This article presents sample results of my ongoing research project on the book of Samuel in the Cairo Genizah. Each of the 650+ manuscripts of Samuel found amongst the documents forthcoming from the Cairo Genizah has been read and analyzed, with the goal of identifying textual variants. Special attention is paid to those variant readings which find parallels in the earlier witnesses (Qumran, Septuagint, Peshitta, Targum, Vulgate).
For the past several years I have undertaken the task of reading every manuscript of the book of Samuel found in the Cairo Genizah. This is a pilot project, for to the best of my knowledge no such parallel work has previously been accomplished.1 We have had the documents in our possession for more than 120 years; it took a long time to produce a reliable catalogue of the exceedingly large number of Cambridge documents;2 and of course a complete registry of all of the documents (regardless of where they are housed) is possible today only via the Friedberg Genizah Project (FGP) module of the Friedberg Jewish Manuscript Society (FJMS) portal available on the internet.
I conducted the first part of this research at the Taylor-Schechter Genizah Research Unit in the Cambridge University Library in summer 2015, reading the documents in situ;3 and then conducted the second part of this investigation assisted by my able student and research assistant Jenna Kershenbaum during academic year 2016–2017, via online access to the documents in other collections at
In addition, Ms. Kershenbaum and I personally inspected the Samuel manuscripts in the collection of the Library of the Jewish Theological Seminary (currently housed temporarily at nearby Firestone Library of Princeton University, while the J.T.S. library undergoes major renovation) in March 2017. Finally, I was able to personally examine many of the Bodleian Library manuscripts during my visit to Oxford in July 2017.
These documents run the gamut. Many contain the full masorah and are written in three columns, as is customary in medieval codices. Others lack the full masorah, but are written in a beautiful penmanship, implying the work of a professional scribe. Some suggest a copyist less skilled in the scribal tradition, and therefore may have less probative value, but even these occasionally contain interesting readings. Most use the Tiberian niqqud, but a few important manuscripts employ the Babylonian system.5 Regardless, I can aver that I have read them all.6
One might object to this approach, with the argument that documents such as the non-Masoretic manuscripts and those copied by less-than-professional scribes have little value, with nothing to offer to the field of biblical textual criticism. By incorporating all possible witnesses in my investigation, however, I have followed the lead of two illustrious predecessors, namely, Benjamin Kennicott and Giovanni de Rossi, who collated hundreds of mostly (or entirely) later manuscripts and early printed editions during the eighteenth century.7 They too recorded every possible variant that they could identify. To be sure, Kennicott for one faced some criticism in this regard, though he laboured on nonetheless.8
A number of colleagues, to whom I have mentioned this project, have asked me about the results of my research, such as: Did you find anything interesting? Are there any significant variants?
The answer is: by and large, and to no surprise, the Cairo Genizah manuscripts adhere quite closely to the major medieval codices: Aleppo (A) and St. Petersburg (L).9 Naturally, I could have predicted this before I commenced the research,10 so the real issue is: when they do differ, what kind of differences are to be found?11
While my notebooks and databases are complete, a comprehensive publication of the results of my investigation would take several years to produce and most likely would entail a very detailed monograph. Accordingly, I have elected to write this short article: a) to apprise interested parties of this research; and b) to provide some examples of the variants identified. The illustrations which follow are selected mostly at random, though hopefully they will provide the reader with a general sense of the research accomplished.
The largest number of differences are, again to no surprise, in the spelling of words (plene vs. defectiva) and in the marking of paragraphs (setumot and petuḥot). I have recorded the latter especially, and perhaps one day I will compile and publish a comprehensive list.
For the nonce, here are two examples of each. First, differences in spelling, with one example in each direction (that is, plene vs. defectiva, defectiva vs. plene):
- Bodleian MS heb. d.49.4, fol. 1r, line 17, reads בָשליִשיִת ‘a third time’ at 1 Sam 3:8, doubly plene,12 whereas A and L have only one yod as mater lectionis, albeit in different places: A—בַּשְּׁלִישִׁת | L—בַּשְּׁלִשִׁית.
- JTS L558 = ENA 2640, fol. 14v, col. 2, line 20, reads וַיְשַׁלְּחֻם ‘and they sent them’ at 1 Sam 6:6, in defectiva manner,13 whereas A and L represent the long vowel at the end of the word in plene orthography, וַיְשַׁלְּחוּם.
The following two examples provide for differences in paragraphing, again, with one example in each direction (that is, presence in A and L, with absence in a Genizah document; absence in A and L, with presence in a Genizah document):
- Both A and L have a setuma paragraph break after 2 Sam 19:8,14 whereas Strasbourg MS 4084.6 1r, col. 1, line 14, has none.15
- Neither A nor L has a paragraph break after 2 Sam 15:29,16 whereas JTS MS L594, fol. 12, 1v, col. 1, line 10, includes a setuma break here.17
Having spent several years reading the Genizah manuscripts, I may observe here, albeit as an impression only, that when A and L differ in the paragraph breaks,18 more often than not, the Genizah documents follow A. To repeat, this is only an impressionistic observation, which still requires statistical analysis in support. Nevertheless, I would suggest that such a finding should not be surprising, given the fact that the Aleppo Codex resided in Cairo during the heyday of scribal activity which yielded the documents forthcoming from the storeroom of the Ben Ezra Synagogue.
Thus, for example, note the following:
- JTS MS L594, fol. 13, 1r, col. 1, line 2,19 includes a setuma break after 2 Sam 3:8,20 as in A, though it is lacking in L.
- RNL (St. Petersburg) III B 50 1b, left page, line 6, includes a setuma break after 1 Sam 14:43,21 again as in A,22 though once more it is lacking in L.
Occasionally one encounters a different vowel, sometimes in surprising fashion:
- OIM (Chicago) A11241 1r, col. 1, line 4, reads מְיהוָה (with shewa) at 1 Sam 26:11,23 instead of the expected form מֵיהוָה, as in both A and L (and presumably in all other witnesses).
- Bodleian MS heb. b.8.6, fol. 1v, col. 1, line 26 (bottom line), reads אֶ֑רֶץ at 1 Sam 2:10,24 even though the form is in pause, as marked by ʾatnaḥ, instead of the expected pausal form אָ֑רֶץ, as appears in both A and L.
But the real goal of my research was directed towards this question: To what extent do any of the variants attested amongst the Qumran manuscripts of Samuel and/or the presumed Vorlagen of the Septuagint, Vulgate, Targum, and Peshitta persist into the Middle Ages?25 While I did not find many such cases, especially due to the very fragmentary nature of the Qumran Samuel material, I am able to report here several notable readings found in the Cairo Genizah materials.
- T-S A8.3, fol. 1v, col. 3, line 8, reads מִשַּׂעֲרַוֹת בְּנֵךְ ‘of the hairs of your son’ (with the key noun in the plural) at 2 Sam 14:11,26 although both A and L read מִשַּׂעֲרַת בְּנֵךְ ‘of the hair of your son’ (with the key noun in the singular). To our good fortune, the end of the key noun is attested in 4QSamc, as ות[—hence we may reconstruct משער]ות [בנך] in this Qumran manuscript. This reading also underlies the Lucianic recension of the Septuagint (ἀπὸ τῶν τριχῶν)27 and the Vulgate of St Jerome (de capillis),28 so that we can affirm it as a valid variant amongst the ancient copies and versions.
- The main text of CUL T-S A8.1, fol. 1v, col. 1, line 5, reads וַתֹּאמֶר אֵלָיו ‘and she said to him’ at 2 Sam 20:17,29 as opposed to the reading of both A and L, וַתֹּאמֶר הָאִשָּׁה ‘and the woman said.’ A later hand crossed out the word אֵלָיו ‘to him’ and replaced it with הָאִשָּׁה ‘the woman’ in the margin. In this case we lack Qumran testimony (4QSama skips from a bit of v. 14 to a single letter of v. 19), but interestingly the original reading of the aforecited Genizah document underlies the Peshitta, which reads ܘܐܡܪܐ ܠܗ ‘and she said to him.’
- The opening words of Hannah’s prayer in 1 Sam 2:1 appear both in A and in L as follows: עָלַ֤ץ לִבִּי֙ בַּֽיהוָ֔ה רָ֥מָה קַרְנִ֖י בַּֽיהוָ֑ה—that is, with the divine name Yhwh in both stichs. The reading of 4QSama agrees, as far as we can determine: [עלץ לבי ביהוה] רמ̇ה קרני בי[הו]ה. The Septuagint, however, reads κυρίῳ ‘Lord’ in the first stich, but θεῷ μου ‘my God’ in the second stich, suggesting a text with Hebrew אלהי in the Vorlage. Such a text exists in a single Cairo Genizah manuscript: AIU (Paris) MS I.A.27, fol. 1v, line 8,30 which reads בֵּאלֹהַי ‘my God’ in 1 Sam 2:1, the second stich.31
- Strasbourg MS 4084.7, fol. 1v, col. 1, line 6, reads הִנֵּה ‘behold’ at 2 Sam 19:38,32 as opposed to both A and L, which read וְהִנֵּה ‘and behold.’ The former reading is paralleled in the Peshitta, which reads ܗܐ ‘behold,’ and in one important manuscript of Tg. Jonathan, namely BL Ms. Or. 1471, which reads הָא; ‘behold,’33 both without the conjunction ‘and.’ Unfortunately, as so often happens in such cases, 4QSama fails us in this instance, for the relevant fragment (frg. 142) yields only ]ה עבד[—with only the final letter of either הנה or והנה present.
Some differences between the Genizah manuscripts and the major codices A and L would not show up in the versions, because they are simple grammatical variants. They would, of course, be reflected in the Dead Sea Scrolls, though alas in the following instances we lack the Qumran testimony.
- Bodleian MS heb. d.49.6, fol. 1v, line 23 (the next-to-last on the page), reads אחריֵ שאול ואחריֵ שמוֻאל ‘after Saul and after Samuel’ (with the same form of the preposition) at 1 Sam 11:7,34 whereas both A and L read אַחֲרֵי שָׁאוּל וְאַחַר שְׁמוּאֵל (with different forms of the prepositions). Most frustrating is the fact that 4QSama preserves only ]שאול וא[, so that we cannot know the form of the second preposition.
- Strasbourg MS 4084.6, fol. 1v, col. 1, line 11, reads וְצָלְחֻו אֵת הַיַרְדֵן ‘and they rushed down to the Jordan’ at 2 Sam 19:18,35 whereas A and L read וְצָלְחוּ הַיַּרְדֵּן. The former text introduces the nota accusativi את, which admittedly is unparalleled, for nowhere else in the Bible does the verb צ-ל-ח (Qal) ‘rush’ govern the particle את. As for the Qumran evidence, 4QSama skips from a few letters in v. 16 to a few letters in v. 25.
- In similar fashion, CUL T-S A8.14, fol. 3r, col. 3, line 15,36 includes an אֶת within the phrase סֹבּוּ וְהָמִיתוּ אֶת כֹּהֲנֵי יְהוָה ‘turn and kill the priests of Yhwh’ in 1 Sam 22:17.37 In both A and L, this particle is not present, as both manuscripts read סֹבּוּ וְהָמִיתוּ כֹּהֲנֵי יְהוָה. The Qumran evidence is lacking, for 4QSama has a 19-verse gap here, with nothing remaining from 1 Sam 22:12–23:7.
- Returning to Strasbourg MS 4084.6, fol. 1v, col. 2, line 26 (bottom line), reads וַיהִי כַאְשַר בָא ‘and it was, when he came’ in 2 Sam 19:26,38 whereas the major codices A and L read וַיְהִי כִּי בָא with essentially the same meaning. 4QSama includes bits of vv. 25 and 27, but nothing from v. 26.
Regarding the above eight examples (nos. 9–16), and others like them not reported here, we turn to the question adumbrated above (see n. 25). When the same variant is reflected in an ancient version and again in a medieval manuscript, are we to reconstruct a genetic relationship between the two, even though the matching testimonies are separated by, say, one thousand years; or are we to assume independent processes at work, with different scribes generating the same reading at different points during the long history of textual transmission? Naturally, I considered this crucial question again and again while collating the Cairo Genizah manuscripts; in addition, this issue was covered by Goshen-Gottstein (see above, n. 10); and then Dr. Kim Phillips (Cambridge) raised the question with me more recently (via email June 2017). I conclude, accordingly, with some general thoughts on the matter.
On the one hand, to be perfectly frank, I see no real way of deciding this issue, and thus I am content simply to record the variants. On the other hand, I would say the following, from the stance of statistical probability. Given the size of the corpus (that is, the book of Samuel), and given the enormous number of potential variants (that is, how many minor issues are patient of modification), when a particular reading occurs at Qumran and/or in the presumed Vorlage of an ancient version, and then again in a Genizah manuscript, my working hypothesis is to assume that the variant is due to the constancy of a scribal tradition, rather than one caused by scribes working independently at great remove from each other.
As such, I am more in agreement with the conclusion reached by John Wevers, based upon his thorough comparison of readings from Kennicott and de Rossi and those reflected in the Old Greek and its recensions, “the [medieval] Hebrew variants have perpetuated pre-Masoretic traditions,”39 than I am with the approach promoted by Goshen-Gottstein, who argued strongly against “a genetic relationship between a MS from the Massoretic period and a pre-medieval source,” even when the readings agree.40
In sum, many more examples of textual variants identified during my research could be described, and I hope to return to this project in the future with further publications, including, perhaps, a full catalogue of differences.41 For the present moment, however, my goal has been more modest, to wit, and as indicated above: to provide a sampling of the variants found amongst the Samuel manuscripts which were discovered within the treasure trove of documents housed in the storeroom of the Ben Ezra Synagogue, known to all as the Cairo Genizah.
Appendix: Kennicott and de Rossi
It is only natural to compare the results of my work, even in its initial stages and as reflected in this interim report, with the prodigious collations accomplished by Benjamin Kennicott (1718–1783) in Oxford and his spiritual disciple Giovanni Bernardo de Rossi (1742–1831) in Parma.42 To be sure, as indicated above, the manuscripts and early editions collated by these two scholars are much later than the Cairo Genizah material which I have studied, but the data should be compared nonetheless.
Herewith, accordingly, a chart to guide the reader (see below), with the following conventions:
Not expected to be found, because: a) de Rossi did not include spelling differences (plene vs. defectiva); and b) neither Kennicott nor de Rossi collated paragraph breaks.
No examples found in either Kennicott or de Rossi; the page number in smaller font indicates the expected page where one would expect to find notations, should said variant have been identified.
The page number in either Kennicott, vol. 1, or de Rossi, vol. 2, where the same variant identified in my research is recorded by these earlier scholars.
As the reader may judge, many of the variants found in the Genizah manuscripts of Samuel are still found centuries later in younger manuscripts and early printed editions. On the other hand, and quite significantly, some are not. The reader also must keep in mind that I selected my sixteen examples mostly at random, and certainly before I turned my attention to Kennicott and de Rossi. Noteworthy are nos. 9, 10, and 12, in which cases Genizah manuscripts retain readings reflected in the ancient versions (Qumran, Vulgate, Targum, Peshitta, Lucianic recension of the LXX), though half a millennium or more later the two great European savants registered no such variants.
Kennicott, vol. 1
de Rossi, vol. 2
1 Sam 3:8
1 Sam 6:6
2 Sam 19:8
2 Sam 15:29
2 Sam 3:8
1 Sam 14:43
1 Sam 26:11
1 Sam 2:10
2 Sam 14:11
xxx (pp. 578–579)
xxx (p. 178)
2 Sam 20:17
xxx (p. 590)
xxx (p. 189)
1 Sam 2:1
2 Sam 19:38
xxx (p. 588)
xxx (p. 188)
1 Sam 11:7
2 Sam 19:18
xxx (p. 186)
1 Sam 22:17
xxx (p. 156)
2 Sam 19:26
xxx (p. 187)
In this particular instance, Kennicott (p. 555) noted one manuscript which reads יהוה and not מיהוה. If this is not a simple mistake, it would turn the divine name into a vocative, “O Yhwh,” as occurs in 2 Sam 23:17. Otherwise, one expects the preposition מ- after the word חלילה with Yhwh or Elohim following, as in 1 Sam 24:7, 1 Kgs 21:3, 1 Chr 11:19.