Devas into Demons: The Rigveda, the Avesta, and the Asura-Daeva Reversal

A word that means opposite things in two scriptures

Open the Rigveda and the word asura is, more often than not, a compliment. Varuṇa is called the “wise Asura” (RV 1.24.14, RV 5.71.2); Mitra and Varuṇa together are “Asuras and imperial Lords” (RV 8.25.4); even Agni and Rudra get the title.[1] Open the Avesta, the scripture of Zoroastrianism composed a few centuries later and a few hundred miles to the northwest, and the cognate word daēva is a term of abuse. Zarathustra tells the daevas to their faces, in one of his own hymns, that they are “the offspring stemming from evil thinking, deceit and disrespect.”[2] Meanwhile the Avesta’s supreme god carries the title ahura, the exact word Indian tradition eventually demoted to mean the enemies of the gods.

Two branches of one family took the same pair of words, deva/daeva and asura/ahura, and assigned them to opposite moral camps. This is not a minor lexical curiosity. It is one of the best documented cases of religious semantic inversion in the historical record, precise enough that specialists can point to the individual verses where the old, shared vocabulary starts to pull apart. The philologist F. B. J. Kuiper, who spent much of a long career on this exact problem, called it a “polarization” of an originally undifferentiated group of powers.[3] Georges Dumézil, Mary Boyce, Paul Thieme, and a long line of Indo-Europeanists after them have each taken a run at explaining why.

This piece follows the reversal from both ends: the Rigveda’s own internal drift from “asura” as an honorific to “asura” as a demon, and the Avesta’s much sharper, more deliberate demonization of the daevas under Zarathustra. It uses the one external, datable anchor both traditions share, the Mitanni treaty of around 1380 BCE, discussed at greater length in the companion piece on Mitra, to show that the shared pantheon was still whole within living memory of the split. And it ends with the genuinely open question: was this reversal a single reformer’s decision, or a slow drift that had already begun before either scripture was composed.

A shared word, a forked meaning

Start with the raw philology, because in this story the vocabulary carries the argument. Vedic devá (देव) and Avestan daēva both continue an Indo-European root meaning “to shine,” the same root that gives Latin deus and, ultimately, English “deity.”[4] Vedic ásura and Avestan ahura likewise share a single Indo-Iranian ancestor, glossed by comparativists as something like “lord” or “possessor of ásu,” vital power. Neither pair is a coincidence of sound; both are inherited, exactly the way “father” and Latin pater are inherited from a common Indo-European source.

What changed is which word got the halo and which got the horns.

The table shows the same handful of Indo-Iranian words landing on opposite sides of a moral line in the two traditions, while a few, like the formula “gods and men,” survive untouched by the split.

Aside. Devanagari and Avestan script are both later inventions; neither the Rigveda nor the Gāthās were written down when they were composed. The forms above are reconstructed and standardized transliterations (IAST for Sanskrit, a Latin-based scholarly transliteration for Avestan), not the original writing systems, which postdate the hymns by well over a thousand years in both traditions. See the essay on Vedic oral transmission for how the Rigveda in particular survived that long a gap without a script.

Did You Know? - The word “deity” and the word “daeva” (demon, in Zoroastrian Iran) come from the exact same Indo-European root. - Varuṇa is called “Father Asura” in RV 10.124.3, using the word that later Hinduism reserves for demons. - Ahura Mazdā’s own title, “Ahura,” is philologically the same word as Sanskrit “Asura.” - The Rigveda still preserves a phrase distinguishing “asuras that have become devas” from “asuras that are non-devas,” a fossil of the moment before the split hardened.[6] - Indra’s title Vṛtrahán, “Vṛtra-slayer,” and the Avestan god of Victory, Vərəθraγna, are the same word, but only one of the two gods actually kills a dragon. - Mitra has only one solo hymn in the Rigveda (RV 3.59) yet becomes one of the grandest deities of the Iranian world as Mithra.

The asuras of the Rigveda

In the Rigveda’s family books, the oldest core of the collection, asura is simply a title of power, applied to whichever god the poet is praising at that moment: Varuṇa most often, but also Mitra, Agni, Rudra, Savitṛ, even Indra. The great hymn to Varuṇa’s mystical vision, RV 7.88, composed by or for the sage Vasiṣṭha, describes a state of intimate closeness with the god, in which the poet is made médhira, “wise,” like Varuṇa himself, gaining “(revealed) insight into the cosmic order” (medhā́ ṛtásya, RV 8.6.10).[7] The god who grants this is repeatedly called Asura, and there is nothing sinister in the term. Consider the dual hymn to Mitra and Varuṇa:

Great Varuṇa and Mitra, Gods, Asuras and imperial Lords, True to Eternal Law proclaim the high decree.

(RV 8.25.4, trans. Griffith 1896)

Here “Asura” sits in apposition to “Gods” and “imperial Lords,” not opposed to them. The philologist Wash Edward Hale, in the fullest statistical study of the word, counted roughly 88 occurrences of asura and its derivatives across the Rigveda, and traced how the term’s sense shifts as one moves from the oldest family books (Maṇḍalas 2 through 7) toward the later books, 1, 8, and 10.[8] In the family books, asura as a divine title appears mostly in the singular or the Mitra-Varuṇa dual, and is applied to gods who are, in the same breath, also called deva. Rudra himself is addressed in one verse as devám ásuram, “the god (who is also) Asura” (RV 5.42.11).[9] There is, in other words, no hard boundary yet between the two categories in the earliest layer of the text.

The boundary appears later, and it appears gradually. By Maṇḍalas 1, 8, and 10, generally agreed to include the Rigveda’s more recent material, asura starts attaching itself to Vṛtra and the dragon’s kin, the cosmic obstructers whom Indra defeats. The text itself preserves the transitional moment in an unusually explicit pair of phrases. RV 8.25.4 and RV 7.65.2 speak of devā́v ásurā, “the two Asuras who have become Devas” (Mitra and Varuṇa), while RV 8.96.9 speaks of ásurā adevā́ḥ, “Asuras who are non-Devas.”[10] The Rigveda is here narrating its own semantic history in real time: some old Asuras crossed over and joined the Devas’ camp, and others did not. Later texts remember only the second group, and asura becomes, for the rest of Indian religious history, simply the word for demon.

Hale’s word-count survey shows the singular, honorific usage dominating the Rigveda numerically; the demonic, plural sense that later Hinduism inherits is a minority pattern within the text itself, concentrated in its younger layers.[8]

pie title Grammatical distribution of "asura" in the Rigveda (Hale 1986)
    "Singular (~71)" : 71
    "Dual (~4)" : 4
    "Plural (~10)" : 10
    "Compound, first member (~3)" : 3

A simple visualization of the same count: the overwhelming majority of the Rigveda’s uses of asura are singular, honorific references to an individual god, not the later plural class of demons.

Key Insight: The deva/asura split is not something the Rigveda received ready-made from a prehistoric Indo-Iranian schism. Hale’s analysis suggests it develops within the Rigveda’s own compositional history, hardening across the roughly two to three centuries the family books and the later books were separated by. The Avesta, on the other hand, shows something closer to a decisive break attributable to one figure. That asymmetry, gradual drift on one side, deliberate reform on the other, is itself the central puzzle this essay is built around.

Meter and vocabulary both support treating Maṇḍalas 2-7 as older than 1, 8, and 10; see the essay on syllable counts and Rigvedic chronology for one line of evidence and the piece on Indra across 250 hymns for how Indra’s profile shifts across the same layers.

Ahura Mazdā and the Iranian reform

Cross into the Avesta and the picture inverts. The oldest Avestan texts, the seventeen Gāthās attributed to Zarathustra himself and embedded within the Yasna liturgy, use daēva almost exclusively as a term of condemnation. Zarathustra does not merely ignore the daevas; he confronts them directly, in the second person, and tells them what they are:

But ye gods (daēva), as well as the one who worships you, all of you are the offspring stemming from evil thinking, deceit and disrespect.

(Yasna 32.3, trans. Insler 1975, punctuation lightly standardized)

This is a startling thing to say about beings whose name simply means “the shining ones,” the exact cognate of the Vedic word for “god.” Zarathustra reserves worship for a single figure, whom he addresses as Ahura, “Lord,” or in the fuller form Ahura Mazdā, “Wise Lord.” His own theology sets this Ahura against a cosmic moral dualism rather than against a rival divine faction as such: the famous “Twin Spirits” verse frames the real opposition as ethical, not tribal.

Yes, there are two fundamental spirits, twins, which are renowned to be in conflict. In thought and in word, in action, they are two: the good and the bad.

(Yasna 30.3, trans. Insler 1975)

How translators render this single, pivotal line is itself a small window into the interpretive stakes:

The differences are subtle but real: whether the “twins” are cosmic principles, personified spirits, or stages of a single primal existence shapes how literally one reads Zoroastrian dualism, a debate that runs through the whole of Gathic scholarship.

The Encyclopaedia Iranica’s overview by the Iranologist Gherardo Gnoli makes the important qualification here: “the fundamental opposition in Zoroastrianism is not between groups of deities but between Aša (truth) and Druj (falsehood).”[11] The daevas are condemned because Zarathustra classes them with the followers of the Lie, not because “daeva” was already, before him, a fixed word for demon. That is precisely where the scholarly argument begins.

Methods note. Almost everything we know about the pre-Zoroastrian, “old” Iranian religion that the Gāthās are reacting against comes from a different, later stratum of the Avesta, the Younger Avesta, especially the Yašts, hymns to individual deities such as Mithra (Yašt 10) and Vərəθraγna (Yašt 14). Reconstructing what came before Zarathustra therefore means reading backward through texts composed largely after him, a genuinely circular problem that specialists such as Jean Kellens have written about at length.[12]

Two rival explanations for the daeva’s fall from grace circulate in the modern literature, and it is worth laying them out side by side rather than picking a winner prematurely.

Both camps agree the Gāthās are the decisive turning point in the textual record; they disagree about how much work Zarathustra himself did versus how much was already in motion.

Almut Hintze’s linguistic dating of the Gāthās, based on their archaic grammar relative to the Younger Avesta, places Zarathustra’s own compositions somewhere in a broad window that overlaps the plausible dates for the Rigveda’s own family books, though neither date is fixed with precision and both remain actively debated.[14] The two reforms, Vedic and Iranian, were therefore probably unfolding at something like the same historical moment, on either side of a linguistic and cultural border that had not yet hardened into “India” and “Iran” as separate worlds.

The Mitanni anchor: a moment before the split

Both traditions agree, structurally, on a shared pantheon before the reversal took hold: some combination of Mitra, Varuṇa/Ahura, Indra, and the divine twins. The one place this shared pantheon appears outside either scripture, dated by an entirely independent method, is a Hittite state archive.

Around 1380 BCE, the Hittite king Šuppiluliuma I concluded a treaty with the Mitanni prince Šattiwaza, and the scribes invoked divine witnesses on both sides. Among the Hurrian and Mesopotamian gods named, four are unmistakably Indo-Aryan: Mi-it-ra, Aru-na (Varuṇa), In-da-ra (Indra), and the Na-ša-at-ti-ya (the Nāsatyas, the Aśvins).[15] This document is discussed in full in the companion essay on Mitra and the Mitanni treaty; what matters here is the chronology it establishes. By roughly 1380 BCE, an Indo-Aryan-speaking elite in northern Mesopotamia was still swearing oaths on a pantheon that pairs Mitra and Varuṇa without any hint of the deva/asura polarity found later in either the Rigveda or the Avesta. The treaty is a snapshot of the shared inheritance before either side’s semantic reversal is legible in writing.

The chronology is deliberately asymmetric in its confidence: the Mitanni date is external and firm, while every purely textual date on either side of the ledger carries real uncertainty, discussed further in the dating-debate essay.

Two hypotheses for the great reversal

Why would two closely related traditions take the same handful of words and send them in opposite directions? The Encyclopaedia Iranica’s synthesis, following Kuiper, states the outcome plainly: “after the Iranians had demonized the Devas/Daēvas, the Indians started to demonize the Asuras.”[16] That sentence describes a sequence, but not a mechanism, and mechanism is where scholars disagree.

graph TD
    PII["Proto-Indo-Iranian pantheon: undifferentiated *asura/*daiva vocabulary"]
    PII --> RV["Rigveda, family books: asura = honorific title (Varuna, Mitra, Agni)"]
    PII --> AV["Old Avesta, Gathas: daeva = condemned, ahura = supreme god"]
    RV --> RVlate["Rigveda, later books: asura drifts toward Vrtra's demonic kin"]
    RVlate --> Brahm["Brahmanas onward: asura = fixed demon term"]
    AV --> YAv["Younger Avesta: daeva fully demonic, old gods reorganized as yazatas"]
    Brahm --> Result["Result: asura/ahura and deva/daeva swap moral polarity across the two traditions"]
    YAv --> Result

The diagram traces both traditions from one shared starting vocabulary to a fully reversed outcome, showing that the Rigveda’s shift happens gradually across its own internal chronology, while the Avesta’s shift is already largely complete within its oldest layer, the Gathas themselves.

One camp, associated historically with Mary Boyce, credits Zarathustra with the decisive act: he judged the daevas, the martial, ecstatic, sacrifice-hungry gods of the old Indo-Iranian warrior culture, and cast them out of legitimate worship as a matter of personal religious conviction.[13] A second camp, whose clearest statement runs through Kuiper’s and Gnoli’s work, treats the demonization as already gathering momentum before Zarathustra, so that his contribution was to sharpen and moralize a divide rather than invent one from nothing.[3][11] On the Indian side, Hale’s philological argument points the same way: no reformer’s name attaches to the Rigveda’s asura/deva split, and the text’s own layered chronology shows the categories separating slowly, book by book, without any single dramatic break.[8]

A short gallery of how specialists have actually framed the reversal:

“In the Rig Veda, asura denotes the ‘older gods’… The emergence of the dualistic cosmos was a process of polarization in which some of the asuras… went over to the ‘younger gods,’ the devas.” (F. B. J. Kuiper, Encyclopaedia Iranica, “Ahura”)[3]

“The demonisation of the Asuras in post-Rigvedic India and the demonisation of the Daevas in Zoroastrian Iran took place so late that the associated terms cannot be considered a feature of Indo-Iranian religious dialectology.” (synthesis of the current consensus, following Kuiper and Gnoli)[16]

“The fundamental opposition in Zoroastrianism is not between groups of deities but between Aša (truth) and Druj (falsehood).” (Gherardo Gnoli, Encyclopaedia Iranica, “Indo-Iranian Religion”)[11]

Neither camp claims the reversal was instantaneous, and neither denies that Zarathustra’s own hymns are where the Iranian condemnation first becomes fully explicit in writing. The disagreement is about how much prior drift his words are building on, a question that, honestly, the surviving texts may never settle completely.

Verethragna and Vṛtrahán: same word, different myths

The reversal is not the only kind of divergence visible in this shared vocabulary. Sometimes a word survives intact while the myth built around it does not. Indra’s most famous epithet, vṛtrahán, “slayer of Vṛtra,” commemorates his defeat of the serpent Vṛtra, who had swallowed the world’s waters; freeing them is arguably the single most celebrated act in the entire Rigveda, retold across dozens of hymns. The Avestan cognate of this epithet, Vərəθraγna, survived as an independent divine name, the god of Victory later known in Middle Persian as Bahrām.[17]

But the myths diverge sharply. Vərəθraγna in the Avesta is not a dragon-slayer. That role belongs instead to a different hero, Θraētaona (Modern Persian Fereydun), who defeats the three-headed dragon Aži Dahāka in a wholly separate narrative.[18] The word for “smashing the obstacle” and the story of smashing an actual obstacle-dragon have come apart in Iran, while in India they remain welded to the same god and the same tale.

The comparison shows a word and a function separating cleanly: the same Indo-Iranian root produces a still-attached myth in one branch and a free-floating abstraction in the other. See the essay on the Indo-European dragon-slaying formula for the wider comparative pattern this single case belongs to.

Textual witnesses: how we know what we know

Neither scripture reached us the easy way. Both were composed and transmitted for centuries as oral, metrically fixed liturgy before either was written down, and the manuscript record for each begins astonishingly late relative to the presumed date of composition.

Both traditions preserved their oldest religious poetry primarily through disciplined oral transmission rather than early writing, which is precisely why the Mitanni treaty’s written, externally dated pantheon is so valuable: it is the one point in this whole story that does not depend on either community’s own memory of itself.

graph LR
    Comp["Oral composition, Bronze Age"] --> Oral["Centuries of memorized recitation"]
    Oral --> Script["Purpose-built script adopted"]
    Script --> MS["Earliest surviving manuscripts"]
    MS --> Modern["Modern critical editions (Aufrecht; Geldner; Jamison-Brereton)"]

The same basic transmission pipeline, oral fixation followed by a late written stage, applies to both the Rigveda and the Avesta, which is why philologists treat internal linguistic archaism, not manuscript age, as the primary tool for relative dating in both traditions.

Aside. It is tempting to read “1323 CE manuscript” as meaning the Avesta is a young text. It is not; the manuscript date tells us only when Zoroastrian priests began writing down the liturgy that they and their predecessors had already recited from memory for well over a thousand years, using an alphabet designed for that purpose in the Sasanian period. The same caution applies, with even greater force, to the Rigveda, whose oral transmission is generally regarded by specialists as the more conservative of the two.

What to notice while reading the asura hymns

• [ ] Track whether asura appears with deva in the same verse (as in RV 8.25.4); co-occurrence marks the older, undifferentiated usage.
• [ ] Watch for the phrase pattern “asuras who have become devas” versus “asuras who are non-devas”; the Rigveda names its own transitional moment explicitly.
• [ ] Note which mandala a hymn belongs to; the family books (2-7) and the later books (1, 8, 10) use asura in measurably different ways.
• [ ] In the Gāthās, watch for direct address to the daevas in the second person; Zarathustra argues with them rather than merely ignoring them.
• [ ] Compare any epithet built on a verbal root (like vṛtrahán) across the two traditions before assuming the underlying myth traveled with the word.

Scholarly perspectives compared

Despite real disagreement about mechanism, all three positions converge on the same negative conclusion: the deva/asura and daeva/ahura reversals were not present, ready-made, in a common Indo-Iranian religion. They are historical developments internal to each tradition’s own transmission.

Glossary

Asura (असुर): In the early Rigveda, an honorific title for powerful gods such as Varuṇa and Mitra; in later Vedic and classical Hindu literature, a class of demonic beings opposed to the devas.

Deva (देव): The ordinary Sanskrit word for “god,” from an Indo-European root meaning “to shine.”

Daēva: The Avestan cognate of deva; in Zoroastrian usage, a demon or false god rejected by Zarathustra.

Ahura: The Avestan cognate of asura; the title of Zoroastrianism’s supreme god, Ahura Mazdā, “Wise Lord.”

Ṛta (ऋत) / Aša: The Vedic and Avestan words, respectively, for cosmic and moral truth or order, guarded by Varuṇa in the Rigveda and central to Zarathustra’s ethical dualism.

Gāthās: The seventeen hymns of the Avesta attributed to Zarathustra himself, embedded within the larger liturgical text called the Yasna.

Yazata: An Avestan term for “one worthy of worship”; the category into which older Indo-Iranian deities such as Mithra were reorganized in the Younger Avesta.

Vṛtrahán / Vərəθraγna: Cognate epithet meaning “smasher of the obstacle/resistance”; Indra’s title after slaying Vṛtra in the Rigveda, and an independent god of Victory (later Bahrām) in the Avesta.

FAQ

Did Zoroastrianism and Vedic religion come from the same source? Yes. Both descend from a shared Indo-Iranian religious tradition of the second millennium BCE, evidenced by cognate deity names, shared ritual vocabulary (Vedic yajña and Avestan yasna, Vedic soma and Avestan haoma), and the external Mitanni treaty evidence discussed above.

Is “asura” always negative in the Rigveda? No. In the Rigveda’s older family books it is frequently an honorific applied to Varuṇa, Mitra, Agni, and other gods. The demonic sense strengthens in the text’s later books and becomes fixed only in post-Rigvedic literature.

Did Zarathustra invent the idea that daevas are demons? This is genuinely disputed. One scholarly position credits him with the decisive reform; another holds that a general distrust of the daevas already existed and that Zarathustra intensified rather than originated it. Both positions agree the Gāthās are where the condemnation first becomes fully explicit in the textual record.

Why is Ahura Mazdā’s title the same word Hinduism uses for demons? Because both words descend from a single Indo-Iranian term for “lord” or “powerful being.” Each tradition kept the word but sent its moral valence in the opposite direction from the other.

Does the Mitanni treaty prove which tradition is older? No. It dates a shared pantheon of divine names to around 1380 BCE, external to both the Rigveda and the Avesta, but it does not by itself fix the composition date of either scripture. See the dating-debate essay for the wider argument.

Are Indra and the Iranian Vərəθraγna the same god? They share an epithet and an Indo-Iranian origin, but by the time of the Avesta, Vərəθraγna has become an independent deity of Victory, no longer tied to a dragon-slaying myth, which in Iran belongs instead to the hero Θraētaona.

Why does this matter beyond etymology? It is a rare, well-documented case in which two closely related, independently transmitted scriptures let historians watch a shared religious vocabulary diverge into opposite value systems, offering a concrete test case for how religious meaning drifts over a few centuries rather than staying fixed.

Reading the reversal today

The temptation, faced with a story this clean, is to tidy it further: to say Zarathustra flipped a switch, or that India and Iran simply chose opposite sides of an argument that was already settled elsewhere. The evidence resists that. What the Rigveda shows, book by book, is a slow migration of a word’s moral center of gravity, still visible mid-transit in phrases like “asuras who have become devas.” What the Gāthās show is a reformer’s voice already speaking as if the argument is decided, even though the pre-Zoroastrian texts that would let us check his premises survive only in a later, already-transformed form. Both are real historical processes; neither is a myth of a single dramatic break.

The wider Vedic worldview this essay’s other pieces keep returning to, the guardianship of ṛta, the watchfulness of Varuṇa, the binding function of Mitra, all sit downstream of this same vocabulary. Read the essay on Varuṇa beside Yasna 44, where Zarathustra interrogates Ahura Mazdā with a string of unanswered cosmological questions, and the family resemblance between the two traditions’ sense of a moral universe under watch becomes hard to miss, even as the cast of characters has been sorted into opposite camps. For a concrete way in, read RV 8.25 once for its praise of Mitra and Varuṇa as unambiguous “Asuras and imperial Lords,” then read Yasna 32 once for Zarathustra’s flat condemnation of the daevas, and notice that both poets are reaching for the same handful of inherited words to describe what a rightly ordered cosmos, and its opposite, actually look like.

Data appendix: the reversal at a glance

This table consolidates the article’s core comparative claims; every entry is supported by the sources listed below.