The Nightmare You Could Give Away: How the Rigveda Handled Bad Dreams

A Debt Collected in the Dark

Somewhere in the Punjab, three thousand years ago or thereabouts, a man wakes up before dawn badly shaken. What he does next is not what you would expect. He does not look for a specialist to tell him what the dream meant. He does not check it against a list of omens. He gathers it.

yathā kalām yathā śapham yatha ṛṇam saṃ nayāmasi

“As we collect the utmost debt, even the eighth and sixteenth part, so unto Āptya we transfer together all the evil dream.”

Rigveda 8.47, verse 17. Translation Griffith (1896).

Read that again with an accountant’s eye. The verb is saṃ nayāmasi, “we lead together, we collect.” The nouns are units of fractional payment: kalā, a sixteenth; śapha, literally “hoof,” an eighth; ṛṇa, debt. The bad dream is being handled the way a creditor handles arrears, swept into a single sum and moved off the books. And it is moved somewhere specific, onto a minor god named Trita Āptya, who appears in this hymn and in a handful of others and who, so far as anyone can tell, never asked for the job.

This is not how the rest of the Bronze Age dealt with dreams. Egypt had a dream manual. Assyria had a dream book running to tablets of paired omens. Greece would give us the gates of horn and ivory. Everywhere the surviving evidence points the same way: a dream is a message, and the skilled thing to do with a message is decode it. The Rigveda, alone among the large early corpora, appears entirely uninterested in decoding. It has no dream interpreter, no omen list, no favourable dream. What it has instead is a removal procedure.

Eight verses in roughly ten thousand carry the word duḥṣvapnya. They are enough to reconstruct a whole theory of what a bad dream is, and that theory turns out to be an argument about guilt rather than an argument about the future. It also produces, almost in passing, one of the oldest recorded formulations of a problem that would still be bothering Augustine two millennia later and philosophers of mind today: are you answerable for what you do while you are asleep?

The Word Is Not “Nightmare”

Start with the grammar, because the grammar is the argument.

Duḥṣvapnya (Sanskrit: दुःष्वप्न्य) is built from duṣ-, “bad, ill,” plus svapna, “sleep, dream,” plus the suffix -ya, which in Vedic forms neuter abstracts, the same suffix that turns sakhi, “companion,” into sakhya, “companionship.” So the word does not mean “a nightmare,” a countable episode with a plot. It means something closer to “bad-dreamness”: the condition, the stuff, the residue. Griffith renders it “evil dream,” which is fine but hides the abstraction.

That matters because of what the poets do with it. Duḥṣvapnya is never interpreted, never narrated, never asked about. It is only ever moved. Look at the verbs across all eight occurrences.

Every single verb is a verb of physical conveyance. Carry, hand over, collect, drive off, impel. Nothing here belongs to the vocabulary of reading, asking, or knowing. You would use the same verbs for moving a sack of grain or clearing a blocked channel.

And notice the company duḥṣvapnya keeps. In RV 10.36.4 it is listed alongside the rakṣases, Nirṛti (dissolution, the pit), and the atrin devourers. In RV 10.37.4 it sits next to amīvā, disease, and anāhuti, the failed or unoffered oblation. The bad dream is filed with sickness, ritual failure and hostile entities. It is a contaminant. The poets are not asking what it portends any more than you would ask what a fever portends.

Aside. The one place duḥṣvapnya meets machinery is RV 10.36.4, where the grāvāṇaḥ, the soma pressing stones, are asked to drive it off with their noise. The stones are the loudest object in the Vedic ritual world, and noise is a standard purgative. On what those stones actually were and how they sounded, see The Soma Workshop.

Everyone Else Was Reading Dreams

To see how unusual this is, put the Rigveda next to its contemporaries.

Papyrus Chester Beatty III, from Deir el-Medina and dating to the Ramesside period around the thirteenth century BCE, is a working dream manual. Each entry follows a fixed template: “If a man sees himself in a dream doing X,” then a verdict, good or bad, written in red for the ominous ones, then the interpretation. The manual carries on the order of two hundred entries, most of them favourable, and the interpretations frequently run on puns, so that seeing something whose name sounds like another word delivers the second word’s meaning [1, 2].

Mesopotamia went further. The compilation Assyriologists call the Assyrian Dream Book, edited and translated by A. Leo Oppenheim in 1956, is a systematic omen series sitting inside a much larger apparatus of dream reports, incubation practices and ritual responses [3]. A Mesopotamian who had a bad dream had two options, and both were technical: find out what it meant, and then undo it. Greece is later but points the same way. When Penelope tells her dream of the eagle and the geese in Book 19 of the Odyssey, everything turns on interpretation and on whether the dream came through the gate of horn or the gate of ivory. The question is always: is it true, and what does it say?

The Rigvedic silence is loud. There is no Vedic word in the Rigveda for a dream interpreter. There is no verse in which a dream is asked what it means, and no verse in which a god sends a dream in order to communicate something. There is not even a good dream. Susvapnya does not appear. The corpus knows only the bad kind, and only as something to be got rid of.

The closest the text comes to reading a dream is a Varuṇa hymn in the second maṇḍala:

“O King, whoever, be he friend or kinsman, hath threatened me affrighted in my slumber: if any wolf or robber fain would harm us, therefrom, O Varuṇa, give thou us protection.”

(RV 2.28.10, translation Griffith 1896)

Someone the poet knows appears in his sleep and threatens him. Even here, though, no one asks what the figure signified. The dream is taken as a live danger and answered with a request for protection, in the same breath as wolves and thieves. The response is security, not hermeneutics.

Methods note. Counting words in the Rigveda by string search is treacherous. A search for svapn- returns twenty verses, but two of them, RV 10.63.3 and RV 10.78.1, contain svapnasaḥ, which is not svapna at all. It is su-apnas-, “having good works, effective,” and Griffith accordingly renders the gods there as “active.” Two of twenty is a ten percent error rate from a single homograph. Every count in this article was checked against a translation before being used.

The Accounting Metaphor

Back to that extraordinary seventeenth verse of RV 8.47. Three fractional terms are stacked in one line:

$$\text{kalā} = \tfrac{1}{16}, \qquad \text{śapha} = \tfrac{1}{8}, \qquad \text{ṛṇa} = \text{the whole outstanding debt}$$

The values are the standard ones in later Sanskrit usage. Kalā is a sixteenth part; śapha means “hoof,” and the usual explanation is that a cow’s four cloven feet give eight hoof sections, hence an eighth. Whitney’s translation of the Atharvavedic reworking of this same verse makes the arithmetic explicit: “As a sixteenth, as an eighth, as a debt they bring together” (AV 6.46.3) [4].

So the image is of a creditor going round collecting arrears in small denominations until the full sum is in hand, and then transferring the whole balance to another party. The bad dream is a liability. It sits on your books. It can be aggregated, and once aggregated it can be assigned.

This is not a stray metaphor. Debt is one of the load-bearing concepts of Vedic religious thought, and ṛṇa language runs through the Rigveda’s dealings with the gods and with guilt. What the dream verses add is the discovery that a liability which is not financial at all can be handled with financial machinery. You do not need to understand the dream. You need to know who will take it.

Which brings us to the strangest verse in the group.

niṣkaṃ vā ghā kṛṇavate srajaṃ vā duhitar divaḥ | trite duḥṣvapnyaṃ sarvam āptye pari dadmasi

“Even if, O Child of Heaven, it make a garland or a chain of gold, the whole bad dream, whate’er it be, to Trita Āptya we consign.”

RV 8.47.15, translation Griffith (1896)

A goldsmith’s ornament and a garland-maker’s wreath turn up in the middle of a nightmare charm. Sāyaṇa, the fourteenth-century commentator, read the verse as the dream threatening the ornament-worker and the garland-maker, with the evil then lodged in Trita and thrown off. Modern translators read the syntax differently, with the bad dream itself producing or being made into the ornament. The Sanskrit is genuinely hard here, and I would not want to insist on either reading.

But both readings share something worth noticing. The bad dream is being attached to an object that can be given away. A necklace, a garland: portable things, things that leave the house on someone else’s body. Whatever the precise grammar, the model underneath is that the immaterial can be loaded onto the material and shipped out, which is the same logic that produces amulets, scapegoat rites and the practice of burying a trouble at a crossroads.

Who Is Trita, and Why Does He Take It

Trita Āptya is one of the Rigveda’s genuine oddities. His name means “the Third,” transparently a numeral, and his epithet Āptya is usually connected with āp-, “water.” He is named in roughly forty verses, often in soma contexts in the ninth maṇḍala, sometimes as a slayer of the three-headed serpent Viśvarūpa, which is the job Indra normally does [5]. He never gets a hymn of praise of his own.

Comparative evidence puts him a long way back. The Avestan tradition preserves Thrita and Θraētaona, the dragon-slayer who becomes Ferīdūn in the Persian epic; the Vedic Trita and the Iranian figures descend from a shared Indo-Iranian original. On the Vedic side he has been reduced to a functionary. On the Iranian side he kept the dragon.

The functionary role is what interests us here, and RV 8.47 shows it plainly. Verse 13, one verse before the dream sequence begins, sets the pattern:

“Whatever ill deed there is, O gods, whether open or concealed, place all of that on Trita Āptya, far away from us.”

RV 8.47.13, after Griffith (1896)

The word for ill deed is duṣkṛta. Notice the structure: first the sins go to Trita, then the bad dreams go to Trita, in consecutive verses of the same hymn, with the same destination and nearly the same syntax. The hymn is not treating dreams and offences as two problems. It is treating them as one problem with two symptoms, and applying one solution.

Maurice Bloomfield, working on the Atharvaveda in the 1890s and 1900s, gave this figure the name that stuck: Trita is the scapegoat of the gods [6]. The Atharvaveda makes the chain explicit, running the guilt from gods to Trita and from Trita on to human beings, so that the whole system of transfer becomes a kind of relay. The Rigveda does not spell that out, but it does not need to. The mechanism is visible.

graph TD
    A[Sleeper wakes disturbed] --> B["Bad dream = duḥṣvapnya"]
    B --> C[Aggregated like a debt]
    C --> D[Dawn asked to carry it]
    D --> E["Trita Āptya, the Third"]
    E --> F[Also receives duṣkṛta, ill deeds]
    B --> G[Later option: send to an enemy]
    G --> H["RV 10.164.5, AV 6.46.3"]
    F --> I[Household declared anāgas, guiltless]

There is a further twist, and it belongs in the “traditional attribution, treat with care” column. The Anukramaṇī, the old index of authorship, assigns RV 8.47 to a poet called Trita Āptya. If you take the attribution at face value, the hymn that dumps every bad dream onto Trita was composed by Trita. Indological convention is to treat such attributions as later inference from names occurring inside the hymn rather than as historical record, and that is almost certainly what happened here. Still, it tells you the tradition had noticed the same peculiarity, and had no better explanation than to make the recipient the author.

Verse 16 adds a second name, Dvita, “the Second.” Trita and Dvita together, the Third and the Second, look like the fossil of a set of three brothers. The Vedic poets seem to have inherited a numbered trio and kept only the function.

Can You Sin in Your Sleep?

Now the part that lifts this material out of folklore.

In the seventh maṇḍala, the poet Vasiṣṭha spends several hymns pleading with Varuṇa, the god who watches, counts and binds. On what Varuṇa is and why his hymns feel unlike anything else in the collection, see Varuṇa and the Invention of Cosmic Guilt. At one point Vasiṣṭha tries to explain how a decent man ends up guilty:

na sa svo dakṣo varuṇa dhrutiḥ sā surā manyur vibhīdako acittiḥ | asti jyāyān kanīyasa upāre svapnaś caned anṛtasya prayotā

“Not our own will betrayed us, but seduction, thoughtlessness, Varuṇa, wine, dice, or anger. The old is near to lead astray the younger: even sleep removeth not all evil-doing.”

RV 7.86.6, translation Griffith (1896)

The list is a defence lawyer’s list. Not my own competence, says the poet, but a slip: liquor (surā), anger (manyu), the vibhīdaka dice, and plain inattention (acitti). Every Rigveda reader knows the dice from the Gambler’s Lament; here they are again, filed under diminished responsibility, alongside drink. (For the dice themselves and the nut they were cut from, see The Gambler’s Hymn.)

Then the last clause, and it is a hard one: svapnaś caned anṛtasya prayotā. Literally, “not even sleep is a warder-off of untruth.” Griffith gives “even sleep removeth not all evil-doing.” Sleep does not clear you. The gates do not close at night.

This is a genuinely startling thing to find in a text this old. The poet has thought through his own excuse and found the hole in it. If drink and anger and dice mitigate guilt because they suspend the will, then sleep, which suspends the will completely, ought to be the perfect alibi. It is not. The wrongdoing gets through anyway.

Roughly two thousand years later, in Book X of the Confessions, Augustine turned the same question over at length, troubled that his continence held while he was awake and collapsed while he was asleep, and asking whether he was the same man in both conditions. The problem has never really gone away. Contemporary philosophers of mind who work on dreaming still argue about whether dream actions carry moral weight, and the arguments track the same fault line: is the dreamer an agent or a spectator?

The Rigveda’s dream-charm answers it in one word, and the word is used exactly once in the entire collection.

ajaiṣmādyāsanāma cābhūmānāgaso vayam | jāgrat svapnaḥ saṃkalpaḥ pāpo yaṃ dviṣmas taṃ sa ṛcchatu yo no dveṣṭi tam ṛcchatu

“We have prevailed this day and won: we are made free from sin and guilt. Ill thoughts, that visit us awake or sleeping, seize the man we hate, yea, seize the man who hateth us.”

RV 10.164.5, translation Griffith (1896)

The word is saṃkalpa, here in the phrase saṃkalpaḥ pāpaḥ, “evil intention,” qualified as belonging to both the waking state and the dreaming state (jāgrat svapnaḥ). Saṃkalpa would go on to an enormous career. It becomes the standard term for the formed intention a Hindu ritualist declares aloud before a rite, and the term the Upaniṣads and the Gītā use for the willing that binds a person to action. In the Rigveda it occurs precisely once, in an anti-nightmare charm, describing a bad intention that reaches you in your sleep.

Whatever else is uncertain, that is a striking place for a word to enter the record. Vedic India’s vocabulary of intention shows up first attached to a nightmare.

The charm’s third verse makes the doctrine as plain as it gets:

yad āśasā niḥśasābhiśasopārima jāgrato yat svapantaḥ | agnir viśvāny apa duṣkṛtāny ajuṣṭāny āre asmad dadhātu

“If by address, by blame, by imprecation we have committed sin, awake or sleeping, all hateful acts of ours, all evil doings may Agni bear away to distant places.”

RV 10.164.3, translation Griffith (1896)

Jāgrato yat svapantaḥ, “whether waking or sleeping.” The formula makes no distinction. Offences committed in sleep are offences. They are removed by the same means as offences committed at noon.

Aside. RV 10.164 has five verses, and the traditional indexes name its deity not as a god but as duḥṣvapnaghna, “the killing of bad dreams.” That is a category of one. Its metre is mostly anuṣṭubh, the short four-by-eight line that dominates spell literature, rather than the triṣṭubh of the grand liturgical hymns. Both facts point the same way: this is charm material that ended up inside the Rigveda’s most miscellaneous book.

Two of these hymns share a line. RV 8.47.18 and RV 10.164.5 both open ajaiṣmādyāsanāma cābhūmānāgaso vayam, “today we have won, we have gained, we have become free of guilt.” That exact line occurs nowhere else in the Rigveda. Two hymns from different books, both concerned with bad dreams, both ending in the identical declaration of acquittal. Whatever ritual these belonged to, the acquittal formula was its punchline, and the point of the ritual was to be able to say it.

The Sleepless Gods

The Rigveda’s wider vocabulary of sleep confirms the picture. Sleep is not neutral in this text. It is a lapse.

Look at the first three rows. The gods who matter most in the guilt economy, the Ādityas, are defined by not sleeping. Asvapnaja is a compliment, and a rather intense one: unblinking, unwearied, always on. Varuṇa’s whole moral apparatus depends on continuous observation, and continuous observation is incompatible with sleep. Divine wakefulness is not a physiological fact about gods. It is a description of their office.

Now note where the dream verses of RV 8.47 sit. That hymn is addressed to the Ādityas. The sleepless watchers are the ones petitioned to take away what sleep left behind, and Dawn, the goddess whose arrival ends sleep, is the courier. The hymn’s architecture is coherent all the way down.

Then row five, which is a small horror. In RV 2.15.9 Indra defeats the demons Cumuri and Dhuni by covering them with sleep, and then kills them. Sleep is a weapon there, and what it does to its victims is make them killable. And row six, RV 10.162.6, is the Rigveda’s nearest approach to what a modern reader would recognise as a nightmare experience: something that “deceives you with sleep and darkness and lies down beside you,” and that means harm to your children. The hymn is a charm to protect pregnancy. The presence, the pressure, the sense of a body next to you in the dark: this is the phenomenology that in Europe would eventually be painted by Fuseli and named sleep paralysis by clinicians.

The Mesopotamians had built an entire protective industry around the same intuition. Amulets and plaques against Lamashtu, the demon who attacks women in childbirth and infants at night, were hung in houses and worn on the body, and they work on exactly the logic of RV 10.162: name the intruder, describe her, and expel her.

Where the Nightmare Ends Up

The most revealing change in this material is not doctrinal. It is a change of address.

In RV 8.47, the bad dream goes to Trita Āptya, a god. Nobody is harmed; a mythological figure absorbs it, which is what mythological figures are for. By RV 10.164.5, the destination has shifted: “let it seize the man we hate, and the man who hates us.” The dream is now aimed at a person.

The Atharvaveda completes the move. Its hymn against bad dreams, AV 6.46, takes the Rigvedic accounting verse almost word for word and redirects it:

“As a sixteenth, as an eighth, as a [whole] debt they bring together, so do we bring together all evil-dreaming for him who hates us.”

Atharvaveda 6.46.3, translation Whitney [4]

Same fractions, same verb of collection, new recipient. Trita has been replaced by an enemy. The same hymn also gives Sleep a genealogy that the Rigveda never bothered with, calling it neither living nor dead, an immortal embryo of the gods, with Varuṇānī for a mother and Yama, the lord of the dead, for a father [4]. Between the two collections, the bad dream has acquired a parentage, a personality and a target.

That table contains the whole arc, and the last two rows are where India catches up with everyone else, then overtakes them.

The Upaniṣadic turn is the more interesting of the two. In Bṛhadāraṇyaka 4.3, the dreaming self is examined as evidence: in that state there are no chariots and no roads, and yet the dreamer produces chariots and roads, which tells you something about what a self is and where its light comes from. The Māṇḍūkya makes svapna one of the four canonical states of consciousness. Dreaming has stopped being dirt on the doorstep and become a laboratory. Wendy Doniger’s study of the Indian material argues that the dream eventually becomes the central locus of Indian epistemology, and the distance travelled from RV 8.47 to that position is enormous [7].

Divination arrives too, just late. The Atharvavedic ancillary literature includes a Svapnādhyāya, a chapter on dreams, and by the medieval period India has a full dream manual in Jagaddeva’s Svapnacintāmaṇi, edited and translated by Julius von Negelein in 1912 [8]. It classifies dreams as auspicious and inauspicious with the thoroughness of the Egyptian scribes, roughly two millennia after the Egyptian scribes.

So the Rigveda is not primitive here. It is doing something else on purpose. A culture that later produced one of the world’s most elaborate dream-omen literatures spent its earliest documented centuries refusing to read dreams at all.

What the Ritual Was Actually For

Strip the theology away and look at what a person does in this rite.

They wake up frightened. They name the fear as a definite quantity rather than a fog. They perform a physical action of gathering. They assign the quantity to a recipient outside themselves. And then they say a fixed sentence out loud: ajaiṣma, asanāma, abhūma anāgaso vayam. We have won. We have gained. We are guiltless.

The Belgian Sanskritist Jan Houben, discussing the Vedic anti-nightmare material, framed the interpretive question sharply in the title of his study: magic, superstition, or psychology [9]. It is a fair question and it does not have a tidy answer, but it is worth noticing what the procedure resembles.

Modern clinical practice for chronic nightmares includes Imagery Rehearsal Therapy, developed and tested by Barry Krakow and colleagues, whose randomised trial in JAMA in 2001 followed 169 women with post-traumatic nightmares and found reduced nightmare frequency and PTSD severity in the treated group [10]. The method is not analysis. Patients are not asked what the nightmare means. They are asked to take the nightmare, change it deliberately, and rehearse the new version until it displaces the old one. Meta-analytic work since has supported the approach for post-trauma nightmares [11].

The temptation to claim that an ancient text anticipated modern medicine is exactly the kind of claim that discredits writing about the Vedas, so let me be plain: the Rigvedic poets were not doing cognitive behavioural therapy, and Trita Āptya is not a therapeutic construct. The structural resemblance is still real. Both procedures treat the nightmare as an object to be manipulated rather than a text to be read, both refuse interpretation, both work by relocating the content away from the sufferer, and both end in a declaration.

Three things follow, I think, for a modern reader.

First, this is very early evidence of a culture treating an internal experience as an external, transferable thing. That is not naive. It is a technique, and the fact that people still use versions of it, in confession, in ritual disposal, in the habit of writing a fear down and putting the paper somewhere else, suggests it does something.

Second, the Rigveda located the problem in the right place. It did not ask whether the dream predicted a robbery. It asked whether the dreamer was clean. Its concern was the sleeper’s standing rather than the sleeper’s future, which is a shrewder guess than the Assyrian one, whatever else it is.

Third, and least comfortable, the text saw the loophole and closed it. Svapnaś caned anṛtasya prayotā. Sleep does not clear you. The Rigveda could have made unconsciousness an alibi, and it declined.

Read RV 8.47 straight through some morning when you have woken badly. It is eighteen verses, every one of them ending in the same refrain about aid that is faultless, and the last five are the dream sequence. What comes through is not superstition and not doctrine. It is procedure, of the sort you follow when something has gone wrong in the night and you want, quite badly, to be able to say by breakfast that you are guiltless. Then read RV 10.164, which takes about ninety seconds, and notice that a text of over a thousand hymns found room for one whose entire purpose is to get a bad dream out of the house.

For related ground, see Varuṇa and Cosmic Guilt on the moral apparatus these charms plug into, Karman in the Rigveda on sukṛta and duṣkṛta before karma became doctrine, and The Physician’s Hymn on the wider Vedic technology of removal.