*DAIVA, Old Iranian noun (Av. daēuua-, OPers. daiva-) corresponding to the title devá- of the Indian gods and thus reflecting the Indo-European heritage (*deiu̯ó-), though the category of divinities to which it referred seems to have dropped completely out of the Iranian religious tradition and even to have become demonized. It is extremely difficult to determine just when this change took place and to understand its significance within the framework of Mazdean theology. The impression garnered from various sources is that the process was a gradual one.

In Avestan. In the Gathas the daēuuas had not yet, in fact, become demons. As Émile Benveniste (1967) clearly established, they constituted a distinct category of quite genuine gods, who had, however, been rejected. They were still venerated by the leaders of the larger Iranian nation (dax́iiu-; Y. 32.3, 46.1) and had formerly been worshiped even by the people who accepted the religion of the Gathas (Y. 32.8); they thus formed part of the Mazdean social and religious sys­tem. That they were national gods is confirmed by the fact that they were invoked by means of the Iranian versions of expressions common in Vedic rhetoric, for example, daēuua-/maṧiia-: devá-/mártya-, vīspa-­daēuua-: víśva- devá-, and daēuuo.zušta-: devájuṣṭa-. The poet of the Gathas reproached the daēuuas for being, through blindness, incapable of proper divine discernment (ərəš vī + ci) and of having as a result accepted the bad religion, characterized by aēnah- ­ (approximately equivalent to “error”), along with the good, characterized by auuah- (approximately equiva­lent to “favor”). It appears from the Gathas that the process of rejection, negation, or demonization of these gods was only just beginning, but, as the evidence is full of gaps and ambiguities, this impression may be erroneous. For example, although polemics against the daēuuas and their followers are a major theme of the Gathas, in the other section of the Older Avesta, Yasna Haptaŋhāiti (Y. 35-41; Kellens and Pirart, pp. 30-32, 133-40), they are not mentioned at all. This divergence is extremely puzzling, espe­cially as the doctrines expressed in the two sections are otherwise quite similar. Even in the Gathas no proper names are mentioned, so that it is not even clear precisely who the daēuuas were. Nor does the Younger Avesta (Vd. 10.9, 19.43) appear to shed any light on this specific problem.

Furthermore, in the Gathas the scope of the word aēnah-, and thus of the error to which it refers, is not precisely understood, and criticism of the daēuuas seems to differ slightly in different contexts. In the passage in which their relation to fundamentally negative abstractions (druj-, aka-manah-, pairimaiti-) is defined (Y. 32.3), a syntactical construction otherwise unknown in Old Indo-Iranian has been adopted: mas­culine plural subject plus verb “to be” plus attributive adjective in the form of a singular neuter noun in the accusative plus ablative. The meanings are thus fun­damentally incomprehensible. The pejorative terms applied to the daēuuas are duždāh- “miserly” and xrafstra-, referring to noxious creatures, depicted as harmful in the Younger Avesta, though the signifi­cance of the term in the Older Avesta is not certain. On the other hand, the daēuuas were never identified as drəguuaṇt- “people of the lie,” which would be very significant if it could be demonstrated that it is not a chance of survival.

In the Younger Avesta the daēuuas were represented as small, wicked genies who disturbed the order of the world, human health, and the regularity of religious life, in contrast to the daēuuaiiasna-, literally, “those who sacrifice to the daēuuas,” adherents of other religions. There are, however, exceptions in several passages. For example, the daēuuaiiasnas who of­fered nocturnal libations to Anāhitā (see anāhīd; Yt. 5.94) seem from the evidence to have been Mazdeans who were thus designated because of their deviation from accepted ritual; this passage suggests that in the period of the Younger Avesta the objects of all reli­gious disapproval, whatever petty event may have inspired it, were identified as daēuuas. In the Vidēvdād (10.9,19.43) Iṇdra (Ved. Índra), Sauruua (Ved. Śarvá), and Nåŋhaiθiia (Ved. Nāˊsatya) are mentioned at the head of a list of daēuuas, immediately after reference to Aŋra Mainiiu (see ahriman); in the Pahlavi books the same three were recognized as the enemies of Aṧa, Xšaθra, and Ārmaiti respectively. There are three possible explanations of why these Iranian equivalents to Vedic gods should be mentioned in the Vidēvdād. First, it is conceivable that they were still worshiped in some Iranian circles, though at the period when the Vidēvdād was compiled that seems very unlikely; furthermore, this explanation does not take into account the probability that there was no word *daiva- “god” in Iranian (see below). If, on the other hand, these gods represented an ancient memory, they would provide clues to the identification of several of the daēuuas in the Older Avesta, but this suggestion raises other difficulties, aside from the apparently miraculous survival of this memory. In particular, it is difficult to explain why only these three gods would have been “demonized,” or perhaps not “undemonized,” while Mithra, Vāiiu, and others were not. Finally, it might be suggested that the three gods were simply avatars of the Indian gods, but it would be surprising to find in the Vidēvdād such close links with the Vedic religion coupled with complete adaptation of the names to Iranian phonology.

In Old Persian. The word daiva- appears three times in the plural in an inscription of Xerxes I (r. 486-65 b.c.e.) at Persepolis (XPh, ll. 36, 38, 39; Kent, Old Persian, pp. 150-52), and daivadāna- occurs once, in the accusative singular, in the same inscrip­tion (l. 37), referring to one of the daivas’ sanctuaries: “And in addition (utā) among these countries there was [a place] where previously the daivas were wor­shipped (yad). Therefore, at the command of Ahura Mazdā, I destroyed this sanctuary of the daivas, and I proclaimed, "The daivas are no longer to be worshiped!" In this place where previously the daivas had been worshiped, there I worshiped Ahura Mazdā with a brazman according to Harmony (artācā brazmaniy).” The possible Avestan parallel to the usage of the word daivas in this text remains entirely unclear. Mary Boyce (Zoroastrianism II, p. 175) has argued that Xerxes, “as a Zoroastrian, was recording the destruc­tion of an Iranian sanctuary devoted to the worship of those warlike beings condemned by the propheṭ . . . ” Muhammad Dandamayev, on the other hand, has sug­gested that the daivas mentioned by Xerxes were Mithra, Anāhitā, and the other gods condemned by Zoroaster (p. 226). Ugo Bianchi cited the use of the word daiva- “with an absolutely negative sense (that is to say, not depending upon an adjective or on the context)” as evidence that the Achaemenids were in fact Zoroastrians (1977, p. 1). In the view of all three of these scholars, Xerxes thus understood the word in the sense “rejected god,” which is identical with the Gathic meaning.

Other scholars have taken into account the broader context in which the inscription was composed, in­cluding what is known of the political history of the reign of Xerxes. Somewhat earlier in the inscription (ll. 13-28) Xerxes listed the countries that were part of the Achaemenid empire and reported (ll. 28-35) that, at the time of his accession to the throne, one of them was in revolt; he put down the rebels and restored the country to its former condition. If the country “in revolt” could be equated with the country where false daivas had been worshiped, it would follow that Xerxes understood daiva- in the Younger Avestan sense “god of another religion.” Hans Hartmann, Henryk S. Nyberg (pp. 364-66), and apparently Jacques Duchesne­-Guillemin (1962, p. 156) all identified the country in revolt as Babylon, but Gherardo Gnoli (p. 79) has expressed reserve about this identification, particu­larly in view of what is known of the Achaemenids’ religious policy. The question of Xerxes’ specific policy toward the Babylonian temples was reviewed by Amélie Kuhrt and Susan Sherwin-White, who concluded that the daivas mentioned in the inscription cannot be identified with the gods of Babylon.

On the other hand, Heleen Sancisi-Weerdenburg insists that inscription XPh does not reflect specific historical events but is merely a proclamation of royal religious ideology, announcing in effect “All rebellion will be punished, and the sanctuaries of the gods of the rebels will be destroyed.” If she is correct, the sense of daiva- in the inscription would be similar to that in the Avestan passage in which the daivas are said to have offered nocturnal libations to Anāhitā (Yt. 5.94). It should be added that Xerxes replaced the rebels’ cult with the official Achaemenid cult, represented by his sacrifice to Ahura Mazdā according to a precise ritual (artācā brazmaniy). From this perspective the daivas appear to have been gods, of whatever origin, who might side with rebels, even hypothetical ones, and thus imperil the political and religious order of the empire, which was rooted in the imperial cult. Consid­ering that the Achaemenids did not prevent subject populations from practicing their own religions, as specified in the treaties of political submission and tribute, it must be concluded that the gods whom Xerxes called daivas could be worshiped but that such worship could not be expressed with the verb ya- “to worship.” In Xerxes’ text yad can mean only “to perform the official worship, the worship of the founder.” Although Xerxes’ mention of the daivas is most logically explained in this way, the available documentation does not permit further clarification of the actual historical usage of yad.

Problems of interpretation. It is difficult to reconcile the fragmentary and chronologically discontinuous information on the daivas provided in the sources. There is an essential contradiction between the testi­mony of the Gathas, which suggests that the rejection of the daēuuas constituted a major crisis in Iranian religious thought, and the fact that no known Iranian dialect attests clearly and certainly the survival of a positive sense for *daiva-. Only four possibilities have been identified. First, if a curious amalgam of signs in the last lines of the inscription from Surkh Khotal could be interpreted as δeiio, then Mithra could be said to have borne the title *daiva- (Humbach), but this reading has not been accepted. Second, proper names in which dēv is the first element are known from Sogdiana (Nöldeke, 1923; Henning), but, as Marijan Molé has pointed out, they are attested only from an area where the population had been converted to Bud­dhism. Third, that the white dīv of Māzandarān, the subject of an episode in the Šāh-nāma, may have been an authentic ancient local divinity (Nöldeke, 1915) is purely conjectural. Fourth, Georges Dumézil’s inter­pretation of Ossetic œvdīv as equivalent to *apa-daiva “godless” (1960) was received with skepticism by Benveniste (1959).

If the pejorative sense of *daiva- was common to all Iranian languages, it could not have resulted from an innovation; it would have had to be an original con­stituent of the Iranian language and religion, but it is impossible to make this fundamental fact of Iranian linguistics accord with the information in the Gathas. Attempts to explain the pejorative connotation of *daiva- thus involve either accepting the linguistic evidence while denying that of the Gathas (the con­stituent hypothesis), accepting the evidence of the Gathas and denying that of the language (the reform hypothesis), or attempting to reconcile the two (the progressive hypothesis).

According to the constituent hypothesis, the *daivas had never been Iranian gods. Martin Haug claimed that a schism had occurred at the time of the Indo­-Iranian migrations, a notion that is now out of date. Thomas Burrow, on the other hand, argued that the *daivas were gods of the Indian tribes who lived in the same territory as the Iranians. The daēuuas named in the Vidēvdād (10.9, 19.43) would thus have repre­sented either a memory of the national past or the specific gods of the ancient enemy. Both of these interpretations are contradicted by the Gathic evi­dence, which cannot simply be passed over in silence.

In the reform hypothesis the rejection of the *daivas is considered to have been the work of Zoroaster (Lommel, pp. 88-92; cf. Gershevitch; Bianchi, 1978, pp. 19-22; Gnoli, pp. 73-83). This interpretation is intimately connected with explanation of Mazdaism by means of the dialectical schema: thesis, antithesis, and synthesis. The rejection of the daēuuas would thus have been the core of the antithesis and the fundamen­tal act of Mazdean monotheism. There are several insurmountable flaws in this hypothesis, however. First, it does not account for the fact that a personal doctrine, which was disseminated gradually by mis­sionaries, ended by predominating in exactly the geographical area where the Iranian languages were spo­ken. Second, it depends heavily on the meager evidence suggesting that *daiva- was originally equiva­lent to “god.” Finally, it does not account for the fact that one man was able to impose his condemnation of the traditional gods upon his entire people. Herman Lommel (p. 91) and Ilya Gershevitch (pp. 79-80), though aware of the force of this last objection, argued that the total rejection of these gods is evidence of the genius of Zoroaster.

According to the progressive hypothesis, the rejec­tion of the *daivas was a gradual phenomenon, rooted in the Indo-Iranian past but evolving during the course of Iranian history, a result in which Zoroaster played a role of greater or lesser importance. This hypothesis has three variants. In that proposed by James Darmesteter and Molé the “rejection” consisted of nothing more than a striking lexical accident to the Iranian word *daiva-. This interpretation, like the constituent hypothesis, is contradicted by the Gathic evidence and also by the onomastic data of the Vidēvdād (10.9, 19.43), which Darmesteter sought unsuccess­fully to explain away. This variant is the most extreme and the farthest from the dialectical hypothesis, for it postulates that there was no religious divergence be­tween India and Iran at all. According to the variant proposed by Boyce (Zoroastrianism I, pp. 85, 197), which Duchesne-Guillemin came close to adopting in 1962, the Iranians’ distrust of the *daivas was general, but Zoroaster was the actual artisan of their demonization. To the extent that it accords to the prophet the primordial role, this interpretation has the same weaknesses as the reform hypothesis. The third variant was implied in a paragraph by Antoine Meillet (p. 62) and in Benveniste’s explanation of the forma­tion of the cult of Mazdā (1929, pp. 35-42), but it was fully developed by Nyberg (p. 96) and Duchesne-Guillemin (1953, pp. 27-28). It reflects an attempt to reconcile all the known evidence in acceptable fash­ion. According to this argument, there were two categories of divinity in Indo-Iranian religion, the *asuras and the *daivas. The Indians demonized the first and the Iranians the second; Zoroaster’s role was at most simply to accelerate the latter process. This hypothesis raises new difficulties, however. First, the use of ásura- in the Rigveda is so unsystematic and inconsistent that it can hardly be said to confirm the existence of a category of gods opposed to the devás. Burrow has clearly shown, furthermore, that the demonization of the ásuras took place so late that the associated terms cannot be considered a feature of Indo-Iranian religious dialectology. To propose a prehistoric opposition *ásura/daivá therefore would involve interminable and entirely conjectural discus­sions of the status of Mithra, Apąm Napāt, and *Ahura (Boyce, 1969; Bianchi, 1978). There is also the question of interpreting the Gathic passages if reference to the rejection of the daēuuas was not what was intended. Dumézil foresaw the difficulty when he defined Zoroastrianism as a reform of a reform (1945, pp. 63-64). If it is assumed that the debasement of the daēuuas and the elevation of Mazdā constituted the primary reform, then it is not clear in what the second­ary reform consisted. Johanna Narten (passim) has demonstrated that it could not have been the formula­tion of the system of beings, as Dumézil believed; on the contrary it appears that the rejection of the daēuuas was the focus of attention in the Gathas. Another obstacle to acceptance of this hypothesis is the reap­pearance of the banished gods in the Younger Avesta. The dialectical schema does not account for the diver­sity of the evidence. It is not clear, for example, where Mithra belonged or why the Mazdeans rehabilitated Vāiiu but not Rudrá, the Nāˊsatyas, or even Indra. The evidence does not support any closely reasoned argument. For example, in contrast to Duchesne-Guillemin, who believed that Mithra was “the god of cruel offer­ings and drunken ecstasies” (1953, p. 40), Boyce declared that “Mithra’s general character as god of justice and good faith accords admirably with Zoroaster’s tenets” (1969, p. 17).


E. Benveniste, The Persian Reli­gion According to the Chief Greek Texts, Paris, 1929.

Idem, Études sur la langue ossète, Paris, 1959.

Idem, “Hommes et dieux dans l’Avesta,” Festschrift für Wilhelm Eilers, Wiesbaden, 1967, pp. 144-47.

U. Bianchi, “L’inscription "des daivas" et le Zoroastrisme des Achéménides,” RHR, 1977, pp. 3-30.

Idem, Mithra and the Question of Iranian Monotheism. Études mithraïques, Acta Iranica 17, Tehran and Liège, 1978.

M. Boyce, “On Mithra’s Part in Zoroastrianism,” BSOAS 32, 1969, pp. 10-34.

T. Burrow, “The Proto-Indoaryans,” JRAS, 1973, pp. 123-40.

M. Dandamayev, Persien unter den ersten Achämeniden, Wiesbaden, 1976.

J. Darmesteter, Ormazd et Ahriman, Paris, 1877.

J. Duchesne-­Guillemin, Ormazd et Ahriman, Paris, 1953.

Idem, La religion de l’Iran ancien, Paris, 1962.

G. Dumézil, Naissance d’archanges, Paris, 1945.

Idem, “€ propos de quelques répresentations folkloriques des Ossètes,” Festgabe für Herman Lommel, Wiesbaden, 1960, pp. 39-40.

I. Gershevitch, “Die Sonne das Beste,” Mithraic Studies 1, 1975, pp. 79-81.

G. Gnoli, Zoroaster’s Time and Homeland, Naples, 1980.

H. Hartmann, “Zur neuen Inschrift des Xerxes von Persepolis,” OLZ 40, 1937, pp. 158-60.

M. Haug, Essays on the Sacred Language, Writings and Reli­gion of the Parsees, Bombay, 1862.

W. B. Henning, “A Sogdian God,” BSOAS 28, 1965, pp. 253-54.

C. Herrenschmidt, “Notes de vieux-perse III (artācā brazmaniy),” IIJ 36, 1993, pp. 45-50.

H. Humbach, “Der iranische Mithra als Daiva,” Festgabe für Herman Lommel, Wiesbaden, 1960, pp. 75-79.

J. Kellens, “Trois réflexions sur la religion des Achéménides,” Studien für Indologie und Iranistik 2, 1976, pp. 121-26.

Idem and E. Pirart, Les textes vieil-avestiques I, Wiesbaden, 1988.

A. Kuhrt and S. M. Sherwin-White, “Xerxes’ Destruction of Babylonian Temples,” in H. Sancisi-Weerdenburg and A. Kuhrt, eds., Achaemenid History II. The Greek Sources, Leiden, 1987, pp. 69-78.

H. Lommel, Die Religion Zarathustras, Tübingen, 1930.

A. Meillet, Trois conférences sur les Gâtha de l’Avesta, Paris, 1925.

M. Molé, Culte, mythe et cosmologie dans l’Iran ancien, Paris, 1963, pp. 5-7.

J. Narten, Die Aməṧa Spəṇtas im Avesta, Wiesbaden, 1982. T. Nöldeke, “Dīv,” ARW 18, 1915, pp. 597-600.

Idem, “Deva,” ZII 2, 1923, p. 318.

H. S. Nyberg, Die Religionen des alten Irans, Leipzig, 1938.

H. Sancisi-­Weerdenburg, Yaunā en Persai. Grieken en Perzen in een ander perspectief, Ph.D. diss., Groningen, 1980.

(Clarisse Herrenschmidt and Jean Kelllens)

Originally Published: December 15, 1993

Last Updated: November 11, 2011

This article is available in print.
Vol. VI, Fasc. 6, pp. 599-602