Abstract
An expedition of the Egypt Exploration Society in 1913–14 discovered four fragments of the Hebrew Bible (from the books of Kings and Job). This article presents the first critical edition of the fragments. With a few minor exceptions, the fragments conform to the Masoretic Text. The possible datings of these fragments range from the third to the early eighth centuries
Introduction
In 1913–14 the Egypt Exploration Society undertook an expedition to El-Sheikh ‘Ibada, the site of ancient Antinoopolis. The expedition, led by John de M. Johnson, uncovered numerous papyri and parchments. 1 Four fragments of parchment, P.Ant. 47–50, held in the Sackler Library Papyrology Room in Oxford, contain passages of the Hebrew Bible: 1 Kings 22:12–18, 28–33; 2 Kings 21: 8–9; Job 1: 19–2: 4; 20: 24–21: 14. But since the unvocalised consonants of these fragments are almost identical to those of the traditional manuscripts of the high Middle Ages, the Masoretic Text, no official edition of the fragments has appeared before. William McHardy, in an appendix to the first volume of The Antinoopolis Papyri, only presented the text of P.Ant. 48, while simply noting that the other fragments were virtually identical to the Masoretic Text, or, in the case of P.Ant. 50, that ‘the text is not decipherable’. 2 Ada Yardeni did manage to decipher the text of the second fragment of Job, and provided a pencil sketch of the fragments in Colette Sirat’s Les Papyrus en caractères hébraïques trouvés en Égypte. 3
Although the text-critical value of these fragments is indeed limited, the fragments are of great historical significance, since there are few textual witnesses from the period after the Dead Sea Scrolls and the manuscripts from the Genizah, the storeroom of the Ben Ezra Synagogue in Cairo. 4 That is why this interval is known as the ‘dark’ or ‘silent’ period in the history of the transmission of the Hebrew Bible. There are a few possible exceptions: the Nash papyrus containing the Decalogue and the Shema (now more commonly dated to the second century bce), a papyrus from Oxyrhynchus inscribed with Exodus 2: 23–25, a scroll from En Gedi with Leviticus 1: 1–2:1, and two leaves of a seventh- or eighth-century Torah scroll (MS London-Ashkar). 5 Mordechai Veintrob recently identified 13 fragments that were part of the same scroll as those two leaves of MS London-Ashkar. 6 Since these fragments are from the Cairo Genizah, the two leaves were probably from there as well. There are two fragments of Genesis from the Cairo Genizah that may be older than most of the biblical manuscripts from the Genizah, whose oldest securely dated manuscript, the Cairo Codex of the Prophets, dates from the 827th year after the destruction of the temple according its colophon. 7
In this article, we will discuss the physical properties of the four fragments from Antinoopolis, compare the text to the traditional text of the Hebrew Bible, attempt to date the fragments, and ask what the fragments can tell us about the history of Jews in Egypt.
Description
The two fragments of Kings, P.Ant. 47–8, are inscribed only on the hair side, which means that these are the remains of a scroll rather than a codex. Biblical scrolls are often written on the hair rather than the flesh side of parchment, for which Maimonides gives a possible explanation: ‘It is a law transmitted by Moses from Sinai that a Torah scroll be written on gevil, on the hair side … Anything written on qelaf on the hair side, or on gevil or dukhsustus on the flesh side, is unfit’ (Mishneh Torah 2.3.1.8). 8 The quality of the parchment of the two Kings fragments is very good. Since the colour of the parchment is pale and its surface smooth, this is probably vellum, parchment made from the skin of a young calf. The present appearance of the parchment may reveal something of its preparation. If a hide is rubbed in with salt, flour, or dung, it turns darker over time, but if it is soaked in a lime solution, it retains its pale colour. The preparation of parchment in a lime solution was typical in the West, whereas the salting, flouring, and tanning were more common in the East, at least in the Middle Ages. 9 However, Ira Rabin recently pointed out that both methods of preparation existed side by side already in Hellenistic Judea. 10 Most of the parchment scrolls from Qumran were probably prepared with salt or flour, the ‘eastern’ method, but the pale parchment Temple Scroll (11QTa) resembles the ‘western’, European parchment of the Middle Ages. 11 The preparation method of the Kings fragments is therefore not conclusive evidence for the dating or the provenance of the parchment.
The margin above the column is 60 mm, and the space between the columns ranges from 17 to 30 mm, the lines of the first column are 85–105 mm wide, the letters are 4 mm high on average. The columns are indicated by vertical, the lines by horizontal ruling. The large leaf of the Kings manuscript, P.Ant. 47, is cut off in a fairly straight line on the right side, there are holes at regular intervals along the cut, and there is a bit of string stuck in these holes. These may be the remains of stitching between two sheets, or may indicate that the Kings fragment was repaired after being damaged or reused as the binding of a codex. 12
The letters are written slightly below the ruling, as if ‘hanging’ from the horizontal lines. The Kings fragments are written in a clearly legible and consistent script, evidently the work of a professional scribe. Analysis of the ink would be helpful in determining the date and the provenance of the fragments, but this is beyond the scope of this article.
If the two fragments were once part of one scroll that contained both 1 and 2 Kings, this must have been a very large scroll. The books of Kings count 35,810 words in the edition of the Biblia Hebraica Stuttgartensia: if we assume that each line contained seven words on average, each column 42 lines, 13 and each column – including one margin – was about 11 cm wide, the scroll must have measured about 13.4 m in length in total. If we assume that each line contained six words, each column 38 lines, 14 the scroll must have been about 17.3 m long in total. By contrast, the 18 sheets of the longest of the Dead Sea Scrolls, the Temple Scroll, measure 8.1 m in total, and the longest biblical scroll found in the caves of Qumran, the Great Isaiah Scroll, is 7.3 m long. 15
The Job fragments, P.Ant. 49–50, are much more difficult to read, since the parchment is dark brown and covered in dust and dirt. Nevertheless, the script is precise and consistent. The lines are 60 mm wide, the letters are 2 mm high on average. Like the Kings fragments, the Job fragments are inscribed on one side only. Since only two lines have been preserved in full (P.Ant. 49.14–5), and since we do not know how many lines would have fit in one column, there is no way of estimating the length of the entire scroll. On the right side of P.Ant. 49, a seam line of 79 mm long is visible, where two leaves of parchment were stitched together.
Text criticism
At two minor points, the Antinoopolis fragments differ from the text of the Leningrad Codex and the Aleppo Codex, the two manuscripts modern editions of the Hebrew Bible take as their starting point. In P.Ant. 47, col. 2.9, there are traces of an ה that has been erased: the ink is smeared out over the parchment. This is interesting because a minority of manuscripts also read הקטן ‘the small’ instead of קטן ‘small’ in 1 Kings 22:31. Benjamin Kennicott, in his Vetus Testamentum Hebraicum cum variis lectionibus, documents nine such manuscripts with הקטן instead of קטן. 16 This reading may well have been influenced by a parallel passage in 2 Chronicles 18:30: ‘And the king of Aram ordered his charioteers, saying: Do not wage war against the small, the great (את הקטן את הגדול), but only against the king of Israel.’ The Antinoopolis fragment is not the only textual witness showing the conflation of these two texts. In the Bodleian Library, MS Huntington 12, a letter heh before גדול in the same verse, 1 Kings 22: 31, has been erased. 17
Interestingly, whereas most manuscripts of 2 Chronicles 18: 30 read את הקטן את הגדול, ‘the small, the great’, some manuscripts and ancient translations reflect the reading את הקטן ואת הגדול, ‘the small and the great’. 18 This reading was probably influenced by 1 Kings 22: 31: את קטן ואת גדול, ‘small and great’. the minor variations in this small phrase in 1 Kings and 2 Chronicles testify to the continued influence the parallel passages exerted on each other. At the same time, the erasure of the ה in the Antinoopolis evinces an effort to keep the text of 1 Kings distinct from the text of 2 Chronicles.
In Job 21: 10, the Leningrad Codex and the Aleppo Codex read יגעל ‘he fails’, but P.Ant. 49.15 has יגעיל. 19 Kennicott documents some 30 manuscripts with יגעיל in scriptum plene instead of יגעל. 20 There is no difference in meaning, only in orthography.
Dating
Johnson’s excavation report mentioned only one Hebrew fragment: ‘A fragment of Hebrew on brown leather attested the presence of the Jewish element.’
21
Fortunately, Johnson mentioned that this fragment was found in mound N, in the lower strata of which several papyri dating from the reign of Justin I, i.e. 518–27
McHardy, who provided some brief notes on the Hebrew fragments, argued: To date such fragments with any degree of assurance is impossible. At most it may be said that no feature of the fragments rules out the possibility of an early date, and they may very well fall somewhere within the period of the Greek papyri with which they were found, i.e. 3rd–6th cent.
22
At the time, in 1950, this dating seemed fully justified, since only papyri from the second to sixth centuries were published in the first volume of The Antinoopolis Papyri. 23 The other volumes also contained papyri that could or should be dated to the seventh century. 24 In his report Johnson also mentions in passing that mound N yielded many Coptic papyri: ‘In quantity Greek fell much below Coptic’. 25 The excavations at Antinoopolis also uncovered a number of Arabic papyri. Unfortunately, only a small number of the Coptic and Arabic papyri have been published to date. However, it should be noted that of the small number of documents in Coptic and Arabic that have been published, 12 papyri in Coptic are possibly from the eighth century, and the four bilingual papyri in Arabic and Greek date from the late seventh and early eighth centuries. 26 This means that a later dating of the deposition of the Job fragment, to the seventh or eighth century, cannot be excluded on archaeological grounds.
It is impossible to date the fragments on palaeographic grounds with any precision, since there are only a few dated manuscripts of the Hebrew Bible from this period. The oldest securely dated medieval manuscript of the Bible, the Cairo Codex of the Prophets, is from 895
The shape of the letters of the Antinoopolis Kings fragment have been described in more detail in Birnbaum, The Hebrew Scripts, I, 223–5, and in Engel and Mishor, IMSA 7, 52–5. Salomo Birnbaum, after comparing the Kings fragments to writing samples from the fourth and ninth centuries, assigned the fragments to the fifth century. 29 Engel and Mishor concluded that the MS London-Ashkar and the Oxyrhynchus Exodus fragment are more similar to each other than to the Antinoopolis Kings fragments. 30 They pointed out the distinct shape of the letters of the Kings fragments: the horizontal strokes are broader than the vertical strokes, the ends of strokes are sharply pointed, the letters are taller and their shape more consistent than in the other two texts (see fig. 1). The variation in thickness, or ‘shading’, is a feature of a more formal hand, 31 and has parallels in the En Gedi Leviticus Scroll and the Oxyrhynchus Exodus fragment (Bodleian Library, MS Heb. d 89 [P] i).

Comparison of a. Oxyrhynchus fragment, b. London-Ashkar manuscript, and c. Antinoopolis fragments (from Engel and Mishor, IMSA 7, 54).
Judith Olszowy-Schlanger has argued for a dating of the Kings fragments to the Islamic period, on palaeographic grounds.
32
Tal Ilan prefers the archaeological evidence, which knows no settlement in Antinoopolis in the Muslim period, and thus opts for an earlier dating in the sixth century. In the end, the extant evidence is inconclusive. What we do know for certain is that the Job fragments were discarded after 518
Jews in Antinoopolis
The publication of the biblical fragments from Antinoopolis is part of a larger project of reconstructing Jewish life in Egypt in the many centuries of Late Antiquity between the crushing of the Jewish revolt in Egypt in 117
Most of the evidence for Jews in Antinoopolis is ambiguous, either because the persons are not clearly identified as Jews, or because the evidence cannot be securely dated. Examples of ambiguous evidence are an undated amphora stopper decorated with a ‘menorah’, a seven-armed candelabrum that was the Jewish symbol par excellence, a papyrus mentioning a Moses and an Abraham, and a pot inscribed with ‘Jacob’. 39 But these names could have belonged either to Jews or to Christians. In addition to the parchment fragments discussed here, there are five less ambiguous pieces of evidence, proving that there were certainly some Jews in Antinoopolis: three papyri and two inscriptions.
A Jewish marriage contract from 417 ce, written in Aramaic with many Greek loanwords, documents the marriage between Samuel and Metra.
40
A Greek contract from 557, P.Ant. I 42 (=CPJ III 508), mentions that Aurelios Iosephios son of Souros, who lived in the village of Lenaios in the nome of Antinoopolis, was a Jew. A lease from 569 or 570
It should also be mentioned here that a fragmentary manuscript of the Samareitikon was reportedly found in Antinoopolis. The Samareitikon, the Greek translation of the Samaritan Pentateuch, has been preserved only in marginal glosses of manuscripts, 44 an inscription of Numbers 6: 22–27 in a synagogue at Thessaloniki, and a fragmentary manuscript of Deuteronomy 24–27, allegedly from Antinoopolis. 45 The editors of that fragmentary manuscript, Paul Glaue and Alfred Rahlfs, noted that it was palaeographically similar to a Coptic manuscript of the fifth or sixth century in the Vatican Library (Borg. copt. 109), but could not agree on the dating of the Samareitikon manuscript: ‘Glaue hält das 5., Rahlfs das 6. Jahrh. für wahrscheinlicher’. 46 Exactly how this manuscript ended up in Antinoopolis, if it did, we do not know: was there a Samaritan community in Antinoopolis as well? No certain conclusion can be drawn from this manuscript.
Interestingly, neither Kings nor Job features prominently in the Jewish reading cycles. Some chapters of Kings were used in the haftarot, passages from the prophets read alongside the Torah in the synagogue, but not the chapters preserved in P.Ant. 47–8. 47 According to a rabbinic tradition, the book of Job was read on the ninth day of the month of Av, the date of the destruction of the Temple, which was commemorated with rites of mourning and the reading of exceptionally depressing texts. 48 Whether these scrolls were meant to be read privately at home, recited in a synagogue, or studied in a school, the fragments from Antinoopolis may indicate that Kings and Job were more widely known than initially thought. 49
Conclusion
Although these sparse pieces of evidence indicate that there were definitely Jews in Antinoopolis in the fifth and sixth centuries
Edition
In this edition, we follow the Leiden Conventions, with a number of exceptions. Instead of indicating the illegible letter with dots, we have transcribed them in order to present an understandable text and because we would not have been able to make sense of bits of the Job fragments (P.Ant. I 49–50) without the help of the Masoretic Text. (The position of words on the line is somewhat conjectural in the case of the Job fragments.) Second, in order to avoid confusion with the hireq, we have indicated unclear letters not with the usual dots below the letters but with empty circles above the letters (א֯). Third, there are no notes at the end, because most of the details have been discussed in the introduction and for the Job fragments it is often impossible to distinguish between dirt and ink traces. This edition differs from Ada Yardeni’s meticulous sketch only at two significant points: in P.Ant. I 49.15, Yardeni read אעיל instead of יגעיל (see earlier section ‘Text criticism’), and in P.Ant. I 50.11, we were able to decipher seven more letters.
P.Ant.
Footnotes
Acknowledgements
We would like to thank Dominic Rathbone and the Egypt Exploration Society for granting permission to publish these fragments, Arkaprabha Chakraborty for carefully analysing the fragments, and the attendants of the Workshop for Manuscripts and Text Cultures at Queen’s College, Oxford for their questions and suggestions, in particular Christelle Alvarez, Ed Love, and Andreas Winkler. We would also like to thank the anonymous reviewers for their meticulous and incisive comments.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
