May I Die Today If I Am a Sorcerer: The Rigveda's Witchcraft Problem

The Last Thing Vasiṣṭha’s Book Says

Book Seven of the Rigveda belongs to Vasiṣṭha, the priest who stood with King Sudās at the Battle of the Ten Kings and whose family produced some of the most controlled liturgical poetry in the collection. It holds the great Varuṇa hymns, the Indra hymns, the dawn hymns. It ends with two of the strangest texts in the Veda.

The second to last is RV 7.103, the frog hymn, in which the monsoon frogs croak like priests reciting around the soma bowl. Readers have argued for a century about whether it is a rain charm, a joke at the expense of brahmins, or both. It is at least funny.

The last hymn, RV 7.104, is not funny at all. Twenty-five verses ask Indra and Soma to boil the wicked in their own guilt, to hurl stone weapons burning with flame, to plunge the eaters into a darkness that has no floor. Halfway through, the tone changes. The poet stops describing his enemy and starts defending himself.

adyā murīya yadi yātudhāno asmi yadi vāyus tatapa pūruṣasya

“May I die this very day if I am a sorcerer, or if I have burned up any man’s life. And may he be stripped of all ten of his sons, the man who falsely says to me: sorcerer.”

Rigveda 7.104, verse 15. Translated from the Sanskrit; compare Griffith (1896) and Wilson. [1][2]

Read that again and notice what it implies. Somebody had called this poet a witch. The accusation was serious enough that the only available answer was a public conditional self-curse, staked on his own life and on his accuser’s children. The hymn is not an attack. It is a defence, and the defence was kept in the canon.

What follows is an attempt to take that sentence seriously: to work out how large the sorcery problem actually was in the Rigveda, what a sorcerer was thought to do, why the vocabulary for it clusters in almost no hymns at all, and what happened to these verses in the centuries after they were composed.

A Vocabulary You Can Nearly Count on Your Hands

The Rigveda has roughly 10,500 verses. Searching the romanised Śākala text for the sorcery vocabulary produces numbers small enough to be surprising.

Yātudhāna (यातुधान), the standard word, is a compound: yātu, “sorcery, spell”, plus a second element from the root dhā, “to hold, to place”. A yātudhāna is a sorcery-holder, someone who carries the stuff around. The word occurs in 26 verses. Nineteen of them are in RV 10.87. Three are in RV 7.104. That leaves four scattered occurrences in the entire rest of the corpus.

Kimīdin (किमीदिन्) is rarer still: three verses, all of them in those same two hymns. The native analysis reads it as a frozen question, kim idānīm, “what now?”, the prowler who goes about asking what is happening in your house. It is a lovely etymology and it may well be a later rationalisation of a word nobody could parse. The term has no secure Indo-Iranian pedigree and could belong to the substrate vocabulary Kuiper collected. [3][4] Suggestive, not settled.

This is not a theme spread thinly across the collection. It is a theme almost entirely contained in two hymns, with a thin scatter elsewhere. Taking the concentration of the yātudhāna vocabulary as the share falling inside RV 10.87 and 7.104:

$$C = \frac{22}{26} \approx 0.85$$

Compare rakṣas, which turns up in roughly sixty different hymns and belongs to the ordinary furniture of Vedic thought. A rakṣas is a thing out there in the dark. A yātudhāna is a person, and the Rigveda talks about people like that in almost exactly two places.

Methods note. These counts come from searching the romanised Śākala Saṃhitā for stems, and they are approximate rather than exact. Vedic sandhi fuses word boundaries, so a handful of forms will be hiding inside compounds my patterns did not catch. One term in particular resists counting: yātu on its own is homophonous with the third-person imperative of yā, “let him come”, so a raw search returns dozens of verses in which nobody is doing any magic at all. Where a number here looks precise, it is precise about the search, not about the language. Grassmann’s Wörterbuch zum Rig-Veda remains the instrument for doing this properly. [5]

What the Sorcerer Was Accused of Doing

The charge sheet is remarkably consistent across both hymns, and it is almost entirely about eating.

Nothing here is exotic. Milk that fails, a beast that sickens, food that goes off, a man who wastes away, a ritual that goes wrong. These are the ordinary disasters of a herding household, and the hymns explain them by pointing at a neighbour who has been quietly eating the family’s substance. The Vedic sorcerer is not a wizard with a system. He is a parasite.

The counter-attack is built out of the same metaphor, only bigger. Agni, invoked as rakṣohan, the demon-slayer, is given iron teeth and told to seize the sorcerer with his tongue, chew him, and put him in his mouth.

“Iron-toothed one, touch the sorcerers with your flame, Jātavedas, when you are kindled. Seize the mūradevas with your tongue; twist the raw-flesh eaters and set them inside your mouth.”

(RV 10.87.2, after Griffith 1896, revised against the Sanskrit)

The logic is symmetrical and rather cold. The accusation is that someone is consuming you. The remedy is a bigger mouth.

Aside. The Rigveda never tells you how to perform sorcery. There is no incantation for wasting a rival, no procedure for stealing milk at a distance, nothing that would let a reader of RV 7.104 become a yātudhāna. Every word on the subject is defensive. That asymmetry is worth holding onto, because the Atharvaveda, which inherits these very hymns, is perfectly willing to supply both halves.

Owl-Sorcery, Dog-Sorcery

The best poetry in RV 7.104 is in the verses about shape. Sorcerers in these hymns do not stay in human form. They travel at night, they hide their bodies, and they arrive as birds.

ulūkayātuṁ śuśulūkayātuṁ jahi śvayātum uta kokayātum |

suparṇayātum uta gṛdhrayātuṁ dṛṣad eva pra mṛṇa rakṣa indra

“Strike down owl-sorcery, screech-owl-sorcery, dog-sorcery and cuckoo-sorcery. Crush the eagle-sorcery and the vulture-sorcery, Indra, as with a millstone.”

Rigveda 7.104, verse 22. Translated from the Sanskrit; compare Griffith (1896). [1]

Six compounds in two lines, each one a different animal fused to the same second member. Sanskrit lets a poet build these as fast as he can think of birds, and the effect in recitation is a rising list that lands on the flat thud of the millstone. Three verses earlier the same hymn describes the sorceress moving khargaleva naktam, “like an owl at night”, hiding her body in deceit, and asks that the soma pressing stones kill the demons with the noise they make.

Two things follow. The sorcerer is imagined as a night creature, which is why the counter-hymns are so heavily fire-based: Agni is the only thing awake and looking. And the Rigveda already knows the association between owls and harm that the bird omens of RV 10.165 will treat as ominous rather than magical. In the omen hymn the owl’s shriek is a message that has to be neutralised. Here the owl is not a messenger at all. It is a disguise.

Dawn, predictably, is on the other side. RV 7.104.23 asks that Uṣas drive off mithunā … kimīdinā, the kimīdins “in pairs”, a small detail that Agni’s hymn repeats at 10.87.24. Whatever these beings were, they came two at a time.

The Oath

Return to verse 15, because it is the hinge of the whole hymn and possibly the most human sentence in the Rigveda.

The utterance has a precise shape. It is a conditional self-curse: if the proposition is false, then let the speaker suffer the stated harm. George Thompson has argued that this class of Vedic utterance, the truth-act, works as a performative in the strict sense; the speaker stakes the order of the world on the truth of a statement, and the statement’s truth is what makes the act effective. [6] The later Sanskrit tradition will name this satyakriyā and use it to stop fires and cure snakebite. Here it is being used to survive an accusation.

The following verse turns the knife.

“He who calls me a sorcerer when I am no sorcerer, or who, being himself a demon, says ‘I am pure’, let Indra strike him with the great weapon. Let him fall lowest of every living creature.”

(RV 7.104.16, translated from the Sanskrit; compare Griffith 1896)

Consider the epistemics. The poet cannot demonstrate that he is not a sorcerer, because sorcery leaves no evidence; that is what makes it sorcery. His accuser cannot demonstrate that he is not one either. So the hymn does the only thing available: it converts an unfalsifiable charge into a wager on divine enforcement, and it adds the observation that the person most likely to be a demon is the person loudly claiming to be pure.

Anyone who has read the anthropology of witchcraft will recognise the structure. Evans-Pritchard’s Azande material, gathered in the 1920s, describes exactly this: a charge that cannot be disproved, a socially available oracle for settling it, and the routine suspicion that the accuser has motives of his own. [7] I am not suggesting the Rigveda is Zande, and comparisons across three millennia and two continents earn nothing on their own. What the parallel does establish is that this Vedic verse is not a piece of exotic demonology. It is a recognisable social document, and it sits inside a collection we normally read for cosmology.

Aside. It is tempting to attach RV 7.104 to the famous feud between Vasiṣṭha and Viśvāmitra, in which each sage is later said to have destroyed the other’s sons. The temptation should be resisted. The feud narrative is post-Rigvedic, elaborated in the epics; the hymn itself names no opponent, no occasion and no family. The Anukramaṇī gives the poet as Vasiṣṭha and the deity as indrāsomau rakṣohaṇī, “Indra and Soma, the two demon-slayers”, and that is all the tradition actually supplies. [8] Reading the epic backwards into the hymn tells us about later Sanskrit literature, not about the Punjab in the second millennium BCE.

Where These Hymns Sit

Placement in the Rigveda carries information. Oldenberg showed in 1888 that the family books are arranged by deity and, within each deity’s group, by descending hymn length, so material that violates the pattern or sits past the end of it falls under suspicion of having been added late. [9] Apply that to the sorcery hymns and something clicks.

The tail of Book 7 is doubly suspicious: the Anukramaṇī itself hedges there, crediting hymns 101 and 102 to a second poet alongside Vasiṣṭha. And the hymn immediately before our counter-charm is the frog hymn, another text that has never sat comfortably in a liturgical collection. Whoever closed Vasiṣṭha’s book put the two least canonical things last, which is roughly what the khila material at the far edge of the corpus would later do on a larger scale.

RV 1.133 spoils any clean story, though. It sits in the middle of a Parucchepa group, and it asks Indra to smash the sorceresses’ heads with his broad foot and throw them into a narrow pit. Nothing about its position marks it as an afterthought. The honest conclusion is the boring one: the material is old, the enthusiasm for collecting it was late, and the editors mostly put it where it would cause least trouble.

The Other Veda in the Room

Nineteenth-century Indology worked with a tidy division of labour. The Rigveda was high religion, cosmology, the solemn liturgy of the śrauta rites. The Atharvaveda was magic: charms for hair growth, curses on rivals, spells against fever. Bloomfield’s own classification of the Atharvan corpus includes the categories ābhicārikāni and kṛtyāpratiharaṇāni, sorcery hymns and counter-sorcery hymns, which tells you where the magic was thought to live. [10]

Our two hymns dissolve that division, because they are in both collections.

Whitney’s note on Atharvaveda Śaunaka 8.4 is unambiguous. The hymn is, in his words, “with no change of order of verses, and with only few and insignificant variants, RV. vii. 104”, and it turns up again in Paippalāda 16. The preceding Atharvan hymn, AV 8.3, carries the first 23 verses of RV 10.87 in slightly shuffled order. [11] Both are filed by the Atharvan tradition under the same technical heading: cātana, the “driving-away” hymns, from cat, to chase off.

graph TD
    A[RV 7.104: Indra and Soma] --> C[AV Saunaka 8.4]
    B[RV 10.87: Agni Raksohan] --> D[AV Saunaka 8.3]
    C --> E[AV Paippalada 16]
    D --> E
    C --> F["Catana class: driving-away hymns"]
    D --> F
    F --> G[Kausika Sutra applications]
    G --> H[Rite when cows give bloody milk]
    G --> I[Rite when fire glows with no fuel]
    A --> J[Anukramani poet: Vasistha]
    B --> K[Anukramani poet: Payu Bharadvaja]

The Rigveda goes further and names the tradition out loud. In the middle of the Agni hymn the poet asks the god to burn the liar

“with divine light, in the manner of an Atharvan.”

(RV 10.87.12, atharvavat jyotiṣā daivyena, translated from the Sanskrit)

Atharvavat is an adverb formed directly on the name of the Atharvan priests. A Rigvedic poet, inside the Rigveda, describes his own operation as being done the Atharvan way. Whether the Atharvaveda as a collection existed at that point is a separate and much harder question; Witzel’s work on the formation of the Vedic canon places the Atharvan collections after the Rigveda while treating much of their content as contemporaneous. [12] What the adverb shows is that the category was already there, and that the Rigvedic poets knew they were borrowing it.

The Anukramaṇī adds a detail worth pausing on. It assigns RV 10.87 to Pāyu Bhāradvāja, who is also the traditional poet of RV 6.75, the hymn to the weapons, the one text in the collection that sings to a bowstring, a quiver and a war drum as though they were gods. Traditional attributions are data about the tradition rather than proof of authorship. Still, it is a suggestive pairing: the same name attached to the Rigveda’s hymn about arming yourself against men and its hymn about arming yourself against what you cannot see.

Who Gets Accused

Modern readers arrive at this material expecting the witch to be a woman, because that is the shape the category took in medieval Europe and in much of later India. The Rigveda does not cooperate.

RV 10.87, the long Agni hymn, is almost entirely masculine. Its sorcerer is yātudhānaḥ, a male noun, verse after verse. RV 1.133 is entirely feminine: Indra is asked to crush the heads of the yātumatī, the sorceresses. RV 1.191 closes with the sun rising as adṛṣṭahā, the slayer of the unseen, destroying “all the sorceresses” of the night. And RV 7.104 deliberately covers both.

“Kill the male sorcerer, Indra, and the woman who exults in her magic power. Let the mūradevas fall apart with their necks twisted; let them not see the sun as it rises.”

(RV 7.104.24, translated from the Sanskrit; compare Griffith 1896)

Four verses in the whole collection use an explicitly feminine form. That is too few to build a demography on, and I would not push it. What can be said is that the Rigveda has not yet settled the gender of the sorcerer, and that when it wants to be comprehensive it names both.

There is also the reverse case, which the counter-sorcery hymns make easy to miss. RV 10.145 is a woman’s charm, spoken in the first person, for digging up a plant that will drive away a rival wife and secure a husband, close kin to the herbal repertoire of the healing hymns. RV 10.159 has a woman announce that the sun has risen and so has her own good fortune, that she has overcome her co-wives and subdued her husband. Nobody calls these women sorceresses. They are the operators, and their hymns were kept. Read alongside the dialogue hymns and the erased female voices, they suggest that domestic ritual power was ordinary, and that “sorcery” was the name for the same power when it was pointed at you.

And then there is Arāyī, who is neither a woman nor quite a demon.

“Arāyī, one-eyed and misshapen, go to the mountain, you constantly-shrieking Sadānvā. With these, the men of Śirimbiṭha, we frighten you away.”

(RV 10.155.1, after Griffith 1896)

The hymn goes on to tell her to seize a log floating at the river’s edge and ride it out of sight. This is not liturgy. It is what you say when something has gone wrong in the house and it needs to leave, the same practical impulse behind the Rigveda’s method for handing a bad dream to somebody else.

What the Verses Did Next

The most instructive thing about RV 7.104 and 10.87 is what the later tradition did with them, because the contrast exposes what the Rigveda is deliberately not doing.

In the Rigveda these are words with no stage directions. There is no instruction about when to recite them, no equipment, no procedure, no occasion. In the Atharvan tradition all of that arrives. Whitney’s notes record that the Kauśika Sūtra, the Atharvaveda’s ritual manual, prescribes verses 15 to 18 of AV 8.3 for an oblation to be made when the cows give bloody milk, and another verse of the same hymn for use when a fire flares with no fuel, when a bright glow appears with no fire in it, and when the fire “puffs”. [11]

That is the whole distance between the two collections in one footnote. The Rigveda hands you a poem about an invisible enemy. The Kauśika Sūtra tells you that the enemy has arrived, and that you will know because there is blood in the milk.

The Rigvedic tradition did eventually build its own version of this. The Ṛgvidhāna, a late text prescribing the practical magical application of Rigvedic verses, treats the hymns as instruments to be deployed for stated ends. [13] And the self-imprecating oath of 7.104.15 has an even longer career: the conditional self-curse becomes the śapatha of the legal literature, and by the time of the dharma texts the courts have a whole taxonomy of ordeals for exactly the situation the hymn describes, an accusation that no ordinary evidence can settle. [14]

Aside on translation. Griffith usually renders yātudhāna as “demon”, occasionally leaving it untranslated as “Yatudhana”. The choice matters more than it looks. “Demon” moves the referent out of the village and into mythology; you picture something with claws. “Sorcerer” keeps it where the Sanskrit puts it, on a person with a name who lives nearby. Verse 15 is close to unintelligible under the first reading, because nobody bothers to deny being a supernatural monster, and perfectly clear under the second. This is the sort of quiet decision that shaped how the nineteenth-century translations taught Europe to read the Veda.

What the Hymn Is Actually About

Strip the mythology and RV 7.104 is a document about a small community in which somebody’s cattle sickened, somebody’s child wasted, and the explanation available was that a neighbour had been feeding on them in the dark. The poet who was named could not disprove it. He answered with the only instrument his culture had that was stronger than an accusation, which was an oath: a sentence built so that if he were lying, the order of the cosmos itself would kill him.

That is a different Rigveda from the one that gets quoted. The collection is famous for the creation hymn that ends in a shrug and for Varuṇa watching the world for breaches of ṛta. Those are the texts that made its reputation as the first sustained metaphysics in an Indo-European language, and the reputation is earned. But the same tradition, in the same metre, addressing the same gods, also preserved a poet’s insistence that he had not eaten anyone.

The connection between the two is not decorative. Ṛta is the principle that the world holds together because true things and false things are actually different, and RV 7.104.12 says as much in so many words: the wise man tells the true from the false because their statements are at odds, and Soma protects the true one. The oath in verse 15 is that principle put under load. If truth and falsehood are genuinely distinct, then a man can stake his life on a true statement and survive. The metaphysics and the witchcraft accusation are the same idea, seen from a comfortable distance and from no distance at all.

The scholarly lesson is smaller and worth stating plainly. The wall between the Rigveda and the Atharvaveda, between religion and magic, was built by nineteenth-century scholars who found the Atharvan material embarrassing and the Rigvedic material sublime. Two hymns, transmitted almost verbatim in both collections and filed by the Atharvan tradition under its own technical heading for expulsion charms, are enough to show that the wall was never in the texts. [15]

Read RV 7.104 straight through, in any translation, and then read RV 10.87 immediately after. The first is a man defending himself. The second is a priest arming a fire. Somewhere between them is a village at night, where everyone knows who is being talked about.