Xerxes I

Xerxes is the Greek and Latin form (Greek Xérxēs) of an Old Persian throne-name written x-š-y-a-r-š-a and read Xšayāršā. Rüdiger Schmitt analyses it as a compound of the verbal stem xšaya-, "ruling", and the noun ṛšan, "hero, man", so that the whole means "ruling over heroes", a formation close in shape to the Vedic kṣáyad-vīra of the same sense.^[1] The Greek and Elamite renderings both reflect a medially shortened variant of the name that, on Schmitt's argument, must already have existed in spoken Old Iranian rather than being coined by the Greeks.^[1] The name passed into every script of the empire, Babylonian, Elamite, Aramaic and Egyptian, and beyond it into the Biblical Aramaic Aḥašweroš, the Ahasuerus of the Book of Esther; the identification of that figure with Xerxes is secure on linguistic grounds, whatever the historicity of the tale told of him.^[2]

The sense of the name is itself a claim of kingship, and it belongs with the titulary the king carved at the head of his monuments: great king, king of kings, king of countries containing all kinds of men. The throne-name was borne after him only by a grandson, Xerxes II, whose few weeks on the throne in 424 are nearly his whole record.

More is known of Xerxes, in one sense, than of any other Achaemenid king, because he is the one who invaded Greece, and the war became the founding event of classical Greek political and literary self-understanding. That prominence is also the central difficulty of the reign, for it is almost wholly a Greek prominence, made by the losers' descendants and shaped to their needs. Amélie Kuhrt states the corrective bluntly: the danger is that of "letting a setback suffered in one campaign along the empire's north-western fringe define an entire reign of twenty-one years", when the administrative documents from Persepolis and the archaeology of the palaces show "no disruption in production or major building activities throughout Xerxes' reign".^[3]

The two great Greek sources are works of extraordinary quality and equally extraordinary bias. The earliest is the tragedy The Persians of Aeschylus, staged at Athens in 472 BCE by a poet who had fought in the war; it sets the defeat in the Persian court at Susa, brings on the ghost of the dead Darius to rebuke his son, and makes Xerxes return in rags. The ghost's rebuke of the son who bridged the strait is the locus classicus of the whole tragic reading of the reign:

He conceived the hope that he could by shackles, as if it were a slave, restrain the current of the sacred Hellespont, the Bosporus, a stream divine; set himself to fashion a roadway of a new order, and, by casting upon it hammer-wrought fetters, made a spacious causeway for his mighty host. Mortal though he was, he thought in his folly that he would gain the mastery over all the gods, aye even over Poseidon.

Aeschylus, Persians 745–750, trans. Smyth^[4]

And the ghost asks: "Must this not have been a distemper of the soul that possessed my son?"^[4] Yet the play fixes its moral less firmly on the bridge than the later shorthand suggests. Its only explicit references to hybris (lines 808 and 821) come in Darius's prophecy of Plataea, and attach in the first instance not to Xerxes or his bridging but to the troops left in Greece, punished for "violating and vandalizing shrines and sanctuaries".^[5] The moral the ghost draws aloud is spoken of Plataea and the shrine-sackers, not of the sea:

heaps of dead shall make known, even to the third generation, a voiceless record for the eyes of men that mortal man needs must not vaunt him overmuch. For presumptuous pride, when it has burgeoned, bears as its fruit a crop of calamity, whence it reaps a plenteous harvest of tears.

Aeschylus, Persians 818–822, trans. Smyth^[4]

The ghost who condemns the bridging of one strait had himself bridged the Bosporus, a starkness that is dramatic requirement rather than remembered difference, and A. F. Garvie's commentary concludes that the play "does not have a single simple 'moral'": Aeschylus sets "side by side amoral and moral explanations of human suffering" and leaves the choice open, so that the king's offence "can be detected only in retrospect. The recognition comes with the failure".^[6] The play is in any case an Athenian construction, not a window onto Persia, though parts of its picture, from the tribute to the opposition of the Persian bow and the Greek spear, are confirmed by the Persepolis reliefs, with the caveat, in Margaret Root's words, that "neither Aeschylus' word-portraits nor the Achaemenids' form-portraits can be taken at face value".^[6] Janett Morgan judges the play "an artificial construct" that "presents the Athenian audience with a view of 'Persia', a mythical place that acts as a foil for Athenian needs and Athenian politics".^[7] Even that judgement marks one pole of a live debate: Garvie holds that the Athenian audience, though victors, were "still able to identify with the losers in their suffering", a genuine tragedy of shared humanity, where Edith Hall reads the play as "the earliest testimony to the absolute polarization in Greek thought of Hellene and barbarian" and Thomas Harrison presses the case until it is "not really a tragedy at all; it is an indictment of Persian monarchy".^[6]

Herodotus, writing a generation later, gives the fuller narrative and the same moralising shape; Maria Brosius notes that he works to "a literary structure", alternating morally good and bad kings, with Cyrus and Darius in the first class and Cambyses and Xerxes in the second, so that the vain and impious Xerxes is a figure the scheme requires.^[8]

Behind them stands Ctesias of Cnidus, whose fourth-century Persica supplies the only continuous ancient account of the reign's end and much of its court colour, and whose reliability is lowest of all. Pierre Briant treats him as the chief author of a distorting Orientalism, a writer who shows the Persian court "through the keyhole" of harem gossip.^[9] The nineteenth-century verdict of Eduard Meyer, that the Persica is a work "breathing seraglio and eunuch perfumes, mixed with the foul stench of blood", captures the whole tradition of the effeminate, capricious oriental despot that the Greek and later European imagination built on Xerxes.^[10] That image was named and attacked from 1978, when Edward Said's Orientalism prompted Heleen Sancisi-Weerdenburg to identify Ctesias as its first proponent, and the inaugural Achaemenid History Workshop at Groningen in 1983, led by Briant, Kuhrt and Sancisi-Weerdenburg, reopened the whole question of whether the empire had really been in decay.^[10] The result is that the modern task with Xerxes is less to add evidence than to correct for the bias of what survives.

Nor is the Greek image one image: the hostile lineage from Aeschylus through Herodotus to Ctesias is itself a simplification. Emma Bridges, in the standard study of the king's reception, separates the strands: the anguished tragic king of Aeschylus, the complex and human king of Herodotus, the panhellenic foe of the fourth-century orators, the pious benefactor of Josephus, the magnificent monarch of the Book of Esther, and, in the court narratives that Herodotus's harem stories opened and Ctesias developed, "the vision of Xerxes as the epitome of the decadent palace playboy which would win out".^[5] Her conclusion is a methodological warning for the whole reign: the successive re-imaginings show that the king "inspired an astonishing array of responses; in this sense the 'real' historical Xerxes has become inseparable from the literary character(s) who developed in his wake".^[5]

Xerxes was not Darius's eldest son, and his elevation turned on the meeting of two royal lines in his person. Darius had older sons by an earlier marriage to a daughter of the noble Gobryas, the eldest being Artobazanes; Xerxes was the eldest of the sons borne to Atossa, and Atossa was a daughter of Cyrus the founder. Xerxes's own account of the succession is inscribed at Persepolis (XPf), and it is unusual precisely in feeling the need to explain the passing-over of his brothers:

Darius had other sons also; (but) thus was the desire of Auramazda: Darius, my father, made me the greatest after himself. When my father Darius went to his (allotted) place, by the favour of Auramazda I became king in my father's place. When I became king, much that (is) superior I built. What had been built by my father, that I took into my care and other work I added.

Xerxes, XPf §4, trans. Kuhrt^[11]

Brosius reads the phrasing carefully: that Xerxes was "chosen as king from among his brothers" implies he was selected by the king to be heir, not that he held a natural right, and there is nothing to show that the choice was made only late in the reign.^[12] The Old Persian term for the choice, mathista, "the greatest", carries no implication that Xerxes had been made co-regent; Kuhrt notes that there is no evidence for the co-regency thesis of Calmeyer, and that the Babylonian version of the text makes clear Xerxes became king only when Darius died.^[3] What is secure is that Xerxes was designated well before Darius's death, most likely before the Egyptian campaign of the last years.^[12] The designation evidence at Persepolis divides into two exhibits of unequal strength. The secure one is labelled: on the east jamb of the south doorway of the Tacara, Darius's private palace, the figure of a crown prince stands opposite his father, and the trilingual inscription on his skirt names him, "Xerxes, son of King Darius, an Achaemenid".^[13] The famous unlabelled audience reliefs, traditionally read as Darius enthroned with Xerxes standing behind him as crown prince, are weaker witnesses: re-examination showed that the panels had been cut for the centre of the Apadana stairways, not the Treasury where they were found, and they may instead show Xerxes as king with his own son Darius as the heir behind him, a re-reading already reported in the 1988 Cambridge Ancient History and material to the end of the reign (see the assassination below).^[13]

The ancient traditions divide on how the choice was made, and the division is instructive. Herodotus (7.2–3) sets the dispute before Darius's death and turns it into a duel between Artobazanes, who claimed the right of the eldest, and Xerxes, who claimed descent from Cyrus and precedence as the first son born after Darius became king (born, in the Roman phrase, in the purple); the exiled Spartan king Demaratus, then at the Persian court, is made to supply the winning argument, and Herodotus adds that Xerxes would have prevailed anyway, "for Atossa was all powerful".^[14] Justin (2.10) and Plutarch instead set the dispute after Darius's death, make it fraternal and non-violent, and have it arbitrated within the family by the king's brother Artaphernes, who decides for Xerxes; the "Ariaramnes" of Justin is a variant of Herodotus's Artobazanes.^[15] A further tale of armed opposition from an Ariamenes of Bactria, ultimately from Ctesias, was already being discounted in the 1988 Cambridge Ancient History.^[13] Modern scholarship keeps the shared kernel and discards the Greek elaboration. The fundamental point, that Xerxes was not the eldest and that his appointment was in some way exceptional, is confirmed by his own inscription; the figures of Demaratus and the kingmaking Atossa are Herodotean invention.^[3]

The role of Atossa is the sharpest case. Herodotus makes her the decisive power behind the accession, but her name is absent from Xerxes's own statement of legitimacy and from the Old Persian royal inscriptions and the published Fortification texts alike, and Briant rejects the constitutional reading outright: there was strictly no Achaemenid queen, the mother of the crown prince held no special right, and Xerxes stresses instead the patrilineal transmission of power.^[9][12] Matthew Stolper draws the distinction that the field now accepts: the "all-powerful" Atossa is the literary Atossa of Herodotus, "a trope about harem power", and is incommensurate with the historical figure, who did not in fact exercise such power at Xerxes's accession.^[16] What decided the succession, on the modern reading, was pedigree: as the son of Atossa and grandson of Cyrus, Xerxes united Darius's line with the founder's, and it was this that made him the physical embodiment of the dynastic claim Darius had built at Behistun.^[2] He came to the crown in November 486, a grown man in his mid-thirties who had governed, which itself tells against the later Greek portrait of an untried and pampered prince.^[17]

Darius died with a revolt in Egypt already burning and a second Greek war unlaunched, and the first business of the new reign was to hold the empire together. Egypt had risen in 486, its burdens of tribute and conscription sharpened, the Greek sources suggest, by the check at Marathon and by Darius's demands for renewed war. Xerxes reduced it in 485 and installed his brother Achaemenes as satrap, an appointment that fits the Achaemenid habit of setting close kin over the great provinces.^[2] Two documents peg the chronology. Trouble was open by June 486, months before Darius died in November: a contemporary letter shows a local Egyptian, Khnumemakhet, ordered by the satrap to bring grain down from Nubia, protesting that it cannot be unloaded without a strong guard, since, in the Cambridge Ancient History's rendering, "rebels are already so bold as to parade within the sight of the authorities during the daytime".^[13] At the other end, the first clear sign of order restored is an inscription recording royal quarrying begun again in the Wadi Hammamat by 9 January 484, which brackets the suppression at up to a year and a half.^[13] Herodotus says the king made Egypt's servitude harsher than under Darius, but Brosius flags that no Egyptian evidence corroborates a genuinely harsher policy, and the claim remains disputed.^[8] The Egyptian record is not quite neutral in that dispute, though: unlike his father, Xerxes built no royal monuments in Egypt and left no evidence of ruling as pharaoh, a silence already read in 1988 as a real change of register, so the argument is properly between silence as accident and silence as policy.^[13]

Babylon then rose twice, under two pretenders, Bēl-šimânni and Šamaš-erība, each of whom took in his dating-formulas the title "king of Babylon and the lands", claiming not merely local kingship but the empire as a whole. One Babylonian tablet is dated to the "accession year of Belšimanni king of Babylon", and he seems to have held the title for only about a fortnight.^[18] The chronology of the two risings is a historiographic problem in its own right, and the successive positions are worth setting out, because the reign's whole shape has been read from them. The old orthodoxy of F. M. Th. de Liagre Böhl placed the revolts in two separate years, Bēl-šimânni in 484 and Šamaš-erība in 482, arguing from a supposed change in Xerxes's titulary.^[3] That two-year split was not a curiosity of the 1930s but the state of the art in the standard reference work for decades. Kuhrt herself, writing in the Cambridge Ancient History in 1988, dated Bēl-šimânni to Xerxes's second year, the autumn of 484, and Šamaš-erība "probably, though not certainly" to his fourth, the autumn of 482, on documentary rather than titulary grounds, noting that fresh material had by then made it impossible to keep the older scheme that had dated both rebels to the same year, 482. The historiography thus ran from both-in-482, through the 484-and-482 split that even the destruction myth's own chief refuter still held, to Waerzeggers's both-in-484.^[13] Briant proposed instead to redate them later, to about 481 and 479, so as to fit Xerxes's movements after the Greek war.^[9] The current consensus, established by Caroline Waerzeggers from the Borsippa archive evidence, places both revolts firmly in Xerxes's second regnal year, 484 BCE; the earlier dating, Waerzeggers argues, had rested less on the evidence than on "the pervasive image of Xerxes as destroyer of sanctuaries".^[19] The revolts were suppressed, in the classical tradition by Xerxes's son-in-law Megabyzus, and Babylonia was later divided, the great satrapy of Babylon-and-across-the-River split into two, a division the recent handbook of imperial administration treats as settled fact, Egyptian officialdom dealing thereafter with a Transeuphratene "independent from Babylonia since the administrative reform of Xerxes I".^[2][20] That redivision the Cambridge Ancient History had already, in 1988, judged more probably the work of Xerxes than of Darius, the combined governorship of Babylon-and-Across-the-River being still attested late in Darius's reign (see Darius I).^[13]

What drove the rising has lately acquired an economic face. Tax documentation is ubiquitous in the Persian-period Babylonian archives, far more so than under the preceding Neo-Babylonian empire, and Michael Jursa and Martina Schmidl reckon the minimum annual load on an elite propertied family, from the best-documented Borsippa archive, at 48 shekels of silver for direct taxation and service obligations, slightly less than the price of a male slave or the annual wages of a well-paid hired worker, bearable for the affluent and beyond the reach of lesser households.^[20] In Xerxes's first two years, letters from the Ezida temple at Borsippa show a royal official stopping the payments owed to the priests for their cultic services, the only plausible reason being that the funds had been requisitioned by the crown; the episode "may have directly preceded the revolts in 2 Xerxes, but this cannot be proven", yet it is at the least "a clear indication of an economic (and resulting social) crisis in northern Babylonia these years, which, if it was not actually caused by Persian demands on Babylonian resources, was certainly exacerbated by them".^[20] The revolts, in the same study's phrase, are "a major watershed in the history of Babylonia under Persian domination", and "it is beyond discussion that the Persian reprisals targeted the Babylonian urban elites, especially priests and families with ties to local government".^[20]

The durable historiographic gain of the modern work is the recovery of what the suppression actually meant. There is no destruction layer at Babylon (see below), but there is a sharp and datable social rupture. A cluster of some two dozen private archives from the north-Babylonian cities, Babylon, Borsippa, Sippar and others, breaks off in Xerxes's second year, and the owners are the prebendary urban elites who ran the temples. Robert Rollinger draws the conclusion that Xerxes "did not just crush the rebellions but grasped the opportunity for taking action in order to deprive traditional north Babylonian elites of their power base".^[21] Brosius reads the same "end of archives" as evidence that Xerxes took severe measures to curb the political influence of the great Babylonian families.^[8] The change in the royal titulary, once read as an abrupt demotion of Babylon after 482, is now understood differently: the element "king of Babylon" was dropped only gradually, remained in occasional use in the post-revolt years, and is even attested under Artaxerxes I, which points to slow cultural change rather than a single act of retribution.^[19] The correction is older than the current literature suggests: tablets dated between Xerxes's fifth and sixteenth years, using both "King of Parsa and Mada" and "King of Babylon", were already known to the 1988 Cambridge Ancient History, which concluded that it was "no longer possible to be sure that a change in titulary early in the reign of Xerxes definitely was ordered", or that any such change could be tied to the rebellion.^[13] Nor was the aftermath only punitive. As Kristin Kleber argues, the fiscal reforms gradually introduced under Darius and before "were established more decisively under Xerxes", a more uniform basis of taxation on land being among the results; the editors of the administration handbook gloss the pattern as the Persians showing "their understanding of the principle of never wasting a good crisis".^[20] That transitional reading has a 1988 pedigree of its own. Kuhrt had already warned in the Cambridge Ancient History that the supposed Babylonian decline after 482 was largely an artefact of text survival, to be "treated with extreme caution", and pointed to a fragmentary family archive, that of the Tattannu of Borsippa, whose members traded on under Xerxes on a scale "comparable in scope and size to the famous Murashu firm"; she proposed that the reign be regarded "as a transitional period in which the reforms initiated by Darius I were beginning to take effect", the very position Kleber now states as consensus. Her argument turned the break in the archives against the decline thesis, where Waerzeggers has since read the same break as the dated fingerprint of Xerxes's reprisals; the two are complements rather than rivals, a targeted rupture of the rebel elite without a general economic collapse.^[13]

The great expedition of 480 was, from the Persian side, an imperial enterprise rather than a civilisational duel, and reconstructing its aims means reading past the Greek framing. Herodotus makes the decision a study in the Persian compulsion to expand and stages an elaborate council of war, but Briant shows that the whole debate is composed after the event and built from Greek literary types: the ambitious counsellor against the wise one, with the effect of making Xerxes indecisive and, by implication, cowardly, which suited the tradition.^[9] George Cawkwell puts the source problem at its sharpest: the reported discussions between Xerxes and his advisers "are almost certainly entirely fictitious and must be rigorously excluded from the mind when one considers Persian aims and strategy".^[22]

What can be reconstructed is a layered rationale. The campaign continued Darius's aim of settling the empire's north-western frontier in Persia's favour, and it took up projects his father had already begun; it also let a new king prove his ability to lead a victorious expedition and, on the ideological plane, renew in person the bonds of loyalty owed to the throne, the empire being put on display at strategic points along the march.^[3] John Hyland's recent study makes this the centre of the reign: Persian policy toward the Greeks arose "not from traditions of East-West conflict or lust for territorial expansion, but from efforts to maintain the imagery of world rule and royal glory through displays of power and patronage on distant maritime frontiers", and Xerxes's own march was "a performative spectacle of heroic kingship which aimed to bolster the empire's internal cohesion".^[23] The oldest Greek explanation already dramatises the same structural pressure on a successor. Only eight years after the event, Aeschylus has Atossa account for the decision as the goading of a new king measured against his father:

This lesson impetuous Xerxes learned by converse with the vile. For they kept ever telling him that, whereas thou by thy spear didst win plenteous treasure for thy children, he, on his part, through lack of manly spirit, played the warrior at home and made no enlargement of his father's wealth. Hearing such taunts many a time and oft from evil counsellors he planned this expedition and armament against Hellas.

Atossa, in Aeschylus, Persians 753–758, trans. Smyth^[4]

The scene is theatre, not testimony, but it shows that the earliest Greek reading of the war's origin was psychological and dynastic, the son taunted with adding nothing to his father's wealth, before it was providential.

This was not incompatible with conquest. Briant insists the military objective was clearly outright conquest of the Greeks, not merely vengeance on Athens, which is what distinguishes the expedition of 480 from the raid Datis had led in 490.^[9] The frame of a clash of civilisations, still common in modern accounts, would have been foreign to the Persians, and it misdescribes the war on its own terms: the Achaemenid empire was multi-ethnic, and Greeks fought on both sides, from Thessaly and Thebes to the Ionian contingents of the fleet.^[21] Not every scholar has abandoned the older frame: Manousos Kambouris deliberately keeps the war as 'A True Cultural World War' and reads Xerxes's aims in terms of spoils and the radiance of a warrior king, a traditionalist pole against which the display-and-consolidation reading defines itself.^[24]

The preparations were on a scale the Greeks never tired of, and their two most famous works, the canal cut through the Athos peninsula and the bridges flung across the Hellespont, became for later rhetoric the very emblems of megalomania. Both were in fact rational engineering. The canal, cut where an earlier Persian fleet had wrecked rounding Athos in 492, has been traced by modern excavation between Nea Roda and Tripiti; it was a real channel, wide enough for two triremes, and its purpose was practical as well as commemorative.^[23] Herodotus himself reports it was dug so that the king might "display his power and leave a memorial", and Benedikt Isserlin, who directed the excavation, reckoned a workforce of about two thousand could have finished it in sixty to eighty days, against Herodotus's three years.^[23] The bridging of the Hellespont was likewise a measure of sense: as A. R. Burn observes, set against the alternative of ferrying a large army and its animals in the boats of the time, "the bridge was a practical measure", and the digging of the canal was cheaper and safer than dragging ships overland.^[25]

Onto these works Herodotus fastened the reign's defining act of impiety. When a storm broke the first bridge, the king had the strait punished:

When Xerxes learnt this, he was furious and ordered three hundred lashes to be laid upon the Hellespont and to lower into the sea a set of fetters.

Herodotus 7.35, trans. Kuhrt^[14]

The scene is the seed of the whole tragic reading, the mortal who would master the sea. It need not be read as the Greek reading takes it. Kuhrt, following Mary Boyce, notes that the whipping of the water may be a ritual act, tied to the Iranian belief that salt water is sweet water contaminated by the forces of chaos, rather than a fit of a mad tyrant.^[3] The recurring image of Persian soldiers driven to fight and work under the lash is, on Kuhrt's analysis, a Greek ideological device meant to stress the difference between free men and subjects, with no other evidence behind it.^[3] The same verdict comes from inside the operational tradition: Lazenby, closing his military study of the war, holds that the explanation of the Persian failure as "the courage of free men pitted against an oriental despotism" can be "confidently dismissed", since "war puts no premium on 'freedom'", and that there is no reason to believe the king's non-Persian subjects fought any less loyally or fanatically, Herodotus himself singling out for praise, against the grain of his own trope, the Egyptians at Artemisium, the Samians at Salamis and the Sacae at Plataea.^[26] The great reviews of the host, at Abydos before the crossing and at Doriscus after it, are best understood in the same key. Briant argues that the catalogue of forty-seven ethnic contingents Herodotus enumerates at Doriscus was never a battle order at all: it was "a review of the Empire", a ceremonial microcosm of the realm's diversity held up as a mirror to the king's own power, distinct from the far smaller fighting force that actually marched on.^[9]

No question about the reign has been more thoroughly revised than the size of the invasion force. Herodotus builds his figures into a total of some 2,600,000 fighting men, doubled with camp followers to more than five million souls, a host he expressly likens to the army of the Trojan War.^[9] The figures are rejected across the modern literature on logistical grounds alone; Michael Dandamaev observes that Herodotus followed an oral tradition "designed to show that a tiny, freedom-loving people was attacked by the huge army of a despotic king", and Hyland reads the grand total as "a quantification of hubris", staged for an audience primed to marvel before celebrating the destruction of the marvel.^[17][23] Manning notes that the neat decimal army of "myriads" is itself a Greek literary construct: no ancient document confirms Persian units of ten thousand, and Herodotus is in effect parodying Persian counting at Doriscus.^[27]

The realistic figures are much smaller and are converging. Briant, reasoning from the only pitched land battle fought on Greek soil, Plataea, where the Persians did not greatly outnumber the Greeks, arrives at about 60,000 for the field army and around 600 triremes for the fleet, against the 1,207 of Herodotus and Aeschylus; Kuhrt adopts the same order of magnitude.^[9][3] The downward convergence is in fact older than the Achaemenid-history school. Charles Hignett, in the classic operational monograph of 1963, preferred Meyer's ceiling of 100,000 for the Asiatic troops and reconstructed a force of 70,000 Asiatic infantry with some 9,000 cavalry, "approximately 80,000 men", which "would have been large enough to overcome any resistance which the Greeks could offer on land in 480".^[28] Lazenby reached the same order by an independent route, from the stockaded camp Mardonius built at Plataea, which Herodotus makes ten stades square: its estimated capacity of 60,000 to 70,000 men yields "a total of something like 70-80,000 men for Xerxes' Asiatic army", with perhaps another ten or twenty thousand Thracian and Greek allies picked up on the march.^[26] Hignett also reports the elegant conjecture, suggested to him by an Oxford scholar, that the Greeks systematically misread a Persian chiliad, a unit of a thousand, as a myriad of ten thousand, which applied to Herodotus's figures produces 170,000 infantry and 8,000 cavalry; the theory sits suggestively beside Manning's point about the myriads, though Lazenby judged it, like the other figure-juggling hypotheses, very difficult to believe.^[28][26] The fleet's canonical 1,207 is epic arithmetic of the same kind as the Trojan comparison: Hignett showed from Herodotus's own narrative that the Persian fleet at Salamis was only slightly superior to the Greek, so can hardly have exceeded about 600 at Doriscus, and that the traditional 1,207, "already known to Aeschylus, must be an invention", arising perhaps from a desire to credit Xerxes with a fleet larger than the one Agamemnon led against Troy, the ship-numbers in our manuscripts of the Iliad's catalogue summing to 1,186.^[28] The upper bound is fixed by water: General Maurice, going over the route after the First World War, reckoned the Hellespont region could sustain at most about 210,000 men and 75,000 animals, and, as Hyland stresses, an absolute logistical ceiling is not a headcount.^[23] At the higher end of the modern range, Duncan Head reconstructs an order of battle of about 146,000 foot and up to 40,000 horse, while still endorsing the logistic argument that even a host of 210,000 would have strained supply past breaking.^[29] The most careful recent scholarship favours "well under 100,000" for the land army; Hyland's own minimum is about 50,000 combat troops with 20,000 non-combatants, and a fleet of about 650 triremes, roughly 140,000 men including marines.^[23] The mechanism behind the two very different pictures is Briant's distinction, which Hyland endorses, between a short-term "parade army" displayed at the reviews and the real combat force that continued to Athens.^[23] The remaining spread among scholars is still wide, from the 50,000 to 70,000 of the tradition Delbrück founded to the 180,000 to 200,000 of the advocates of larger figures, but the direction is settled: the invasion army was large for its day and essentially Iranian at its fighting core, not the numberless motley horde of the tradition.^[23][22]

Read from the Persian side, the campaign of 480 was, until its last act, a sequence of successes. At the pass of Thermopylae a Spartan-led stand held the narrows for some days but was turned and overwhelmed; the encounter cost the king enough that Herodotus has him leap in fear from his watching-seat:

During these encounters in the battle, they say that the king, as he watched, leaped up three times from his throne afraid for his army.

Herodotus 7.212, trans. Kuhrt^[14]

The Greek tradition made Thermopylae a moral victory of free men over slaves driven by whips, but on the evidence it was a Persian victory. Hyland argues that Herodotus has distorted a plausible case of Achaemenid ideological display, and that the viewing of the enemy dead and the treatment of Leonidas's body correspond well with internal Persian representations of royal triumph; Lloyd Llewellyn-Jones puts it plainly, that in under a week Xerxes had broken the last barrier before Athens.^[23][30] Athens itself was then taken and the temples on the Acropolis burned, which completed the punishment of the city that the campaign of 490 had failed to inflict.^[8] The Persian record of the event is telling in its own way: Xerxes reported his "good fortune" to Susa and had the Athenian exiles in his train go up and sacrifice on the citadel after their own custom, a detail that later bears on the reading of the daiva inscription.^[3]

The reverse came at sea, in the narrows off Salamis. In the tradition the battle opens with a ruse: Themistocles sent word to the king that the Greeks meant to slip away, and the fullest Persian-angle account, Diodorus's, which Kuhrt prints, has Xerxes press the attack on the strength of it, into water that cancelled his numbers:

the king gave orders to his admiral to advance against the enemy, while he himself moved down the coast to a spot directly opposite Salamis from which he could watch the course of the battle... in the struggle forty ships were lost by the Greeks, but more than two hundred by the Persians.

Diodorus 11.18, trans. Kuhrt^[31]

Whether the famous message was ever sent is itself disputed, and the two classic monographs on the invasion stake out the poles. Charles Hignett rejected the story as one of the legends that gathered round the battle: the Persians are asked to have accepted "without question the eleventh-hour assurances brought by Sikinnos", to have acted on them without any attempt at confirmation, and not even to have detained the messenger, a credulity "too staggering to be accepted as fact"; on his reading no bait was needed, for once the Greeks stood at Salamis the Persian admirals were "faced with a deadlock" and "had no choice but to retreat or to introduce their fleet by night into the straits".^[28] J. F. Lazenby defends the tradition: the story was current within eight years of the event, for Aeschylus already has it, Herodotus supplies circumstantial detail and Thucydides accepted it, and "that a message of some sort was sent, there is no good reason to doubt", though its content was probably no more than the warning of a Greek flight that Aeschylus gives it.^[26]

For the fight itself the earliest voice is again Aeschylus, whose messenger-speech is the oldest surviving description of the battle, composed within a decade by a probable eyewitness. The Greek line came on with a cry the play made immortal:

On, ye sons of Hellas! Free your native land, free your children, your wives, the fanes of your fathers' gods, and the tombs of your ancestors. Now you battle for your all.

Aeschylus, Persians 402–405, trans. Smyth^[4]

And the same speech gives, from the Greek side, the mechanism the modern accounts reconstruct analytically, a fleet too large for the water it fought in:

but when the mass of our ships had been crowded in the narrows, and none could render another aid, and each crashed its bronze-faced beak against each of its own line, they shivered their whole array of oars; while the Hellenic galleys, not heedless of their chance, hemmed them in and battered them on every side. The hulls of our vessels rolled over and the sea was hidden from our sight, strewn as it was with wrecks and slaughtered men.

Aeschylus, Persians 413–420, trans. Smyth^[4]

The watching king on the shore is the counterpart of his throne at Thermopylae, the Persian idea of valour proved under the royal eye; among the dead was the fleet's commander Ariabignes, Xerxes's own half-brother.^[3] The one captain said to have counselled against fighting in the strait was Artemisia of Halicarnassus, whose advice the king valued.^[3] Aeschylus adds a coda on the islet of Psyttaleia, where the picked Persians the king had posted to kill shipwrecked Greeks were themselves slaughtered; the play makes the episode outweigh the sea-fight because, as Garvie puts it, "the sea-battle involved mainly the allies, but on Psyttaleia it was the noblest and most faithful of the Persians themselves who died", once more by "Xerxes' misjudgement", though the exaggeration of the islet's importance is, Garvie judges, "not for political, but for dramatic reasons".^[6] Why Xerxes forced a naval battle in so poor a position, throwing away his advantages of number and speed instead of trusting to a blockade, is the crux of Salamis; Hyland frames it as the central historical problem, and the older political history of A. T. Olmstead read it flatly as the king's own blunder, a spectacular victory sought where patience would have served.^[23][32]

Herodotus presents the aftermath of Salamis as a panicked, starving flight, and this too is overdrawn. Xerxes withdrew to Asia with part of the army, but he left his general Mardonius in Greece with a picked force, the best of the Iranian troops and the Guard, to finish the war the next year, and the northern route home stayed loyal, Thessaly, Paeonia and Macedon caring for his soldiers and Thracian and garrison posts holding for him.^[3] Kuhrt takes the loyal north and the quality of the force left behind as decisive against the picture of a decimated rout; Hyland doubts the king viewed the journey back to Sardis as a retreat in any sense.^[23] Burn made the same point earlier: Xerxes's return to show himself safe at Sardis was "no more than Darius had done after the Scythian campaign", and Mardonius kept the main force, including all the good Iranian troops.^[25]

The war was nonetheless lost in 479. Mardonius's army, which Herodotus numbers at 300,000 but which a realistic estimate puts at about 35,000 to 40,000 combat troops, was beaten and Mardonius killed at Plataea, and a second Persian force was destroyed at Mycale across the Aegean.^[23] Briant cautions against the tradition's scapegoating of Mardonius: any Persian general in his position would likely have made the same tactical choices, which followed from the armament and combat style of the royal army, and the deeper failure was that the Persians had not drawn the lessons of the hoplite revolution.^[9] The defeats stopped the attempted extension of Persian power into Europe and cost the empire real losses, but they were losses on a frontier, not at the centre.^[3]

The scale of the setback has to be measured, not assumed. Hyland reckons the combined death toll of Salamis, Plataea, Mycale and the attendant weather and disease at tens of thousands, "a painful blow to the empire", comparable with Athens's Sicilian disaster or Rome's early losses to Hannibal, yet not high enough to limit the empire's manpower or to reach beyond localised mourning.^[23] The territorial loss was real but bounded: the naval defeats freed the Aegean for the Greeks, but the losses in Asia Minor proper were minimal, the mainland Ionian cities stayed under Persian control, and Xerxes continued to take an interest in the western front.^[9] Briant's summary is that in 479 "the book was not yet closed", and there is no reason to think Xerxes had reached any turning-point or written off the west.^[9]

The empire also had means of absorbing the news. Kuhrt notes that the effects of the defeats were softened by distributing looted statues among the satrapal centres, that regular commerce with Greek centres continued, and that a later pseudo-Persian account preserved by Dio Chrysostom hints at how the campaign could be presented at home as a success.^[3] Dandamaev records the official Persian version, that Xerxes defeated the Spartans at Thermopylae, destroyed Athens, imposed tribute and returned, and notes that on Persian seals the Greeks are shown fallen to the ground or on their knees.^[17] Olmstead, no revisionist, reached a striking verdict: taken by itself Salamis "was a check to the Persian frontier advance and nothing more", the recently won territory not lost, the army intact, the fleet still powerful, so that in a real sense it was "actually a Persian blessing".^[32] Brosius states the political result flatly: the defeats "did not lead to any repercussions in the empire", no rebellion and no challenge at court, and Xerxes ruled on until 465.^[8]

Against the Greek record of a destroyer stands the monumental record of a builder, for it was under Xerxes that Darius's new dynastic capital at Persepolis (Old Persian Pārsa) reached its greatest splendour. Ernst Herzfeld's excavator Erich Schmidt credited Xerxes with completing the great Apadana and his father's residential palace, and with newly raising the Terrace Gate, his own palace and the block long called the Harem, altering the Treasury, and beginning the Throne Hall.^[33] The most famous of these is the monumental western gate, the Gate of All Nations, its portals flanked by colossal human-headed winged bulls in the Assyrian manner, through which delegations of the subject peoples passed toward the king. Over it Xerxes set a trilingual inscription (XPa), quoted in full in the entry on Persepolis, naming the building the "Gateway of All Lands" and closing on a double credit, "whatever work appears (to be) good, all that we have done by the favour of Auramazda", the work his father's and his own together.^[34]

The building inscriptions carry a textual crux that scholars call the continuator-versus-self-assertion problem. On the glazed bricks of the Apadana (XPg) Xerxes credits his father with the work and adds only that he "added to that work and built more":

By the favour of Auramazda, much that is good did Darius the king, my father. And also by the favour of Auramazda, I added to that work and built more.

Xerxes, XPg, trans. Kuhrt^[35]

Yet on the Apadana stairways (XPb) he claims the same building alone, stating in the Babylonian version that "by the grace of Ahuramazda have I built this house", without a word of Darius.^[33] The deliberate alternation, shared credit on the Gate and the Apadana bricks, sole credit on the stairways, is the epigraphic form of the reign's whole ambiguity between the pious heir and the self-asserting king. His own residence he called a hadish (XPd), and the vast Throne Hall, the Hall of a Hundred Columns, was founded by Xerxes and finished by Artaxerxes I, its foundation record naming both.^[33] He also commemorated and extended Darius's work at Susa, and set an inscription on the cliff at Lake Van where his father had prepared the surface but left no text.^[3] The Treasury tablets of these years show workers from Caria, Ionia, Egypt and Babylonia moved around the building sites, the administrative underside of the great programme.^[36]

The art-historical record deepens the same point. Margaret Cool Root and Ann Farkas treat Xerxes essentially as the completer of a canon fixed under Darius: by the end of Darius's reign the formulae of royal inscription and the repertoire of imperial art had been codified, and later representations of the king are adapted copies of motifs devised earlier.^[37] Farkas judged Persepolis "mostly completed" by the end of Xerxes's reign, and read Xerxes as continuing the sculptural practices his father had established.^[38] The Achaemenid canon was also deliberately impersonal. Root's deeper argument is that the reliefs subsume the personal image of a king in a dynastic one, so that it is hard to find any feature meant to distinguish one king from another, and "Xerxes the innovator" is nearly invisible by design.^[37] The completion of the grandest architectural programme of the ancient world is not the act of a decadent.

The image of Xerxes that emerges from his own inscriptions is consistent: the heir worthy to continue the work his father began and entrusted to him. He re-inscribed at Mount Elvend a text set beside Darius's and echoing it almost word for word, and he re-issued under his own name the great statement of the ideal king that Darius had placed on his tomb (DNb, as XPl).^[3][39] Kuhrt's judgement is that with Xerxes "the imperial ideology crystallised, becoming clearer and sharper".^[3] The clearest sign of the sharpening is the peoples-list of the daiva inscription (XPh §3), which repeats Darius's tomb list but with a significant change of key: where Darius introduces his subjects as countries he "seized", Xerxes emphasises a smooth take-over of an existing and diverse empire, and his list, far from admitting the losses of 479, actually expands the standard roll by adding new peoples, giving the impression that he had not merely kept but enlarged his paternal heritage.^[3] The paired father-and-son inscriptions cut into the doorways of the palaces, Darius named on one jamb and Xerxes mirroring him on the other, make the same claim of seamless dynastic, political and ideological continuity in stone.^[3]

Of the inner life of the court the sources are late, Greek and lurid, and they have to be handled as such. Xerxes's queen was Amestris, named by Ctesias as a daughter of Onophas and by Herodotus as a daughter of Otanes; the two names are a warning in themselves, and Briant judges the identification of her father with the Otanes who was one of Darius's Seven "totally hypothetical", so that no descent from the conspirators of 522 should be asserted for her.^[9] The court she presided over is known through the standard Greek image of exacted magnificence, the king's table supplied with the first fruits of every land, which Briant reads as the counterpart of the harem, the empire's best required to come to the centre.^[3][9]

The famous story of the reign's late court is the Masistes affair (Herodotus 9.108–113), the tale of Xerxes's passion for the wife and then the daughter of his brother Masistes, of the robe Amestris wove and took, of her savage mutilation of Masistes's wife and of Masistes's attempted revolt and death. It is read across the modern literature not as reportage but as a Greek court-novella. Daniel Potts records that Erwin Wolff placed it in the genre of "harem love stories", and that the story most likely embellishes a genuine attempt by Masistes to raise a rebellion in Bactria after Salamis.^[40] The recurring scholarly key is that the royal robe Amestris wove was a symbol of kingship, so that the demand for it encodes a dynastic power-play, not a love triangle; Llewellyn-Jones reads Amestris throughout as an "imperial matriarch" defending the succession of her son, a dynastic guardian rather than the cruel intriguer of the Greek reading.^[41] The companion anecdote, that Amestris in old age had fourteen noble Persian youths buried alive, Briant warns need not be taken literally, still less in Herodotus's moralising frame.^[9]

The wider Greek picture of the sequestered harem is now largely dismantled, and here the reign is a test case for a live debate. Briant speaks of "the myth of the harem": the sequestered Ottoman-style seraglio is a modern projection, the very presupposition that led the excavators of Persepolis to label a building the Harem, and the Fortification tablets show royal women travelling freely and managing their own estates, land and staff, not living cloistered.^[9] Brosius shows the same from the documents: over nearly a century royal women appear as substantial landholders, and their actions, Amestris's included, work within the rules of the court rather than against them.^[12] Against the deflationary camp, Llewellyn-Jones argues that the harem was a real and central institution, stripped of its Orientalist gloss: polygyny was a political instrument that tied the empire together, and royal women's prestige lay in a chosen invisibility, not in oppression.^[41] The two camps agree on the substance, mobility, property, political weight, and divide only on whether to keep the word.

The most striking of Xerxes's own texts is the trilingual proclamation found in the garrison quarters at Persepolis and known as the daiva inscription (XPh), which gives the sharpest statement of royal religion any Achaemenid left and has generated more argument than any other. Its titulary is more emphatic than the Gate's, adding to the usual styles a claim of pure Iranian descent:

I (am) Xerxes, the great king, king of kings, king of countries containing all kinds of men, king on this great earth far and wide, son of Darius the king, an Achaemenid, a Persian, son of a Persian, an Aryan, having Aryan lineage.

Xerxes, XPh §2, trans. Kuhrt^[42]

The passage that gives the text its name reports the suppression of a cult:

And among those countries there were (some) where formerly the daivas had been worshipped. Afterwards by the favour of Auramazda I destroyed that place of the daivas, and I gave orders: 'The daivas shall not be worshipped any longer!' Wherever formerly the daivas have been worshipped, there I worshipped Auramazda at the proper time and with the proper ceremony.

Xerxes, XPh §4b, trans. Kuhrt^[42]

The daivas are the false gods of the Iranian tradition; the Old Persian word is cognate with Sanskrit deva, "god", but in Old Persian and Avestan it means "demon", and daivadāna, the word Kent renders "sanctuary", carries the derogatory sense of a "den".^[42] Like his father's, the inscription binds right worship to well-being in this life and the next:

You, who shall be hereafter, if you shall think: 'Happy may I be (while) living and (when) dead may I be blessed', obey that law, which Auramazda has established!

Xerxes, XPh §4d, trans. Kuhrt^[42]

What this records is genuinely disputed, and the debate has two poles. The older reading, which Briant traces to Herzfeld and reproduces from Mayrhofer in order to demolish it, took the text as evidence of "a profound but intolerant religiosity", Xerxes the destroyer of the Marduk temple and the enemy of subject cults, the very symbol of religious intolerance.^[9] A related strand, associated with R. C. Zaehner, read a real suppression of a surviving daeva-worshipping cult; Albert de Jong reports the reading and faults it for taking the ritual texts too literally.^[43] Against all such historicist readings stands the deflationary consensus. Kuhrt asks the decisive question, which country: Babylonia is ruled out, since it is now certain there was no destruction of its temples or cults; Egypt shows continuity of its cults; and the Greek destructions do not fit a text about replacing the daivas with Ahura Mazdā, given that Xerxes had Athenian exiles sacrifice on the Acropolis and there is, she notes, no way to identify the burnt Athenian temple with the daiva sanctuary.^[3] Briant presses the methodological point that the country "in commotion" is left deliberately vague, that the syntax does not even fix the daiva sanctuary in that same country, and that XPh is "first and foremost" an ideological assertion, situated in the king's time rather than history's.^[9] The Companion reaches the same end from the parallel structure of the text's clauses: rebellion against the king is against the will of Ahura Mazdā, and a rebel country's gods are by definition daivas, so the inscription "can only be taken as a political document" about loyalty, saying nothing about religious tolerance or intolerance.^[21] The deflationary reading has a pedigree and a cleanest formulation. Heleen Sancisi-Weerdenburg's 1980 dissertation first set out the case against connecting the inscription with specific historical events, and Henkelman has observed that XPh, as "an ideological manifesto dealing with the eternal order", has parallels in earlier Neo-Elamite inscriptions; Bridges states the conclusion with precision: the omission of any named country suggests a text, "although perhaps originally conceived as a response to a particular incident", that was "constructed as a timeless and universal declaration of the king's will and ability to maintain order among subject peoples", and the impression it creates "is one which Xerxes chose deliberately to project".^[5] The one substantial dissent is Mary Boyce's: she rejects both the Babylonian and the Athenian identifications on the evidence, but reads the text as recording a genuine religious act by Xerxes as a Zoroastrian, the destruction of an Iranian daeva-shrine, most likely an open-air one.^[44] Her reading is now a minority position, but a serious one, and it has not always been the minority. Writing in 1996, Klaus Koch could report that, where the Xerxes passage had once been referred to a destruction of the Marduk temple and a repudiation of Babylonian polytheism, most Iranists then thought of a violent confrontation with groups in Iran who had rejected Mazdaism; the deflationary consensus is a settlement of the last generation, not the verdict of a century.^[45] A dated hinge sits between the poles. The 1988 Cambridge Ancient History floated the historicist identification only conditionally: the "famous, but somewhat vague" inscription might be connected with the Babylonian reprisals, Egypt could be the country in commotion "when I became king" and Babylon the second, "where previously false gods were worshipped"; but whether the change under Xerxes "was a political response to rebellion in Egypt and Babylon" or "a reflection of changing religious beliefs and practices in the Achaemenid court" it declared unclear. The refusal to decide was in the standard handbook a decade before the deflationary consensus formed.^[13]

The daiva inscription is the pivot of the older question whether Xerxes was a Zoroastrian reformer who hardened his predecessors' open religious policy. Boyce is the strongest voice for the affirmative: she reads the text as showing "Xerxes concerned with morality as well as with doctrine and observance", acting as a Zoroastrian against the warlike beings the prophet condemned.^[44] The affirmative once had Cambridge behind it too, and on different grounds: testing the kings against four defining beliefs, the 1988 Cambridge Ancient History judged that Xerxes in the daiva inscription "displays his belief in the negative aspects of Zoroastrianism by being vidaeva, or against the evil spirits", and that "in their theology and basic beliefs, therefore, Darius and Xerxes were Zoroastrians"; the same pages, curiously, read the mention of Arta beside Ahura Mazdā in XPh as "the return to pre-Zoroastrian polytheism", one measure of how unsettled the text's religious reading has always been.^[13] The 2021 Companion gives a qualified yes of a different kind: it treats Achaemenid religion as broadly Zoroastrian on the strength of the shared vocabulary of arta and drauga, truth and the Lie, and of artavan, "blessed", but insists that the kings were not monotheists, Ahura Mazdā being the highest god and never the only one, and that to speak of Achaemenid "tolerance" or "intolerance" is a misnomer, since the harm done to a cult was always a response to political disloyalty, never to belief.^[21] Briant and de Jong hold the sceptical line: Briant makes daiva "the starting-point for endless argument on the Zoroastrianism of the Achaemenids" and reads XPh not as doctrinal reform but as a firmer codification of the official cult of the great Persian god, whose guarantor Xerxes presents himself to be even more emphatically than Darius.^[9] The dispute over the man's religion, then, largely dissolves into the dispute over the one text, and that text will not by itself carry a persecution.

One of the most durable stories about Xerxes is that after the Babylonian revolts he sacked the city, razed its temples, and carried off or melted down the golden statue of Marduk, ending Babylon's sacred kingship in an act of vengeful impiety. The fullest late statement of it is Dandamaev's: the ramparts demolished, the course of the Euphrates changed, priests executed, the Esagila temple and the ziggurat Etemenanki severely damaged, and the golden Marduk statue, weighing some 600 kilograms, brought to Persepolis and melted down.^[17] This picture is now rejected. Waerzeggers narrates the demolition of the orthodoxy: the maximal story, expressed most clearly by George Cameron in 1941 and Böhl in 1962 and embroidered later with the diverted Euphrates and the felled ziggurat, was deconstructed by the mid-1990s as "a chimaera without substance".^[19] The turning-point was the 1987 study of Kuhrt and Susan Sherwin-White, which showed that four of the load-bearing clues had been misread: the statue Herodotus describes Xerxes taking was a man's, not the god's; the Akitu New Year festival, supposedly suspended by Xerxes, had in fact long been in abeyance; the shortening of the titulary was gradual, not abrupt; and the title "king of Babylon" continued in use.^[19][46] The 1988 Cambridge Ancient History catches the orthodoxy at the moment of collapse: its narrative chapter still relayed Ctesias's tale that the great statue of Bel-Marduk "is said to have been taken off and melted down", while cross-referring in the same sentence to Kuhrt's chapter later in the same volume, where the refutation stood; the story and its demolition were bound together into one book.^[13] Kuhrt's own chapter did the demolishing in her own voice. The whole classical destruction dossier, she wrote, the notices in Ctesias, Arrian, Strabo, Diodorus and Plutarch, "stem from one misunderstood mention in Herodotus" of the removal of a statue, so that "Xerxes introduced no sudden, drastic or brutal changes" in Babylonia. She pressed two arguments the later summaries tend to drop. The first is a control-case from the Ptolemaic Canon, which marks the years of Sennacherib, who had really destroyed Esagila and the statue of Marduk, as a "kingless" period, yet gives Xerxes his full complement of regnal years, so that Babylonian scribal tradition itself did not remember him as the statue's destroyer. The second is a precise titulary datum: far from vanishing after the revolts, the element "king of Babylon" is attested for Xerxes as late as his sixteenth year, about 470, and reappears even for Artaxerxes I, evidence of "how poor a guide the titles of Persian kings are to historical realities". The later portrait of Xerxes the temple-wrecker, she judged, survives only in "somewhat later classical sources which may reflect a particular bias against the invader of Greece", the very Alexander-era mechanism the Briant and Bridges material sets out below.^[13] The archaeology has since confirmed that the sites of Babylonia's major cities show no destruction layer under Xerxes, and the cult of Esagila continued into Hellenistic times.^[21]

The crux is Herodotus 1.183, the one passage cited for the sacrilege. Briant shows that it will not bear the weight: Herodotus distinguishes the cult-statue of the god (Greek agalma), which in his own day still stood in the temple, from a separate golden statue of a man (Greek andrias) that stood outside the sanctuary and that Xerxes may have taken; the text does not say Xerxes carried off Marduk's cult-image at all.^[9] The Herodotus commentators reach the same point: Herodotus "must have confused two different statues", and what he preserves is at best "a hostile local tradition" that Xerxes the hybristes stole a statue.^[47] The lurid destruction narrative, Briant adds, was chiefly developed by the Hellenistic authors drawing on Alexander's historians, who set the Macedonian's piety in restoring Babylon's temples against the Great King's impiety in destroying them, a contrast whose propagandistic character is plain.^[9] Bridges supplies the mechanism's concrete anecdotes: the later texts "bear the traces of the propaganda which packaged the expedition as a means of punishing the Persians for the sins of their ancestors". Arrian has Alexander order rebuilt at Babylon "the sanctuaries which Xerxes had destroyed", the shrine of Bel among them, and undertake to send back to Athens the bronze statues of Harmodius and Aristogeiton that Xerxes had removed; Quintus Curtius has him destroy the town of the Branchidae and massacre its people for their ancestors' betrayal of Apollo's temple at Didyma to Xerxes; and Arrian frames the burning of Persepolis itself as punishment for the Persians' destruction of Athens, the reception loop closing on the builder's own monument.^[5] Kuhrt and Sherwin-White drew the methodological rule: Arrian's notice of the temples Xerxes destroyed should not be taken "as hard and fast evidence for any real destruction by him".^[5] What survives the critique is not nothing: the revolts of 484 were real and were followed by targeted reprisals against the specific rebel elite families, the "end of archives", but not by a sack of the city, a razing of Esagila, or the melting of Marduk.^[19] Kuhrt states the residue with care: "it is now certain that there was no destruction of Babylonian temples or cults by Xerxes".^[3]

Xerxes was murdered in his bedchamber in the late summer of 465, in the twenty-first year of his reign, and the hard fact is anchored by a near-contemporary Babylonian astronomical tablet, the nearest thing to a documentary record. Under the month Abu it notes, of Xerxes, that "his son killed him", a notice Kuhrt dates to within a few days of 4 August 465; an Aramaic document from Elephantine registers Artaxerxes's accession by that winter.^[48] The anchor is stronger than a bare date, because BM 32234 is a compilation of eclipse observations and the death notice, in Huber's rendering "Month V 14[+x] Xerxes was murdered by his son", sits among datable lunar-eclipse records, so the year can be checked against the sky. It lately was. Gérard Gertoux argued from Thucydides's Themistocles chronology and from the same tablet that the murder belonged ten years earlier, on 24 August 475, with a ten-year co-regency of Darius and Xerxes behind it; Peter Huber then recomputed every eclipse on the tablet's reverse and found all of them "compatible with calculations for the low chronology (death of Xerxes in 465 BC)", while "at least five of them flatly contradict the calculations for Gertoux's higher chronology".^[49] The traditional date stands, verified rather than merely assumed, and the tablet fixes the day to the fourteenth of the month or shortly after.^[49] Beyond the bare fact the record scatters into irreconcilable Greek versions, which Kuhrt prints side by side. In Ctesias the moving hands are the courtier Artapanus and the eunuch Aspamitres:

Artapanus, a very influential man with Xerxes and Aspamitres, the eunuch, who was also very influential, resolved to kill their master. Having done so, they persuaded Artaxerxes, his son, that Darius, Xerxes' other son, had murdered him.

Ctesias, Persica F13 §33, trans. Stronk^[50]

Diodorus (11.69) makes Artabanus the commander of the royal bodyguard and the eunuch Mithridates his accomplice; Justin (3.1) opens with the topos of the king "once the terror of the nations around him" grown "an object of contempt even to his own subjects", and gives Artabanus seven grown sons.^[31][15] The versions disagree on the eunuch's name, on whether Artabanus was a eunuch or a general, and on who instigated the deed; the assassin Artabanus son of Artasyras must in any case be kept apart from Xerxes's uncle of the same name, the sage counsellor of Herodotus's Greek-war narrative.^[3] What the versions share is that the designated heir, Darius, was killed in the coup, framed for his father's murder, and that a younger son survived to take the throne as Artaxerxes I, a name Brosius glosses as "he who reigns through Arta", through the truth.^[8] Modern scholarship treats the elaborate palace-coup story as unhistorical in its detail: Brosius points out that had Artabanus truly aimed at the throne he would have faced the opposition of Xerxes's other sons, the satraps of Bactria and Egypt, and of queen Amestris, which "renders the entire episode as presented in the Greek sources difficult to accept as historical fact".^[8] One piece of material evidence may nevertheless corroborate the kernel the versions share. The audience reliefs found in the Persepolis Treasury, long read as Darius enthroned with Xerxes as crown prince, were shown on re-examination to have been cut for the centre panels of the Apadana stairways, and "they may well show Xerxes as the Great King with his son Darius, the crown prince, behind him"; their removal from that position of honour to the Treasury's seclusion was read in the 1988 Cambridge Ancient History as an act of Artaxerxes I, "who had good reason not to show that the murdered Darius had been crown prince", though on the reliefs' deliberately dynastic idiom the identification of the pair cannot be certain.^[13] The one clear political aftershock is that the turmoil of the succession encouraged the Egyptians, under the Libyan Inaros, to expel the Persian tribute-collectors and rise in revolt.^[9]

The traditional verdict on Xerxes, inherited whole from the Greeks, is that his reign began the empire's long decline: after the founder, the conqueror and the great organiser, a fourth king who inherited everything, was broken at Salamis, and sank into the intrigues of the harem, the first of the soft successors. Briant traces the modern form of this to Rawlinson's Five Great Monarchies of 1867, where "the character of Xerxes sank below that of any of his predecessors" and "all this explains the decline of the Empire", and its ancient root to Plato, who wrote in the Laws that after Xerxes "hardly any king of the Persians has been genuinely 'great,' except in style and title".^[9] Aeschylus is still misused to the same end, his Salamis, "the tomb of Persian power", read as the historical ruin of an empire.^[9]

Between Rawlinson and the dismantlers stands the tradition's most instructive exhibit, because it annotates itself. As late as 1988 the Cambridge Ancient History was still writing the received portrait: a Xerxes who, disheartened by the failures in Greece, "withdrew to the centre of the empire and licked his wounds in comparative isolation", and an Achaemenid history that from 479 onwards was "a holding operation in the face of challenges from rebellious subjects, ambitious satraps and external enemies". Yet the same chapter conceded that the withdrawal "may be a false impression based on a general lack of source material", almost nothing being known of the king's doings after Mycale from any good source; and its own footnote named the trap. In criticising Xerxes, it wrote, "we do so within a received tradition which has roots in antiquity", descending from Plato's contrast of the self-made Darius with "foppish, court-bred, Xerxes", a theme "perpetuated by western scholarship to this day"; and "one wonders how Xerxes would appear to history if we had better, particularly Iranian, sources for his reign". The decline thesis footnoted with its own refutation is the neatest illustration the historiography affords of a verdict read backward out of Greek silence.^[13]

This whole narrative has been dismantled by the modern synthetic scholarship, above all by the Groningen school of Briant, Kuhrt and Sancisi-Weerdenburg. Briant's charge is that the decline is a Hellenocentric teleology, a construct that lets Greek narrators explain Alexander's conquest backward, and that it rests on the accident that the Greek narrative sources fall silent after 479: to deduce from their silence that Xerxes lost interest in ruling and gave himself to Persepolis and the harem "is the result of a peculiar sort of historical method based, one might say, on the fact that the narrative evidence falls silent".^[9] Matt Waters puts the counter-case at its plainest: the empire's "downfall" was 150 years in the making, and to link Xerxes's defeat in Greece to Alexander's conquest is "nonsense", a connection made only in propaganda; Salamis, Plataea and Mycale were neither the first Persian setbacks nor the last.^[51] Llewellyn-Jones agrees that the reading of Plataea as the beginning of the end "makes no sense", the empire being still strong and, in Xerxes's own post-479 inscriptions, still claiming expansion.^[30] The empire that Josef Wiesehöfer's readers were once invited to see declining under Xerxes in fact endured, governed and prosperous, for another century and a half, to 330 BCE.^[52]

Two after-images of Xerxes ran in parallel through later Europe, and both are Greek or biblical in origin. The first is the archetypal oriental despot of tragedy and history, the barbarian king whose pride free men break, the name Aeschylus made a byword for overreaching; it was fixed by the Greek victory and, in the West, never revised.^[7] Jane Grogan traces the type to Aeschylus's Persae and Thucydides, the founding texts in which Xerxes's 'Persian vices' become the fixed content of the European barbarian, even as Aeschylus's sympathetic ghost of Darius disrupts the indictment.^[53] In Rome the type became a stock moral exemplum: Pompey derided the marine engineering of Lucullus by calling him Xerxes togatus, "Xerxes in a toga"; Lucan set Caesar's siege-works at Brundisium beside the Hellespont bridge and the Athos canal, "a familiar pairing which was by now a well-established paradigm of arrogance"; and Caligula's biographers reached for the same comparison to damn the emperor's bridge of boats.^[5] The line runs unbroken to the present, ending for now in Zack Snyder's 300 films, "our own generation's most widely known response to antiquity's most notorious Persian king".^[5] The second is the magnificent monarch of the Book of Esther, whose Ahasuerus is Xerxes: the king of the lavish feast, who became in European art an emblem of princely splendour, his wedding to Esther a model of the sumptuous courtly marriage.^[21] Brosius suggests that the historical murder of 465, and Herodotus's romance of Xerxes and the wife of Masistes, may themselves lie behind the plot of Esther, so that the biblical magnificence-image and the Greek tyrant-image draw on the same well.^[8] A third after-image, the least remembered, is wholly positive: the pious benefactor of the Jewish tradition. In the eleventh book of Josephus's Jewish Antiquities, Xerxes takes over with the kingship his father's "piety and honour towards God", is "exceedingly generous towards the Jews", lets Ezra carry the dedicated gold and silver from Babylon to Jerusalem with expenses from the royal treasury, makes his own offering of wheat and exempts the returning priests from tribute, the Ezra and Nehemiah stories transposed to Xerxes from the biblical Artaxerxes. This picture of "piety, munificence, tolerance and respect for a culture other than his own is wholly at odds with the image of Xerxes which prevails in the Greek-derived traditions", and, like the Persian sources themselves, it is a measure of how far the mainstream portrait was made by Greeks who bore the scars of the invasion.^[5] In the Persian and later Iranian tradition the figure is far dimmer, for the reasons that dim every Achaemenid there, the loss of the native narrative and the eclipse of the dynasty's memory after Alexander.

What survives most durably of Xerxes is neither the war nor the court intrigue but the stone: the Gate of All Nations on the terrace at Persepolis, the daiva inscription with its hard claim of the one true worship, the titulary that names him son of Darius, an Aryan of Aryan lineage. Between the raging tyrant of the Greek theatre and the careful builder of the royal inscriptions the sources leave a figure who cannot be fully resolved. Nor should the opposition be drawn as image against reality, for the symmetry cuts both ways: the Xerxes of the homeland, his image on the palace walls and his words cut in stone, is in Bridges' judgement "as much an artificial construct as the character we encounter in the Greek literary texts", each retelling shaped by the agenda of its creator, and the silence of the Persian sources on the Greek campaign is "a powerful reminder" that western perspectives on this king are overwhelmingly the inheritance of those his invasion scarred.^[5] The honest course, which the modern scholarship has taken, is to hold the two portraits apart rather than blend them, and to recognise that the confident old verdict of decadence is in large part the victors' story, told at the expense of a king whose empire wrote almost no history of him and whose enemies wrote nearly all of it.