A trilingual sales contract on papyrus from Roman Arabia (P.Yadin I 22)

Michael Zellmann-Rohrer
Freie Universität Berlin, Institut für Wissensgeschichte des Altertums,
Published under the Creative Commons Attribution-ShareAlike 4.0 license.
Abstract: This contribution considers the context, textual content, and means of textual division in a trilingual sales contract from Roman Arabia. The text, P.Yadin I 22, formed part of the so-called Babatha archive, the family papers of a Jewish woman who later took refuge in the Judaean Desert, where the documents were later discovered. Despite the absence of formal punctuation marks or word division comparable to modern conventions, the writers of this papyrus used various subtler means to highlight divisions in sense and content, related also to their choices among the three languages at work in the contract (Greek, Nabataean Aramaic, and Jewish Aramaic). This highlighting will have served in the first instance to ease future reference in a document meant to achieve maximum legal precision and authority, but the interplay of languages may also be seen to have ideological implications in the history of the region.

Keywords: papyrus, contract, trilingual, Greek, Aramaic.


Papyrus is an ancient writing material prepared from the fibrous pith of the papyrus plant (Cyperus papyrus), laid out in two-ply sheets, and broadly comparable in physical character to modern paper. Developed in ancient Egypt, papyrus as manuscript substrate was used from the pharaonic through the medieval Islamic periods (ca. 3000 BC–AD 1000), where its survival in the archaeological record is also the best, due to dry conditions in desert areas. It also came into broad use across the Mediterranean area, and even beyond, but survivals are limited by less favourable, that is, wetter, climate conditions. The texts written on ancient papyrus ranged widely, from literary manuscripts through technical treatises, manuals for the performance of cult, and students’ textbooks, to official documents and deeds and receipts from the legal and economic dealings of everyday life.


The papyrus examined here is part of an archive, that is, a collection of documents deliberately kept together in antiquity, belonging to a Jewish woman named Babatha, hence the ‘Babatha archive.’On the archive see recently Czajkowski (2017); Esler (2017). Babatha, born at the beginning of the second century AD in a prosperous agricultural community at the southern shore of the Dead Sea (modern Jordan), later fled with her archive in the unrest of the Bar Kokhba Revolt in AD 132–35. She eventually took refuge in a cave in the Judaean Desert, and presumably died there, where, along with the documents, still enclosed in their ancient leather pouch, other artefacts such as a key and two key-rings, which probably belonged to properties mentioned in the archive itself, were found in controlled excavations in 1960–61. The site took the name the Cave of Letters after its modern discovery. A central concern of the archive is an ancestral estate, consisting in particular of date-palm orchards. The date-palm (Phoenix dactylifera), now extinct in the region, proved well-suited to cultivation there in antiquity in areas where irrigation was available, yielding a valuable cash crop for export. This estate was in the family’s home in Maōza (Greek Μαωζα from Aramaic maḥôzâ ‘harbour’), assembled by the father of the principal in carefully-documented purchases of parcels of his neighbours’ land from AD 99. At that time the area still lay in the kingdom of the Nabataeans, whose ruler had however from 62 BC been reduced to a client of Rome, which controlled the bordering province of Judaea. Internal references in the documents bear witness to the transition to Roman control of Maōza too, after the annexation of the Nabataean kingdom in AD 106 and the formation of the Roman province of Arabia. The estate, known as ‘the plantation (gannat) of Nikarchos’, a former owner, belonged to Babatha’s second husband. After his death, she seized the property in an attempt to pay off some of his debts, setting off litigation between her and his heirs, which lies beyond the scope of this study.


The present document (P.Yadin I 22; Beyer 2004: 236–37 nV22) is written in black ink on a rectangular sheet of papyrus, cut from the roll on which it would originally have been sold after manufacture.Jerusalem, Israel Antiquities Authority storerooms, Dead Sea Scrolls inv. 5/6Hev 22; the edition of reference for the text is Lewis, Yadin, and Greenfield (1989) (henceforth P.Yadin I), pp. 98–101 no. 22; cf. Beyer  (1984: 319) and Beyer (1994: 181–82 no. *nV 22); and further the entry in the Trismegistos database, text no. 23502. The text is a sales contract for a date-palm crop in advance of its harvest, transacted at Maōza on 11 September AD 130. It is the work of no fewer than six, and perhaps as many as eight hands writing in three languages. The papyrus was later cached along with other documents in the ‘Cave of Letters’ in the Judaean desert, owing this disposition to the circumstances of the Bar Kokhba Revolt described above, and is now kept in Jerusalem. This contract is paired with another found in the same context, P.Yadin I 21 (Beyer 2004: 235–36 nV21),Jerusalem, Israel Antiquities Authority storerooms, Dead Sea Scrolls inv. 5/6Hev 21; edited as P.Yadin I 21; see further the entry in the Trismegistos database, text no. 23501. which relates to the same sale: both the buyer and the seller affirm the terms of the sale in separate, witnessed documents, of which we happen to have both versions. The buyer, Simon son of Iēsouos (in Greek; in Aramaic, Šimʿôn son of Yešūʿa) made his affirmation in P.Yadin I 21, and the seller, Babatha, in P.Yadin I 22. Both documents are multilingual: P.Yadin I 21 is primarily in Greek with paratexts in Jewish Aramaic, while P.Yadin I 22 has the same and adds Nabataean Aramaic. The sale is of the current year’s crop of three separate date-palm orchards in the vicinity of Maōza, which Babatha had recently inherited after her husband’s death, and perhaps represents an underlying arrangement of labour-lease, in which the buyer in effect leases the orchards for the purpose of harvesting the crop, receiving a portion in return; Simon’s version may have remained with Babatha due to a dispute over her claim to the orchards, which resulted in the cancellation of the deal.See recently Radzyner (2005: 145–63).

Text structure

The structure of the text and its paratexts may be divided by content into four main sections: A–D. The first and most extensive (A) has further sub-sections; the shorter paratexts (B–D) consist of additions aimed at giving precision and confirmation to the legal force of the document. I outline the structure first in prose, then present a schematic form below. The body of the document (A), in Greek, is concerned first to state a precise date (A.a) and location (A.b) for the transaction. The text then couches itself as a first-person (‘subjective’) declaration from the seller to the buyer (despite the former’s choice to employ a professional scribe to draft it, as we will see), beginning with a formulaic expression of greetings (A.c) and proceeding through the acknowledgment itself (A.d), the terms including price and delivery of payment (A.e), and penalties for the buyer and seller, including in the latter case a failure in an additional duty of indemnification (A.f–g). The section closes with provisions for the disposition of any excess from the crop of dates, which is to go to the buyer in consideration of his efforts in harvesting them (A.h). Since the seller and issuer of the document was a woman, the intervention of a legal guardian was required for the instrument to be legally valid, and that this was duly done is stated in an appendix (A′). As the seller had further elected to use the services of a professional scribe in drafting the document, it was also necessary for a recapitulation to be given in her own hand (or rather that of her legal guardian) as a fuller form of signature: this text comes in B, in Nabataean Aramaic. The professional scribe responsible for A has, perhaps at the same time as writing A but certainly before the signatures of the witnesses in D, added a note affirming his role (C), offset some vertical distance from the end of A to accommodate the paratexts: he identifies himself as Germanus, λιβλάριος (liblarios: ‘scribe’, a loanword from the Latin librarius). Last come the signatures of witnesses to the transaction, in Jewish Aramaic: their number is uncertain due to the damaged condition of the papyrus, but there appear to have been either four or six, who fitted their names with varying degrees of competence in and around the space left between B and C.

Schematic version

A. Body (Greek), lines 1–30

  1. Dating by regnal year of the Roman emperor (14th year of Hadrian); eponymous Roman consuls (Marcus Flavius Aper and Quintus Fabius Catullinus); Roman month (three days before the Ides of September); Roman provincial era (25th year of the Province of Arabia); and Macedonian month (24th of Gorpiaios)
  2. Location: Maōza in the territory of Zoara
  3. Salutation: the seller, Babatha, addresses greetings to the buyer, Iēsouos, and states both of their domiciles (Maōza)
  4. Subjective declaration of sale: of the current year’s crop of dates from three date-palm orchards, named as Gannath Pherora, Gannath Nikarkos, and ‘Molchaios’ (orchard)’, of which the seller has taken possession in lieu of the recovery of her dowry from the family of her late husband
  5. Terms
    1. Price: in cash, 42 talents; and in kind, two kor and sata (of dates)
    2. Delivery: to be measured out on the local standard in the buyer’s house
  6. Buyer’s penalty: two denarii per talent in cash with an additional penalty in kind
  7. Seller’s duty of indemnification and seller’s penalty: twenty silver denarii in cash
  8. Provision for excess produce: all excess will go to the buyer in consideration of his expenses

A’. Statement of intervention of legal guardian (Greek), lines 28–30: a guarantee has been asked and acknowledged via the seller’s legal guardian, Iōaanēs Macchouthas

B. Seller’s recapitulation (Nabataean Aramaic), lines 31–35

C. Subscription by the scribe of A–A′ (Greek), lines 39: ‘It has been drafted by Germanus, librarius.’

D. List of four witnesses (or six, depending on whether the now illegible 36 gave a heading or a witness name) (Jewish Aramaic), lines 36–40 (fitted around the pre-existing part of line 39)

Internal highlighting of textual divisions

The preceding analysis has been based on a modern reading of the text from an edition printed with modern conventions of orthography and punctuation, informed by expectations of the form based on comparable documents. The sense that could have been made of the document in its ancient context was probably also determined by internal highlighting and other deliberate markings of textual divisions. The main body (A), the work of the professional scribe Germanus, has the appearance of continuous text. Formal punctuation is noteworthy in its absence. The same goes for B and D, as much as their shorter length and the abilities of their less experienced writers allow us to determine. Yet some subtler means of highlighting divisions can nevertheless be found, of six types.

I) Spaces between words

Close inspection of A in particular shows that the lettering is not so continuous as it first appears. Without direct access to the thought-processes of the ancient writer, we cannot be sure to what extent such spacing was intentional; in fact, it is not consistent, and is sometimes even present by coincidence in a place where it is not required, such as in the midst of a single Greek word. The following examples nevertheless tend to show a correlation between spacing and sense division:

  1. A.d (line 9), a space between [Μα]ωζᾳ (Maōzā) ‘(in) Maōza’ and κατέχω (katekhō) ‘I possess’ marks the beginning of a parenthetical statement in which the seller details how she took possession of the orchards whose crops she is selling.
  2. A.e (line 15), a space between σου (sou) ‘(in) your (house)’ and ζύγῳ (zugō) ‘(on) the weight-standard (of Maōza)’ marks a sense-pause between the specification of place of payment and the standard on which it will be counted.
  3. A.e (line 17) a space between δύω (duō) ‘two (kor)’ and σάτα (sata) ‘(five) sata marks the transition between specification of two different measurements for payments in kind.
  4. A.e–A.f (line 17), a space between Μαωζας (Maōzās) ‘(on the volumetric standard) of Maōza’ and ἐάν (eān) ‘if’ marks a new clause and the transition from the subsection on terms (e) to that on the buyer’s penalty (f).
  5. A.f (lines 18–19), a space between προγέγραπτε (progegrapte) ‘(if you do not do ... as) aforewritten’ and δώσις (dōsīs) ‘then you will pay’ marks a sense-pause between two limbs of a conditional (condition and consequence) in the formula for the buyer’s penalty.
  6. A.f (line 19), a space between μοι (moi) ‘(you will pay) to me’ and ἑνί (heni) ‘for (each) one (talent)’ marks a sense-pause between the specification of payment and the beginning of the specifics of its terms.
  7. A.f–g (line 20), a space between μίαν (mian) ‘one (unit of payment in kind in dates)’ and ἐμοῦ (emou) ‘while I (will indemnify you)’ marks a new clause and transition from the section on the buyer’s penalty (f) to that on the seller’s obligation to indemnify and the related penalty (g).
  8. A.g (lines 21–22), a space between ἀντιποιουμένου (antipoioumenou) ‘(from every) counterclaimant’ and ἐάν (eān) ‘if (someone makes a counterclaim)’ marks the transition from the statement of the seller’s obligation to indemnify to the penalty for failure to do so.
  9. A.g (lines 23–24) a space between προγέγραπτε (progegrapte) ‘(if I do not do ... as) aforewritten’ and ἔσομαι (esomai) ‘then I will be (in debt to you)’ marks the division between two limbs of a conditional (as in 5 above) specifying the seller’s penalty for failure to indemnify.

This form of highlighting was not confined to the Greek text drafted by the professional scribe (A), for the Nabataean Aramaic of B even in its much shorter compass and as the work of a competent though not professional writer shows at least two examples:

  1. B (line 31), spaces between the first writing of šmʿwn ‘(I, Babatha daughter) of Šimʿôn’ and zbnt ‘I have sold’, and between the second writing of šmʿwn ‘(to you,) Šimʿôn’ and ʿllt ‘the produce (of the date-palm orchards)’ mark sense-pauses respectively between the clauses identifying the seller and fact of sale, and buyer and item sold.

II) Spaces between lines

Normal practice in copying texts on papyrus was to leave some vertical interlinear space, which professional scribes will as a rule have striven to make as even as possible; variations can therefore draw the eye to internal division. In A.a–b (lines 1–6), the formulaic opening clauses specifying the date and place of the transaction have rather tight vertical spacing, and could even have been drawn up in bulk, in advance of any particular transaction; the substance in the following sections, giving the names of the principals and the details of the transaction, is spaced more generously (A.c–h), corresponding to the most individualized portion of the document that will also have benefitted from maximum clarity and legibility. The appendix (A′) reverts to tighter spacing and smaller letter-size. There are also two corrections by the scribe in A (line 16), which he elected to make directly above the line.

III) Vertical separation of sections

Among the most readily apparent features of the text is its vertical segmentation into distinct blocks, each of which, with one important exception that tends to prove the rule, is begun on a new line (A/A′, 1–30; B, 31–35; C, 39; D, 36–40). This disposition not only facilitates the cooperation of multiple writers on the same document by keeping their lines of text separate, but also aids the reader’s eye in distinguishing particularly the main body from the paratextual additions. The Greek (A, C) sections with the greatest official weight enclose those in the local languages (Nabataean Aramaic and Jewish Aramaic) at the top and bottom (see further below), but for the last of the witnesses signing in Jewish Aramaic in D who, whether from graphic inexperience or even a subconscious tendency towards subversion, begins his line (40) on a level with the professional subscription in C (39) but then lets it dip beneath and, in effect, have the last word.

IV) Letterforms

The relationship between a change in content and a shift in letter-size observed in Section II) can be seen in some cases where the size of a letter might highlight a textual division.

  1. At the beginning of the text (A.a, line 1), the first word of the dating clause, and of the document, ἔτους (etous) ‘year (14 of the reign of Hadrian)’, has its first letter enlarged.
  2. In the specification of terms (A.e, line 17), the numeral in a sum to be paid in kind, δύω (duō) ‘two (kor of dates)’ is written with an enlarged initial letter.
  3. Within the dating clause (A.a, line 3), the forename of one of the eponymous Roman consuls, Κοείντου (Koeintou) ‘Quintus’, is offset from the following word by means of an enlarged and flourished final letter.

As with the method described in I), however, in each of the phenomena illustrated in 2. and 3., counter-examples can be found showing inconsistency or even coincidence. One can therefore speak only of a general correlation between the highlighting and the associated meaning.

V) Internal headings

The enlarged initial letter of the first word of the document discussed in IV.1 cannot properly be considered a heading, since it is already integral to the sense of its clause, but practically it serves some of the same functions. It immediately identifies the genre of the document, a contract, which will invariably have opened with such a dating clause. A more transparent instance of an internal heading occurs at the end of A′ (line 30), where the professional scribe closes the appendix to the main body of the contract with ἐπιγραφή (epigraphē) ‘additional writing,’ to introduce the paratextual additions of the seller’s recapitulation (B), his own signature (C), and those of the witnesses (D).This interpretation of ἐπιγραφή seems preferable to that advanced in P.Yadin I, as the term surely has this sense in P.Yadin I 5 a ii 14, already cited there, while its sense to mark a scribal correction preceding in the document is unparalleled (and the scribe will then have neglected to signal the other supralinear correction in this same line 16); the following ἔτι δέ will then be a phonetic spelling for ἔστι δέ, meaning ‘(There is additional text), and it is (as follows)’, another well-known formula in headings, not a stipulation to the supralinear addition ἔτι δέ in line 16.

VI) Horizontal alignment and justification of lines

The text is a multilingual amalgam of one Indo-European language with a left-to-right writing direction (Greek), and two Semitic languages (or more precisely, dialects of a single language with widely divergent writing systems) with right-to-left writing directions (Nabataean Aramaic and Jewish Aramaic). The horizontal disposition of the text will thus be partly conditioned by language choice, but scope nevertheless remains for highlighting textual divisions. The professional scribe of the Greek section in A naturally takes a left justification, but aims also, via word-breaks across lines, to keep the right margin fairly tidy as well; when the same scribe roughly centre-justifies C with respect to A, the two sections are united across their vertical separation by B and part of D. In the Nabataean Aramaic recapitulation (B), the justification will naturally be to the right, but the writer chose not the right end of the main Greek text (A) as his standard, but rather that of A′, somewhat inset from that, which is the closest point at hand and also happens to be that in which the writer of B himself is named, for he is the legal guardian of the seller whose intervention as such is acknowledged in A′. This writer is either unwilling or unable to break words across the line, as Germanus the professional scribe of A readily does—indeed, no parallels for such word-breaks are found in contemporary Jewish and Nabataean Aramaic—so the left edge of B is much less even than the corresponding (right) edge of A. The witnesses who signed in Jewish Aramaic (D), in turn, also took up a right-justification, but set as their standard the right edge of the main body (A) rather than A′ or for that matter the proximate B, in keeping with the function of their signatures as witnesses to the transaction as a whole, more than to any one of its parts.

Language and script

Some words, finally, on script choice. Script, tied closely to language, cannot be considered as deliberate a means of division-highlighting as those just described, but it is nevertheless a choice, which may be understood in the same conceptual space. The Greek of the main body is consistent with the position of the Greek language as that of administration in the Roman Near East: it is no surprise that the professional scribe chose it, and that in turn the parties to the transaction engaged a professional with this script- and language-competence, which they apparently lacked. The second Greek intervention, the professional scribe’s signature in C, effectively brackets the non-Greek paratexts, by this (near) circumscription authorizing and subsuming them within Hellenophone officialdom. Elsewhere, language choice probably has to do with the limiting factor of the competence of the writers, in this case the legal guardian and the witnesses, locals without Greek-language education in a region where even its official use was only a quarter of a century old. It is noteworthy however that the locals may use either Nabataean, as the seller’s legal guardian does (B), or Jewish Aramaic, as the witnesses (D). The writer of B was certainly also Jewish, and it is difficult to imagine that he would have been unable to pen his portion in Jewish Aramaic too, but he chose Nabataean instead. One may conjecture that this choice was aimed at increasing the official impression and legal force of the document, since Nabataean had been the official language of local administration until recently, and in fact some documents in the Babatha archive dating to that period are drafted entirely in Nabataean. This writer, then, will have acquired competence in this former official language, probably when it was still official, and reverted to it as the next best option besides Greek when called upon to discharge his present role as legal guardian. The signatures in D, finally, may follow an internal vertical organization with respect to literacy, as they seem to be in roughly descending order of competence in forming letters regularly and in a straight line. The most irregular and apparently unpractised of these writers, in line 40, may, as noted already, nevertheless have put his sloping line to work in letting the local Aramaic circumscribe the official Greek and bringing the document to its conclusion.


The papyrus document P.Yadin I 22 is a legal instrument concerned with establishing precisely and durably the terms of a private transaction. It became part of the papers retained by the seller, while further copies, now lost, surely will have done for the buyer. Such documents were likely to be consulted in the first instance in case of a breach of terms or later dispute between the parties. The primary constituent of the meaning that such future consultations might have hoped to extract would have been of course the written text itself. For ease of reference and precision, however, various means of highlighting divisions within the text would have been of service. These devices would, for example, have assisted anyone who wished to pick out a particular specification from amongst the lengthy legalistic formulae, such as the location or kind of property concerned, the date on which the sale took effect, the names of the parties, and the details of the penalties to which they subjected themselves.

At a higher level, the different writers were clearly concerned when disposing their texts with the need to align the different writing modes associated with the different languages used in the same document. While language choice and script disposition themselves should not be pressed too hard for meaning, absent more discursive testimony from the original writers, ideological factors cannot have failed to leave some impression on physical form. P.Yadin I 22 is a witness to an unusually complex interaction between different linguistic and scribal communities: Greek, the contemporary language of administration, with its official position and prestige, which would have sufficed for the entire document; Nabataean Aramaic, the recent predecessor to Greek in this role in the area; and Jewish Aramaic, the mother tongue of the transacting parties and witnesses and soon to have its day, however short-lived, at the official centre of the Bar Kokhba Revolt in nearby Judaea-Palestine. Closely linked with choices of language and script are divergent ways of interacting with existing or anticipated text on the part of each of the parties. The professional scribe of the Greek text sets out the framework in which the other writers are to operate, at the graphic as well as the legal level. The writer who summarizes and subscribes in Nabataean Aramaic aligns closely to that program, whereas the witnesses, signing in Jewish Aramaic, may be seen to assert a certain independence.

Further reading

  • Bagnall, R. 2011: Everyday writing in the Graeco-Roman East. Sather Classical Lectures 69. Berkeley: University of California Press.

  • Crisci, E. 1996: Scrivere greco fuori d’Egitto: ricerche sui manoscritti greco-orientali di origine non egiziana dal IV secolo a.c. all’VIII d.c. Papyrologica Florentina 27. Florence: Gonelli.

  • Gascou, J. 2009: ‘The papyrology of the Near East’. In R. Bagnall (ed.), The Oxford handbook of papyrology. Oxford: Oxford University Press, 473–94.

  • Macdonald, M.C.A. 2009: Literacy and identity in Pre-Islamic Arabia. Collected Studies 906. Farnham: Ashgate.

  • Schiffman, L.H. and VanderKam, J.C. (eds) 2000: Encyclopedia of the Dead Sea Scrolls, 2 vols. New York: Oxford University Press.


Beyer, K. 1984: Die aramäischen Texte vom Toten Meer samt den Inschriften aus Palästina, dem Testament Levis aus der Kairoer Genisa, der Fastenrolle und den alten talmudischen Zitaten, vol. 1. Göttingen: Vandenhoeck & Ruprecht.
Beyer, K. 1994: Die aramäischen Texte vom Toten Meer samt den Inschriften aus Palästina, dem Testament Levis aus der Kairoer Genisa, der Fastenrolle und den alten talmudischen Zitaten, vol. 1: Ergänzungsband. Göttingen: Vandenhoeck & Ruprecht.
Beyer, K. 2004: Die aramäischen Texte vom Toten Meer samt den Inschriften aus Palästina, dem Testament Levis aus der Kairoer Genisa, der Fastenrolle und den alten talmudischen Zitaten, vol. 2. Göttingen: Vandenhoeck & Ruprecht.
Czajkowski, K. 2017: Localized law: the Babatha and Salome Komaise archives. Oxford studies in Roman society and law. Oxford: Oxford University Press.
Esler, P.F. 2017: Babatha’s orchard: the Yadin papyri and an ancient Jewish family tale retold. Oxford: Oxford University Press.
Lewis, N., Yadin, Y., and Greenfield, J.C. 1989: The documents from the Bar Kokhba period in the Cave of Letters, vol. 1: Greek papyri. The finds from the Bar Kokhba period in the Cave of Letters 2. Jerusalem: Israel Exploration Society.
Radzyner, A. 2005: ‘P.Yadin 21–22: sale or lease?’ In R. Katzoff et al. (eds), Law in the documents of the Judaean Desert. Supplements to the Journal for the Study of Judaism 96. Leiden: Brill, 145–63.
Figure 1. P.Yadin I 22.
A. Main body of the contract (Greek)
A′. Statement of legal guardian (Greek)
B. Seller’s recapitulation (Nabataean Aramaic)
C. Subscription by the scribe of A–A′ (Greek)
D. List of witnesses (Jewish Aramaic)