arrives, it is arriving very late.

Almost every reader who has heard asura explained has heard it as a-sura, “not a god”, a negation. In the forty-fifth canto of the Ramayana’s Bala Kanda, Vishvamitra tells Rama the story of the churning of the ocean. Among the things that rise from the churned sea is Varuni, goddess and personification of sura, meaning strong liquor. The sons of Aditi accept her. The sons of Diti refuse. And so, says the text, the sons of Aditi became suras and the sons of Diti became asuras.

It is a pun about drink. Desiraju Hanumanta Rao, annotating the passage, quotes the Sanskritist Pandit Satya Vrat calling it a case of etymology by sound-resemblance, guided by similarity of sound “more than anything else”. The joke did not cause the shift; the shift was complete centuries before this passage was composed. What the joke supplied was a memorable back-story for a change that had already happened, and it has been repeated as philology ever since.

The philology says something else. There is no negative prefix in asura. The Indian grammatical tradition offers several derivations, including one from asu, breath or vital force, which the Brahmanas and Puranas favour, and Yaska’s Nirukta already records the a-privative reading as a folk etymology. Comparative reconstruction points elsewhere, to an Indo-Iranian word meaning lord, the same word as Avestan ahura and, on the usual account, cognate with Hittite hassu, king. Hale’s own conclusion is that asura meant something like “lord”. The question is not fully closed, and it does not need to be for the load-bearing point to hold: whichever derivation is right, nothing in the word negates anything.

The clincher is that sura, meaning god, is a later coinage. Monier-Williams says so in as many words in his dictionary of 1899: sura is formed from asura as though from a-sura, and it first appears in the Upanishads and the epics. The feminine sura, liquor, is old and Rigvedic. The masculine sura, god, is not. You cannot negate a word that did not yet exist.

Two families, one father

By the time of the Upanishads the two camps are fixed, and the way the oldest of them opens the story repays attention. The Brihadaranyaka Upanishad begins its third section like this:

द्वया ह प्राजापत्या देवाश्चासुराश्च । ततः कानीयसा एव देवा ज्यायसा असुरास्त एषु लोकेष्वस्पर्धन्त ।

dvayā ha prājāpatyā devāś cāsurāś ca. tataḥ kānīyasā eva devā jyāyasā asurās ta eṣu lokeṣv aspardhanta

“Two were the offspring of Prajapati, the devas and the asuras. Of these the devas were the fewer, the asuras the greater; and they contended over these worlds.”

Bṛhadāraṇyaka Upaniṣad 1.3.1 #

They have one father. Prajapati, the lord of creatures, generates both lines, and the word for what they then do is aspardhanta, from a root meaning to compete, the ordinary word for a rivalry. Two branches of one family, quarrelling over an inheritance.

The word the Upanishad uses for the asuras is jyayasa, and it is genuinely ambiguous. It can mean greater, more numerous, or elder. Shankara, as Swami Madhavananda renders him, takes it as more numerous and supplies a reason: the asuras are the ones absorbed in thoughts and actions directed to visible ends. The seniority reading has to come from elsewhere, and it does, in a place worth noticing. In the twelfth book of the Mahabharata, after the war is over and the ground is covered with the dead, Vyasa is trying to talk Yudhishthira out of his despair, and reaches for a precedent.

इदं च श्रूयते पार्थ युद्धे देवासुरे पुरा । असुरा भ्रातरो ज्येष्ठा देवाश्चापि यवीयसः ॥

idaṃ ca śrūyate pārtha yuddhe devāsure purā asurā bhrātaro jyeṣṭhā devāś cāpi yavīyasaḥ

“This too is heard, Partha, of the ancient war of devas and asuras: the asuras were the elder brothers, and the devas the younger.”

Mahābhārata 12.34.13 · Critical Edition; the vulgate numbers Śānti Parva differently #

The Puranic genealogies fill in the family, and they belong to a later layer than the Upanishad’s Prajapati, which is worth saying rather than smoothing over. The sage Kashyapa marries the daughters of Daksha. Diti bears the daityas; Danu bears the danavas; Aditi bears the adityas, who are the gods. One father, sisters for mothers. The Shiva Purana makes Diti the eldest of Daksha’s daughters, though the standard genealogies in the Mahabharata’s first book and the Vishnu Purana enumerate Aditi first and say nothing about relative age.

This is not the architecture of a war between good and evil. It is the architecture of a succession dispute.

The line that brahmins crossed, and what that proves

If the deva-asura conflict were a matter of species, or of blood, nobody could change sides. People do.

Three verses after the line about the elder brothers, still in the same speech, Vyasa continues the story of that ancient war. Brahmins who had mastered the Vedas, he says, “deluded by pride”, attached themselves to the danavas and fought for them. They became known through the three worlds as shalavrikas, literally house-wolves, a word that also does duty for jackal and wild dog. Eighty-eight thousand of them were killed by the gods.

Veda-knowing brahmins, the text’s own highest human category, went over to the asura side. Whatever else the passage shows, it shows that the boundary could be crossed.

It is worth being careful about what it does not show. The very next verse says that those who promote adharma must be slain by the gods as the fierce daityas were, which is a moral argument rather than a categorical one, and the brahmins in the story are annihilated for crossing rather than vindicated by it. A reader could fairly take the passage as boundary enforcement: a warning about the price of stepping out of alignment. That reading has force. What it cannot do is make the camp a function of birth, because in the story birth is exactly what failed to hold these men in place.

There is a stranger witness still, from a text nobody would call soft on hierarchy. The Manusmriti, enumerating the three states of being, places the daityas in the first and highest, the sattvika, in the same breath as ascetics, hermits and brahmins (12.48, in Ganganatha Jha’s rendering). The lawbook that is most often quoted for its rigidity puts the demons in the good column.

The traffic runs the other way as well, and more famously. Vibhishana walks out of Lanka and takes refuge with Rama, and is crowned king in his brother’s place. Prahlada, son of the greatest asura enemy the Vaishnava tradition knows, is a devotee of Vishnu from childhood. If the line were a bloodline these stories would be impossible. They are among the tradition’s favourites.

The demon who could not be talked out of God

Take Prahlada seriously, because he is the case that breaks every comfortable category.

His father, Hiranyakashipu, has extracted a boon of near-invulnerability through tapas, austerity so severe the worlds shake, and this is itself worth pausing on. The asuras get their power by the same means the sages do. Austerity generates force, and the force is issued to whoever generates it, with no regard for whose side he is on. The cosmos of the Puranas runs on law rather than on partiality. That is precisely what makes the asura dangerous: he has played by the rules and won.

He is also a monster, and the Bhagavata Purana does not spare the reader. When his son will not stop chanting the name of the god he has declared an enemy, Hiranyakashipu has the boy poisoned, thrown from a cliff, trampled by elephants, bitten by serpents, starved and burned. The child is five years old. Nothing in the reading offered here should soften that, and the tradition never asks it to be softened.

What it does ask is that the reader notice who is speaking the great devotional verses of the seventh canto. They are in the mouth of a boy of the demon race.

The same canto contains a passage stranger still. Narada, asked how the enemies of God so often achieve union with him, replies that the mind fixed on God through hatred reaches him as surely as the mind fixed on him through love. The text lists dvesha, enmity, alongside bhakti, devotion, among the states that hold the divine constantly in mind, and it has a word for what Hiranyakashipu practised: vairanubandha, the bond of hostility. Hatred, sustained without interruption for a lifetime, is a form of unbroken attention. Attention is what the path requires.

This is a genuinely difficult doctrine, and it should not be smoothed over. It says the asura who spends his existence obsessed with God is closer to him than the lukewarm devotee thinking about nothing much at all.

It is also not unique to India. Islamic mysticism produced a strikingly similar reading of Iblis, the angel or jinn who worshipped God before the fall: in Hallaj, in Ahmad Ghazali, in Attar, Iblis is the truest monotheist, damned for refusing to bow to anything but God, and his exile is the proof of his love. Two traditions that share almost nothing else arrived at the same uncomfortable thought.

The killing that the texts call a grace

Which brings us to the moment Hiranyakashipu dies.

Vishnu takes the form of Narasimha, neither man nor beast, on a threshold that is neither inside nor out, at a twilight that is neither day nor night, and tears the asura apart with claws that are no weapon. Every clause of the boon is honoured, and the boon fails anyway. It is one of the great scenes in Indian literature, and the avatara doctrine turns on it.

What happens immediately afterwards is the part that matters here. Vishnu’s own attendants speak.

सोऽयं ते विधिकर ईश विप्रशप्तस्तस्येदं निधनमनुग्रहाय विद्मः ॥

so ‘yaṃ te vidhikara īśa vipra-śaptas tasyedaṃ nidhanam anugrahāya vidmaḥ

“He was your own servant, O Lord, cursed by the brahmins. This killing of his, we understand, was for his grace.”

Bhāgavata Purāṇa 7.8.56 · second hemistich #

Nidhanam anugrahaya. The death was a favour. In the frame the Bhagavata builds around it, Hiranyakashipu is one of Vishnu’s two doorkeepers, serving out a brahmin’s curse in demon bodies, and the claws that open him are the door home.

The pattern recurs, with limits worth naming. Bali, the asura emperor whom Vishnu dispossesses with three steps as the dwarf Vamana, is spared and installed as ruler of the netherworld, and the Bhagavata promises him the office of Indra in a coming age. Vrikasura, granted the power to kill anyone whose head he touches and immediately trying it on the god who granted it, is destroyed by an outright trick: Vishnu arrives as a young brahmin, tells him Shiva is not to be trusted, and suggests he test the boon on his own head. The texts are not squeamish about divine deception when it suits them, and it would be dishonest to pretend the asura always falls purely by his own flaw.

Nor does the frame extend to everyone. The doorkeeper story covers one pair of beings across three births. It says nothing about Vritra, Namuci, Shumbha and Nishumbha, or Kalanemi, who are simply enemies and are simply killed.

The hard line

An account that stopped here would be tidy and would be dishonest, because there is a text in the tradition that overrides all of this in authority, and it takes the opposite view.

The sixteenth chapter of the Bhagavad Gita divides dispositions into the daivi sampad, the divine endowment, and the asuri sampad, the demoniac one, and describes the second without mercy. At 16.19 Krishna says that the hateful and cruel, vilest of men, he hurls ajasram into demoniac wombs.

Every major commentator renders ajasram the same way. Shankara glosses it santatam, “for ever”. Ramanuja’s bhashya has him hurled “ever” into the current of existence, and the Shrivaishnava word-for-word gives ajasram as “continuously”. Sridhara has him kept “incessantly revolving in samsara”. There is no dispute here to exploit: the word means unceasingly, and what it attaches to is the hurling, birth after demoniac birth.

Whether anyone ever gets out is a separate question, and on that the commentators do not go as far as they are often reported to. Ramanuja, three verses later, writes that the man freed of lust, anger and greed comes to the supreme goal. The doctrine of a class of souls permanently unfit for release belongs to Madhva, whose Dvaita system sorts souls into those fit for liberation, those who revolve for ever, and the tamo-yogya, fit only for darkness. Madhva did not comment on 16.19. His hard line comes from his system, not from this verse.

This is the honest state of the record. The Puranic narratives keep dissolving the boundary, and the Gita keeps insisting on it, and both are scripture. Any account that reports only the first has chosen a side and should say so.

Cousins across the mountains

There is one more fact about the word, and it needs care, because it has been fought over for two centuries.

The Avestan language of ancient Iran has ahura, which is the same word as Sanskrit asura by regular sound correspondence, and its supreme god is Ahura Mazda, the Wise Lord. Avestan also has daeva, the same word as Sanskrit deva. In the later Avesta the daevas are small wicked spirits who disturb the order of the world, and the Videvdad, at 10.9 and 19.43, names Indra, Sauruua and Nanghaithya at the head of the list, immediately after Angra Mainyu. Those three are recognisably the Vedic Indra, Sharva and the Nasatyas. Somewhere in the reign of Xerxes I, between 486 and 465 BCE, the Achaemenid king left an inscription at Persepolis recording that he destroyed a daivadana, a sanctuary of the daivas, and forbade their worship. It is the only datable anchor anyone has in this argument, and scholars have disagreed for a century about which sanctuary and whose gods.

The cognation is uncontested. The rest requires more care than it usually gets.

Kellens and Herrenschmidt, writing “Daiva” for the Encyclopaedia Iranica in 1993, point out that in the Gathas, the oldest Zoroastrian stratum, the daevas have not yet become demons: following Benveniste, they are genuine gods who have been rejected. The demonisation belongs to the Younger Avesta. To speak flatly of “Zoroastrian scripture” is to flatten a development, which is the same error as flattening the Rigveda into the Puranas.

The nineteenth-century explanation, that a religious schism split a single Indo-Iranian people who then demonised each other’s gods, is associated with the German Indologist and Iranist Martin Haug. The Encyclopaedia Iranica calls the notion out of date in a single line. Its own objection, that the Rigvedic use of asura is “so unsystematic and inconsistent that it can hardly be said to confirm the existence of a category of gods opposed to the devas”, is aimed at a related hypothesis rather than at Haug directly, but it cuts the same way. And Thomas Burrow, in “The Proto-Indoaryans” in the Journal of the Royal Asiatic Society for 1973, showed that the demonisation of the asuras happened so late that the vocabulary cannot be an inherited Indo-Iranian feature at all. If Burrow is right, the Iranian material cannot explain the Sanskrit shift, because the Sanskrit shift had not happened yet when the two traditions parted.

Nor is the mirror as neat as it is usually drawn. Kellens and Herrenschmidt think it probable that there was no Iranian word daiva meaning “god” at all for anyone to have inverted, and whether ahura ever named a class in Iran rather than one god and a companion or two remains disputed, turning on the awkward cases of Mithra and Apam Napat. Two traditions each innovated, in different directions, over centuries. Neither is the baseline from which the other deviated, and anyone who presents this as a clean inversion is selling a conclusion.

One thing the Iranian material does establish, and it cuts against the argument this article has been building. Zoroastrianism inherited the same vocabulary and the same cosmology and produced Angra Mainyu, a genuine adversary principle in opposition to the good. The shared inheritance clearly did not prevent a Devil. Whatever kept one out of Hindu thought, it was not the raw material.

The asuras have descendants who are voting

There is a reason this word cannot be discussed as a purely antiquarian matter, and it is that a large number of Indians alive today claim the asuras as ancestors.

The reading begins in 1873, with Jyotirao Phule’s Gulamgiri. Phule, working in Pune, inverted the Vamana story: Bali, or Baliraja, is the just indigenous king of a prosperous people, and Vamana’s three steps are conquest. Kerala’s Onam festival, which mourns Bali’s dispossession and celebrates his annual return to a people who still remember his reign as a golden age, had preserved something of that sympathy for centuries without the political frame. The frame turned it into an argument.

The argument grew. Periyar in Tamil Nadu read Ravana the same way. Since 2011, Mahishasura Jayanti or Mahishasur Shahadat Divas has been observed at Jawaharlal Nehru University and then well beyond it, mourning Mahisha as a martyred indigenous king and reading Durga’s spear as conquest. The Asur community, an Adivasi people of Jharkhand, Chhattisgarh and West Bengal, claim descent from Mahishasura and observe Durga Puja as a period of mourning. The city of Mysuru carries his name and a state-erected statue of him.

Scholars dispute the historical claim, and the case that the deva-asura myths encode a memory of conquered indigenous peoples is far weaker than its popularity suggests; the colonial-era race theory that first suggested it is discredited on its own terms. But the political fact does not depend on the historical one. For millions of readers the asura is not a monster in a story. He is a grandfather.

An article that celebrated the porousness of the boundary while saying nothing about the people standing on the far side of it would be missing the largest thing in the room.

What wears the shape of what

Return to Vyasa, standing on the field at Kurukshetra with a king who cannot stop grieving.

Seven verses after telling Yudhishthira that the asuras were the elder brothers, and three verses after the eighty-eight thousand brahmins who died fighting on their side, he says this:

अधर्मरूपो धर्मो हि कश्चिदस्ति नराधिप । धर्मश्चाधर्मरूपोऽस्ति तच्च ज्ञेयं विपश्चिता ॥

adharmarūpo dharmo hi kaś cid asti narādhipa dharmaś cādharmarūpo ‘sti tac ca jñeyaṃ vipaścitā

“There is a dharma that wears the shape of adharma, O ruler of men, and an adharma that wears the shape of dharma. The discerning must know this.”

Mahābhārata 12.34.20 · Critical Edition #

The text has just spent a dozen verses establishing that the two camps are brothers and that learned men crossed between them in both directions. Then it says the categories cannot be read off the surface. The sequence is an argument, and it is deliberate.

This is why Hindu thought had so little use for a Devil. A cosmos organised around a single adversary works best when the line between the sides holds still. The Sanskrit narrative texts kept refusing to let it: the beings on the wrong side are family, the boundary is crossed in both directions, and their defeat is frequently their release. Where the tradition locates the more serious obstacle, it locates it inside, in avidya, ignorance of what one is, in the veiling work of maya, and in the inner enemies of desire, anger and greed. Sri Aurobindo made that move explicit. In The Secret of the Veda he read the two mothers as two conditions of awareness: “Aditi the infinite, the undivided, is the mother of the Gods, Diti or Danu, the division, the separative consciousness the mother of the Titans.” In his letters on yoga he kept the asuras as real beings while placing them on a map of consciousness rather than of history, on the mentalised vital plane, defined by an egoistic strength that refuses a higher law. Read that way the ancient war is not over, and it is not somewhere else.

None of that abolishes the asuras as beings. The Bhagavata that calls Hiranyakashipu’s death a grace also catalogues twenty-eight hells, and the Gita means what it says. The texts hold both, and so should any honest reading of them.

The asuras were the elder brothers who wanted the inheritance, and lost. Some of them were carried home in the losing.

And the ones who were not still have people who remember their names.