An Ancient Scroll n May 2007, on the eve of the Jewish festival of Shavuot (Pentecost), a rare fragment of of the Book of Exodus: Ian ancient Torah scroll was placed on display The Reunion of Two at the Shrine of the Book in The Israel Museum, Jerusalem. Known as MS Ashkar- Separate Fragments Gilson No. 2 (hereafter MS Ashkar; figs. 2a–c, 4), the manuscript contains a passage from the book of Exodus (13:19–16:1) that , includes the “Song of the Sea.” When visiting Edna Engel Hebrew Palaeography the exhibit, Dr. Mordechay Mishor, a member Project, Israel Academy of Sciences of the Academy of the Hebrew Language, was and Humanities struck by the similarity between MS Ashkar and another manuscript, formerly known as Mordechay Mishor, Academy of the MS London, Jews’ College 31 (hereafter MS Hebrew Language London; figs. 1, 3), also a fragment of a Torah scroll containing a text from Exodus (9:18– 13:2). Suggesting that the two might be parts of the same original scroll, Mishor turned to Dr. Edna Engel of the Hebrew Palaeography Project at the Israel Academy of Sciences and Humanities for paleographic verification. Engel’s preliminary paleographic examination revealed a perfect match between the manuscripts; the two researchers decided to collaborate on a detailed study, and with the enthusiastic support of Dr. Adolfo Roitman, Director of the Shrine of the Book, they began reconstructing the scroll’s remarkable story. Engel and Mishor set out to establish contact with the custodians and owners of the two manuscripts, beginning with Duke University’s Manuscript Department, which gave them permission to work with facsimiles of MS Ashkar and to publish the results of their study. The current owner of MS London was located as well; he, too, readily consented to cooperate on the challenge of reuniting the two manuscripts. The result was the joining of two ancient gvil1 fragments from opposite ends of the world and the reconstruction of a section of one of the oldest surviving Torah scrolls.2 Fig. 1 Until the late 1970s, MS Ashkar was Detail of MS London, 7th–8th century, part of the Hebrew manuscript collection of gvil. Courtesy of the Loewentheil Family, New York Dr. Fuad Ashkar, a Lebanese-born American
24 physician. Dr. Ashkar was unaware of the script.10 They are first-hand evidence of the importance of the partially blackened sheet religious and cultural traditions of the Jews of gvil in his possession until he made contact living in the late Second Temple Period with Professor James H. Charlesworth (first century BCE – first century CE) and of Duke University,3 who commissioned the time of the Bar Kokhba Revolt (early Carbon‑14 tests to confirm his suspicion second century CE). Indeed, there are dated that the manuscript was written in the Jewish manuscripts, including splendid seventh or eighth century.4 The document biblical codices, from the tenth century on, was placed in Duke University’s Rare Book, but almost no dated or datable texts have Manuscript, and Special Collections Library survived from the intervening eight hundred for safekeeping. In 2004, Charlesworth years. Consequently, any written material apprised Roitman of the existence of the that can be assigned to that “silent era” is of fragment and consented to having it brought inestimable value. to the Israel Museum for examination and The combined London-Ashkar man- display. After long months of conservation uscript is one such scroll. Although it is and restoration by Michael Maggen, Head undated, Carbon‑14 testing and paleographic of the Paper Conservation Laboratory at the analysis leave no doubt that the scroll was Israel Museum, it was placed on exhibition in written in those interim years, specifically, in the Shrine of the Book.5 the seventh or eighth century; closer exami- MS London was formerly preserved in nation suggests an Egyptian provenance.11 the library of Jews’ College, London,6 but it is The significance of the reunited not known how and when it wound up there. manuscript lies not only in the fact that it The librarian, Ms. R. P. Lehmann, asked comes from the above-mentioned “silent era,” paleographer S. A. Birnbaum to appraise but that after the first- and second-century the gvil fragment, and after a thorough scrolls from the Judean Desert, the London- examination Birnbaum published an extensive Ashkar scroll is the earliest biblical text description of the manuscript, which he containing significant extant sections. There assigned to the eighth century.7 Rare books are reasonable grounds for assuming that it and manuscripts dealer Stephan Loewentheil, was read regularly in a synagogue,12 but – as proprietor of the 19th Century Rare Book and we shall see below – there are arguments for Photograph Shop, bought the manuscript at a and against this conjecture. public auction and still owns it.8 The scroll was a well-known literary In what follows, we maintain – based format in ancient Jewish tradition. Much has on textual and paleographic evidence – that been written about its antiquity as well as the the two fragments are part of a single scroll shift from scroll to codex, the forerunner of (hereafter MS London-Ashkar; fig. 5).9 the modern bound volume. This transition took place among non-Jewish cultures in the I. Introduction early centuries of the Common Era, but was delayed somewhat in the Jewish world. In The Dead Sea Scrolls found in the Qumran fact, we find no reference in Jewish literature caves and the documents and letters from to the writing of codices before the end of the caves of Wadi Murabba‘at and Nahal the eighth century.13 Ancient scrolls were Hever in the Judean Desert are the oldest written on gvil – pieces of animal skin that datable manuscripts written in the “Jewish” are thicker than parchment and written on
IMSA 7 2015: 24–61 25 26 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments Figs. 2a–b MS Ashkar before and after conservation, 7th–8th century, gvil. Duke University Rare Book, Manuscript, and Special Collections Library, Durham, North Carolina
IMSA 7 2015: 24–61 27 28 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments only the hairy side.14 Jewish religious sources Column length reveal a preference for the use of gvil for Each column in MS London-Ashkar has Torah scrolls, but do not prohibit the use of 42 lines. According to Tractate Soferim parchment instead.15 2:11, there may be 42, 60, 72, or 98 lines Although the MS London-Ashkar per column.18 The column length of MS scroll is written on gvil, it does not necessarily London-Ashkar affords the Song of the Sea mean that it was used in the synagogue. The section an elegant symmetry (five lines at the scroll format continued to be used for various top of the column, a blank line, the 30 lines kinds of texts until the early Middle Ages, of the body of the song, another blank line, though it remains obligatory for liturgical and five more lines below that; (figs. 2b, 4). use in the synagogue until today. The scribal tradition of Torah scrolls supports the idea Ruling technique that MS London-Ashkar was intended for MS London-Ashkar complies with the liturgical use. Like the Torah scroll, this halakhic guidelines – horizontal rules mark manuscript was written in an elegant hand the lines and vertical rules indicate the by a skilled professional scribe who complied margins. Both are aligned with “pricking with a uniform tradition of complex rules and marks” on either side of the sheet. According detailed textual prescriptions that has been to Tractate Soferim I 1:15, “If he wrote on a observed meticulously in every region of sheet that was not ruled, the scroll is invalid.” the Jewish Diaspora.16 MS London-Ashkar adheres to this tradition and is characterized Section breaks (parashiyot) by the following main features in compliance MS London-Ashkar closely adheres to the with rabbinic literature. original Tiberian tradition. On the different types of section breaks, see Talmud Bavli Number of columns Shabbat 103b: “An open section may not be There are seven columns in MS London and written closed, nor a closed section open.”19 at least four in MS Ashkar. According to the Talmud Bavli Menahot 30a: “A man should Format use sheets [of gvil or parchment] that contain The format of the Song of the Sea and the between three and eight columns; he should not verses before and after it complies with use one that contains fewer or more columns.” the guidelines in Tractate Soferim and the Masorah. Line length In MS London-Ashkar, the number of letters Text per line ranges from 16 to 27.17 The Talmudic The text of the scroll is compatible with passage quoted above states that the width of a the detailed orthography prescribed by the Masoretic guidelines. There are six columns למשפחתיכם column should not exceed “such as three times” (i.e., 27 letters in biblical spelling, in the Torah that must begin with a speci- as in Exodus 12:21 and Numbers 33:54, or, fied word.20 One such instance occurs in who“) הבאים according to Maimonides, 30 letters in plene MS Ashkar, with the word .(see Mishneh Torah, were coming”) in Exodus 14:28 (figs. 2b, 4 ;למשפחותיכם :spelling Hilkhot Sefer Torah 7:4). According to ‘Adat Devorim, all other columns in a Torah scroll must begin with Fig. 2c the letter vav (a tradition referred to as vavei Detail of MS Ashkar
IMSA 7 2015: 24–61 29 ha‘ammudim [a word-play on “the vavim of scribe of an ancient Torah scroll filled out a the columns” and “the hooks of the pillars” line with dots.24 Accordingly, she maintained in the Tabernacle]). MS London-Ashkar that this use of symbols raises the question does not follow this rule; to the best of our of what is deemed to be an addition to the knowledge, there is no evidence that this consonantal text, which alone is permitted in custom was observed before the thirteenth a liturgical scroll.25 She concluded that not century either in Torah scrolls or in codices. everything considered an inviolable rule today Even after the tradition became widespread, was considered as such in the early Middle it was regarded in the Middle Ages as exces- Ages; and we should not infer that those sive elegance, to which some authorities were rules, which were established in a later period, strongly opposed. were known to the scribes of ancient scrolls. Sirat’s conclusion suggests, therefore, that Left justification the technical means employed by the scribe According to Talmud Bavli Menahot 30a–b, “if of MS London-Ashkar to justify his lines do [at the end of a line] he has [to write] a word not preclude the possibility that the scroll was of five letters he must not write two letters in considered fit for use in the synagogue. the column and three outside [in the margin], In either case, even if MS London- but three in the column and two outside.”21 Ashkar had not been part of a scroll The scribe of MS London-Ashkar was careful for liturgical use, its uniqueness and to left-justify his lines and employed several importance remain undiminished. Its standard methods of medieval Hebrew scribes: highly precise dating and presumed place (1) He used a variety of graphic forms to fill out of writing make it an extremely important the line – one or several short wavy lines (e.g., link to the “silent era.” MS London 10:13), a circle, or several small circles (e.g., MS London 10:14). Circles drawn * * * in the middle of the line (e.g., MS London may have served The second verse of the Song of the Sea (את and נשא between ,10:13 the same purpose. (2) The scribe sometimes in MS Ashkar includes the line, “This is ;ואנוהו) ”writes the last letter, or letter and a half, in the my God and I will adorn Him margin, leaving space between this letter and Exodus 15:2) – “adorn” being used here the end of the line. (3) He stretches the last metaphorically as “glorify.” However, the letter in the line horizontally so that its roof sages understood “adorn” literally to support reaches the left margin.22 the idea of attaining an aesthetic ideal in the Given the fact that the scribe of MS observance of the precepts: “‘I will adorn London-Ashkar scrupulously complied with Him’: Adorn yourself before Him in the the other laws for writing a Torah scroll, his fulfillment of precepts. Make in His honor methods for observing the last one – keeping … a beautiful Torah scroll, and write it with the left margin straight – are somewhat fine ink, with a fine reed pen, by a skilled unexpected.23 Indeed, even though nothing penman” (Talmud Bavli Shabbat 133b). in the laws of writing Torah scrolls relates to Looking at the clean and elegant filling out the line using graphic elements, it design of the Torah scroll of which these would seem to be invalid under the guidelines two manuscripts are a part, we cannot help in force today. The acceptability of this method but marvel at the work of that ancient skilled was discussed by Sirat, who noticed that the penman who labored painstakingly to be
30 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments faithful to the significance that the sages first surviving column of MS Ashkar (from attached to this verse of the Song of the Sea. the end of 13:19 to the middle of 14:13) contains 244 words (quite coincidentally, II. The Textual Connection between the average figure for the columns of MS MS London and MS Ashkar London). Between the end of the text of MS Mordechay Mishor London (Exodus 13:2) and the beginning of the damaged column at the beginning of MS MS London and MS Ashkar: Two Ashkar (Exodus 13:19) there are 263 words Consecutive Sheets missing. This number clearly corresponds to one column (for comparison’s sake, columns MS London is a complete gvil sheet contain- two and three of MS London have 261 and ing seven columns of biblical text, from the 266 words, respectively). Thus, the first middle of Exodus 9:18 until one word before column of the sheet on which MS Ashkar is written is the direct continuation of MS .(באדם ובבהמה לי ]הוא[) the end of Exodus 13:2 In contrast, MS Ashkar is damaged at both London (fig. 5). ends; only four columns have survived, and With this data we can reconstruct MS the two outermost ones are severely damaged. Ashkar as follows (fig. 4): Its text is also from Exodus, from the middle • Column One, completely reconstructed: until 16:1 (fig. 4). 263 words, from the end of Exodus 13:2 to )]את בני[ ישראל לאמר( of 13:19 There is a gap of 16 verses and two the middle of 13:19; partial verses – from 13:2 (end) until 19 – • Column Two, severely damaged: 244 between the end of MS London and the words, from the middle of 13:19 to the middle first surviving column of MS Ashkar. The of 14:13; number of words in this missing text would • Column Three, almost perfectly preserved: fill one column; thus, we may infer that the 223 words, from the middle of 14:13 to the damaged column at the right end of MS middle of 14:28; Ashkar was preceded by another one, which • Column Four, containing the Song of the constituted the first column of the original Sea, is double-width and written according gvil sheet and the direct continuation of MS to the Masoretic guidelines for the layout of London. The top of that column completed that passage and the verses preceding it; Exodus 13:2, which began at the very end of • Column Five is calculated to have contained The end some 260 words, but only the top of the. באדם ובבהמה לי | ]הוא[ :MS London of the column led directly into the middle of column has survived; On the assumption that the sheet was the • (כי השבע השביע | ]את בני[ ישראל לאמר) verse 19 of the surviving column. (fig. 5) same length as MS London (which has seven The numerical data are as follows: MS columns), and given the fact that the column London is written in seven columns. There with the Song of the Sea is double-width, it are a total of 1,708 words on the sheet. The seems likely that MS Ashkar originally had number of words per column ranges from one more column on the left, making a total 208 to 266, or, consecutively: 237, 261, 266, of six. 246, 241, 249, and 208 (the 208‑word column is particularly narrow, perhaps because it was the last one on the sheet). The average number of words per column is 244. The
IMSA 7 2015: 24–61 31 32 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments Fig. 3 MS London
IMSA 7 2015: 24–61 33 The Text mid-third century CE; Talmud Yerushalmi Megillah 3:6 [74b]; Tractate Soferim 12:9).31 Having established that MS London and MS The song itself is written on thirty Ashkar are two consecutive sheets of a single lines (Tractate Soferim 12:10); lines 29–30 Torah scroll, they can now be treated as one: are written according to the conjectured MS London-Ashkar. In order to categorize text of A: בא סוס פרעה ברכבו ובפרשיו בים וישב יהוה עליהם MS London-Ashkar, its text was compared את מי הים ובני ישראל הלכו ביבשה בתוך הים to the reconstructed text of the Aleppo Codex (hereafter A), after J. S. Penkower,26 and with This layout has typological value since is not the Masoretic guidelines in MS Leningrad found in all manuscript types. b19A (hereafter L).27 After the Song of the Sea there is again the traditional blank line followed by five Spelling, section breaks, layout of the more lines that run until the bottom of the Song of the Sea column. Their arrangement is not identical As Birnbaum has noted,28 the text of MS with those in A, but the first line begins with and the ,אחריה the next with ,ותקח London is identical to the Masoretic Text the word ,as in A; the last two lines ,סוס hereafter MT), and this is true of MS third with) unlike – ולא יכלו and מדבר Ashkar as well. however, begin with It .ויבאו and ויצאו Here we should note a distinctive A, where they begin with orthographical feature of the conjoined forms should be noted that the scribe apparently – was not constrained to write the lines this ;אחרי כן =( אחריכן 29,)10:11 ;לכו-נא =( לכונא – According way. There is a closed section break in the .(11:8 ;ואחרי כן =( ואחריכן and ,(11:1 to the traditional cantillation marks, each middle of the third line (the one that begins the scribe simply could have made ;(סוס pair is linked by a maqqaf (hyphen). Note with that these words were not squeezed together it a bit wider, adding approximately seven due to a lack of space at the end of the line; more letter widths, so that the next line would and the following one with ויצאו the absence of a space between them was begin with .as in A ,ויבאו .intentional With regard to the section breaks, Birnbaum noted one divergence: the Masorah Ty polog y has a parashah setumah (closed section) in To determine the typology of MS London- Exodus 12:1 whereas MS London is marked Ashkar I relied on Penkower,32 who described by a parashah petuhah (open section).30 a sheet from a Torah scroll known as TSC The Song of the Sea and the lines (Torah Scroll Christie’s), dated to the tenth before and after it are written in accordance or eleventh century, containing the text of with the scribal tradition: the first line at the Exodus 10:10–16:15. Until the exposure of MS Ashkar in 2007, TSC was the oldest ,הבאים top of the column begins with the word followed in the next line surviving Torah fragment containing the ,ביבשה the next with This Song of the Sea, and because TSC overlaps .במצרים ,and, finally ,מת then ,יהוה by is followed by a blank line, after which the MS London-Ashkar to a great extent (Exodus song is written “blank over text and text over 10:10–16:1, except for the damaged portions in of MS Ashkar), we can rely on Penkower’s ,(אריח על גבי לבינה ולבינה על גבי אריח) ”blank keeping with the principle handed down in description of TSC and the comparative rabbinic literature in the name of Rav (died information provided in his article.33
34 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments where L=TSC. In any case, MS ,(22 :14) חמה ,Comparing TSC with the Oriental Sephardic, Ashkenazic, and Yemenite London-Ashkar corresponds to the text of A. traditions, Penkower found that it matched the Oriental type, which is represented by Section breaks (parashiyot) five medieval codices: the Aleppo Codex The manuscripts examined by Penkower (A); MS Sassoon 507 (now MS Jerusalem diverge significantly both from A and among 24° 5702) (S); MS British Library Or. 4445 themselves.36 The closest to A is TSC, which (B); MS Cairo 3 (C3); and MS Leningrad seems to have only two deviations: one can b19A (L). MS London Jews’ College 31 was be reconstructed in the damaged portion at not among the manuscripts examined for the beginning of the manuscript, 10:12 (an 34 comparison. open parashah instead of a closed one, with no indication of an emendation); the other Textual variants at 13:11, where the hand that emended the מימים Penkower’s examination of the textual manuscript scratched out the words at the end of the line and rewrote ימימה (variants (mostly plene and defective spellings in the Oriental manuscripts he examined35 them, squeezed together; on the next line, he revealed that MS London-Ashkar is the only and rewrote it in והיה scratched out the word one that corresponds fully to A (except for 37 stretched letters to fill the entire space. All the few places already noted, where spacing of the other manuscripts have deviations from was not observed) as well as to the marginal A – two in S, seven or eight in B, nine in C3, Masoretic notes in L, even in those places and four in L, none of which was emended.38 where the text of L itself does not comply. As noted above with regard to section TSC, which is a fragment of a Torah scroll breaks, MS London-Ashkar deviates from A intended for liturgical use and thus should in only one place (12:1), where, unlike A, it have been faithful to the guidelines for writing has an open parashah. This variant is shared Torah scrolls, deviates from A in nine or ten with L and C3, which also have an open places. Eight or nine of these places have been parashah here. emended, generally by inserting or removing a mater lectionis but also by supplying an The Song of the Sea and the preceding and which ,(10:13) את entire missing word, such as was inserted between the lines. The words following lines are written The layout of 15:19 – lines 29–30 of the (3:12) אל כל עדת ישראל לאמר בעשר over an erasure; Penkower suggests that the Song of the Sea in MS London-Ashkar had been omitted by the (in which line 30 begins with the words בעשר or עדת word את מי הים original scribe. In one place (11:1) a letter has ) – is characteristic of the Oriental ,and it is possible type, and we find the same layout in A, B ,אחרי כ )ם( ]ן[ ,been emended (again according to Penkower) that the scribe C3, and L. (In S, which is also Oriental, the verse is written without a break.) Following .אחריכם had originally written We cannot know if these are scribal Maimonides’ Hilkhot Sefer Torah, this errors or evidence of a different scribal tradition was preserved in the Yemenite type. tradition, especially in the case of variants In the Sephardic type, line 30 begins with In the Ashkenazic type, the .הים the word והנ)ו(תר common to two manuscripts, such as 12:15b(, where verse is written on two or three lines without( הראש)ו(ן ;where S=C3 ,)12:10) where B=TSC; and a break.39 ,(13:6) מצ)ו(ת ;L=TSC
IMSA 7 2015: 24–61 35
Fig. 4 Transcript of MS Ashkar (missing parts restored in gray)
36 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments
IMSA 7 2015: 24–61 37
Fig. 5 Reconstruction of the sequence of the sheets: the last two columns of MS London, the seam, and the first three columns of MS Ashkar (missing parts restored in gray)
38 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments
IMSA 7 2015: 24–61 39 שלחנא – All the types agree with regard to rabbinic literature.42 For example )=שלח נא(: ביום השביעי אמ’ לו שלחנא ביד תשלח .the lines preceding the Song of the Sea ,is (Exodus 4:13, in Midrash on Psalms 18 הבאים Beginning the column with the word attributed to an ancient tradition, “according 22 [ed. Buber, 75b]; MS Cambridge Or. בשעה שאמרתה לי לך אל פרעה אמרתי אליך ;(to the Torah scroll revised by Ezra Hasofer” 786 Exodus 4:11, in Midrash) שלחנא ביד תשלח Adat Devorim, MS Firkovich 22a). In many‘) manuscripts, the width of the last lines at the on Proverbs, fragment [ed. Ginzberg, I, end of the preceding column is modified so 165]). This tradition is also documented in that the column with the Song of the Sea can several of the Dead Sea Scrolls, including six ;(שפטו-נא :MT) שפוטונה :The scribe of TSC had places in 1QIsaa: 5:3 40.הבאים begin with התערבונא :36:8 ;(ויגידו-נא :MT) ויגידונא :to add an extra line to the previous column.41 19:12 ;(זכר-נא :MT) זכורנא :38:3 ;(התערב-נא :The scribe of MS London-Ashkar calculated (MT הבטנה :64:8 ;(עמדי-נא :MT) ]ו[עמודינא :the previous column precisely and did not 47:12 This is also the version of a .(הבט-נא :have to modify the letters or spaces in order (MT to comply with this tradition. parabiblical text (4Q382 f. 9.6 [DJD XIII, In the next four lines, MS London- 368]) that includes an excerpt from 2 Kings Ashkar conforms to the tradition (= A) 2:4 (reflecting the version of the Septuagint): ויאמר לו :MT) ]ויואמר אליה[ אל אלישע שיבנה פה that is also reflected in TSC, L, and other .(אליהו אלישע שב-נא פה manuscripts. Manuscripts display variants as one word in Exodus אחריכן in the lines below the blank line that follows L also has and אחרי-כן the Song of the Sea; the uniformity in the 11:1. The conjoined spelling of is found in several old manuscripts ואחרי-כן Yemenite type, following A, seems to be based on Maimonides’ guidelines. This is also and printed texts,43 including L (except for ,only אחרי the practice of TSC but not of C3, L, and S. line-ends, where there is room for Note that ‘Adat Devorim specifies the first as in 11:8); the scribe of L writes the words words of the five lines preceding the Song of as a single unit even where the combination the Sea, but not the lines following it. is not hyphenated, as in Genesis 45:15. The phenomenon is also attested in biblical Distinctive textual variants in MS London- passages quoted in manuscripts of rabbinic ואחריכן יצאו ברכוש גדול :Ashkar literature, such as We have already noted a distinct feature of (Genesis 15:14, in Mekhilta, Pisha 14 [ed. ויקרא ;(MS London-Ashkar, i.e., the conjoined letters Horowitz-Rabin, 49], MS Oxford 151 אל משה וישבו אליו אהרן וכל הנש’ ואחריכן נגשו לכו-נא =) לכונא :of three hyphenated words – .Exodus 34:32, in Sifre Numbers 73 [ed) ’וגו ;ואחרי כן and ,(11:1 ;אחרי כן =( אחריכן ;(10:11 ואחריכן יצא ;(Horowitz, 68], MS Vatican 32 .(11:8 ;ואחרי–כן =) Genesis 25:26, Sifre) אחיו וידו אחזת בעקב עשו is also found לכונא The closed-up form in L, where the spaces between the words are Deuteronomy 343 [ed. Finkelstein, 397], MS Genesis) ואחריכן יצאו ברכוש גדול ;(sometimes negligible, especially at the end of a Vatican 32 -Generally 15:14, Genesis Rabbah 44:20 [ed. Theodor .(15:22 ;ויצאואל =( ויצאו אל line, as in speaking, the scribe of L seems to have taken Albeck, 442], MS Vatican 30). the liberty to dispense with the space in As for the section breaks, we found one deviation from A: in Exodus 12:1, MS ;בחרי-אף=) בחריאף hyphenated words such as 11:8). The omission of the space before the London-Ashkar has an open parashah, like L appears, albeit only occasionally, in and C3, whereas A has a closed one. These נא word
40 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments few differences between MS London-Ashkar the Middle Ages. The Hebrew script did and A indicate the former’s possible link to not undergo major changes between its the tradition reflected in L. emergence from the Aramaic script and its development as a square script. Nevertheless, General Evaluation over the centuries there were variations in its styles and features, with a gradual progression We now can offer a general evaluation of from the “proto-square” script to the mature the work of the MS London-Ashkar scribe square script that lends elegance to the in light of the various manuscripts of the dated manuscripts of the tenth and eleventh Oriental type. All of the manuscripts centuries. available to Penkower, including TSC, show The laboratory dating of MS Ashkar, traces of emendations meant to harmonize identical to Birnbaum’s dating of MS London, with the MT (the conjectured text of A), allows us to assign MS London-Ashkar to the although most of them still have variants that seventh or eighth century. Here, then, in the were not emended. There were nine or ten context of the evolution of Hebrew script in orthographical variants in the original text of this period, I focus on the script that is used TSC, eight or nine of which were emended; in the two fragments of this scroll in order at least two in S (and perhaps three more), all to corroborate the dating and try to identify of which were emended; four in B, three of where the scroll was written. which were emended; ten in C3, all of which To demonstrate the affinities of the were emended; and nine in L, three or four of script used in the two manuscripts that which were emended. Most divergences from comprise MS London-Ashkar, their common the text of A with regard to section breaks characteristics were placed on an axis of were not emended (two in S; seven or eight development between two historical poles for in B; nine in C3, and four in L); but note the which we have datable texts – the latest of the emendation in TSC, where a somewhat crude Dead Sea Scrolls and the Wadi Murabba‘at technique was used to open a break that was texts (first and second centuries CE) and originally written as closed (13:11). In contrast fragments from the Cairo Genizah (eighth to all these manuscripts, MS London-Ashkar and ninth centuries CE). After having shows no signs of emendations.44 The scribe located the reunited manuscript on this time exhibited an impressively diligent adherence axis, the paleographic description focused to the MT. on the similarities between the script of MS London-Ashkar and that of a manuscript of III. The Paleography of MS known provenance. London-Ashkar Edna Engel The Paleographic Features of MS London-Ashkar45 Considering the paucity of extant texts from the second through ninth centuries, any The following form-related parameters manuscript that can confidently be dated to have been drawn to describe the script of that period is extremely valuable, constituting MS London-Ashkar: (1) textural features, an important link between the Hebrew script including the proportions of the letters in a first adopted as the Jewish form of writing word and a line, joined letters, and so on; (2) and the script commonly used by Jews in elements shared by several letters, including
IMSA 7 2015: 24–61 41 the common upper horizontals of the open a. Recurring fixed features letters bet, dalet, heh, het, kaf, kof, and resh; • Spacing between rows and between words the right verticals of dalet, heh, het, and resh; • Spacing between unjoined letters the shape of the connection between the • Broad, soft, and unpressured quality of the horizontal and vertical lines; the base of the line, with a constant width and no shading48 letters bet, kaf, mem, nun, ayin, peh, zadi, and • Visual links between letters, such as the tav; the narrow “roof” of the letters gimel, elongation of the neck of the lamed toward the zayin, nun, and final nun; and the left legs letters in the previous row, or the “nesting” of tet, ayin, zadi, final zadi, and shin;46 (3) of two adjacent letters having opposite slants, a description of several individual letters.47 such as alef followed by lamed • Relatively heavy letters produced by broad Texture (fig. 6) pen strokes, which contrast with their small Letters measuring approximately 3.5 mm internal spaces wide hang from the ruled line and are • Marked squatness of the letters, with a written close together within a word, fixed ratio of width that exceeds the height imparting a distinctive texture to the script. Despite the many variations in letter size and b. Recurring variant features shape, the two manuscripts have a regular • Uniform dimension of some letters versus and harmonious texture manifested by the varying dimensions of others (e.g., a final various fixed features. This regularity is mem is always larger than a heh) produced in part by the consistent repetition • Straight verticals in some letters (e.g., dalet of these features and in part by the consistent and heh) versus curved verticals in others (e.g., Fig. 6 Texture: a. MS recurrence of dissimilar ones that could be resh) London; b. MS Ashkar expected to disrupt the uniform rhythm. • Contrast between the slants of some a letters (e.g., between the rightward slant of the alef and the leftward slant of letters such as dalet and heh) versus the straight vertical orientation of most letters • Contrast between the centripetal motion of the alef versus the centrifugal motion of the other letters
Joined letters (fig. 7) “Nested letters” refer to the phenomenon b wherein a letter encloses the following one in a word to produce a predictable configuration. The slant of the second letter is modified, as if it is trying to accommodate the letter that encloses it. Examples are a bet inside an ayin, a yod inside a nun, and an alef inside a gimel or nun. Similarly, a letter that touches the next one produces a predictably configured pair of letters, as when a nun touches a vav that follows it, or a bet touches a resh.
42 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments a Elements shared by several letters varies independently. The variation increases Bases (fig. 8) with changes in the slant of the letter as a b Elongated and diagonal bases of letters are whole, from a straight vertical movement Fig. 7 conspicuous in both parts of the scroll, thus to a sharp angle. Nevertheless, as with the Joined letters: a. MS London; b. MS Ashkar enhancing the phenomenon of nested letters. texture, the variations in this letter do not mar the rhythm of the script because the a Long upper horizontals (fig. 9) variants are repeated with regular frequency. b The left ends of the long upper horizontals Fig. 8 (as in heh or resh) turn upward in a prominent The right head (a) is produced by a Bases: a. MS London; vertical or diagonal tag. The horizontals double pen stroke, inward then outward. b. MS Ashkar themselves show a light downward slant to The length of the long diagonal (b) varies a the left. from letter to letter. The left leg (c) ends in a double pen stroke, inward and outward. The b Right verticals (fig. 10) connecting point between the three elements Fig. 9 The right verticals have a slight curvature that varies from letter to letter. The diagonal and Long upper horizontals: a. MS London; b. MS becomes convex toward the bottom. the left leg may be connected anywhere Ashkar between the top and the middle of the left a Connections between horizontals and verticals leg (c). The head (a) is attached from the (fig. 11) midpoint to the lower part of the diagonal (b). b The connection between the horizontals Fig. 10 and verticals may be sharp or rounded. The a Right verticals: a. MS leftward slant and the shortened right leg London; b. MS Ashkar create a right-angle connection in the heh a and tav. b b
Long verticals (descenders, fig. 12) Fig. 11 The long verticals (descenders) range from Connections between horizontals and straight lines, as in kof, to lines with a slight verticals: a. MS curvature, as in final kaf and final tsade. c a London; b. MS Ashkar Occasionally the line displays a conspicuous a convexity, as in the final nun in MS London. b
Short upper horizontals (fig. 13) b Fig. 12 The short upper horizontals of the letters Long verticals (descenders): a. MS gimel, zayin, nun, and final nun and the left Fig. 14 London; b. MS Ashkar legs of tet, ayin, tsade, final tsade, and shin Alef: a. MS London; b. MS Ashkar are diamond-shaped and connect to a short a vertical of uniform thickness. b
Particular letters Fig. 13 Short upper Alef (fig. 14) horizontals: a. MS The letter alef has the greatest variety. In fact, London; b. MS Ashkar it is hard to specify a uniform morphology for this letter because each of its components
IMSA 7 2015: 24–61 43 Lamed (fig. 15) Most of the stone texts were discovered The body of the letter (a) is small and in the Land of israel, but the Hebrew and angular. Because of its small size the letter is Aramaic papyri, and a few fragments on elevated above the baseline to create a height gvil, came to light in the Egyptian Desert,50
a b differential with the adjacent letters. The mainly in excavations at Oxyrhynchus and long neck (b) encroaches upon the letters of Antinoöpolis. The Greek papyri with which the line above it. they were bound helped date the fragments to the Roman-Byzantine Period,51 with a b Mem (fig. 16) lower terminus in the third century CE and Component a of this letter, produced like the an upper terminus in the late seventh or letter kaf, connects at its left edge to element early eighth century, when the use of Greek a b, which is shaped like the letter yod. This was already in decline. The papyri include Fig. 15 Lamed: a. MS London; ancient form of mem resembles that found in fragments of Torah scrolls and a variety of b. MS Ashkar the earlier Dead Sea Scrolls. texts and documents, such as letters.52 a Among the oldest material in the Final mem (fig. 17) Cairo Genizah are individual sheets bearing What is unique about the final mem is biblical, halakhic, liturgical, and other b the sharp angle of the base, sometimes Hebrew texts, all invaluable aids in tracing a accompanied by a tag that rests on the the development of Hebrew script in the b baseline. eighth and ninth centuries. These texts, most of which were written on rough parchment, Fig. 16 Mem: a. MS London; Shin (fig. 18) are vocalized using the ancient Babylonian or b. MS Ashkar The right head (a) is produced by a rightward Palestinian system that preceded the spread of pen stroke. The head (b) is formed by a thick the Tiberian vocalization signs. Along with rightward pen stroke. The central head (c) is poetic compositions written on palimpsests diamond-shaped, resembling the short upper over sixth-century Greek, Latin, Palestinian a b horizontals of other letters. The triangular Syriac, and other texts, they reflect writing connection between the two large elements traditions that predate the tenth century. Fig. 17 Final mem: a. MS creates the base of the letter. Components b The diversity of writing materials, as London; b. MS Ashkar and c meet near the top of the former. well as the transition from gvil and papyrus to a parchment, certainly influenced the evolution On the Development of the Hebrew of the script. At the same time, my research Script from the Second to Ninth has identified characteristics common to all b Centuries49 types of script in this era, regardless of the
c surface and writing implement employed. b a Except for the scrolls from Qumran and Wadi These have made it possible to construct a Murabba‘at, very few fragments featuring typology of scripts for the period between Hebrew script written in pen and ink on the second and ninth centuries, which can papyrus, gvil, or parchment have survived be divided into three main stages: the first from the eight centuries constituting the to third century CE; the fifth to seventh Fig. 18 Shin: a. MS London; b. Roman, Byzantine, and early Islamic Periods. century; and the eighth and ninth centuries. MS Ashkar Far more numerous are the inscriptions inlaid The beginnings of the Jewish script – in mosaics or those carved or engraved in, the forerunner of the square Hebrew script – or painted on, limestone, basalt, and marble. can be seen in the Dead Sea Scrolls and stone
44 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments Fig. 19a The formal script of the first main stage of the Jewish script as exemplified by the Thanksgiving Scroll (1QHa), Qumran, 1st century CE, parchment. The Hebrew University of Jerusalem, exhibited in the Shrine of the Book, Israel Museum Jerusalem
Fig. 19b The semi-formal script of the first main stage of the Jewish script as exemplified by an Exodus Scroll fragment (Mur 1:4–5 Exodus), Wadi Murabba‘at, 2nd century CE, parchement. Israel Antiquities Authority
inscriptions from the Second Temple Period.53 The interaction between the square The Jewish script evolved in several similar script and a more cursive hand, and the styles, whether written in ink or engraved increasing influence of cursiveness on the in stone. In all styles there is a distinction formal script, gave rise to a representative between a representative calligraphic hand semi-formal script as early as the Second and a common hand.54 Here I will relate to Temple Period. This variant, well represented the representative calligraphic script only. in the Wadi Murabba‘at scrolls (first and From the earliest appearance of the second centuries CE), continued to develop representative calligraphic script, there has in parallel to the more conservative formal been a clear distinction between the formal script. As with the latter, it has the same traits and semi-formal scripts (figs. 19a–b). Over whether engraved in stone or written in ink.56 the generations, the formal script preserved This semi-formal script was more widespread many of its features without change, because it was executed with broader and including: the scrupulous attention paid to freer pen strokes and made writing easier. In shading; the formation of letters with straight general, it is larger and more rounded than and vertical pen strokes; tags adorned with the formal script. Whereas the verticals in the a reserved elegance; and diamond-shaped square script are straight-up-and-down, those heads on the narrow “roofs” and diagonal of the semi-formal script are slanted. It is lines. These formal characteristics are very more compressed and cursive, and sometimes common in the calligraphy of the scrolls the elements of a letter are detached. Letters and stone engravings of the Second Temple are broad, with soft, rounded, outlines. Period.55
IMSA 7 2015: 24–61 45 Fig. 20 Fragment of a and many other letters, which slant to the ketubbah (MS left. Most of the verticals have a similar Köln, Inv. 5853), inclination to the left. The upper horizontals Antinoöpolis, 417 CE. Universität slant toward the base line and end with a zu Köln. Institut prominent tag at a slight angle. The tops of für Altertumskunde Papyrussammlung the short upper horizontals have a reverse curvature. The semi-formal style is also evident in the shapes of particular letters. The compressed form and slant of the alef
Fig. 21 (opposite) Although ornamented, the letters flow and distinguish this letter from the elongated Munich palimpsest lack shading. alef of the formal script. As for mem, the tall (Ms. München, Palimpsest Clm Texts preserved from the period body of the formal script is replaced by a squat 29416), 8th century immediately following the Bar Kokhba form. The detached lines of some letters, like CE, parchment. Revolt show a decline in graphic quality mem and alef, highlight the soft character of Bayerische Staatsbibliothek that attests to stagnation in the development the script in this ketubbah. München of the representative formal scripts. In all Texts that exemplify the script styles
Fig. 22 (opposite) scripts, the characteristic calligraphic features of the sixth to eighth century include mosaic Genizah fragment used in the scrolls disappeared. Unlike the inscriptions from the Land of Israel, an (MS Cambridge, University Library T‑S Second Temple Period, when many stone engraving from southern Italy, as well as A38.15), ca. 8th–9th inscriptions are distinguished by their amulets, papyri, and palimpsests. Here, too, century, parchment. Cambridge University calligraphic excellence, texts of the second there is a distinction between the formal and Library and third centuries exhibit a much simpler, semi-formal scripts, but the interrelationship semi-cursive script. between the two is also evident. Most texts The strong influence of the common from the sixth to eighth centuries, whether script on the formal script may explain written in ink or engraved in stone, appear as why the semi-formal script comes into its what may best be described as a cross between own as the representative script, replacing the formal and semi-formal styles. the increasingly rare formal script. Indeed, An example of this hybrid is provided a high proportion of the material that has by three fragments of a palimpsest scroll survived from the third and fourth centuries overwritten with a Latin text from the (early Byzantine Period) through the eighth first half of the eighth century (hereafter century (early Islamic Period) appears in the the “Munich palimpsest”; figs. 21, 24). semi-formal script. These fragments, discovered and published A papyrus fragment of a ketubbah from by Malachi Beit-Arié, contain the prayer Antinoöpolis, written in Aramaic and Greek service and liturgical poems for the Day of in 417 CE, is representative of the semi- Atonement and were written, judging by the formal script of that period (fig. 20). date of the Latin text, in the sixth or seventh The flexible texture of the script century.57 Fig. 23 (opposite) emphasizes broad, compressed letters A broad and compressed script with Comparison of MS London-Ashkar with written on widely spaced lines. The strokes, variations in the angle, contiguous letters (e.g., the semi-formal without shading, are undulate and of uniform nun and waw), and short upper horizontals script of the lower thickness. that curl backward are characteristic of terminus: a. Exodus Scroll fragment; The inclination of the script fluctuates the semi-formal script of the Munich b. MS London-Ashkar between the alef, which slants to the right, palimpsest. Here, however, the letters do
46 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments not have uniform lines or rounded shapes. On the contrary, the horizontals and straight verticals are connected at an angle, and the verticals show slight shading. These features evidently reflect the influence of the formal script on the semi-formal one.58 The Hebrew script of the Islamic Period is found in the oldest texts from the Fig. 21 Cairo Genizah (figs. 22, 25). While it still exhibits characteristics of both earlier styles and is not too far removed from the script of the sixth and seventh centuries, its affinity with the elegant square script of Oriental manuscripts of the tenth and eleventh centuries is unmistakable. Nevertheless, it does not contain a clear square script and should best be defined as “proto-square.” Fig. 22 The strokes of the letters are still of uniform thickness with no shading. The letters in a a single line are not all the same size and bear random ornamental characteristics, such as stylized tags on the horizontals and decorative curls on the verticals.
Locating MS London-Ashkar in the b Range of Scripts from the Second to Ninth Centuries
The broad and compressed letters, the curvature of the lines, and the fluctuating angle of some letters associate the script of MS London-Ashkar with the semi-formal style. If we place it between the two extremes a b a b of the second–third century and the eighth– ninth century, we discover that some of its characteristics are already present in the semi-formal script of the Wadi Murabba‘at scrolls, are found to some extent in the fifth and sixth centuries, and are still present in the older Genizah fragments.
Fig. 23
IMSA 7 2015: 24–61 47 Fig. 24 Münich palimpsest
48 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments Fig. 25 Genizah fragment (MS Cambridge, University Library T‑S A38.15), ca. 8th–9th century, parchment. Cambridge University Library
IMSA 7 2015: 24–61 49 Fig. 26 Comparison of MS London-Ashkar with the semi-formal script of the upper terminus: a. MS London-Ashkar; b. Genizah fragment a b
Comparison with the semi-formal script of both have a homogeneous line, the broader the lower terminus: The Wadi Murabba‘at pen strokes in the Wadi Murabba‘at texts script (fig. 23) highlight the differences between the two As in the Wadi Murabba‘at script, the scripts. Some of the changes in the shapes handwriting of MS London-Ashkar employs of the letters are a result of the evolution of the same homogeneous and solid horizontal the script over the intervening centuries, line, no shading, a similar flatness of the such as the size of the lamed in comparison letters, and similar letter pairs. The common to the other letters. The small-bodied and features of the two scripts include: high-necked lamed of the Wadi Murabba‘at • Alef: A broad letter with a long diagonal texts was rendered lower and wider in MS that tilts the letter sharply to the right London-Ashkar, even though it remains • Final mem: Lower left corner with a tag relatively smaller than the other letters. resting on the baseline • Heh: Leftward and downward slant with a Comparison with the semi-formal script of thick tag on the horizontal; the right vertical the upper terminus: A Genizah fragment slants to the left; the (left) leg of the letter (fig. 26) touches the roof The common features of the two scripts • Lamed: Small angular body; long neck that include: encroaches upon the previous line • Alef: A broad letter with a long diagonal • Alef and tav touch that tilts the letter sharply to the right • Nesting of yod in the nun; contact between • Final mem: Lower left corner with a tag the bet and nun resting on the base line Despite these similarities, there are • Heh: Elongated horizontal that tilts the clear differences between the hand of MS letter to the left London-Ashkar and the Wadi Murabba‘at • Lamed: Small angular body; long neck script, evident both in the shapes of many that encroaches upon the line above it letters and in their texture. Even though • Nesting of letters in the preceding gimel
a b c d
Fig. 27 Comparisons of texture: a. Exodus Scroll fragment 2nd century; b. Fragment of a ketubbah; c. MS London-Ashkar; d. Genizah fragment
50 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments Placing MS London-Ashkar between the The short upper horizontals evolved from two termini the backward curving form of the Wadi a b c Texture (fig. 27) Murabba‘at script into the diamond-shaped Fig. 28 The closely spaced letters, compressed top of the nun in MS London-Ashkar, and Features common to shape of the script, curvature of the lines, then into a more calligraphically stylized a group of letters: a. Exodus Scroll and fluctuation of the slant are common to diamond in the Genizah fragment. fragment ; b. MS all three stages mentioned above: The alef London-Ashkar; remains broad and compressed and its long Alef (fig. 29) c. Genizah fragment diagonal tilts it to the right. By contrast, Despite the alef ’s width, flatness, and the final mem, slanted to the left, shifts rightward slant, it, too, exhibits gradual the angle of the script to the left. The roof change. In the Wadi Murabba‘at texts and a b c of the heh also slants the letter to the left. in MS London-Ashkar, the connecting point Fig. 29 The short-bodied lamed, with its long neck between the diagonal and the left leg ranges Alef: a. Exodus Scroll encroaching upon the line above it, is also a from the apex of the letter to somewhere fragment ; b. MS London-Ashkar; common characteristic of all three stages, as lower down. In the Genizah fragment, the c. Genizah fragment are the nested and contiguous letters, primary connection is positioned at the top of the features of the Jewish script in the texts from letter. Alongside the heads of the short upper Qumran and Wadi Murabba‘at. What, then, horizontals, the base of the left leg of the alef has changed? The main development in the splays to the right in the Wadi Murabba‘at a b c texture of the script seems to involve the texts and becomes more complex in both MS Fig. 30 Heh: a. Exodus Scroll nature of the pen strokes, the evolution of London-Ashkar and the Genizah fragment. fragment; b. MS which has created a less crowded texture. A The latter displays the beginning of shading, London-Ashkar; c. Genizah fragment gradual change in the width of the strokes making the left leg appear thinner than the has produced thinner lines in the letters other two elements. and the beginning of shading. Straighter verticals, horizontals, and bases of letters Heh (fig. 30) have produced a more uniform texture. The Even though the transverse line of the heh still variations in the slant of certain letters, such slants to the left in the Genizah fragment, as alef and final mem, are less extreme, as is the letter is now open, the right leg is more the contrast between the slants of the various elaborate, and the left leg has a different letters of the alphabet. The wider spacing is shape and design. facilitated by the reduced frequency of joined letters, both contiguous letters and nested In sum, the placement of MS London-Ashkar pairs. Changes over time are also evident between the two termini – the first-second in the ornamentation, as manifested in the century and the eighth–ninth century – short upper horizontals, the left leg of the alef, corroborates the dating of MS London- and the trunk of the peh. The simple heads Ashkar to the seventh–eight century. of the Wadi Murabba‘at script have turned into diamond-shaped heads in MS London- An Attempted Identification of the Ashkar, and especially in the Genizah Provenance of MS London-Ashkar fragment.59 During my search for written texts from Features common to a group of letters and this period, I came across a tiny fragment examples of particular letters (fig. 28) measuring about 3 cm long with a maximum
IMSA 7 2015: 24–61 51 width of 1.5 cm. Although the letters are MS London-Ashkar. In both manuscripts, smaller than those of MS London-Ashkar, the the letters are suspended from the ruled similarity in the writing caught my attention. line and their proportions and spacing are The fragment was found with the Hebrew similar. Even though the shapes of individual papyri at Oxyrhynchus, brought to the letters are close in all three fragments, the Bodleian Library in Oxford, and published clear differences between the Antinoöpolis by C. Sirat (hereafter the “Oxyrhynchus scroll, on the one hand, and the Oxyrhynchus fragment”; fig. 31). It contains six truncated fragment and MS London-Ashkar, on the lines from Exodus 2:23–25. Despite its small other, chiefly in terms of texture, make size, it was hard to avoid the hypothesis that the resemblance between the Oxyrhynchus it was part of the Exodus scroll that was fragment and MS London-Ashkar even more torn from the sheet preceding MS London. distinctive.64 However, Michael Maggen’s examination The uniform thick line, drawn without of the high-resolution photograph I exerting pressure on the pen, is strikingly received from Oxford determined that the similar in the Oxyrhynchus fragment and fragment is on parchment rather than gvil in MS London-Ashkar, but different in and has a different color and texture than the Antinoöpolis scroll. In contrast to the MS London.60 continuous line bearing a constant thickness Furthermore, the letters in this in the first two manuscripts, the thickness fragment are about forty percent smaller of the line in the Antinoöpolis scroll varies, than those in MS London.61 Despite the as can be seen chiefly in the rounded points obvious differences, only a close examination of contact between its broad horizontals by a trained eye allowed for the secure and narrower verticals. Similarly, whereas determination that the fragment is of the ends of the strokes of the letters in MS parchment rather than gvil and thus not London-Ashkar are truncated, they are part of MS London-Ashkar. Even if the sharply pointed in the Antinoöpolis scroll. Oxyrhynchus fragment is not part of the The angular variation that results same scroll, the unmistakable similarity in chiefly from the contrast between the the handwriting strongly suggests that both rightward slant of the alef and the leftward were written by the same scribe or, at the slant of the other letters is also evident in very least, by members of the same scribal the Antinoöpolis scroll, where the rightward school, and thus come from the same period slant of the alef is more pronounced, thereby and region. highlighting its difference from MS London- I have juxtaposed the Oxyrhynchus Ashkar and the Oxyrhynchus fragment. fragment and MS London-Ashkar with There is also a small and inconspicuous a parchment Torah scroll excavated at difference in the proportions of the letter. Antinoöpolis in Egypt (hereafter the The breadth and flatness that are prominent Antinoöpolis scroll)62 in order to show the characteristics of the scripts in MS London- similarity between the first two (fig. 32). Of Ashkar and the Oxyrhynchus fragment are all surviving examples of script from this less evident in the Antinoöpolis scroll, where period, that of the Oxyrhynchus fragment is the letter seems to be somewhat taller. closest to the script of MS London-Ashkar.63 One obvious feature of MS London- The texture and shape of the letters in Ashkar also found in the Oxyrhynchus the Oxyrhynchus fragment resemble those of fragment is variation in the shapes of certain
52 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments Fig. 31 העבדה וי זעקו ותעל שועתם אל Oxyrhynchus fragment MS Oxford, Bodleian) האלהים מן העבדה וישמע אלהים Library, Oriental את נאקתם ויזכר אלהים את בריתו Collections Heb.d. P]), ca. 7th–8th] 89 את אברהם את יצחק ואת יעקב century, parchment. Oxford University וירא אלהים את בני ישראל וידע Library אלהים ומשה היה רעה
Fig. 32 Comparison of MS London-Ashkar with contemporaneous Egyptian fragments: a. Antinoöpolis scroll (MS Oxford, Sackler Library [formerly in the Ashmolean Museum] Papyrology Room, Antinoöpolis frag. 47), ca. 7th–8th century, papyrus, Collection of the Egypt Exploration Society; b. MS London-Ashkar; c. Oxyrhynchus fragment
a b c
IMSA 7 2015: 24–61 53 a
b
c
letters. The Antinoöpolis scroll, by contrast, Bet shows great uniformity in the letter forms. In the first two manuscripts, the base juts out Despite the many changes in the to the right of the body of the letter; this is Fig. 33 Comparison of form of the alef, there are letters in MS not evident in the third manuscript. individual letters: London-Ashkar that resemble those in the a. Oxyrhynchus Oxyrhynchus fragment, as demonstrated by Heh fragment; b. MS London-Ashkar; the comparison of individual letters in all The letter in the third manuscript has a c. Antinoöpolis scroll three manuscripts (fig. 33). shorter roof with a sharper downward slant than that found in the first two. Alef In the Oxyrhynchus fragment and MS Lamed London-Ashkar, the elements of the alef The more angular body in the third are connected at the same point: The long manuscript distinguishes it from the first and curved diagonal slants to the right, two. meeting the left leg high up, beneath the top of the letter; the right element in both Nun meets the diagonal near its lower end. In the The diamond-shaped head in the first two Antinoöpolis scroll, the diagonal approaches manuscripts is rendered as a vague shape in the midpoint of the left leg whereas the right the third. In the third manuscript, the vertical element reaches the midpoint of the diagonal. line and the base of the letter meet higher up. The fading of the ink at the bottom of the left leg in the first manuscript disguises the Resh similarity in its shape to that in the second. Clear shading of the letter in the third Nevertheless, there is a clear difference in manuscript is not visible in the first two. the shape of the double pen stroke that forms the base of the left leg of the alef in the first Tav two manuscripts as opposed to the third. The In the third manuscript, unlike the first two, head of the letter is different as well – thick in the left leg extends beyond the roof. the first two and thinner in the third. There is also a distinct difference in the quality of the Final mem line used to construct the letter. In the first The roof of the final mem in the third two examples, the brush strokes are uniform, manuscript ends with an extra stroke that whereas in the third the thickness of the line creates an additional decorative element not varies. The alef in the third manuscript has a found in the first two manuscripts. sharper slant than in the others.
54 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments Letter pairs of the seventh–eighth century, confirming The alef is angled more sharply away from the Carbon‑14 testing. the tav in the third manuscript than in the Comparison with the fragment from other two. Oxyrhynchus in Egypt allows us to address the provenance of MS London-Ashkar. Thus, in an attempt to identify MS London- Although the resemblance between the Ashkar’s provenance, it can be concluded scripts of the Oxyrhynchus fragment and that the close affinity of MS London-Ashkar the scroll can hardly be coincidental, the with the Egyptian Oxyrhynchus fragment, poor physical state of the fragment and as demonstrated by the difference of both the damaged elements of some letters these manuscripts from another manuscript preclude an unequivocal determination. of the same period (the Antinoöpolis scroll), These reservations notwithstanding, confirms MS London-Ashkar’s affiliation there is a distinct relationship between the with an Egyptian school of writing. Oxyrhynchus fragment and MS London- Ashkar. The demonstrable Egyptian Conclusion provenance of the Oxyrhynchus fragment corroborates the hypothesis that MS London- MS London-Ashkar was reunited on the Ashkar was written in Egypt as well,65 and basis of textual criteria; the two sheets were that both the fragment and the scroll were reconstructed as part of a single scroll. The written either by the same hand or by scribes paleographic analysis describes the scribal of the same school.66 characteristics of each part of the scroll, corroborating what is visible to the eye. * Photos: MS Ashkar © The Israel Museum, Paleographic parameters were employed Jerusalem, by Ardon Bar Hama; MS London © The Loewentheil Family; Thanksgiving Scroll to position the scroll on the axis between the (1QHa) © The Israel Museum, Jerusalem, by first–second and eighth–ninth centuries, David Harris; Exodus Scroll fragment (Mur1 by which time Hebrew writing had evolved [Wadi Murabaat]) © Israel Antiquities Authority; Fragment of a ketubbah (MS Köln, Inv. 5853), into a proto-square script on the way to the Antinoöpolis © By permission of the Universitaet mature square script of the tenth and eleventh zu Köln; Munich palimpsest © By permission of centuries. A comparison between the script of the Bayerische Staatsbibliothek München; Genizah fragment © By permission of the Syndics of the MS London-Ashkar and the scripts known Cambridge University Library; Oxyrhynchus in the three main stages between the first fragment: photo and transcription: © by permission and ninth centuries allows us to characterize of the Bodleian Libraries, University of Oxford; Antinoöpolis scroll (MS Oxford, Sackler Library the script of MS London-Ashkar. By locating [formerly in the Ashmolean Museum] Papyrology it on the axis of development between two Room, Antinoöpolis frag. 47) © Photograph historical poles I found that traces of the courtesy of the Egypt Exploration Society earlier Jewish script attest to the relative ** The authors presented their research in two lectures antiquity of MS London-Ashkar, whereas its at the Shrine of the Book, The Israel Museum, Jerusalem, in March-April 2008. The manuscript of affinity with the proto-square script of the this article was delivered to the editors of IMSA in Genizah fragment indicates its closeness to 2009. the script of the eighth–ninth century. Thus, 1 Owing to the difficulty in translating the Hebrew the script of MS London-Ashkar can be term gvil, the original term is used here in transliteration. For an explanation of the term, see defined as a representative semi-formal script Appendix 1.
IMSA 7 2015: 24–61 55 2 Another stage in the reconstruction of the scroll 10 The Jewish (or Ashuri) script developed from the came when Engel recognized the similarity between Aramaic script and is the prototype of the modern the script of a tiny parchment fragment from square script. Oxyrhynchus in Egypt and that of MS London- 11 In his initial study, Charlesworth suggested that Ashkar (see below). MS Ashkar came from Egypt. Similarities between 3 James H. Charlesworth is currently the the Oxyrhynchus fragment and the manuscript, to George L. Collord Professor of New Testament be discussed below, corroborate his hypothesis. Language and Literature at Princeton Theological 12 In recent years, Colette Sirat has discovered a Seminary. number of biblical fragments that she believes date 4 See Appendix 2, as well as interviews with from the first millennium CE. In addition to the Prof. Charlesworth cited in the bibliography three manuscripts discussed here (MS London, (Charlesworth 1980a, 1980b, 1980c). MS Antinoöpolis, and MS Oxyrhynchus), she 5 mentions two other small parchment fragments For a physical description of MS Ashkar, see of biblical scrolls (Sirat 1985): 1. Two fragments Appendix 1. from the book of Job (MS Oxford, Ashmolean 6 A partial photograph appears in Encyclopedia Museum, Antinoöpolis Papyri 49 and 50); and 2. Biblica 5, pl. 14. Two half-columns from the book of Deuteronomy 7 See Birnbaum 1959. (Berlin, Staatliche Museum P 10598). In an article co-authored with Ada Yardeni and M. Dukan (Sirat 8 MS London is a gvil strip with seven columns. The et al. 1994), she describes six biblical palimpsests, gvil is brown or light brown, the ink dark brown. which she maintains were written before the year The strip is between 75.5 and 76 cm wide, has a 1000. Judging by the known scripts of that period maximum height of 51.3 cm (on the undamaged (see Engel 1990), it is clear that none of them right side), and a minimum height of 46.0 cm (along comes from a Torah scroll written before the ninth the torn left edge). The left and right edges of the century. Fragments of a Torah scroll, including sheet are torn along the perforations of the stitching. Cambridge T‑S NS 3.21, were found by Sirat Although there are tears in the middle of the strip, among the Cairo Genizah fragments and published the only textual lacuna in the seven columns is a by Ada Yardeni (1990), who assigns this scroll to hole in the last column, where several words are the Byzantine Period on paleographical grounds. missing on four lines. Despite several moisture 13 stains of various sizes, all of the letters are intact and For a summary of the change from scroll to codex, the entire text can be read without difficulty. see Beit-Arié 1992–93, 10–12; see also Haran 1987–88. 9 The authors would like to thank Dr. Adolfo 14 Roitman for his assistance, initiative, and Unlike gvil, parchment is processed so that both encouragement. Thanks also to Michael the flesh side and the hair side can be written on. Maggen and Irene Levitt for their assistance On the preparation of gvil and a summary of the and cooperation. Stephan Lowentheil and his literary sources about it, see Glatzer 1989, 185–88. associate, Thomas Edsall, provided invaluable help 15 Maimonides, Mishneh Torah, Hilkhot Tefillin 1, 8–9: by sending us a facsimile of MS London, thereby “It is a halakhah from Moses at Sinai that Torah enabling us to make a close comparison between scrolls be written on gvil and written on the hair the two manuscripts, as well as by supporting the side... Even though this is a halakhah from Moses publication of this article. We would like to thank at Sinai, a Torah scroll written on parchment is Prof. Eric M. Meyers, Director of the Center for valid.” Jewish Studies, the American Schools of Oriental 16 Unlike the precise traditions regarding the Research (ASOR), Duke University, Durham, material from which Torah scrolls can be made North Carolina, as well as Prof. Linda McCurdy, and the layout of the text, in the Middle Ages Director of Research Services at the Rare Book, there was no uniform tradition regarding the Manuscript, and Special Collections Library shapes of the letters. The different styles of script of Duke University, for allowing us to use the that prevailed in different geographical regions, enhanced facsimiles of MS Ashkar. Thanks also to as well as the historical evolution of the shapes Prof. James H. Charlesworth for his crucial part in of letters, meant that each region and period our project. Many thanks to all of those who edited had its own typical Torah hand. Later, however, and prepared the article for press: Mike Rogoff, who a strict tradition developed with regard to the translated it into English, Silvia Rozenberg, editor shapes of the letters, obliging scribes to follow of IMSA, Nancy Benovitz, and Hani Davis. Finally, specific rules for their lines and strokes. The thanks are due to Prof. Malachi Beit-Arié, who read two Talmuds (the Yerushalmi, completed by a draft of the article and offered helpful suggestions. 400 CE, and the Bavli, completed around 500 CE)
56 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments enumerate some of these rules, some of which were employed in medieval codices; see Glatzer can be traced back to the tannaim of the first 1989, 215–21. and second centuries, such as the long passage in 22 On the various techniques for left-justification of Bavli Menahot 30a that is introduced by the words Hebrew manuscripts, see Beit-Arié 1993, 82–84; “our sages taught,” which is the standard incipit for an updated summary of the topic see Beit-Arié for a tannaitic quotation. Other rules, such as 2009, chapter 7. those found in Yerushalmi Megillah 1:8, 71d, are 23 designated “halakhah of Moses at Sinai,” which Birnbaum (1959, 123) noted the unusual methods means statutes in force from time immemorial employed by the scribe of MS London to fill out the for which the sages no longer knew the source lines by means of graphic elements. or rationale. Treatises summarizing the laws 24 Sirat et al. 1994, 864. concerning Torah scrolls were composed at various 25 So, too, according to Glatzer (1989, 220), who notes dates after the canonization of the Talmud; these that filling out the line is not mentioned in the include Tractate Sefer Torah (ca. 600 CE), Tractate halakhic literature on Torah scrolls. Soferim (ca. 800 CE), and various other works 26 ascribed to the geonim. One of the most important Penkower 1992. later treatises is ‘Adat Devorim, attributed in all 27 The text of L was examined using the facsimile probability to the eleventh-century Karaite author published by Makor. Joseph al-Kostandini. Hilkhot Sefer Torah (8:5) 28 in Maimonides’ Mishneh Torah (twelfth century) Birnbaum 1959, 123. written as two words, and (שא נא) But see 10:17 is very important since Maimonides had access 29 .appearing on two separate lines (דבר נא) to and relied on the manuscript known today as 11:2 the Aleppo Codex (Keter Aram Zovah); in Hilkhot 30 Birnbaum 1959, 123. We have evidence for the Sefer Torah he also listed all of the open and closed division into parashiyot from the Dead Sea Scrolls. section breaks (petuhot and setumot) in the Torah In the fragments parallel to our text, published by (8:6–10). Today’s rules come from the standard J.E. Sanderson in DJD XII, there are two places halakhic code in Joseph Caro’s Shulhan Arukh (first with an open parashah; in both, the Qumran scroll printed in venice in 1565/66), Hilkhot Sefer Torah, corresponds to the Masoretic Text (and our scroll): Yoreh De‘ah 270–84). Detailed instructions about the beginning of Exodus Chapter 14 (4QExodc, textual variants are listed in the Masoretic notes frg. 14; p. 114, pl. XIX) and Chapter 14, before found all along the margins of ancient codices verse 26 (4QExodg, frg. 18, p. 145, pl. XXI). These (references below are to the earliest source). fragmentary texts are ascribed to the first century 17 Except for the column containing the Song of the BCE (DJD XII, 100 and 145). Sea, which was subject to special rules. 31 The Dead Sea Scrolls show no evidence of such 18 Some Torah scrolls do not comply with this a layout for the Song of the Sea and the verses specification; see Penkower 2002, 263–64. before and after it. Parts of the song have survived c 19 See also Tractate Soferim 1:15. A closed section in 4QExod (mentioned above), but the text is (parashah setumah) means that the new section not divided as in the Masorah. Nevertheless, the begins on the same line as the one just completed, layout is distinctive by virtue of the small spaces between the verses and within them: in v. 14, before עד יעבר חיל אחז following a blank space, or on a new line, after an indentation. An open section (parashah petuhah) ; in the middle of v. 16, before before v. 19; in the same verse, before ;עם זו קנית ובני ישראל means that the new section begins at the start of a new line, after an incomplete or blank line. ; and before v. 20 (on these spaces see 20 DJD XII, 99–100). The same practice was found This is already mentioned in the Masoretic notes in 4Q365 (Reworked Pentateuchc), fragment 6b, and in ‘Adat Devorim, MS Firkovich 22a. which contains remnants of verses 17–20 of the 21 According to Halakhot Kezuvot, attributed to Song (published by E. Tov and S. White in DJD eighth-century Yehudai Gaon: “The writing should XIII, 268, pl. XXII): there are small spaces in the ,before v. 19; in v. 19 ;מכון not go more than two letters beyond the incised middle of v. 16, before ,and before v. 20 ;את מימי>!< הים rule” (Margulies 1942, 158); note that as late as the before and after sixteenth century such precision was considered to after an addition in v. 19, following Exodus 14:29 be an optional refinement: “All of these are merely (see DJD XIII, 269). This fragment is attributed excellent precepts, but if one deviated [the scroll] to the first century BCE (DJD XIII, 260), c c is valid” (Shulhan Arukh, Yoreh De‘ah 273:5). Today like 4QExod . Both fragments – 4QExod and the custom is to left-justify the lines in a Torah 4Q365 – were mentioned by Prof. Charlesworth scroll by stretching certain letters. Other methods in his report (see Appendix 2). There is no way of knowing whether a format closer to that of the
IMSA 7 2015: 24–61 57 Masorah also existed in that period, perhaps within 51 For a summary of the scripts in papyrus texts, see a different circle of scribes. Note that two copies Sirat 1985. of the book of Psalms were discovered on Masada; 52 The papyrus and gvil fragments from Antinoöpolis one of them (Mas 1e [Mas Psa]) is written in two are housed in the Sackler Library in Oxford, the parallel columns, like the codices of the Middle Oxyrhynchus papyri in the Oriental collections of Ages, while the other (Mas 1f [Mas Psb]) has small the Bodleian Library in Oxford. spaces, like the Song of the Sea in 4QExodc and 53 4Q365 (Masada vI; Psa, 76–90; Psb, 91–97). Both On the script of the Dead Sea Scrolls, see Avigad scrolls are attributed to the first half of the first 1961. c century CE and are therefore later than 4QExod 54 The common script has an irregular texture because b a and 4Q365; Ps seems to be older than Ps and may of the variable spaces between letters and words date from the end of the first century BCE (Masada and because of the variety of forms for each letter. vI, 77 and 92). It should be noted that, contrary to The individual characteristics of each writer are in the consensus of scholars, the Song of the Sea in the evidence here. traditional version of the Samaritan Pentateuch is 55 not written in columns (Prof. Abraham Tal, personal See, for example, the Uzziah inscription – a stone communication). carving on a tomb assigned to the end of the Hasmonean Period (Sukenik 1931). The horizontal 32 Penkower 2002. and vertical pen strokes are straight and vertical, 33 Penkower 2002, 237ff. the tags are carefully formed, and each letter stands out individually but does not interfere with the fine 34 Penkower 2002, 237 n. 5, relying on Sirat’s texture of the line. objection to Birnbaum’s dating of MS London (Sirat 56 et al. 1994, 861 n. 3). A comparison between stone engravings and texts written with a calamus (reed pen), as well as the 35 Penkower 2002, 238–42. textual distinction between papyrus and parchment, 36 Penkower 2002, 248–50. support the assumption that the mode of script 37 Penkower 2002, 248–89, 251, pl. 1. employed was not determined by the writing 38 material (or type of composition); the semi-formal Penkower 2002, 249–50. script was used both on parchment and papyrus, in 39 Penkower 2002, 255–61. documents as well as in biblical or liturgical texts. 57 40 Penkower 2002, 255. See Beit-Arié 1967–68. 58 41 Penkower 2002, 256. Note that the diverse features of these two 42 examples – the ketubbah from Antinoöpolis and The rabbinical sources here and below are cited the Munich palimpsest – may reflect regional according to the databases of The Hebrew characteristics that distinguish European scripts Language Historical Dictionary Project of the from those originating in Egypt and the Near East. Academy of the Hebrew Language (Ma’agarim). The small number of examples of script from this 43 See The Old Testament, note on Genesis 6:4. period makes it difficult to explore this hypothesis. ומשמאלם 44 The final mem of the word s (14:29) is 59 An even more elaborate shape is found in elegant damaged because the gvil has peeled off; in its place manuscripts of the tenth and eleventh centuries. there is a circular shape whose nature is unclear. 60 I would like to thank Prof. Joseph Geiger, who 45 The facsimiles of MS Ashkar are not very clear examined the original fragment at my request. because of the poor physical condition of the According to his description, it is of medium original. In the facsimiles of MS London, by thickness (thicker than that of the adjacent contrast, the letters appear in fine detail. fragments in the same display). The front is see Yardeni 1990. yellowish-brown with darker patches; the back, a ;שעטנ”ז ג”ז That is, the letters 46 dark reddish-brown. The ink is dark brown. 47 For details on my analytical method, see Engel 61 1999. It is possible that the gvil shrank significantly, as leather is apt to do. On this phenomenon, see Reed 48 Shading may be seen on a letter where the thickness 1972, 317–23; see also Glatzer 1989, 188–90. of the horizontal lines contrasts with the thinness of 62 the vertical lines, and vice versa. See Sirat 1985, 35–37. 63 49 See Engel 1990. It is important to emphasize the extremely poor quality of the extant letters in the Oxyrhynchus 50 A few examples of Hebrew script have also survived fragment. The fading of the letters and their tiny from southern Italy, Spain, and southern Yemen. size, as well as the shrinkage of the leather, all
58 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments make it difficult to identify the precise elements, Encyclopedia Biblica 5 Encyclopedia Biblica, strokes, and slant of the letters. This hampers any Vol. 5. Jerusalem, 1968. comparison and certainly detracts from the precision of the findings. Enlarging the three fragments to Ginzei Schechter L. Ginzberg, ed. Ginzei make their lines correspond clearly demonstrates Schechter I. New York, 1928 (Hebrew). the difference in the size of the letters in the Halakhoth Kezuboth M. Margulies, ed. Oxyrhynchus scroll and especially the relatively large size of the letters of the Antinoöpolis scroll. Halakhoth Kezuboth Attributed to R. Yehuday Gaon. Jerusalem, 64 Whereas the shape of a letter is easy to imitate, the quality of the stroke and its movement are more Ma’agarim The Hebrew Language Historical characteristic of the scribe’s handwriting than of the Dictionary Project. The Academy of script typical of a particular period or place. the Hebrew Language.
IMSA 7 2015: 24–61 59 Manuscript and the Habakkuk Commentary. Glatzer, M. New Haven, 1950. 1989 The Aleppo Codex: Codicological and Palaeographical Aspects. Sefunot (New Avigad, N. Series) 4 (19): 167–276 (Hebrew). 1961 The Palaeography of the Dead Sea Scrolls and Related Documents. In Haran, M. Essays on the Dead Sea Scrolls in Memory of 1987–88 The Codex, the Pinax and the Wooden E.L. Sukenik, ed. Y. Yadin and Ch. Rabin, Slats. Tarbiz 57: 151–64 (Hebrew). 107–139. Jerusalem (Hebrew). Penkower, J.S. Beit-Arié, M. 1992 New Evidence for the Pentateuch Text in the 1967–68 The Munich Palimpsest: A Hebrew Aleppo Codex. Ramat Gan (Hebrew). Scroll Written before the Eighth Century. 2002 A Sheet of Parchment from a 10th or Kirjath Sepher 43: 411–28 (Hebrew). 11th Century Torah Scroll: Determining 1992–93 Hebrew Manuscripts of East and West: its Type among Four Traditions (Oriental, Towards a Comparative Codicology. The Sefardi, Ashkenazi, Yemenite). Textus 21: Panizzi Lectures 1992. London. 235–64. 1993 The Makings of the Medieval Hebrew Book. Reed, R. Jerusalem. 1972 Ancient Skins, Parchments and Leathers. London and New York. 2009 Hebrew Codicology: Historical and Comparative Typology of Hebrew Medieval Sirat, C. Codices Based on the Documentation of the 1985 Les papyrus en caractères hébraïques trouvés Extant Dated Manuscripts in Quantitative en Egypte. Paris. Approach. Internet version 1.0 (2008). Sirat, C., P. Cauderlier, M. Dukan, and Linked to SfarData online database M.A. Friedman, eds. of the Hebrew Palaeography Project 1986 La Ketouba de Cologne: un contrat de of the Israel Academy of Sciences and marriage juif à Antinoopolis. Opladen. Humanities (under construction at
60 An Ancient Scroll of the Book of Exodus:The Reunion of Two Separate Fragments IMSA 7 2015: 24–61 61