Document of the Month 5/26: Trade Letters from the Indian Ocean
Jewish Trade Letters with Blessings for Muḥammad
by Alan Elbaum
This Document of the Month discusses a surprising phenomenon among the Cairo Geniza documents: the appearance of Muslim blessings for the prophet Muḥammad (the taṣliya) in two mercantile letters sent by twelfth-century Jewish traders in the Indian Ocean.1
I identified this phenomenon through my research with the Indian Ocean Documents team of the Princeton Geniza Lab as we work toward completing the publication of the hundreds of Cairo Geniza documents known collectively as the India Book.2 These documents, dating mainly to the eleventh and twelfth centuries, offer rare and invaluable evidence about the mercantile networks linking the Mediterranean and Egypt with the Red Sea, South Arabia, the west coast of India, and Southeast Asia.3 The India Book was the name given by Shelomo Dov Goitein both to the corpus itself and to his projected seven-volume series of editions. Goitein did not complete the project in his lifetime, but the first four volumes have since been published in Hebrew by Mordechai Akiva Friedman and Amir Ashur, along with an English translation of the first three volumes.4
Blessings for Muḥammad
The puzzle begins with ENA NS 48.2, a fragment of a Judeo-Arabic5 letter sent by Maḍmūn ibn Ḥasan, a Jewish communal leader and wealthy merchant based in Aden, to Avraham ibn Yijū, a North African Jewish merchant living on the Malabar coast of India at the time of the letter (1136–39) (Figure 1).6 The sender can be identified based on the handwriting—belonging to one of Maḍmūn’s secretaries—and the reports about current affairs in Aden: The economy is in a slump, and the governor Bilāl is taking the first pick of imported merchandise, including an unidentified commodity called דרכי (DRKY). Ibn Yijū later reused the blank verso of the letter for a business account; as he had the letter in his possession, he was most likely the intended recipient.7
Fig. 1: New York, Jewish Theological Seminary, ENA NS 48.2 verso. The commodity name DRKY is the first word on line 5. Courtesy of The Jewish Theological Seminary.
As I was recently studying AIU XII.116 (Figure 2), an unedited mercantile letter written in difficult Arabic script, I noticed the same mysterious word دركي (DRKY).8 The Arabic letter lacks an address but turns out to be a nearly verbatim version of the Judeo-Arabic letter ENA NS 48.2.9
How did two copies of one letter sent from Yemen to India, one in Judeo-Arabic and one in Arabic, end up in the Cairo Geniza? Both copies presumably formed part of the personal archive of Ibn Yijū, which he carried back with him from India to Yemen and from Yemen to Egypt. Indian Ocean traders regularly sent multiple copies of the same letter due to the precariousness of travel; there is another letter, also from Maḍmūn to Ibn Yijū, of which fragments of at least four different copies ultimately wound up in the Cairo Geniza.10
Beyond the difference of script, the two copies of the letter diverge in another key way. Whereas the Judeo-Arabic version moves directly from the end of one sentence (“You should not believe that I have neglected your need and that which would produce something beneficial to you”) to the beginning of a new sentence (“By God, this year there was in the city a great fall (in the market)”), the Arabic version inserts a conclusion consisting of blessings for the addressee (“abundant peace upon you and your intimates”), for God (“praise be to God alone”), and—unexpectedly—for Muḥammad (“may His blessings be upon His messenger our master Muḥammad the prophet and his pure family”).11 The Arabic version then continues in the margin with the same text as in the Judeo-Arabic version (“By God, this year…”). The presence of this premature conclusion in the Arabic version suggests that Maḍmūn originally intended the letter to end after the first sentence and that the remainder is a postscript. We can further speculate that the Arabic version preceded the Judeo-Arabic version and that the scribe of the latter omitted these blessings in the process of copying from the Arabic original.
Fig. 2: Bibliothèque et archives de l’Alliance Israélite Universelle (AIU), AIU XII.116 recto, with the bottom lines containing the taṣliya circled. The commodity name DRKY appears in lines 16 and 19. Courtesy of Bibliothèque et archives de l’Alliance Israélite Universelle.
The blessing for Muḥammad (the taṣliya) routinely appears at the conclusion of medieval Arabic letters and government correspondence written between Muslims.12 It also comprises part of the formulary of petitions to the Fatimid state, regardless of the religion of the petitioner.13 By contrast, medieval letters written between Jews in the Islamicate world typically end with wa-l-salām or ve-shalom (“and peace”). Rarely, influenced by Muslim models, they include blessings for the prophet Moses.14 To the best of my knowledge, AIU XII.116 is the first known appearance of the taṣliya for Muḥammad in a letter sent from one Jew to another.15
The presence of the taṣliya led me to wonder if the identifications of the sender and recipient may have been mistaken. Perhaps this letter was originally addressed to a Muslim trader instead of Ibn Yijū, who merely came into its possession after the fact? This is possible, but another recent discovery points to a more persuasive explanation: that Jewish merchants sometimes hired Muslim scribes to write their Arabic correspondence, and those scribes, accustomed to ending letters with blessings for Muḥammad, did not refrain from doing so merely because they were writing on behalf of Jews.
The second recent discovery is T-S Ar.38.88 (Figure 3), an unpublished Arabic letter composed circa 1167 by a Jewish merchant named Abū ʿAlī ibn Abī ʿUmar, who had recently crossed the Red Sea from ʿAydhāb to Aden, conveying news to his family members in Fustat (Old Cairo). The sender and the recipients can be confidently identified based on comparison with a published Arabic letter (T-S Ar.40.56).16 Each letter covers the same subjects in the same order: how the sender lost everything in a raid by the Turkish Ghuzz in Upper Egypt; how he received 30 dinars from Ḥalfon ibn Maḍmūn of Aden; and how Abū ʿImrān, the husband of a young female relative, converted to Islam. However, the letters are clearly written in two different Arabic hands. Goitein suggested that Abū ʿAlī may have written the earlier letter (T-S Ar.40.56) in his own hand, but Friedman has argued that it is more likely written in the hand of a professional Muslim scribe.17 The ending of Abū ʿAlī’s second letter (T-S Ar.38.88) strongly supports Friedman’s argument; between the ḥamdala (blessings for God) and the ḥasbala (“our sufficiency is God, and He is the best trustee”), the scribe has inserted the taṣliya for the prophet Muḥammad.18
Fig. 3: Cambridge University Library, T-S Ar.38.88 verso, with the concluding formulae circled, including the taṣliya. Courtesy of the Syndics of Cambridge University Library.
With the examples above, we now have two instances of letters sent by twelfth-century Jewish traders in Aden to fellow Jews overseas that were most likely written by professional Muslim scribes who incorporated the conventional taṣliya for the prophet Muḥammad. Why does this matter? In the remainder of this Document of the Month I will make the case that these Arabic letters written by Muslim scribes for Jewish traders offer an entry point for understanding the embeddedness of Jewish Indian Ocean traders in their wider interreligious milieu.
The survivorship problem
Until recent decades, the vast majority of scholarship on the Cairo Geniza was based on Hebrew-script fragments, mainly written in the languages of Hebrew, Judeo-Arabic, and Aramaic. This trend in scholarship reflects both the statistical fact that Hebrew-script fragments comprise the great majority of the approximately 400,000 Geniza fragments and the disciplinary division that has historically situated Geniza studies as a subfield of Jewish studies.
Studies in the documentary Geniza—the corpus of approximately 40,000 documents of everyday life, such as letters, legal deeds, and account ledgers—share a similar bias. Given that we do not yet understand the criteria by which the Jews of Fustat deposited some papers but not others in their genizot,19 we are faced with difficult questions of representativeness. Do the Geniza documents allow us to write histories only of the Jews of medieval Egypt, or do the findings apply more broadly? Even within the Jewish communities, how representative are the Geniza fragments of the Jews’ total written production?
Goitein addresses these questions when he asserts in the introduction to the India Book, “Since the Geniza is essentially a repository of papers written with Hebrew characters, it is natural that it should deal mainly with the commercial activities of Jews and between Jews. It seems also that business was conducted to a large extent along denominational lines, simply because this was the practical thing to do.”20 There is a telling slippage here. Goitein first draws attention to the possibility that the wealth of information about intra-Jewish commercial affairs might reflect selective preservation of sources rather than a historical trend. Yet he then resolves the dilemma by implying that Jewish commercial affairs were indeed largely conducted within the Jewish community, a plausible hypothesis but one that strictly Jewish sources cannot prove.
Similar assumptions shape the sole dedicated study of Arabic-script Geniza letters, the 1992 dissertation of Sabih Aodeh, Eleventh Century Arabic Letters of Jewish Merchants from the Cairo Geniza.21 Due to his perception of the intrinsically Jewish nature of the Geniza, Aodeh takes as a (debatable) axiom that almost all of his 72 letters were written by Jews and for Jews.22 And in his discussion of why relatively few Arabic letters survive in the Geniza, Aodeh emphasizes that Jews simply preferred Hebrew script for reasons including the “cursiveness and difficulty” of Arabic script, a desire for secrecy, and perhaps “national identity and love of the Hebrew language”.23 Other scholars, including Goitein himself in other places, and Jessica Goldberg in her monograph on eleventh-century Mediterranean trade, are far more careful to delineate what can and cannot be known about intra-religious versus interreligious commercial dealings on the basis of predominantly Hebrew-script material.24
Script choice
I would like to suggest that the use of Arabic script both in letters amongst Jews and between Jews and non-Jews in the medieval Islamicate world was far more widespread than scholars have tended to believe. The Arabic documents that do survive in the Geniza—the tip of an Arabic iceberg—allow us to trace the outlines of a much larger corpus, now lost. Rather than avoiding Arabic script due to its difficulty, many of the Jews of the community who deposited papers in the Geniza were able to switch between Hebrew script and Arabic script with relative ease.
This suggestion is in part based on the increasing pace of discovery of Geniza documents in Arabic that were written, read, or used by Jews, including the two Arabic letters with blessings for Muḥammad discussed above. In recent years, thanks to the turn in Geniza research toward the neglected Arabic fragments, including an exhaustive cataloguing effort at the Princeton Geniza Lab, it is no longer tenable to consider the documentary Geniza (as opposed to the Geniza as a whole) “essentially a repository of papers written with Hebrew characters”. The Princeton Geniza Project database currently lists 20,037 documents written primarily with Hebrew characters (mainly in Judeo-Arabic, Hebrew, and/or Aramaic) and 8,645 documents written primarily with Arabic characters (mainly in Arabic).25 The Arabic-script documents, far from a negligible fraction, are almost one-third of the total.26
A striking number of these new Arabic finds relate to the India Book. Out of the original corpus of ~430 documents, only ~10 documents (~2 percent) are written in Arabic.27 Out of the ~250 new India Book documents subsequently identified by Mordechai Akiva Friedman, Amir Ashur, and myself, over 20 documents (~8 percent) are in Arabic. Although the number of Arabic India Book documents remains small in absolute terms at some 30 documents altogether, the sample size is now large enough to give a sense of the sorts of Arabic documents that once circulated on Indian Ocean trade routes.
If many Jews were indeed perfectly comfortable with the use of Arabic for correspondence with fellow Jews, can the study of new Arabic documents shed light on the question of interreligious business dealings? I believe the answer is yes. Documents in Judeo-Arabic were written by Jews for Jews; there is little evidence of non-Jews gaining proficiency with the Hebrew script in the medieval period.28 By contrast, documents in Arabic are ambiguous. If a writer literate in Judeo-Arabic chose to use Arabic instead, or hire an Arabic scribe, they intended the resulting letter to be legible to Jews and non-Jews alike. Medieval Islamicate letters have been described as “semi-public, semi-private”; they were rarely sealed texts composed by a single author and read by a single reader.29 Long-distance mercantile letters passed through the hands of multiple bearers, including trusted friends and unwanted snoops. They could be read aloud to assembled listeners, and recipients sometimes needed to collaborate—like modern paleographers—to decipher them.30 Each of the writer’s choices (hire a scribe or write the letter oneself? employ a high or low register? write in Hebrew or Arabic script, or one copy in each?) provides meaningful evidence about the social worlds they navigated.
An interreligious milieu
Another India Book letter illuminates how Arabic served as a lingua franca amongst not only Jewish and Muslim traders but also their Indian associates in the twelfth-century Indian Ocean.31 Bodl. MS heb. b 11/22 is a Judeo-Arabic letter dating to 1145–48, sent by the Jewish shipowner Maḥrūz ibn Yaʿaqov, in Mangalore, to his brother-in-law Abū Zikrī Kohen, at that time in Bharuch, Gujarat. Maḥrūz refers in this letter to his dealings with both Jewish and non-Jewish colleagues, in particular a Muslim named Abū l-Qāsim and an Indian shipowner named Tinbū: “If, my lord, you need any gold, please take it on my account from the shipowner Tinbū, for he is staying in Tāna (i.e., Thane), and between him and me there are bonds of inseparable friendship and brotherhood.” He adds, “Attached to this letter is another one in Arabic characters of the same content as this letter.”32 The Arabic copy, as Roxani Margariti suggests, may have been intended to be presented “to Tinbū and perhaps to other non-Jewish merchants as a means of introduction”.33 The same idea might explain the existence of both Judeo-Arabic and Arabic copies of the letter from Maḍmūn to Ibn Yijū with which we started.
Such examples led Goitein to modify his claim (quoted above) that “business was conducted to a large extent along denominational lines, simply because this was the practical thing to do”. He adds, referring specifically to the Indian Ocean context, “[T]he same Geniza letters reveal an astonishing degree of interdenominational cooperation, matched by almost complete absence of animosity against other communities. Partnerships and other close business relationships between Jews and Muslims, or Hindus, or Christians were commonplace.”34 Indeed, while there is minimal evidence of Jewish ship owning in the Mediterranean in this period, the India Book documents attest to numerous ships in the Indian Ocean owned by Jews, some owned in partnership between Jews and non-Jews. Goitein and Friedman have persuasively suggested that the “multinational ambience” in the Indian Ocean may have fostered such collaboration in the Indian Ocean to a greater extent than in the Mediterranean.35
With the new Arabic addenda to the India Book, we can see clearly that the same traders who produced the Hebrew-script documents also conducted much of their affairs using Arabic documents, including many of their dealings with fellow Jews and, presumably, all of their dealings with non-Jews.36 However, due to the limitations of our primary sources, we still do not know the relative proportions of intra-Jewish versus interreligious business dealings. Did the Hebrew-script letters make up the great majority of letters produced by Jewish merchants? How would our understanding of trading patterns shift if we found out that the Hebrew-script letters comprised less than half of the original total? Or only 10 percent?
Some of these questions will likely remain unanswered, but we now have promising paths forward for future inquiry. Marina Rustow has effectively used the idea of a “phantom archive”—the study of vanished documents which survive as mentions in extant documents—to trace the contours of Fatimid state document production beyond the remnants preserved in the Geniza and elsewhere.37 A similar methodology allows us to explore the phantom archive of Arabic mercantile documents that once circulated amongst Jewish and non-Jewish traders and which seem to have been deposited in the Geniza at much lower rates than Hebrew-script letters. In some cases, as with Maḥrūz’s mention of an Arabic letter above, the documents explicitly refer to accompanying Arabic documents.38 In other cases, we can deduce that a Judeo-Arabic text is in fact a copy of a lost Arabic original, or vice versa.39 Lastly, we can take a more capacious approach to the phantom archive by considering that each Arabic document that did find its way into the Geniza points to the existence of a multitude of similar documents that did not.
Concluding remarks
Let’s return to the two letters containing the taṣliya for Muḥammad, sent by Jews to fellow Jews but written by Muslim scribes. These co-produced letters challenge the essentialist notion that documents, like people, could be “Jewish” or “Muslim”. They also raise important questions about the relationships between the Jewish senders and the Muslim scribes: Who were these scribes, and what were their terms of employment with the Jewish merchants for whom they wrote? How extensively did scribes shape the content of letters with their own formulaic repertoires? The fact that Jewish traders seem not to have objected to blessings for Muḥammad in letters sent in their names points to a social world in which the boundaries between Jewish and Muslim mercantile networks were more permeable than the Hebrew-script Geniza documents alone could have revealed.40 In this light, our initial surprise at finding the taṣliya in “Jewish” trade letters says more about modern assumptions than about medieval realities.
Notes
1 On the taṣliya, see Annemarie Schimmel, And Muhammad Is His Messenger: The Veneration of the Prophet in Islamic Piety (Chapel Hill, NC, 1985), 92.
2 Amir Ashur, Alan Elbaum, Pratima Gopalakrishnan, and Marina Rustow, India Traders of the Middle Ages, vols. V–VII (in progress).
3 Marina Rustow, “Goitein, the Indian Ocean, and a Missed Opportunity to Make Medieval History Global”, Speculum 101, no. 1 (2026).
4 Shelomo Dov Goitein and Mordechai Akiva Friedman, India Book, vols. I–IV (in Hebrew) (Ben-Zvi Institute, 2009–2013). Shelomo Dov Goitein and Mordechai Akiva Friedman, India Traders of the Middle Ages: Documents from the Cairo Geniza (‘India Book’) (Leiden, 2008).
5 For the purposes of this article, “Judeo-Arabic” refers to Arabic language written in Hebrew script (rather than any specifically Jewish dialect of Middle Arabic).
6 ENA NS 48.2 verso (India Book II, 27). Ed. Goitein and Friedman, India Traders, 357. On Maḍmūn ibn Ḥasan, see ibid., 37–47. On Avraham ibn Yijū, see ibid., 52–89; Elizabeth Lambourn, Abraham’s Luggage: A Social Life of Things in the Medieval Indian Ocean World (Cambridge, UK, 2018); and Amitav Ghosh, In an Antique Land: History in the Guise of a Traveler’s Tale (New York, 1994).
7 ENA NS 48.2 recto (India Book III, 28). Ed. Goitein and Friedman, India Traders, 672. Ibn Yijū wrote these accounts, for the sale of Indian goods on behalf of a merchant in India, during his stay in Aden in 1141–44. Ibn Yijū often wrote accounts on the backs of letters in his archive, including letters he had received years prior; compare T-S 8J7.23 (India Book III, 2 and III, 28a) and T-S 8.19 (India Book III, 14 and III, 25).
8 Lines 16 and 19 in Figure 2.
9 There are a few scattered characters preserved on verso of the Arabic letter, which may have been part of an address, now almost completely effaced. It is also worth noting that the Arabic fragment is larger and contains more text than the Judeo-Arabic fragment.
10 Goitein and Friedman, India Traders, 9. Four (or possibly five) copies of one letter: ibid., 337 (India Book II, 21–24).
11 Translations of the Judeo-Arabic text are from Goitein and Friedman, India Traders, 358. The additional Arabic text reads والسلم الجزيل عليه وعلى خاصته والحمد لله وحده وصلواته على رسوله سيدنا محمد النبي واله الطاهرين
12 Private letters: T-S NS 163.73, T-S NS 150.127, T-S AS 179.33, ENA NS 50.19, ENA 2738.34. State documents: Geoffrey Khan, Arabic Legal and Administrative Documents in the Cambridge Genizah Collections (Cambridge, UK, 1993), 317.
13 Petition from a Jewish man to the caliph al-Āmir, ending with the taṣliya: T-S Ar.30.273 (ibid., 334).
14 T-S Ar.41.53.
15 Muslim-influenced prayer formulae do appear in Geniza fragments other than letters, including T-S NS 110.26 and Moss. IV,56.1, published in Shelomo Dov Goitein, “Prayers from the Geniza for Fatimid Caliphs, the Head of Jerusalem Yeshiva, the Jewish Community and the Local Congregation”, in Studies in Judaica, Karaitica and Islamica: Presented to Leon Nemoy on His Eightieth Birthday, edited by Jonas C. Greenfield, Shaul Brunswick, and William G. Braude (Ramat-Gan, 1982), 47–57.
16 India Book II, 64, ed. Goitein and Friedman, India Traders, 496.
17 Ibid.
18 The concluding blessings seem to read والحمد لله وحده وصلواته على رسوله سيدنا محمد وسلم تسليما وحسبي الله ونعم الوكيل. The words على رسوله سيدنا محمد are quite cursive and difficult to make out, but the context makes it clear that this must be the taṣliya. On the scribal habit of writing the most formulaic phrases with the least care, see Marina Rustow, The Lost Archive: Traces of a Caliphate in a Cairo Synagogue (Princeton, 2020), 312–13. On the ḥasbala (Qurʾān 3:173), see Daniel Potthast, “Qurʾān Quotations in Arabic Papyrus Letters from the 7th to the 10th Centuries”, in Qurʾān Quotations Preserved on Papyrus Documents, 7th–10th Centuries, edited by Andreas Kaplony and Michael Marx (Leiden, 2019), 42–85.
19 Genizot is the Hebrew plural of geniza. On the difficult questions of provenance, and for a more nuanced picture than the traditional narrative that all of the “Cairo Geniza” fragments were found in a single geniza, see Rebecca Jefferson, The Cairo Genizah and the Age of Discovery in Egypt: The History and Provenance of a Jewish Archive (London, 2022).
20 Goitein and Friedman, India Traders, 25.
21 Sabih Aodeh, Eleventh Century Arabic Letters of Jewish Merchants from the Cairo Geniza (PhD diss., Tel Aviv University, 1992). Almost all of Aodeh’s letters are from the Egyptian and Mediterranean contexts. An exception is T-S NS 327.133 (no. 37), an Indian Ocean letter addressed to Avraham ibn Yijū, misidentified by Aodeh as a letter to the eleventh-century Mediterranean merchant Nahray ibn Nissim.
22 Ibid., i. Aodeh notes that T-S Ar.40.126 (no. 69) is addressed to a Muslim qāḍī.
23 Ibid., 1. All of these claims have some supporting evidence, but it is a fallacy to extrapolate that Jews as a rule preferred the Hebrew script in their commercial correspondence, or that they were not as proficient as Muslims in the use of Arabic.
24 Jessica Goldberg, Trade and Institutions in the Medieval Mediterranean: The Geniza Merchants and Their Business World (Cambridge, UK, 2012), 10, 24–28, 141, 354–55.
25 These numbers are current as of 3 May 2026. This calculation is inevitably somewhat imprecise. Approximately 10,000 documents do not have a primary language labeled in the metadata, and many documents list both Arabic and a Hebrew-script language as primary languages. The “literary” PGP records are excluded. A small number of Judeo-Persian and Persian documents are included.
26 Another caveat: the Arabic fragments tend to be smaller and preserve fewer words than the Hebrew-script fragments, so this may somewhat overestimate the proportion of Arabic text.
27 These figures are calculated from the Princeton Geniza Lab’s internal India Book database. The index to Goitein and Friedman, India Traders lists ~430 documents belonging to volumes I–VII as of 2008.
28 With the possible exception of government inspectors; see Goitein and Friedman, India Traders, 465n12.
29 Goldberg, Trade and Institutions, 63–67 and 83–84. On the issue of “sealing” letters, see Oded Zinger, “The Friedberg Revolution in the Study of the Genizah: The Material and Physical Aspects of Documents”, Ginzei Qedem 21 (2025), 107–16.
30 Preoccupation with unwanted readers: T-S 13J24.8. Assistance with reading: T-S 8J16.19 + T-S NS 323.13.
31 Roxani Margariti, Aden & the Indian Ocean Trade: 150 Years in the Life of a Medieval Arabian Port (Chapel Hill, 2007), 158. On the Arab dominance of western Indian coastal trade in this period, see V. K. Jain, Trade and Traders in Western India (AD 1000–1300) (New Delhi, 1990), 81–83.
32 Goitein and Friedman, India Traders, 477–78.
33 Margariti, Aden, 158.
34 Goitein and Friedman, India Traders, 25.
35 Ibid., 133–37. The greater distances and risks involved in Indian Ocean trade may have also played a role.
36 A possible exception: there is one Geniza fragment (T-S AS 159.248) written in Devanagari script in an as-yet unidentified language, dating unknown.
37 Rustow, Lost Archive, 90–95.
38 See also T-S NS J44 (“Together with this letter I have sent you a copy […] of an account in Arabic and Hebrew”).
39 Arabic copy of a Judeo-Arabic original: p. Heid. Hebr. 12 (India Book II, 50). Judeo-Arabic copy of an Arabic original: T-S K15.98 (India Book VI, 16).
40 There is an important methodological implication: many of the letters listed in footnote 12 above are identified as letters between Muslims simply by virtue of the presence of the taṣliya. If the taṣliya does not identify the religious identity of sender or addressee, but only that of the scribe, then prior assumptions need to be revisited. The same applies to T-S Ar.35.14, another India Book addendum in Arabic from the circle of Maḍmūn ibn Ḥasan.
