skip to content

Writing Lines: T-S D1.108 and the Song of Moses

by Kim Phillips


When a medieval Jewish scribe sat down to write the Torah, he[1] had a lot on his mind. Of primary importance, of course, was the accurate copying of each and every letter: even words for which multiple different spellings were possible (a vast number!) had to be copied so that the precise spelling for that word, in that location, was diligently preserved. In addition, he had to accurately insert paragraph breaks at the traditional points. These paragraph breaks were of two types: the Closed break (Setuma) and the Open break (Petuḥa), and the correct type of break had also to be preserved in each location. Failure to preserve these paragraph breaks correctly could render the entire scroll unsuitable for public use. He may have heard the words of Masseḵet Sofrim (a kind of talmudic ‘how to’ guide for writing Torah scrolls) echoing in his mind: פתוחה שעשאה סתומה וסתומה שעשאה פתוחה הרי זו תיגנז “An open paragraph that has been written as a closed, or a closed paragraph that has been written as an open – that Torah scroll must be stored away”.[2]

When he reached Exodus 15 (The Song of the Sea) and Deuteronomy 32 (The Song of Moses) his task became even more complicated. Each song had to be arranged in a distinct manner, with each new line starting with a particular word. Again, Masseḵet Sofrim was useful to him: in chapter 11 of that work the initial words of each line of every song are specified.

These traditional stipulations, as demanding as they were, were at least well documented. The minor talmudic tractates Masseḵet Sofrim and Masseḵet Sefer Torah laid out these, and many other, rules concerning the correct production of a Torah scroll. Nonetheless, from the mediaeval evidence it would seem that, in addition to these clearly laid out stipulations, there were additional traditions regarding the correct writing of the Torah – additional traditions that were not written in either of these minor tractates, but which were passed down from scribe to scribe over the generations.

As an example, let us consider the paragraph of prose text immediately preceding The Song of Moses. This is how the paragraph is laid out in the Leningrad Codex (B19a):[3]


Leningrad Codex (B19a)

The first three lines of the Song of Moses are visible at the bottom of the image, each line written in two parts, with a central division. However, it is the prose preceding the song that interests us. The first six lines in the image above are the first six lines of the folio. The text itself appears to have been stretched out, with dots added between words to fill out the extra space, and fillers added to the end of each of the six lines. It is as though the scribe knew he needed to fill these six lines, but did not have sufficient text to do so comfortably. Curiously, the Aleppo Codex reveals an identical layout. Again, precisely six lines of prose precede the opening of the poem. Moreover, the first word of each line in the Leningrad Codex is the same as the first word of each line in the Aleppo Codex:







Again, just as in the Leningrad Codex, the text of each of these six lines in the Aleppo Codex is stretched, with multiple part letters used to fill the many gaps between words.

Why did the scribes behind the Leningrad Codex and the Aleppo Codex go to such extraordinary efforts at this point in the biblical text? It would appear that they were following an unwritten tradition regarding the layout of the six lines before the Song of Moses. Maimonides, in his Mishneh Torah (written towards the end of the twelfth century), comes to a similar conclusion:


יש דברים אחרים שלא אמרו אותן בתלמוד ונהגו בהן הסופרים וקבלה היא בידם איש מפי איש… ויהיה בראשי השיטין למעלה משירת האזינו ואעידה אחרי הדרך באחרית להכעיסו קהל שש שיטין


“There are other matters [concerning the correct writing of Torah scrolls] about which they did not speak in the Talmud, yet which the scribes nonetheless practiced; it was a tradition in their possession, handed down person to person… such as: that the following words should begin the six lines immediately preceding the Song of Moses: I will call to witness… after… the way… in latter… to provoke him to anger… congregation.[4]


So, the situation is as follows: some of the earliest and best mediaeval Bible codices (such as the Aleppo and Leningrad codices) appear to have adhered to a particular tradition regarding the writing of the six lines before the Song of Moses. One searches in vain through the talmudic literature trying to find written record of this tradition. The phenomenon simply appears, as it were out of thin air, in these codices.[5] The best part of two centuries later, Maimonides openly discusses the phenomenon for the first time, noting that this particular tradition was not codified in the Talmud, but simply handed down, scribe to scribe, through the generations.

Here is where T-S D1.108 comes in:

T-S D1.108 verso

T-S D1.108 P1 verso



Two small scraps of parchment, probably from the eleventh century. Superficially rather unimpressive, perhaps. However, tucked away in the bottom left hand corner of the parchment is the following text, in characteristic masoretic shorthand:


[           ]           ידעתי סוף

אחרי ר׳ ש׳          מן סוף שט׳

[           ]ש         הרעה סו׳ ש׳

באחרית ר׳ש׳        ייי סוף שט׳

להכעיסו ר׳ש׳       כל סוף ש׳

קהל ר׳ש׳             תמם סו׳ ש׳


The text above stipulates the first and last words of six lines of text - the six lines immediately preceding the Song of Moses. Sure enough, these words match precisely the textual layout found in the Aleppo and Leningrad Codices, and the tradition spoken of by Maimonides. Despite the fragmentary nature of the extant text, there can be no doubt that here is a written record of that elusive, non-talmudic stipulation regarding the correct way to write the six lines immediately preceding the Song of Moses.

Once again, the Genizah offers us privileged access to a geographically and culturally remote past. In this case, access to a particular scribe’s notes. This scribe may or may not have known of the Bibles we now refer to as the Aleppo and Leningrad codices. He almost certainly lived long before Maimonides and the writing of the Mishneh Torah. Yet this scribe, and his notes, embodied something of which Maimonides would later write. This scribe knew that, when copying the Song of Moses, he had to abide not only by the talmudic stipulations, but also by additional traditions, traditions not found in the Talmud, traditions handed down, scribe to scribe. T-S D1.108 thus becomes, from our perspective, the ‘missing link’ between the evidence of the Aleppo and Leningrad codices, and Maimonides’ later explanation.[6]

(Edited on 6 April 2017 to improve clarity.)


[1] See Masseḵet Sofrim 1:14 (Higger).

[2] Masseḵet Sofrim 1:15 (Higger).

[3] The rules for writing a Bible codex were by no means as strict as those for writing a Torah scroll (presumably because the former served no halakhic purpose). Nonetheless, many of the stipulations specified for Torah scrolls were carried over, to some extent at least, to the writing of Torah codices.

[4] Mishneh Torah, Ahava, Sefer Torah, 7:10.

[5] The original text of Or. 4445 is not extant at this point. The tradition discussed here is not adhered to fully by the later hand, but it is clearly known: ואעידה begins a column, and each of the five remaining traditional words begins every other line thereafter. The tradition does not appear to be reflected whatsoever in the text of the Damascus Pentateuch.

[6] For further details, see: Dominique Barthélemy, Studies in the Text of the Old Testament: An Introduction to the Hebrew Old Testament Text Project (Indiana: Eisenbrauns, 2012) 313-323; Sarah Lind, “The Layout of the Song of Moses (Deut 32) in Masoretic Manuscripts and Biblia Hebraica Quinta”, The Bible Translator 64(2), 159-172: Mordechai Breuer, The Aleppo Codex and the Accepted Text of the Bible [Hebrew] (Jerusalem: Mosad Harav Kook) 171-183.



Cite this article

Phillips, K. (2017). Writing Lines: T-S D1.108 and the Song of Moses. [Genizah Research Unit, Fragment of the Month, March 2017].

If you enjoyed this Fragment of the Month, you can find others here.

Contact us:

The zoomable images are produced using Cloud Zoom, a jQueryimage zoom plugin:
Cloud Zoom, Copyright (c) 2010, R Cecco,
Licensed under the MIT License