in symmetrical panels of architectural ornament, on stone plates and vessels, or on cylinder seals. When the versions appear on the same surface, they are normally in a ranked array. At Persepolis, when three versions of an architectural inscription are displayed side by side, the Old Persian is commonly central, the other versions on either side (e.g. DPc, XPcb, XPdb, but not DPa, DPb, XPe). When three versions are displayed in vertical array, the Old Persian is normally uppermost, the Elamite below it, and the Babylonian below that (e.g. DPh, XPda, XPcb). The same order is normal in trilingual royal inscriptions in vertical array on stone weights (e.g. Wa–Wf) and on cylinder seals (e.g. SDa–SDe, SXd); it is also found in quadrilingual inscriptions on stone vessels (adding an Egyptian hieroglyphic version below the Babylonian, e.g. DVsc, XVs, AVs) and on one face of the quadrilingual Suez stele of Darius I (DZb‐c), where the Egyptian hieroglyphic counterpart is on the opposite face.
The same ranking of versions also appears in the unique suite of four distinct but interrelated foundation inscriptions of Darius I (DPd–DPg) that were laid out in non‐symmetrical array and carved all at one time in side‐by‐side panels on a single stone block, 7.2 m wide, on the south face of the terrace at Persepolis. The two Old Persian inscriptions (DPd, DPe) are at the viewer's left, the Elamite inscription (DPf) to the right of them, and the Babylonian inscription (DPg) on the far right, that is, with the left‐to‐right sequence corresponding to the common vertical arrangement (and also found in the mirrored horizontal arrays of DPa and DPb). Each is a distinct text, but they are rhetorically connected as a series to be read from left to right, the language of each adding nuance to its specific contents. While the two Old Persian texts, in an archaizing version of the language of the empire's rulers, focus on royal and divine protection of the Persian land and people, and the Akkadian text, in a venerable language of the larger empire, asserts the geographical scope of the realm, the Elamite inscription (DPf), in the ancient indigenous language of the region, focuses on the construction of the “fortress” at Persepolis itself (see Lincoln 2008: pp. 223 and 231–233). The arrangement of the versions reflects a social and political nuance as well as a decorative choice.
Royal inscriptions on cylinder seals begin with the pronoun “I” and state the king's name and short title as appositions. Hence, unlike inscriptions on non‐royal seals, they are self‐predications rather than statements of ownership. Multilingual examples, with Elamite versions, come from the reigns of Darius I and Xerxes I. Only one actual seal of this kind is extant, said to have come from lower Egypt (Schmitt 1981: p. 19 SDa; Merrillees 2005: p. 52 and pl. vii, no. 16); eight others are attested indirectly by impressions on clay tablets, bullae, and sealings from Persepolis (Schmitt 1981; Garrison 2014; seals with royal name inscriptions impressed on sealings from Daskyleion, in western Anatolia, do not have Elamite versions: Schmitt 2002).
Administrative Texts
Most Achaemenid Elamite administrative texts belong to two archives discovered at Persepolis. The 20 000–30 000 excavated tablets and fragments of the Persepolis Fortification Archive are the remains of about 15 000–18 000 original documents, about 70% of them bearing Elamite texts (Jones and Stolper 2008). About 2400 of the Elamite texts have been published; a publication of about 2500 others, widely cited in preliminary form, is forthcoming; and editions of about 1000–2000 more are in preparation (Henkelman 2013; Henkelman infra, with references). They are dated between 509 and 493 BCE. They record the storage and distribution of food (cereals and cereal products, beer, wine, oil, etc.) and livestock in the region around Persepolis, to support workers, administrators, religious personnel, courtiers, and others.
All but one of the 746 excavated tablets and fragments of the Persepolis Treasury Archive bear Elamite texts, 138 of them published (Azzoni et al. 2017, with references). Most are dated between 492 and 457 BCE. A few are as old as 507‐505 BCE (Stolper, Henkelman, and Garrison 2020). Most deal with payments of silver from a treasury in Persepolis to workers at or near Persepolis.
A single Achaemenid Elamite administrative text believed to come from Susa closely resembles texts from the Persepolis Fortification Archive, even bearing a seal found on Persepolis Fortification tablets (Garrison 1996). The sign forms, text layout, and few preserved words on a fragmentary tablet excavated at Old Kandahar, in Afghanistan, closely resemble those of Persepolis Fortification texts (Fisher and Stolper 2015). These isolated remains of lost archives suffice to show that the use of Elamite for state administrative recording was widespread in Achaemenid Iran.
Of other known isolated Achaemenid Elamite administrative documents, some originated in the Persepolis Fortification Archive (Vallat 1994: pp. 264, 272; Jones and Stolper 2006; Henkelman et al. 2006; probably Grillot 1986). Others, of uncertain provenience, may be survivors of lost archives (Jones and Stolper 1986: pp. 247–253; Garrison, Jones, and Stolper 2018).
Like Elamite versions of Achaemenid royal inscriptions, Elamite administrative records at Persepolis were products of a multilingual environment, in two senses.
First, though dominated by Elamite documents, the archives also included texts in other languages: more than 840 monolingual Aramaic Fortification texts, as well as single Greek, Old Persian and Phrygian Fortification texts, and two in Demotic Egyptian dealing with the same or similar matters as the Elamite records; and Aramaic epigraphs, aids to document handling, on more than 250 Elamite Fortification tablets. Archive keepers had to be passively proficient with information written in several languages, and actively proficient enough to mark their handling of Elamite records with notes written in Aramaic (Azzoni and Stolper 2015).
Second, the many transcriptions of Iranian words in Elamite administrative texts (Tavernier 2007) and the frequent calques on Old Iranian syntax reveal that the records were composed by and for speakers of Old Iranian languages and dialects; for many or most of them, Elamite was an acquired written language.
Seal Inscriptions
About 5% of the seals impressed on Persepolis administrative texts had inscriptions, about half of the monolingual seal inscriptions (more than 90) are in Elamite, and about a quarter are in Aramaic. The seal inscriptions normally give the name, and sometimes the patronym, of the seal's original owner or his superior. A few of the Elamite seal inscriptions name the current seal users, who figure in the transactions recorded on the sealed tablets; they are sometimes figures of high social status or administrative rank (e.g. the seals of Šuddayauda [Garrison and Root 2001: pp. 268–270] or Ašbazana [Garrison 1998]). In many cases, however, the seal is associated with an office, and not with an individual user, and the name in the seal inscription is not connected with the personnel of the archives. Some of the Elamite‐inscribed seals may have been heirlooms (e.g. PFS 93*, inscribed with the name of Kuraš [Cyrus], son of Šešpeš [Teispes], and others, see Garrison 2011), but others were transmitted by official succession or delegation. For users of the documents, the inscriptions alone – even if legible in fragmentary impressions – cannot have been either necessary or sufficient to identify the user and confirm the validity of the sealing.
Texts on heirloom seals are presumably pre‐Achaemenid and early Achaemenid. Conversely, some seals with similar inscriptions, sometimes classified as pre‐Achaemenid Neo‐Elamite, may also be of Achaemenid date (e.g. Amiet 1973: pl. vi–ix; Garrison 2002 [2006]).
Elamite Sources of Old Iranian
Elamite transcriptions of Old Iranian words make Achaemenid Elamite texts the largest single source of Old Iranian lexicon in indirect transmission. Spelling variations, especially in administrative documents, reflect phonological and dialect variation in the forms of Old Iranian spoken around the Achaemenid courts and its differences from the archaic Old Persian of the royal inscriptions (Henkelman 2011a: p. 614f. n. 105). Similar variations in royal inscriptions also reflect editorial or stylistic choices (e.g. miššadanaš [DNa] and mišbadanaš [DSe], mišbazana [DPa], representing
