SCRIBAL LAYOUT • SCROLLS & CODICES • 7TH–13TH CENTURIES
The arrangements in which certain poems and lists of the Hebrew Bible are written, not as running prose but in staggered blocks or divided lines, as prescribed for the Torah scroll by tractate Soferim and by Maimonides, and realised with striking variety in the great Masoretic codices.
Exodus 15; Deuteronomy 32
Judges 5; 2 Samuel 22
Joshua 12; Esther 9
Ashkar-Gilson scroll, 7th–8th c.

The line counts are those of the halakhic sources; the codices do not always follow them, as section III explains.
I. The Rules in the Halakhic Sources
The Babylonian Talmud states the principle in a single sentence: all the songs are written in the pattern of a half-brick over a whole brick, except the list of the sons of Haman and the list of the kings of Canaan, which are written half-brick over half-brick and whole brick over whole brick, so that one course of a wall stands directly above another[5]. Tractate Soferim, a collection of scribal rules frequently dated to the eighth or ninth century, though composed of parts of different ages[6], gives the details. It assigns thirty lines to the “Song of the Sea” and sixty-four to the Song of Deborah in Judges 5, with sixty-five in some versions, and requires a blank line before and after “Haazinu”[2]. For the Song of David in 2 Samuel 22 and Psalm 18 it sets no measure, saying that a skilled scribe spaces the lines by the beginnings, middle pauses and endings of the verses, as throughout Psalms, Job and Proverbs[3].
The layout was a condition of validity, not an ornament. Soferim rules that a scroll in which Haazinu is written like a song, or a song like Haazinu, may not be used for the reading[4], and Maimonides counts a change in the form of the songs among the faults that disqualify a Torah scroll[1]. He also explains where his own figures came from. Having seen great confusion in the scrolls he examined, he relied on the codex, then in Egypt and formerly in Jerusalem, that Ben Asher had corrected, the codex now identified with the Aleppo Codex[1][6]. The legal side of these rules, including the lines that frame each song, is followed through a later code in the journal essay on Meiri’s Kiryat Sefer.

II. Is Haazinu a Song?
The sources do not treat the two poems of the Torah alike. The Talmud gives no instruction for Haazinu, and Soferim specifies only the first word of each of its lines, seventy in number, so that with the blank lines before and after the poem occupies seventy-two lines[6]. Mordechai Breuer argued from the Soferim ruling on Haazinu and the “song” that Haazinu was not reckoned among the songs at all; the true songs were the Song of the Sea, the Song of Deborah and the two lists, while Haazinu was to be written like the poetical books, one verse to a line with a gap dividing each verse[6]. The encyclopaedic overview of Haazinu, by contrast, describes its two-column form simply as the song’s prescribed layout[11].
The difference matters for what the gaps mean. In the true songs the spaces are part of a fixed pattern and help the reader through the text; in the poetical books and Haazinu, Breuer concluded, medieval scribes knew that a gap was required but often placed it to make a visual pattern rather than at the sense division[6]. A later ruling of the Shulḥan Arukh reflects that situation: a scroll is invalid if the song is written as prose, but not if the spacing merely departs from the customary places[6]. One encyclopaedic summary of the Song of the Sea goes further astray and counts the Song of Deborah as the other patterned song of the Torah[10]; it belongs, of course, to the book of Judges.
III. The Codices
The Masoretic codices were not Torah scrolls, and their scribes were not bound by the law of the scroll, but they were influenced by it to varying degrees. Dominique Barthélemy, summarised by Jonathan Lind, held up a Pentateuch copied by Solomon ben Buyaʿa, the manuscript known as B17, as an almost perfect realisation of the scheme in codex form: six prose lines before Haazinu, a blank line, and a marginal ʿayin signalling seventy lines[6]. In the Aleppo Codex the same scribe wrote the poem in sixty-seven lines, because, Barthélemy suggests, his list of line openings omitted four, and the wider columns that resulted forced him to change the prose that follows; Maimonides then adopted the sixty-seven lines and the five-line epilogue of Aleppo, a tradition that few later scribes reproduced exactly[6]. The journal essay on the Bologna Torah Scroll describes one scroll that does.
The Leningrad Codex is the extreme case. Its scribe fitted Haazinu into two columns on facing pages, as in a scroll, but in thirty-seven lines with line openings that seldom agree with Soferim; Breuer said it had “no system at all”, while Lind observes that thirty-five of its half-lines still begin where the traditional ones do[6]. Breuer also noted that the main codices, Aleppo, Leningrad, B17, the Damascus Pentateuch and Sassoon 1053, each divide the poem differently, and that no Tiberian Masorah addresses the layout of the songs at all; the layout notes in later Yemenite and Sephardi manuscripts do not come from the Tiberian Masoretes[6]. The layout of the songs was a scribal tradition carried beside the Masorah, not part of it.

IV. The Oldest Evidence and the Modern Editions
The staggered pattern is older than any of these codices. A fragment of a Torah scroll in the Rubenstein Library at Duke University, the Ashkar-Gilson manuscript, preserves part of Exodus and lays out the Song of the Sea in brickwork without any deviation from the later arrangement; it was found in Beirut in 1972 and may come from the Cairo Genizah[8][9]. Paul Sanders, who first published it, dates it to the seventh or eighth century, joins it with a second fragment known as the London manuscript, and argues that the scroll was of exceptional quality and was probably consulted by the Tiberian Masoretes when they prepared their model codices[7]. The two fragments were shown together at the Israel Museum in 2010[9][12].
Whether the pattern goes back to the Qumran period is answered differently by the sources. The Center for Online Judaic Studies states that in the Dead Sea Scrolls the Song of the Sea does not yet have the brickwork layout[8], while Lind points out that a mid-line gap dividing the verse, the device later used for Haazinu and the poetical books, already appears in at least one Qumran manuscript of the Psalms[6]; the two statements concern different arrangements and are compatible. For modern editions based on Leningrad the choice is practical. Biblia Hebraica Quinta, the successor of Biblia Hebraica Stuttgartensia, is a diplomatic edition, yet its Deuteronomy volume abandons the diplomatic layout of Leningrad for Haazinu in favour of the traditional arrangement of the Torah scroll, a decision Lind questions on the ground that there is no Masoretic intention for the layout to recover[6]. Other divisions of the written page are treated on the page on the sedarim and in the journal essays on open and closed sections and the unusual letters.
V. Maimonides’ Figures
The chapter of the Mishneh Torah in which Maimonides set out the open and closed sections ends with a description of the two Torah songs that is more exact than the Talmud’s. Haazinu, he writes, is written in seventy lines, each with a single space in the middle like the space of a closed section, so that every line is divided in two; he then lists the words that open each line and the words that open each second half-line. The Song of the Sea is written in thirty lines, the first in the ordinary way, and in the rest one line has a single space in the middle and the next has two, so that the line is divided into three and space stands opposite writing and writing opposite space[1]. The same chapter gives the totals of the sections for the whole Torah, 290 open and 379 closed, 669 in all, and requires the letters throughout, songs included, to stand a hair’s breadth apart[1].
The figure of seventy lines for Haazinu does not sit easily with the account of the Aleppo Codex given above, according to which Maimonides adopted the sixty-seven lines of that manuscript[6]. The printed text of the code gives seventy and lists seventy opening words[1]; Barthélemy’s reconstruction, reported by Lind, starts from the codex itself[6]. The translator’s note to the same chapter warns that the commentators debate the exact text of the Mishneh Torah at this point and that different traditions exist for several of the passages, and it advises scribes to copy from a scroll accepted as correct rather than from the list[1]. The discrepancy may therefore lie in the transmission of the code as much as in the scribe’s practice, and no single figure can be called Maimonides’ without stating which witness it comes from.
VI. The Lists in Esther and Joshua
The two lists that the Talmud exempts from the ordinary pattern have their own rules, and the passage that states the exemption also records how the first of them was to be read. The names of the ten sons of Haman, together with the word “ten”, are to be said in a single breath, and the vav of the last name, Vaizatha, is to be drawn out long like a pole[5]. The same page lays down that the scroll of Esther requires ruled lines like a Torah scroll, so the brickwork of the list was set within columns ruled as carefully as those of the Pentateuch[5]. Tractate Soferim, discussed above, gives the measures; here the Talmud adds conventions of reading and of the shape of a letter that belong to the same scribal tradition, and they show that the layout of a list was tied to the way it was read aloud.
The list of the kings of Canaan defeated by Joshua is the other exception. Joshua 12 enumerates thirty-one kings, each followed by the word “one”, and in the layout that the Talmud prescribes each name and each number stands directly above the one before, half-brick over half-brick and whole brick over whole brick, so that the list reads as two narrow columns of writing separated by a continuous channel of space[5]. The rule thus produces the opposite effect to the staggered songs: where the Song of the Sea alternates writing and space from line to line, the lists align them. The sources used on this page state the principle for both lists but give no line counts for Joshua 12 comparable to those of Soferim for the songs, and the codices must be consulted one by one to see how closely their scribes kept to it. The Talmud also gives a reason for the unusual pattern of both lists; that reason belongs to interpretation and is not pursued here.
VII. The Poetical Books and Their Columns
Beyond the songs, the Masoretic codices used a distinct page for the three poetical books. In the Leningrad Codex the text is arranged in three columns of twenty-seven lines, except in Psalms, Proverbs and Job, which are written in two columns[13], and the Aleppo Codex likewise uses three columns for most of its text[14]. The narrower lines of the poetical books follow the verse rather than the width of the column, and the mid-line gap dividing the verse, which Lind found already in a Qumran manuscript of the Psalms[6], became their characteristic form. Soferim, as noted above, refers the scribe of the Song of David to the same practice, spacing the lines by the openings, pauses and endings of the verses as throughout Psalms, Job and Proverbs[3]. The layout of the codices therefore groups the poetic sections of the prose books with the three books of Writings.
The distinction was audible as well as visible. The three poetical books have their own system of accents, and Abū al-Faraj Hārūn, writing in Jerusalem in the eleventh century, records that a congregation would declare a reader in error if he chanted the three books with the cantillation of the other twenty-one or the reverse[15]. In the codices the two-column page coincided with the change of accent system, just as the brickwork of the songs marked the beginning of a song within prose. The written layouts on this page, the accents of the poetical books, and the open and closed sections were maintained side by side by the same scribes, which is why the Masoretic codices remain the main evidence for all three. The accent system itself is described on the page on the Guide for the Reader and in the course lesson on cantillation.
Selected Bibliography
The poetic accents that accompany these layouts in Psalms, Job and Proverbs are treated in the journal essay on the poetic accents; the layout of the page and the accentuation of the text were separate crafts, practised by scribes and by Masoretes respectively.