The Daiva Inscription (XPh)

Where Behistun was cut once into a cliff, the daiva inscription was made to be multiplied. It survives on squarish limestone tablets, described in Kuhrt's edition as a "Trilingual text inscribed on limestone slabs (approx. 50 x 50 x 10 cm), found in the Persepolis garrison quarters; one copy was found in the Tall-i Takht, Pasargadae".^[1] Wiesehöfer gives the sequence of the finds: it is "an inscription that came to light in 1935 in two Old Persian copies as well as in a Babylonian and later an Elamite version in Persepolis (and in 1963 in a third Old Persian copy in Pasargadae)".^[2] The 1935 discovery fell to Ernst Herzfeld's Persepolis excavations, which published the text in the Archäologische Mitteilungen aus Iran; the Pasargadae copy, from the Tall-i Takht citadel platform, is sometimes separately designated XMa. The tablets belong to a recognised family of near-identical trilingual stone tablets of Xerxes recovered at and around Persepolis and Pasargadae (XPf, XPh/XMa and XPl), with find-spots in the Garrison Quarter, the so-called Harem, the northern quarter of Persepolis and Pasargadae.^[3] Those are not primary depositional contexts. At Persepolis, four XPh tablets had been reused to face a mud-brick bench, while the Elamite exemplar survived in fragments on a wall or bench; the Pasargadae copy had been reused in a channel.^[4] Kuhrt interprets the text as a commemorative foundation document while stressing that the find-spots cannot be the settings for which it was intended.^[1] Its original placement and intended audience remain unknown.

The text carries no date. Kuhrt observes that "it is likely that the inscription dates after the Greek campaign" of 480–479, since its list of subject peoples answers the standard lists of the royal corpus; the inference is drawn from the rhetoric, and, as will be seen, an influential line of scholarship holds that the text was composed precisely so as to be undatable.^[1]

The stones also preserve their engravers at work, in two errors that Kent's edition records. In the second Persepolis copy the crucial word itself is miscut: "The daivadāna- of copy a is daivadava- in copy b", the engraver's eye misled by the sign ending the previous line. And copy b was never finished: it "ends midway in this line, with the left-hand part lacking; the engraver was accustomed to Aramaic writing, which went from right to left", so that a craftsman trained in a right-to-left script worked the cuneiform backwards and abandoned the tablet half-cut.^[5] Both observations go back to Herzfeld's discovery publication, and the second, as it happens, would acquire a curious modern afterlife (see How we know).

The Old Persian version runs to sixty lines, which Kent's edition divides into five movements. It opens with the creation-formula of the dynasty, here in Kuhrt's rendering:

A great god is Auramazda, who created this earth, who created yonder heaven, who created man, who created blissful happiness for man, who made Xerxes king, one king of many, one lord of many.

Xerxes, XPh §1, trans. Kuhrt^[6]

There follows the titulary (§2), in which Xerxes styles himself great king, king of kings, son of Darius the king, "an Aryan, having Aryan lineage" (see Xerxes I), and then the catalogue of subject peoples (§3), the dahyāva list that Darius had made a fixture of the genre. Kent translates the roll:

Media, Elam, Arachosia, Armenia, Drangiana, Parthia, Aria, Bactria, Sogdiana, Chorasmia, Babylonia, Assyria, Sattagydia, Sardis, Egypt, Ionians, those who dwell by the sea and those who dwell across the sea, men of Maka, Arabia, Gandara, Sind, Cappadocia, Dahae, Amyrgian Scythians, Pointed-Cap Scythians, Skudra, men of Akaufaka, Libyans, Carians, Ethiopians.

Xerxes, XPh §3, trans. Kent^[6]

The list admits no losses on the north-western front, although the text probably postdates the Greek war; it instead expands the standard rolls by adding the Dahae and the Akaufaka, presenting the paternal heritage as enlarged rather than merely kept.^[1]

With §4 the inscription turns to deeds, and the first is a suppressed rebellion, left carefully unnamed:

Saith Xerxes the King: When that I became king, there is among these countries which are inscribed above (one which) was in commotion. Afterwards Ahuramazda bore me aid; by the favor of Ahuramazda I smote that country and put it down in its place.

Xerxes, XPh §4a, trans. Kent^[6]

Then comes the paragraph that named the inscription, the most disputed sentence in Old Persian:

And among these countries there was (a place) where previously false gods were worshipped. Afterwards, by the favor of Ahuramazda, I destroyed that sanctuary of the demons, and I made proclamation, "The demons shall not be worshipped!" Where previously the demons were worshipped, there I worshipped Ahuramazda and Arta reverent(ly).

Xerxes, XPh §4b, trans. Kent^[6]

A third act of correction follows, vaguer still: "And there was other (business) that had been done ill; that I made good."^[6] Kuhrt reads §4c as the reign declaring itself a good one, Darius's empire in safe hands.^[7] The fourth movement (§4d) turns from the king's deeds to the reader's soul, binding obedience to happiness in life and blessedness in death; it is treated in its own right below. The text closes (§5) with the protective prayer: "Me may Auramazda protect from evil, and my (royal) house and this land! This I pray of Auramazda; this may Auramazda grant me."^[6]

The philology of §4b carries much of the argument, and begins from a paradox of comparative linguistics. Old Persian daiva- is glossed in Kent's lexicon as "'(false) divinity, demon'", cognate with Avestan daēva- and Sanskrit devá- "deity": the same Indo-Iranian word that in India remained "god" had become, in Iran, the name of the gods one must not worship.^[8] The word occurs three times in XPh (lines 36, 38 and 39) and nowhere else in the whole royal corpus. The thing destroyed is a daivadāna-, in Kent's lexicon a "'sanctuary of false divinities'", compounded of daiva- and dāna-, which Herzfeld took "in derogatory sense, 'den'": a den of demons.^[8]

Because so much hangs on so few words, the translations diverge instructively. What Kent renders "that sanctuary of the demons" is Kuhrt's "that place of the daivas", Boyce's "that Daiva-sanctuary", and, in the translation used by the 2021 Companion, an "altar": "by the will of Auramazda, I uprooted that altar to the daivas".^[6][9][10] Even the number of offending countries is a translator's choice: Kent supplies "(a place)" in the singular, Kuhrt "(some)" in the plural, the Companion's version "one". The final clause diverges furthest of all. The Old Persian artācā brazmaniya, describing how the king worshipped Ahura Mazdā where the daivas had been, is Kent's "and Arta reverent(ly)", making Arta, Truth, almost a second divine object of worship; Kuhrt's "at the proper time and with the proper ceremony"; Boyce's "with due order and rites"; and in the Companion's text "being reverent, I worshipped Auramazda and arta ('truth')".^[6][9][10] Lincoln renders the same phrase "with ritual speech accompanied by truth".^[11] Whether the king says he worshipped the god and Truth, or worshipped him correctly, cannot be settled from the grammar, and the uncertainty feeds directly into the question of the text's Zoroastrianism.

The language versions also diverge at the prohibition itself. Henkelman shows that the Elamite §4b is neither a translation nor a close rendering of the Old Persian "I proclaimed" and Akkadian "I commanded": it makes Xerxes place his kiten, divine protection and god-given royal power, upon the daiva-worshippers. That specifically Elamite statement has no Old Persian or Akkadian equivalent. Henkelman therefore treats it as an authorised strand of the king's polyglot message, not as an expendable translator's flourish.^[12]

What kind of composition XPh is can be established from the standard edition, because Kent prints most of it as equations rather than text. His §1 is given simply as "(= XPa 1-6)": the creation-formula repeats Xerxes's own gate inscription, itself modelled on Darius's tomb text. §2 is "(= XPa 6-11, DNa 13-5)": the titulary is lifted from the tomb of Darius at Naqš-e Rostam. The framing of the lands-list in §3 equates to DNa 18–22, and the closing prayer, §5, is "(= DNa 51-5)", Darius's own words to the letter.^[13] The opening, the titulary, the catalogue and the closing of the inscription are thus inherited wholesale. Within §4, the genuinely composed portion, the first and third members are themselves Darian commonplaces: a country "in commotion" put back "in its place", and unspecified wrongs set right, echo Behistun's idiom of restoration. The afterlife promise itself is not new: Behistun V §§73 and 76 already promise Ahura Mazdā's blessing to the worshipper while living and when dead.^[14] XPh's particular additions are the threefold daiva vocabulary, the otherwise unattested Old Persian artāvā, and their linkage to obedient worship and royal order in a single exhortation.^[15] In a text of some sixty lines, the distinctive matter amounts to a few sentences, and the celebrated paragraph sits inside an envelope of the father's formulae.

The novelty of the daiva itself can be counted. Lincoln's table of Old Persian evil-terminology, eleven terms across the corpus, shows daiva "evil god, demon" used never by Darius, three times by Xerxes, and never by any later king; Darius's own enemies are the Lie (drauga, six times), the liar (draujana, four) and the rebel (hamiçiya, forty-eight).^[16] "Ten of these terms are found in Darius's inscriptions, Xerxes used four", Lincoln observes of the wider vocabulary.^[16] The demon, in other words, enters the royal lexicon with Xerxes and leaves with him; the cosmic enemy Darius had called the Lie acquires, for one reign, a population of false gods (see Religion and the Lie).

The derivativeness was once read as a verdict on the man. The parade case was XPc, a Xerxes text that proved to be Darius's lower tomb inscription reissued under the son's name, and such duplicates, Wiesehöfer reports, "have been considered as proofs of Xerxes's 'lack of independence' (Hinz)".^[2] The psychologising inference has been abandoned. The repetitions, Wiesehöfer argues, "were not due to a 'lack of imagination', but were meant to emphasize the validity of generally accepted principles for his own reign as well": in a genre whose business was the permanent order of things, saying what one's father said was the point.^[2]

If the king destroyed a sanctuary, where was it? The text does not say, and a century of scholarship has occupied the silence. Three proposed locations—Babylon, Egypt and Athens—can be tested against other evidence and fail those tests. Boyce's Iranian shrine is different: an unlocated minority hypothesis that the absence of an excavated site neither proves nor disproves. The programmatic reading asks a prior question, whether the passage requires a recoverable location at all.

The oldest and most consequential answer was Babylon. On this reading the country "in commotion" was Babylonia, in revolt in 484, and the daivadāna was Esagila, the temple of Marduk. Waerzeggers's and Seire's historiographical survey states the once-standard package:

The orthodoxy, most clearly expressed by Cameron (1941) and de Liagre Böhl (1962), held that Xerxes punished Babylon severely after the uprisings of Šamaš-erība and Bēl-šimânni, by taking away the statue of Marduk from its sanctuary, by preventing further celebration of the Akitu (or new year) festival, by destroying the city, by eliminating the element 'King of Babylon' from his official titulature, and by splitting the satrapy of Babylon-and-across-the-River into two smaller units.

Waerzeggers & Seire, Xerxes and Babylonia^[17]

In that reconstruction "the Daiva inscription was used as evidence of Xerxes' supposed policy of intolerance".^[17] The whole edifice has since come down. Kuhrt and Sherwin-White showed in 1987 that its clues were misread, beginning with the classical anchor: "the passage in Herodotus about Xerxes' removal of a statue from the temple of Babylon concerns the statue of a man rather than of Marduk" (Herodotus 1.183 speaks of an andrias, a man's statue, where the divine image is an agalma); the Akitu festival had lapsed long before Xerxes; the title "King of Babylon" faded gradually and recurs under Artaxerxes I; and the excavated sites show no destruction layer. "By the mid-1990s, Xerxes as the destroyer of Babylonian temples had been 'laid to rest'": the image had been "deconstructed as a 'chimaera without substance'", a figure "conjured up by the uncritical acceptance and misreading of classical sources".^[17] What archival evidence does show is targeted reprisal in 484 against the rebel cities' old élite families, with no sack and no razed temples (the case is set out under Babylon). Kuhrt's summary in her sourcebook is categorical: the Babylonian identification "cannot be upheld as it is now certain that there was no destruction of Babylonian temples or cults by Xerxes".^[18]

Egypt, which had revolted at the end of Darius's reign, fares no better: the meagre evidence for the revolt's suppression "says nothing about the elimination of any local cults", and the cults demonstrably continued.^[18]

Athens might seem the strongest candidate, since Xerxes really did burn the temples on the Acropolis in 480, and Persian religious polemic could plausibly reach the gods of a state at war with the king. Herodotus himself supplies the refutation. The day after the burning, as Wiesehöfer notes, "Xerxes ordered that sacrifices again be offered to the (Athenian) gods", by the Athenian exiles in his own train: conduct impossible for a king who had just replaced those gods' worship with that of Ahura Mazdā.^[2] Kuhrt is blunt: "There is no way (contrary to some suggestions) that the destruction of the temple on the Athenian acropolis can be identified as the sanctuary of the 'demons', referred to in Xerxes 'daiva' inscription."^[18] The Companion draws the same conclusion:

if Xerxes had considered the religious practices in Athens against his own religious tradition, he would never have allowed Athenians to start offering to the goddess Athena again one day after the destruction of the Acropolis.

A Companion to the Achaemenid Persian Empire (2021)^[19]

The fourth answer takes the text at its word as religion. Boyce, its most careful proponent, rejects Esagila and the Parthenon alike, observing that "Xerxes' lack of precision as to where this Daiva-sanctuary was has left scope for much scholarly debate", and reads the passage as an act inside Iranian religion: "With the Daiva-sanctuary, Xerxes' words make it clear, there was a religious motive for the destruction", the target an Iranian shrine of the old warlike gods condemned by Zarathustra's tradition. Since no Persian temples are attested so early, "It was presumably in the open... it could have consisted of altars on a plain (as at Pasargadae), or on an artificial mound".^[20] Her reading remains a serious minority position. No site confirms it, and she rejected the one archaeological candidate, the infilled Median structure at Nush-i Jan; but a text that supplies no location cannot make the absence of such a site a refutation.^[20]

The programmatic alternative does not propose a fifth location; it holds that no location is required. Briant's analysis concludes that "None of the proposals regarding the identification of the country in commotion is genuinely justifiable", presses the point that the syntax does not even place the daiva-sanctuary in the country that rebelled, and asks why, if Xerxes had meant a particular country, he did not name it: "Xerxes' phrase is thus deliberately imprecise, both chronologically and geographically".^[21] His verdict makes the inscription a specimen of its genre:

They are first and foremost ideological assertions that, situated in the King's time, do not fit into History's time.

Briant, From Cyrus to Alexander^[21]

Wiesehöfer states the same conclusion as a principle of method, answering the question of the missing place-name from the "timeless" character of Achaemenid royal art and writing:

Xerxes had no specific event in view, he proclaimed once and for all: 'Whoever defects from the king will be punished, and the holy places of rebels will be destroyed.' This is an 'ideological and programmatic' declaration, not a reaction of the king that can be historically pinpointed.

Wiesehöfer, Ancient Persia^[2]

The vagueness, on this reading, is the genre speaking: the inscription describes a type of royal action valid for every rebellion and never needed a recoverable referent. That economy is a strength of the programmatic interpretation, though it does not independently falsify Boyce's minority reading.

What, then, does the inscription attest about Achaemenid religious policy? The current scholarship's first move is to reject the frame in which the question was long asked. In the Companion's formulation, "speaking of Achaemenid 'tolerance' or 'intolerance' against other religions in the empire is a misnomer", because "the reason why the Achaemenids acted in such a harmful way for a religion was never the religious belief itself but a political 'disloyalty' of the given community".^[19] Kuhrt reaches the same ground from the empirical side: "Depredations of local temples, when they occurred, were clearly calculated punishments for perceived rebelliousness, a tactic already used by Darius."^[18] Boyce, though she reads §4b itself as religiously motivated, describes the general policy in matching terms: "Persian tolerance of 'anarya' religions appears to have been the beneficent aspect of a fundamental indifference to them, tempered by pragmatism; and destruction of temples is recorded only as a punitive measure after political provocation."^[22]

The Companion grounds the point in the inscription's own architecture. Paragraphs 4a and 4b run strictly parallel, a rebellious country subdued and a daiva-place destroyed, each followed by the restoration of right order under Ahura Mazdā, and the parallelism is the argument: "If a rebellious country worships gods who back such a rebellion against the Achaemenid reign and Auramazda's will, these gods are daivas, opposing Auramazda and the (political) truth."^[19] On this reading the text defines the daivas relationally, as whatever gods stand behind revolt; it neither names a persecuted cult nor articulates a doctrine about foreign religion, and "it does not say anything about religious tolerance or intolerance of the Achaemenids".^[19]

Lincoln's study of the royal ideology gives the passage its largest frame. The inscriptions understand the Wise Lord's creation as originally happy, and disturbance, famine, war and the Lie, as corruptions the king exists to repel; the vocabulary of evil that Table 13 itemises is the map of that battlefield, and the daivas are Xerxes's contribution to it.^[16][11] Kuhrt's reading of XPh belongs to the same family: "what Xerxes does in this text is to elaborate his father's statement about the debt of loyalty owed by subjects to the empire, by emphasising the centrality of the royal role in mankind's redemption".^[7] Within the Companion's interpretation, the paragraph is a theological statement about kingship: gods understood to back rebellion are classified as daivas.^[19]

Because the daēvas are the repudiated gods of the Avesta, XPh has always been the central exhibit in the debate over whether the Achaemenids were Zoroastrians, and the positions divide over this text more sharply than over any other.

The affirmative case reads §4b as Zoroastrian practice in action. For Boyce the king acts as a Zoroastrian in suppressing a shrine of the very beings the prophet condemned, and the inscription shows him concerned with doctrine and observance together.^[20] The 1988 Cambridge Ancient History reached a like verdict by checklist, testing the kings against four defining beliefs (reverence for the ahuras, rejection of the daēvas, belief in the prophet, worship of Ahura Mazdā) and finding that "the Achaemenid kings can be shown to be Zoroastrians, given these definitions": in the daiva inscription Xerxes is precisely "vidaeva, or against the evil spirits", and the royal silence about Zarathustra himself, on this view, reflects the political purpose of the genre rather than ignorance of the prophet.^[23]

The 2021 Companion gives a qualified yes on different evidence: the shared technical vocabulary, arta and drauga answering Avestan aša and druj, "shows beyond doubt that the Achaemenid rulers are well versed in the tenets of Zoroastrianism", yet "the Achaemenids' Zoroastrianism is not identical with the religion we can reconstruct from the texts of the (Younger) Avesta": Ahura Mazdā heads a plural pantheon, and the kings were never monotheists.^[24]

The sceptics narrow what the text will bear. Briant notes that "the daiva sectaries, in the Gathas, are accused specifically of 'error, breach' (aenah) because of the form of their sacrificial practice", and concludes that "Xerxes principally had in mind the method of sacrifice": XPh would then attest a tightening codification of the official cult of Ahura Mazdā, with Xerxes its guarantor, rather than a doctrinal reformation.^[25] And the once-popular strong form of the reformer thesis, in which a living daēva-worshipping sect survived to be persecuted, rests on Zaehner's reading, which de Jong reports and rejects: "The Achaemenian king Xerxes perhaps tried to remove the last remnants of this ancient faith (as could be deduced from his daiva-inscription), but he was not entirely successful, as Plutarch's description of the sacrifices to Ahreman shows"; the construction, de Jong objects, takes its texts too literally.^[26]

The balance of the field treats XPh as evidence within the Zoroastrianism question rather than an answer to it. The word daiva and the phrase artācā brazmaniya prove contact with the Avestan conceptual world; whether they prove the religion of the Avesta is exactly what the translations cannot agree on (see Ahura Mazdā for the wider question). What would settle the matter, an inscription naming the prophet, a datable sanctuary, an Achaemenid liturgical text, does not exist.

The closing exhortation (§4d) has generated less noise and arguably deserves more, but it is not the first royal promise extending beyond death. In Behistun V §§73 and 76, Darius promises divine blessing to whoever worships Ahura Mazdā, both while living and when dead.^[14] Briant describes XPh as a further step: it introduces artāvā and makes conduct, correct worship and the posthumous condition part of one explicit sequence.^[15] Xerxes's formulation runs:

Thou who (shalt be) hereafter, if thou shalt think, "Happy may I be when living, and when dead may I be blessed," have respect for that law which Ahuramazda has established; worship Ahuramazda and Arta reverent(ly). The man who has respect for that law... he both becomes happy while living, and becomes blessed when dead.

Xerxes, XPh §4d, trans. Kent^[6]

The two loaded words are šiyāta, "happy", the happiness Darius's texts make the created condition of mankind, and artāvā, Kent's and Kuhrt's "blessed", the state of the righteous dead. The second is Avestan to the root: "The Old Persian word artavan, meaning 'blessed', corresponds to Avestan ašauuan", the term for the possessor of aša, Truth, and the same element that yields the throne-name Artaxerxes.^[24] Lincoln translates the promise's fulfilment as "he becomes happy when living and righteous when dead", artāvā rendered as "righteous" rather than "blessed", and brings out how far the paragraph goes beyond its Avestan parallels:

Both Avestan and Achaemenian texts entertain happiness as a possibility and a desire that the righteous can realize in both the present and the heavenly future... both the Avesta and the inscriptions identify two means to win happiness, truth and ritual speech, to which the inscriptions add a third, obedience to the law.

Lincoln, Religion, Empire, and Torture^[11]

Building on Herrenschmidt's study of Old Persian šiyāti, Lincoln reads §4d as the point where royal ideology annexes personal eschatology: the law is the law "which Ahuramazda has established", but the king is its promulgator, so that right worship, obedience and salvation form one circuit.^[11] Kuhrt draws the political conclusion:

The passage, he says, from the world of the living to the hereafter is mediated through the person of the king – only through fidelity to the Persian monarch and his order can the individual be saved.

Kuhrt, The Persian Empire^[7]

It is the most Zoroastrian-sounding sentence any Achaemenid ever published, which is why the reformer thesis clung to this text; it is also, read with §4a and §4b, the completion of the political theology, in which the same act of obedient worship secures the empire's order and the worshipper's soul.