Slavery and Dependent Labour

A Babylonian sale contract was drawn up to establish title. It names the seller, the buyer, the price and the competing claims against which the seller stands warranty; it distinguishes a saleable slave from a free citizen or a person belonging to another institution. A Persepolis ration memorandum was drawn up to account for provisions. It names the quantity, the group, the month and the official who authorised the issue, usually without specifying the recipients' legal condition. These different documentary purposes limit comparison. The absence of a status clause from a ration account cannot supply the warranty that a sale contract makes explicit. Greek narratives also extend the vocabulary of servitude to political relationships beyond transactions in people.^[7][6][8]

The Persepolis Fortification archive covers a limited period and region. In Kuhrt's survey, the Treasury texts run from 492 to 458 BCE while the much larger Fortification archive spans 509 to 494, with around half of its texts falling in 500–499 and two-thirds of the Treasury texts clustering in 466. The documented activity takes place largely within Fārs, from the region of modern Nīrīz into eastern Khūzestān. Kuhrt accordingly warns that "there is no way in which we could say that the material provides a window on the empire as a whole".^[6] Foodstuffs dominate the accounts; textiles and metals are poorly represented, and agricultural production remains difficult to reconstruct: "We hear nothing of the peasants working the land, for example. How did the administration extract the fruits of their labour?"^[6] The surviving holding includes hundreds of Aramaic tablets and epigraphs alongside the Elamite records. These must be distinguished from the lost documents once written on perishable materials, including waxed boards and skins. Kuhrt's estimate of five hundred to one thousand Aramaic documents refers to surviving evidence, not a vanished archive. The archive’s selectivity leaves agricultural extraction much less visible than the distribution of food and supplies.^[6]

The Babylonian material is biased in the opposite direction. It survives as the archives of prosperous urban houses and of great temples, so it shows ownership at the top of the social range. Gian Pietro Basello, drawing on Michael Jursa's study of the first-millennium economy, notes that the Egibi family owned at least a hundred slaves while most wealthy Babylonian families owned fewer than five, and that the poor attestation of slaves as agricultural labour is a problem of documentation rather than a finding about the economy.^[1] The Aramaic legal traditions of Egypt and Samaria share a chancery language without sharing a law of status: the formularies, the standard words for a dependant and even the vocabulary of body-marking differ between the two regions.^[9] Documentation of slavery in the eastern satrapies is much thinner; Basello observes that "less is known about slavery in eastern Iran and in Central Asia".^[1]

The fourteenth section of the Behistun inscription provides a multilingual test of workforce terminology. Darius lists what he restored to the people after killing the usurper Gaumāta:

The kingdom which had been taken away from our family, I re-established it, I put it back in its place. In accordance with what had been previously, I made the cult-centres, which Gaumata the magus destroyed. I restored to the people, the pastures and herds, the household slaves and the houses/domains, which Gaumata the magus took away from them.

Darius I, DB §14, trans. Kuhrt^[10]

Roland Kent's older rendering gives the same clause as "the household slaves and the houses which Gaumata the Magian took away from them", and his lexicon glosses the Old Persian word behind it, māniya-, as "household slave(s)".^[11] That gloss is a translator's decision, and the inscription's other versions show how much rests on it. Where the Old Persian has māniya-, the Elamite version has kurtaš, the ordinary administrative word for a worker in the Persepolis tablets; the Babylonian version has agru, "hired worker".^[12][13] The versions draw on their own administrative vocabularies: the Babylonian text has Darius restore the property to the army (uqu) rather than to the people, and summarises it as "cattle, sheep, fields, hired workers, bow lands", a different list from the Old Persian one.^[13]

The Babylonian rendering can suggest a legal distinction which the context does not establish. Matthew Stolper, working from Elizabeth von Voigtlander's edition, warns against the obvious inference. At the time the inscription was composed the Iranian loanword gardu was not yet in common Babylonian use, and the Babylonian version generally avoids loanwords, so an approximate equivalent had to be found for the Elamite kurtaš. In that context, he writes, agru "cannot imply free wage laborers"; he takes the term to have been selected "to indicate workers drawing periodic payments as distinct from those remunerated with income-producing property".^[13] The versions select categories from different administrative vocabularies. Stolper argues that the usage favours workers generally and, more particularly, workers assigned by the crown to estates. His explanation depends on the form of remuneration, rather than a demonstrated right to leave or choose employment.^[13]

The Aramaic version preserves too little to resolve this comparison. Christopher Tuplin notes that the corresponding passage is almost entirely lost; what survives includes the phrase for "their property and houses", which does not map directly onto the better-preserved versions.^[14] Even the Elamite is contested: the words following kurtaš have been read as "estate artisans" or "house/estate subjects", and Tuplin considers whether its phrasing dissociates kurtaš from estates rather than attaching them.^[15] Basello prefers "a general meaning of 'household worker'" to the translation "slave" for Old Iranian *gṛda- and its loanwords.^[12] Jan Tavernier's lexicon gives the reconstructed word the gloss "domestic staff, workman" and compares Old Indic gṛhá-, "house"; Tuplin accordingly compares Greek oiketēs, a member of the household.^[16][15] Etymology identifies the household association, while the obligations of particular workers still have to be established from context.

A second Old Persian word appears in the royal inscriptions for relations of political dependence. Bandaka, conventionally understood as a bondsman or servant, describes a person tied to the king by loyalty and service. Darius applies it, again and again in the Behistun narrative, to the generals who put down the rebellions of 522 and 521 for him.^[17] Kuhrt glosses it as "a term expressing the mutual lines of trust and loyalty linking subject to king", and observes that it means literally "bondsman" or "servant": men whose names were famous and whose families were honoured held, in relation to the king, no special rights.^[17] The word appears in the roll of subject peoples in the same sense:

These are the peoples who obeyed me; by the favour of Auramazda, they were my faithful subjects; they brought me 'tribute'; what was said to them by me, whether by night or by day, that they did.

Darius I, DB §7, trans. Kuhrt^[18]

The Babylonian Behistun renders bandaka by qallu, a word widely used in Akkadian for a chattel slave. Basello and Matarese discuss the resulting range of meanings across political and household relationships.^[19][20] The Old Persian text supplies named examples of senior royal service: Dādaršiš, satrap in Bactria, in §38, and Vivāna, satrap in Arachosia, in §45. Kuhrt translates Darius's description of each as "my subject".^[21] The Aramaic version uses 'lym, a word that elsewhere in Egyptian Aramaic documents denotes slaves, of the general Vahumisa.^[22] At Persepolis the Elamite word that corresponds to bandaka is lipa-, applied to low-ration personnel who, in Basello's phrasing, were "probably slaves in some cases".^[12] These renderings connect the vocabulary of royal service with the vocabulary used of people receiving provisions in a work group.

The Akkadian vocabulary was itself graded rather than binary. Qallu means etymologically "of low standing, of little value"; from the fifth century it gives way to ardu (feminine amtu), which in Kathrin Kleber's account, followed by Matarese, "rather indicates social subordination in general", so that lower officials call themselves the ardu of their superiors and the words appear as elements in personal names expressing subordination to a god.^[20] Elsewhere in the royal inscriptions the Old Persian counterpart of qallu is marīka, roughly "young man", a person of low rank.^[20]

The Greek letter attributed to Darius and addressed to Gadatas requires a separate judgement of authenticity. Its description of the royal correspondent as the king's doulos, "slave", cannot be used without that judgement. Basello follows Briant's argument that the inscription is a Hellenistic or Roman forgery and treats its language as a Greek construction of friendship with a monarch.^[23] Xenophon's account of the condemnation of Orontas by Cyrus the Younger provides a different narrative context: Darius II is said to have "given" Orontas to Cyrus. Basello interprets this as a transfer of allegiance, while the seizure of Orontas by his belt expresses a bond of service. Neither occurrence by itself supplies the legal incidents of ownership found in a slave-sale contract.^[19]

The Persepolis tablets repeatedly record allocations to workers without defining their legal rights. Kuhrt sets out the competing possibilities:

What more can we say about the people who made up the workforce? The all-embracing Elamite term is kurtaš, rendering Old Persian *garda-. It appears in Babylonian and Aramaic documents too, and has given rise to considerable scholarly debate (Stolper 1985: 55–9). Were they slaves? Serfs? Deportees? Prisoners-of-war? Forced labour gangs? In fact, as Hallock has already noted (PFT, Glossary, s.v.), it means workers in the widest sense, including specialist craftsmen (16, no.27) and élite servants (El. libap, see Henkelman 2003a: 129–37; 11, no.68, n.6). In the eyes of the administration dealing with their movements and emoluments, they all fell into one and the same category.

Kuhrt, The Persian Empire^[2]

Richard Hallock's glossary, on which she draws, is the foundation of the modern position. He defines kurtaš as "worker(s), work group", an Old Persian loanword corresponding to Akkadian gardu and Aramaic grd, and finds that it "means 'worker(s)' in the widest sense": it takes in marrip, artisans, and conscripts on one side, and libap, servants and possibly slaves, on the other.^[4] A category that contains conscripts and probable slaves at the same time is an administrative class and not a legal one.

The empire's own scribes glossed it neutrally. On the reverse and margins of several Fortification tablets an Aramaic clerk summarised the Elamite entry, and where the Elamite says kurtaš the Aramaic says gbrn, simply "men".^[4] Hallock's translation of the gloss on one such tablet reads: "In the year 23, in the month Elul. Rations received for 75 men."^[24]

Hallock also raised a difficulty for identifying every kurtaš as a slave. Several texts deal with conscripted persons, rabbap, and two of them call the same people kurtaš. He asked: "why should slaves be conscripted?"^[25] A Treasury tablet sharpens it further by ordering payment to Syrian kurtaš who had been conscripted and were now released, which implies a term of service with an end to it.^[26] So does a Babylonian contract in which a man undertakes a full year's labour service in Elam on someone else's behalf.^[26]

Numerous work groups bear designations drawn from outside Persis. Briant lists Bactrians, Sogdians, Babylonians, Assyrians, Elamites, Arabs, Syrians, Egyptians, Lycians, Carians, Ionians, Sardians, Cappadocians and Thracians among the labelled groups, and the tablets record their movement in large parties: 547 Egyptians from Susa to Persepolis, 108 Cappadocians from Persepolis into Elam, 1,500 men from Persepolis to Susa, and parties of 150 Thracians, 980 Cappadocians and 303 Lycians moved within Fārs.^[27] Work groups were attached to the local treasuries, which functioned as processing centres and workshops: 47 at Shiraz in 504 rising to 231 in 500, 677 at Matezziš in 497, 544 at Uranduš in 500.^[27] Kuhrt adds that the children in these establishments were classified by sex and payment, their declining ration scale suggesting age-sets "sometimes defined by height".^[2]

Hallock identified one case involving Persians in his 1969 edition. He records that PF 871, an account for Persian puhu copying texts at Pittannan, "provides the only known case in which Persians are referred to as kurtaš".^[28] The same file shows how little the Elamite grade-words mean chronologically: "The term 'boy' is not to be taken literally", Hallock writes, and in a companion text the same "boys" are called "men".^[28] On the travel rations he is blunter still: the frequent "boys" "are not necessarily boys in a chronological sense; they are subordinates".^[29]

Wouter Henkelman uses the rendering "(dependent) workers" in his administrative study, retaining the broad workforce category without equating every occurrence with a single legal condition.^[30] Tuplin reports that Henkelman's later work stresses differentiation within the category and doubts that its members should properly be described as slaves. Tuplin himself finds substantial coercion compatible with the records and leaves the terminology open.^[31]

Because the archive is an account of issues, it records with unusual precision how these people were fed, and the figures show fine internal stratification. Briant's analysis of the published ration records finds that 83 per cent of men received 30 qa of grain a month, about 16.5 kg, with the rest between 11 and 25 kg, while 87 per cent of women received between 11 and 16.5 kg.^[32] Those averages conceal wide spreads. In a single account for 702 workers at Matezziš in March 498, allocations run from 27.5 kg at the top down to 2.75 kg at the bottom, across men, women, "boys", girls and a group of eight libap.^[32] The travel texts show the same hierarchy in miniature: "boys" and "servants" at the lowest grade, "gentlemen" (šalup) at a higher one, and "men" and "workers" at either, "men" being, as Hallock notes, "a word which conveys no implication of status".^[29] The libap terminology is itself disputed: Hallock’s category of servants, possibly slaves, contrasts with the later interpretation of élite servants adopted by Kuhrt. The same translated English label can consequently conceal different interpretations of the Elamite word.^[4][2]

From 492 the Persepolis treasurer could authorise silver instead of food. Kuhrt observes that these substitute or equivalence payments were "finely calculated to tiny fractions of a shekel", that how the rate was reached is unknown, and that the surviving documentation of the practice does not predate that year, which raises the possibility that the administration was struggling to meet its obligations from its own productive base, or was disposing of surplus by compelling workers to buy from its depots.^[33]

Briant presses the second reading. The minuteness of some allocations makes it unlikely that silver was actually weighed out; more probably the workers held credit redeemable in the administration's own warehouses, and if so "it was a win–win situation for the administration, since it could fix the prices and require the kurtaš to shop there".^[34] He then points to a documented consequence. Between December 467 and August 466 the Persepolis grain price rose to five times its normal level and beyond, and did not return to normal until the end of that summer; whatever the cause, "the quality of life of the kurtaš eroded drastically because of the administration monopoly".^[34] Money payment, on this account, was a change in administrative technique rather than in status.

Whether the recorded rations represent the whole of a worker's income is the question on which much of the status debate now turns, and it is unresolved. Gerassimos Aperghis argued from the grain figures that adult male kurtaš were deliberately underfed; Tuplin judges the argument "perhaps not quite cogent in its own terms", while allowing that Aperghis's general sense that kurtaš "were more like slaves than anything else is understandable". Henkelman replies that the archive's rations "do not represent the entirety of their remuneration" and that workers were expected to find at least a third of their income elsewhere, which would make the calculation invalid at its foundation.^[35]

Women and children are everywhere in these accounts. One annual grain account from Hadaran records both an ordinary work group and a separate childbirth payment:

5 boys 2, 7 boys 1.5, 4 boys 1, 1 woman 3, 18 women 2, 3 girls 1.5, 3 girls 0.5; total 41 workers … 2 women (who) had borne boys 2 each. 3 women (who) had borne girls received 1 BAR each.

PF 1943, trans. Kuhrt^[36]

In the class of childbirth allocations studied by Hallock, the issue for a male child is twice that for a female child. He interpreted the differential as a reward for increasing the workforce:

The N texts record single payments to mothers who have just delivered children. Mothers of boys invariably receive twice as much as mothers of girls. This fact indicates that the payments are made for the purpose of reward, not to replace earnings lost because of time off from work. Obviously the payments reflect an interest of the state in the increase of the work force.

Hallock, Persepolis Fortification Tablets^[37]

The accounts establish the allocations but do not record a period of leave. Hallock's total of 449 offspring, of whom 247 were boys, combines the N texts with entries in the V texts and a further tablet known to him through Cameron’s transliteration. Briant notes that no twins appear in this evidence.^[37][38] The recorded allocations include support for female as well as male children.

Women could occupy high grades. A wine account of 504/3 records entries in which two men receive two units each and a single woman three.^[39] Hallock identified thirty-seven grain texts with exceptionally high allocations to women: thirty-five of those women received more than the men in their group and nine the same.^[28] Some high-grade women are called "chiefs", Elamite araššara, which Hallock treats as a variant of iršara, and the corresponding wine entitlement could reach three marriš.^[28] Maria Brosius connects female leadership with the organisation of pašap work groups and distinguishes such officers from ordinary workers. In her analysis, men and women doing the same skilled work received equal rations at the relevant grade, while ordinary male and female grain allowances differed, conventionally thirty and twenty quarts respectively. Occupational grade and responsibility are therefore central to her reconstruction.^[40] Kuhrt adopts Brosius's identification of the women receiving high wine rations as team leaders.^[41]

Whether these people lived in families is disputed, and the dispute is instructive because both sides argue from the same tablets. Mikhail Dandamaev inferred family life from the roughly equal overall proportions of men to women and boys to girls across the archive.^[38] Briant rejects the inference on the ground that aggregate percentages assume the kurtaš were a fixed and homogeneous community, and that individual groups have markedly unequal compositions: 70 men, 95 women and 20 boys with no girls at all in one; 103 men, 364 women, 122 boys and 84 girls in another; 250 men, 220 women, 18 boys and 32 girls among a party of Thracians; 501 men and nobody else on a Persepolis building site in 466. At the Shiraz treasury, after two rounds of recruitment, women and children came to make up some 90 per cent of the group, which was then reorganised into eleven subgroups, six of men only and five dominated by women, sometimes accompanied by a few men.^[38] Briant singled out the reference to a worker's wife in PF 999 and argued that the compositions of groups reflected administrative decisions capable of breaking family units. This was his interpretation of the relationship between household life and the organisation of labour:

Even though the (sole) reference to the wife (irtiri) of a kurtaš (PF 999) might possibly lead us to believe that marriages were recognized by the administration, the notion of kurtaš living in families must be given up entirely. … In order to increase productivity, it broke family units or forbade their creation (if any were actually recognized by the authorities).

Briant, From Cyrus to Alexander^[38]

Tuplin treats the existence of family groups as an inference requiring caution and adds that, even if it is accepted, it "does not much affect" the status question.^[31] Kuhrt points to a workshop with fifteen men, sixty-eight women and twenty-one children, including a married pair, while retaining the broader uncertainty: "How the labour force was reproduced is a further unanswered question."^[42] The presence of women and children does not demonstrate stable families in every recorded work group, and the organisation of a ration list need not describe a residential household.

One Akkadian tablet was found among the Elamite documents of the Persepolis Fortification archive, and it is a slave sale:

Bel-iddin, son of Niq[udu(?) of his own free will] sold his slave, Tab-pashir-Bel, whose right hand is inscribed with the names of Shulumme, son of Ahimasu, and Bel-iddin, son of Niqu<du?>, to Marduk-belshunu, son of Bel-shum-ishkun, for the full price of 2 minas 25 shekels of white, refined silver. … Bel-iddin guarantees against suits (brought) by improper or proper claimants (to the slave and against suits claiming) the status of servant of the crown, free citizen, temple slave, [ … ] (for the slave).

Fort. 11786, trans. Kuhrt^[8]

The warranty clause identifies competing legal categories. To sell a person safely, a Babylonian seller had to guarantee that the buyer's title could not be defeated by a claim that the person was a servant of the crown, a free citizen or a temple oblate. These were distinct and enforceable categories, and the contract exists because confusing them was expensive. Tuplin draws the inference for the Persepolis workforce: "Anyone selling a royal kurtaš would surely have infringed this provision."^[43] That inference concerns royal workers specifically. The contract distinguishes private title from claims of royal or temple service and free status; it does not establish that every ration recipient enjoyed the same protection.

Related warranty formulae occur in three Babylonian sale deeds which Kuhrt brings together. BM 30126, dated to Darius II, warrants against claims of free, royal or šušānu status and a further status lost in a gap; it requires the conveyance to be cleared through the royal tax office. The Royal Ontario Museum deed, volume 2 no. 35 in its publication, lists free, royal, temple and šušānu claims, adds a hundred-day guarantee against flight, and records transfer in that office. The temple category is translated "temple slave" by Kuhrt and is also described in modern discussion as an oblate. BM 62588 records another sale and tax-office conveyance, with a flight warranty whose duration of a hundred days is partly restored; its preserved text does not supply the same four-status list.^[44] Kuhrt describes the office, Akkadian bīt miksu, as "a new public institution in Achaemenid Babylonia", while the šušānu category is "fluid, not to say nebulous".^[44] The deeds place transactions in enslaved people within the royal tax administration. For the Royal Ontario Museum tablet, dates under Artaxerxes I, II or III remain possible. Kuhrt regards the latter two identifications as more likely on formal grounds and because of the Greek style of one seal impression; the reign therefore remains uncertain.^[44]

One of those deeds sells a father with his two sons: "Patiridata, their slave, whose right wrist is inscribed with the name of Nana-iddin, his former owner, and whose other wrist is inscribed with the name of Zabuduya his owner, the seller", together with Kidinnu-Bēl and Sananiqu, "a total of three slaves".^[44] The seller's name is Babylonian, the slave's is Iranian, and his sons' names are Babylonian again; whether the name indicates ethnicity or was an adopted name is not known.^[44]

Within these households enslaved people did skilled and responsible work. Basello lists barbers, textile workers, smiths and bakers, notes that some managed their owners' assets and acted as independent entrepreneurs, and records one who represented his master at Susa.^[1] Dandamaev and Lukonin set out the legal limit that constrained such careers: a slave could pledge his peculium but not his own person, and the master could lawfully strip him of it, so that well-to-do slaves could not buy their own freedom because everything they held was in the last analysis their master's.^[45] The same authors record that sixth-century Egyptian documents preserve free people selling themselves, with their existing and future children, into perpetual slavery, a regional practice that prevents the wider decline of debt slavery from being treated as the disappearance of every form of self-sale.^[45] Basello therefore uses a qualified formulation: "Debt slavery was no longer widespread in the Achaemenid period."^[1]

Babylonian slave-trading also operated inside Persis. Of eleven Babylonian texts connected with a place called Ḫumadēšu, identified with the Matezziš of the Persepolis tablets, four record slave sales, mostly under Cambyses; in two of them both the slaves and their owners bear Iranian names while the buyers bear Babylonian ones.^[1]

The Persepolis sale describes a man whose right hand carries the written names of two owners. Related practices are documented in several regions and scripts. A tablet of Cambyses' reign records a slave belonging to Itti-Marduk-balāṭu with inscriptions in Akkadian and Aramaic on his hand.^[46] Temple marks could instead identify a deity: Briant notes that the Eanna temple's flocks and its širku dependants bore the goddess's star.^[7] At Elephantine, Cowley 28 describes Petosiri and Belo, slaves divided between heirs of Mibtahiah. Each bears a yod on the arm to the right of the Aramaic ownership inscription. Cowley regarded the reading of the letter as secure but its significance as uncertain; the description does not specify a cut or place the letter above the inscription.^[47] In the fourth-century Samarian deeds from the Wadi Daliyeh the mark is called rwšm, whose cognates, Tuplin observes, "are apt to entail wounds or incisions". He considers these cognates to favour branding slightly, while leaving the distinction between burning and tattooing unresolved.^[48]

Whether the mark was burned or pricked is genuinely unresolved, and the current editors say so. Assessing the whole body of Babylonian, Egyptian, Aramaic and Greek evidence, the editors of the Aršāma letters conclude: "It cannot be said that a clear conclusion emerges from these data. We have opted for branding in our translation without complete confidence."^[49] Some Babylonian texts mention actual branding-irons and figurative marks such as a star, which favour burning; others speak of writing on the body in Aramaic or Akkadian letters, and a text of any length is hard to produce with an iron.^[48] Herodotus uses stizein, at first sight "tattoo", in three Persian contexts, and a Megarian epitaph calls Persians stiktai; Quintus Curtius describes Greek captives at Persepolis marked with the characters of barbarian letters, but the parallel accounts in Diodorus and Justin omit the detail.^[48]

The clearest Greek instance concerns prisoners of war. Herodotus reports that after Thermopylae most of the surrendering Thebans "were branded by Xerxes' command with the king's marks, from their general Leontiades downwards".^[50] Kuhrt qualifies the usual translation: the Greek phrase "is usually translated as 'branded with the royal brand'. In fact, it is not known how the mark was applied; tattooing is equally possible."^[51] She reads the relationship expressed by the marking as one of ownership, parallel to the estate-owner's mark in the Aramaic evidence, and suggests that the Thebans so treated became slaves, as captives usually did.^[51] The motif also ran loose in Greek writing: Herodotus had already reported the story that Xerxes sent branders to brand the Hellespont.^[52]

Babylonian temples held land, flocks and workshops, and the people attached to them occupied a status of their own. The oblates, širkū, were dedicated to a god and could not simply leave; Kleber's definition, adopted by Matarese, describes them as institutional dependants whose freedom was restricted, by comparison with the free citizens of a Babylonian town, because they were socially subordinated to a temple household.^[53] Calling them slaves flattens a distinction that the sale contracts themselves take care to preserve.

Dar. 43, a tablet drawn up on 27 September 520 BCE at Bīrtu-ša-Kīnāya in the Ebabbar temple's rural territory, regulates the lives of women and children placed under temple supervision. The tablet is broken, so the surviving list need not preserve the whole group. Louise Quillien identifies three women described as widows and five children, together with another woman whose father is named. The women are designated amīltu, a term which Quillien renders neutrally as women; they are not called oblates. The restrictions are specific: they must not reside with a notable, give a son or daughter in adoption to a notable, or move to another town. Three named women, Imattu, Mīstā and Bazītu, are to deliver three gulēnu cloaks annually to Šamaš as an iškaru work assignment, from their own production. The surviving text assigns responsibility for the group to Marduka, an oblate of Šamaš and canal inspector.^[54] Quillien interprets this as a loss of autonomy for previously free women left without adequate family support. The temple provides housing and probably rations; restrictions on residence, remarriage and adoption prevent the women and their children from leaving its dependent personnel. Their poverty and initial freedom are part of this reconstruction, while the obligations and restrictions are recorded in the document itself.^[55]

Temple dependants could also be moved at royal order. In the seventh year of Cambyses the administrator of the Eanna temple at Uruk and the royal commissioner made over forty farm workers to a man who was himself an oblate:

40 farm workers (ikkarātu) [of the Lady of Uru]k, whom Nabû-mukīn-apli, the administrator (šatammu) of Eanna, son of Nādin, descendant of Dābībī and Sîn-šarru-uṣur, Inninur, the royal intendant of Eanna (ša reš šarri bēl piqitti Eanna) gave to Lâbâši, son of Nanāja-ahu-iddin, the farm worker, the oblate (širku) of Ištar of Uruk, these 40 workers he will take to perform the work at the palace of the king, which (lies) in the town of Matnānu. In case anyone of them should go to another place, Lâbâši will bear the punishment of the king.

YOS 7 187, trans. Henkelman & Kleber^[56]

The penalty clause falls on the supervisor, and the workers travel from Babylonia into Persis. Matarese notes that the Achaemenids inherited the practice of requisitioning temple labour from their predecessors: a royal officer, the qīpu, planned the temple's corvée for the king, and the šatammu and the overseer of the oblates organised it.^[57] The requisition was organised through temple officers and a named supervisor whose responsibility extended to preventing flight.

A single career shows how far coercion could run in both directions inside such an institution. Between 539 and 520 BCE a temple dependant of Eanna named Gimillu rose from overseer of the temple livestock to collector of rents and lessee of hundreds of hectares of temple land. In 538 a court of fifteen freemen, convened by six men including the governor of Uruk and recorded by four scribes, convicted him on twelve counts of plundering temple property. In 530 he caught a temple slave who had been in hiding for two years and, instead of returning him, hired him out privately and pocketed the money, then contracted a second hiring of the same man. In 524 he was confined in a warehouse before trial, broke away from his guard and was caught at the temple gates; in 520 he was convicted again.^[58] Dandamaev and Lukonin themselves warn against generalising from him: Gimillu belonged to a privileged group, set against the mass of temple dependants who lived on rations and were ready to flee at any moment.^[58]

A fourth kind of compulsion fell on people who were not dependants at all. Households liable to labour duty owed the crown a term of work, and could discharge it by hiring someone else. A tablet from Dilbat dated 21 March 510 BCE records the whole mechanism in four lines: to fulfil the call-up duty of Uraš-le'i, a man named Uraš-ana-bītišu will go to Elam, and "For three months Urash-ana-bitishu will carry out the labour instead of Urash-le’i." If he fails to go, his own father must pay the cost of the hired workers who will do the service instead.^[3] The duty was compulsory for the liable household and contractual for the substitute, and the guarantee stood behind both. Another Babylonian was sent on his turn of labour service in Elam for an entire year under Darius I.^[26]

Public works were distributed through institutions. A tablet of Cyrus II’s eighth year records the measurement of a stretch of royal road against temple land at Sippar, named officials and the ropes and beams used; Kuhrt's note draws the implication that maintaining sections of the road was a duty of the temples, as work on the canals demonstrably was.^[59]

Where land carried service, credit followed. The Murašû archive at Nippur preserves a loan of 422/1 BCE to the holders of a bow-land: eighty gur of dates are owed to the firm, secured on the field, and "the dates are equivalent to the silver which has been lent to them for clothing and military equipment for presenting themselves at Uruk".^[60] Bow-land, horse-land and chariot-land were grouped in organisations called ḫaṭru; when there was no call-up the holders worked or leased the land and paid tax, and the military obligation did not lapse (see the Murašû archive).^[60]

Stolper identified a further distinction in the Babylonian evidence. Babylonian gardu appear as ration recipients and as holders of land allotments organised in a ḫaṭru, so that "they drew on both of the classic remunerative devices of ancient Near Eastern state economies: ration payments and allotments of income-producing property".^[61] He also asked, in 1985, whether the scattered mentions of gardu implied a royal economic apparatus in Babylonia comparable in type, scale and complexity to that at Persepolis; Tuplin judges that the Egyptian evidence for grd' reinforces the belief that such arrangements existed outside Fārs, while the question of scale remains hard to assess.^[61][62]

How most people entered the Persepolis workforce is a live and named disagreement. Bruno Jacobs holds that the Persepolitan kurtaš were characteristically the product of corvée systems; Henkelman restricts that to the sub-group additionally designated rabbap.^[35] The Aramaic documentation is less explicit about taxation through labour: Tuplin's survey of it concludes that corvée arrangements are elusive there and that "there are never any plain indications that the workers involved are giving their labour as a form of taxation".^[62]

The letters of Aršāma, satrap of Egypt in the later fifth century, bring acquisition, marking and threats against an estate manager together in a single document. Writing to his estate manager Nakhtḥor during a period of disturbance, Aršāma quotes his own earlier order back at him:

Guard [my] personnel and goods forcefully, so that there shall not be a[n]y loss from my estate. Also, from elsewhere, personnel of artisans of every kind, seek (in) sufficient (numbers), and bring (them) into my courtyard, and mark (them) with my brand, and make (them) over to my estate, just as the [pre]vious officials were doing.

Aršāma, TADAE A6.10, trans. D. G. K. Taylor^[5]

The order does not state whether the personnel were to be bought, captured or acquired by some other means. Porten and Yardeni render the opening of the instruction "seek domestic staff of craftsmen of all kinds" and the marking clause "mark them with my brand".^[63] Taylor’s "personnel" and their "domestic staff" render the same workforce designation. Both translations connect the personnel with an estate or household; neither specifies how those people had been recruited. Briant, who calls the newly recruited people "slaves", also acknowledges that the means of acquiring them is unknown.^[7]

Coercion in the letter runs upwards as well as down. Aršāma closes with a threat to his manager and colleagues: if losses continue or additions are not made, "you will (all) be questioned forcefully, and a severe sentence will be produced for you (Nakhtḥor)".^[5]

Tuplin's assessment of the status of these people is carefully bounded. The grd' of this letter are "at least potentially branded or tattooed workers, and nothing in other Bodleian letters really contradicts what that implies about their (possible) status"; but, he adds, "the Persepolitan and Babylonian evidence affords no comparably direct view of the status of kurtaš/gardu".^[31] He warns explicitly against reading the Egyptian case back into Iran: the people seized, marked and apparently permanently assigned to Aršāma's estate "hardly sound like ordinary corvée products", and "assimilation of the recruitment process of A6.10 to the Persepolitan and Babylonian situation begs a lot of questions".^[31][64] Tuplin explains Aršāma's word choice in terms of estate ownership: *gṛda- is a generic term, and "Aršāma uses it precisely because it captures his sense of ownership".^[64]

Ḥinzani the image-maker is an unusually individualised member of this estate personnel. He appears as Aršāma's 'lym, receives rations on the footing of the satrap's other garda, and is associated with people of his own household. Briant interprets the orders governing his movement between Susa and Egypt as evidence of dependence rather than a freely chosen itinerary. Tuplin discusses the long-term personal relationship with Aršāma, distinguishing his named career from the often anonymous groups of institutional workers.^[7][64] The letters also describe a party of Cilicians in terms which, for Briant, permit either slaves or servants, and show that estate garda possessed goods of their own, some of which Nakhtḥor is accused of taking. Briant therefore concludes that "the word garda is more a label than a technical term whose judicial status can immediately be defined".^[7] Briant nevertheless concludes that these estate personnel were dependants or slaves, rather than free men, working for provisions supplied by the stewards.^[7]

Josef Wiesehöfer catalogues twelve cases of Persian deportation reported in ancient tradition: Egyptian craftsmen to Susa under Cambyses, Barcaeans to Bactria, Paeonians to Phrygia, Milesians to Ampe, Eretrians to Cissia or Media, the resettled Anaspastoi on islands in the Persian Gulf, mutilated Greeks encountered near Persepolis, the Branchidae to eastern Iran, Boeotians to Media, prisoners from Sidon to Babylon and Susa in 346/5, Jews to Hyrcania, and Carians to Borsippa.^[65] Wiesehöfer's list carries its own warning. Within that narrative dossier, the Sidonian case is distinguished by attestation in a Near Eastern source, a Babylonian chronicle; every other case rests on Greco-Roman or Christian writers, and in several the historicity is doubtful or the figures plainly exaggerated.^[65] Of the western testimony as a whole he writes that "we should be highly skeptical about the western testimonies, for they tend to associate deportees with slaves too easily".^[65]

From Herodotus a procedure can be reconstructed in five stages: the issue of a royal order, the capture and preparation of the group, their conveyance to the king, the king's determination of new settlements, and the transfer.^[66] Deportees were settled collectively, in family or ethnic groups, in regions whose economic development or military protection made the settlement useful, and were permitted to keep their own traditions, which did nothing to make the movement voluntary.^[66] On status Wiesehöfer is careful in both directions: the Greek terms doulos and andrapodon and the oriental ardu, garda and kurtaš are all vague, "there is not much to be said for the assumption that the Persian deportees had the status of slaves", and yet "there is no doubt that the deportees were significantly restricted in their movement and freedom of action".^[67] On scale he corrects the received picture: he favours smaller, manageable groups over the mass transfers implied by some narratives and questions Herodotus' picture of a depopulated Miletus.^[67] The comparison with Assyrian mass deportation cannot be run on the surviving material, since the Assyrian claims may themselves be ideological, and the evidence he surveys is heavily concentrated on the west.^[67]

John Hyland proposes a more direct administrative test of the deportation narratives. He has published journal entries from the Fortification archive recording the travel of Lycian workers to Persis in 502 and 501 under authorisations issued by Artaphernes, satrap of Sardis; one party, conducted by two men with Anatolian names, Hizalla and Mida, drew a single day's flour at a rate from which Hyland calculates approximately 1,633 travellers.^[68] Most of these people are labelled not kurtaš but marataš, a word that appears elsewhere as an adjectival qualifier of kurtaš and so marks a sub-category inside the labour system rather than a separate one; marataš also occur from Carmania, Drangiana, India and Arachosia.^[68] Hyland proposes that Elamite marataš represents an Old Persian *varata- cognate with Avestan varǝta-, "confined, captive", continued in Middle Persian wardag and New Persian bardeh, "slave", and argues that these entries "provide the first direct administrative evidence for the Persian practice of deportation described in Classical authors".^[69] This interpretation depends on the proposed etymology and the administrative context: Henkelman, excerpting another Fortification journal in 2017, recorded only that the term "is probably Iranian, but as yet unexplained", while noting that its distribution makes an ethnonym or toponym unlikely.^[70]

A demographic argument points the same way without relying on any etymology. A text of spring 506 records rations at Rakkan for more than 1,100 Greeks, of whom 253 were children, a composition Hyland reads as indicative of involuntary travel.^[71] Briant, noticing the reverse pattern, the overall shortage of youths in the archive, suggests that some may have been sent to court as palace slaves, on the model of the tribute in children that Herodotus records.^[38] That tribute is itself attested only in the Greek account: the Colchians and their neighbours as far as the Caucasus, Herodotus writes, rendered gifts "every four years and are still so rendered, namely, an hundred boys and as many maidens", while the Ethiopians beyond Egypt sent five Ethiopian boys every other year (see tribute and taxation).^[72]

Greek writing about Persian servitude is a body of political rhetoric, and it describes free Greeks quite as much as Persian subjects. Its most quoted moment puts the charge in a Spartan's mouth and aims it at a Persian general. When Hydarnes advised the envoys Sperthias and Boulis to submit to Xerxes, they answered that his counsel was half knowledge and half ignorance: "you know well how to be a slave, but you have never tasted of freedom, to know whether it be sweet or not".^[73] Nothing in the exchange concerns saleability, inheritance or the capacity to contract. Herodotus' recurring image of subjects working and fighting under the whip belongs to the same register; Kuhrt observes that the motif serves "to stress one of his points, i.e. the difference between free men and subjects", and that "There is no other evidence for Persian soldiers fighting under physical duress".^[74] Dandamaev and Lukonin state the old synthesis that grew from this material: that all the king's subjects, satraps included, were regarded as his slaves, which is why the Greek authors wrote that the whole Persian people were a mass of slaves deprived of civic virtues.^[75]

Herodotus also describes an identifiable commercial practice, involving a Greek trader who supplied markets within the empire. Herodotus reports that Hermotimus of Pedasa, the most honoured of Xerxes' eunuchs, had been "taken captive by enemies and exposed for sale" and bought by Panionius of Chios. Panionius, he writes,

would procure beautiful boys and castrate and take them to Sardis and Ephesus, where he sold them for a great price; for the foreigners value eunuchs more than perfect men, by reason of the full trust that they have in them

Herodotus 8.105, trans. Godley^[76]

Kuhrt distinguishes the conflicting classical stereotypes of treachery and devotion from the problems of identifying individuals. Several high-ranking men described as eunuchs bear Persian names and belong to prominent families:

There is a further problem as to whether the term translated 'eunuch' should always be understood to mean a castrated man, or whether it is a more general term defining royal body servants generally, only some of whom were castrati.

Kuhrt, The Persian Empire^[77]

The Alexander historians supply accounts whose detail must be weighed against their literary construction. Quintus Curtius describes Greeks at Persepolis working in slave prisons, mutilated, speaking of the women "whom chance and necessity have joined to [them]" and of children "whom slavery has compelled [them] to acknowledge".^[38] Matarese explains the construction: the Greeks who collaborated with the Persians are contrasted sharply with these figures, who are presented as innocent victims to justify Alexander's role as avenger, and Philostratus' still more lurid account of the Eretrians conflicts with Herodotus, who repeatedly insists that the deportees were not harmed.^[78] The Greek narrative is evidence for what Greeks believed and wanted to say about Persian power. It is not a register of legal status.

A deed from Elephantine illustrates the obligations that could survive manumission. Meshullam son of Zakkur issued it for a woman and her daughter on a date which Kraeling converts to 12 June 427 BCE. The woman appears as Tpmt in Kraeling's transcription of this deed and as Tamut in his treatment of earlier documents concerning the same person. Kraeling identifies her with the handmaiden whom Meshullam had previously given in marriage to Ananiah son of Azariah. That marriage, in his reading, had not itself freed her.^[79] The release document states:

… said Meshullam, son of Zakkur, the Jew of Yeb the fortress of the degel of Arpaḥu to the woman Tpmt by name, his handmaiden, upon whose hand is a mark, on the right, thus: "(belonging) to Meshullam" saying: I have taken thought for thee in my life. I have gone and released thee (effective) at my death, and I have released Yehoyishma by name, thy daughter, whom thou didst bear to me.

Kraeling 5 (TADAE B3.6), trans. Kraeling^[79]

The following clauses exclude claims by Meshullam's family and other potential successors, prohibit renewed marking or sale, and impose a fifty-karsh fine on anyone asserting power over the women. Kraeling reads the release as taking effect at Meshullam's death. He also notes that the phrase "but thou art freed to the god" leaves uncertain whether a regular liberation ritual is meant or protection against a later challenge to the deed.^[79] Continuing service is express: "We will serve thee as a son or daughter provides for his (or her) father in thy lifetime and unto thy death, (and after thy death) we will provide for Zakkur as the son provides for his father". In Kraeling's interpretation this binds mother and daughter to support Meshullam and then his son. Their refusal would incur a separate fine of fifty karsh and a restriction on bringing proceedings. The protection against sale and marking therefore coexists with a continuing obligation to the former owner's family.^[79] Babylonian arrangements could likewise combine release with a continuing payment. The possibility of a slave accumulating a peculium does not imply an enforceable right to purchase freedom: the master's power over that property constrained the bargain.^[1][45]

The Greek text of the trilingual decree from the Letoon at Xanthos connects manumission with a civic cult's revenue. Under the satrap Pixodarus, the citizens established a cult and provided that those manumitted should pay two drachmas to the god. Metzger's Greek edition reads δύο δραχμάς, with the same amount in his French translation.^[80] The identification of the king named in the Aramaic date and the resulting absolute chronology are disputed; Kuhrt places it under Artaxerxes IV, whereas Marek discusses an earlier attribution. The amount of the levy does not settle that chronological question.^[81] The Aramaic text is shorter and does not include the manumission clause; the Lycian wording has its own interpretative difficulties.^[82]

Beyond Egypt, a substantial Persian-period corpus of Aramaic slave sales comes from the Wadi Daliyeh, fourth-century Samarian deeds in which 'lym is the standard word for the person sold and in which a slave can be warranted as "unimpaired".^[9] Their formulary differs from the Elephantine documents in significant respects, illustrating the persistence of regional contractual traditions within a shared administrative language (see the Elephantine papyri).

The surviving records do not provide a defensible proportion of enslaved people within an Achaemenid population. Totals derived from administrative texts must first be understood as counts assembled from records. Dandamaev's tabulation, reproduced by Briant, gives 21,576 recorded individuals divided among men, women, boys and girls. Briant challenges the use of its overall proportions as a picture of stable households: the groups differed in composition and size, moved between places, and could be reorganised.^[38] Before any archive total could become a demographic estimate, repeated appearances and the archive's coverage would have to be established. The demographic problem thus depends both on identifying recurrent recipients and on establishing the status of the persons recorded.

No surviving empire-wide code resolves the differences between their status terms. Conversely, slave sales continued under Achaemenid kings, and public offices participated in recording them. The documentation therefore supports neither a claim that Persian rule abolished slavery nor the assumption that every person under royal authority was a chattel slave.^[44][17]

Briant offers a strong interpretation of the coercion evident at Persepolis. He argues that the transfer and deployment of foreign work groups denied them autonomy, producing "an undifferentiated labor force that the administration intended to exploit unhindered". He compares this condition with the rural dependence of Hellenistic laoi, who retained villages, families and property: "All in all, this was a situation much closer to slavery than the 'helot' type of rural dependency".^[83] This interpretation concerns the control exercised over labour, while the conscription and release evidence points to variation in the duration and conditions of service.^[25][26] The central distinction remains between an institution's power to deploy a worker and a legally enforceable right to sell that person; the sources do not always allow both to be established for the same individual.