Chapter 9 Recreating Single Columns Based on Fragments and Parallels

In: Material and Digital Reconstruction of Fragmentary Dead Sea Scrolls
Jonathan Ben-Dov
Search for other papers by Jonathan Ben-Dov in
Current site
Google Scholar
Asaf Gayer
Search for other papers by Asaf Gayer in
Current site
Google Scholar
, and
Eshbal Ratzon
Search for other papers by Eshbal Ratzon in
Current site
Google Scholar
Open Access

The reconstruction of a fragmentary scroll is a difficult task, similar to putting together a jigsaw puzzle with only a few of the pieces, and with the complete picture unknown. After collecting the various data described in the preceding chapters, we now reach the essential unit for the reconstructed canvas: the column. The work of material reconstruction involves establishing the borders and measurements of all columns, both those preserved intact and the fragmentary ones. Some reconstructions also require positing blank, hypothetical (“dummy”) columns, whose existence is proven by the reconstruction procedure. The best way to assemble the data is to recreate them on a digital canvas.1 Such a canvas and the way to produce it is the heart of the discussion in the present volume. The canvas should contain accurate measurements for all columns and their margins throughout the sheets of the scroll. The length of the scroll is then determined based on the number of fragments and other available information.

In Qumran, scrolls are constructed from a series of leather or papyrus sheets, stitched or glued to each other. Each sheet is divided into columns, i.e., writing blocks. A column is an inscribed surface, limited on four sides by un-inscribed surface, i.e., top, bottom, and side margins. Being the essential unit for digital restoration, the definition of a column involves such concepts as width, number of lines, distance between lines, and the size of margins (top, bottom, intercolumnar, and the intercolumnar margin at the seams between sheets of leather). These factors provide the skeleton of a given scroll, which translates into a two-dimensional canvas.

The column is the essential building block of the reconstruction process. Since column features often vary throughout a scroll, each column ought to be treated separately. Reconstruction begins by determining the width and height of the writing block, which is then translated to a square box drawn on the digital canvas (figure 20). Thereafter, one should turn to the reconstruction of the top, bottom, and intercolumnar margins. Seams between sheets should be clearly marked on the canvas by means of a broken or colored line.

Figure 20
Figure 20

Two consecutive columns of 4Q418. Borders are marked in boxes. The broken line indicates the seam between sheets.

© IAA, LLDSSDL, Shai Halevi/Najib Anton Albina

1 Width

The width of columns in the Qumran scrolls is greatly variable, although it correlates in a general way to the height of the scroll.2 The last column in a sheet will often have a different width than the preceding ones, in order to fill out the remaining space. The methods for establishing the width of a column in a fragmentary scroll include the tracking of dry rulings, of complete lines of text, and of parallel text. The creation of hypothetical columns using material markers is also required. These parameters are discussed below, arranged from the simplest to the most difficult for reconstruction.

1.1 Dry Rulings or Complete Lines of Text

When dry rulings are preserved in both right and left margins as for example in 1QM, 1QpHab, and 11QTa, they indicate the width of the column. The same pertains to scrolls where complete lines are preserved, even without the dry rulings. Note, however, that while all lines begin flush on the right ruling, they deviate from the left ruling by as much as 1–2 words on either side.3 Tov addressed this issue by recording the width in the form of a range of numbers rather than in concrete terms.4 However, defining the width in the form of a range is unsuitable for the purpose of digital reconstruction, whereby a definite line is required. One should therefore measure the width of as many lines as possible, and find the mode number of line width, i.e., the most frequent figure for the width of a single line among the lines of a given column.5 If only a few lines are preserved without a substantial demonstration of width, the average or median width of lines can also be selected.

1.2 Using a Parallel Text

When neither dry rulings nor complete lines are preserved, the width of the column should be deduced from the text of parallel copies. Most biblical scrolls fall into this category (notwithstanding the doubt with regard to the nature of the biblical text represented in them), and many non-biblical scrolls as well. One has to fill in the missing text based on the parallel, breaking the lines according to the available place.

Using a parallel text for establishing column width requires the positioning of a fragment within the typed text (see figure 20). This fragment then functions as a textual anchor, pinpointing the text to a specific layout. A fragment whose place in the column is verified by material indicators, such as the existence of margins, is preferable for this purpose. Such a fragment would provide both the width of the column and the exact position of the text within it. A less-securely placed fragment would not allow the secure location of the text in the exact layout. After placing the fragment, the parallel text would provide the number of letters and points of break between the lines. This parallel text is cast in the layout of the fragmentary scroll (figure 21). If carried out with appropriate methods, this action should provide a solid estimation of the width of the column.

Figure 21
Figure 21

Estimation of the width of a column of 4Q82 by casting the text of Micah 1:12–3:36 in the layout and script of 4Q82, fragments 91, 93, 94. Reconstruction and Graphics: Joshua Matson

© IAA, LLDSSDL, Shai Halevi

The number of characters in a line and its width in centimeters are not mutually indicative, as there are other factors to consider, such as the sizes of letters and of the space separating them, and size of the inter-word space. Most scholarly reconstructions of texts are made with standard Hebrew computer fonts such as SBL Hebrew or David. Since these fonts reproduce neither the exact size of each individual letter, nor the relationship between the letters in the manuscript, they cannot be used for estimating the width of the line in centimeters.

A method developed by Edward Herbert in 1997 for the reconstruction of long textual units may be used for determining the width of a column.6 Herbert’s method, however, is highly demanding. He suggested six-steps for conceiving the width of letters and columns. Each step is in turn developed into several sub-steps and work stages.7 They involve not only measuring each letter but also calculating the mean width and the standard deviation, as well as establishing an intricate statistical method for the compilation of this aggregate of data. We find this method inconvenient for use by textual scholars and paleographers in their everyday work.

Instead, the particular characteristics of each handwriting can be represented on the canvas with a custom computer font designed especially for each scroll, which mimics the hand of the scribe, thus sparing the need for a concrete measurement of each letter. This requires some technical skills. By using the excellent new images of the LLDSSDL and the DSSDP and with current computer software, reliable figures of column width can be reached and applied to entire lines. Such fonts are a steady, reproducible tool, easy to measure and compatible for statistical analysis. Carefully designed fonts can account not only for the widths of individual letters, but also for the letter’s interaction with neighboring letters (kerning), and for the distance between words and between lines. Chapter 10 discusses the creation of such fonts. Quite surprisingly, we will show that reconstructing a column with a custom font gives a good approximation of its width even when the size and location of vacats is unknown.8

1.3 Hypothetical (Dummy) Columns

The method delineated here requires in some cases the reckoning of “hypothetical” columns, for which neither fragments nor parallel text survived. Concrete numbers for such columns are dictated by the trial-and-error procedure of the material reconstruction, as explained in chapter 12. The starting point of this process should be the mean number of letter spaces per line (i.e., letters and spaces between letters) of the known neighboring columns.

In her work on the Masada copy of the Songs of Sabbath Sacrifice, Carol Newsom suggested the concept of “corrected letter-spaces” for estimating the column width. According to her method, narrow letters, such as ו, ‪ז‬, and י, as well as the spaces between words, are counted as half a letter.9 Newsom’s method became the standard method for letter counting, usually without adjustments or considerations.10 However, according to our investigation, the number of spaces and narrow letters in various lines is not significantly different, and its effect on the overall reconstruction is marginal. Furthermore, in many cases the width of such letters – these so-called narrow letters – hold the same width as other regular letters and vice versa. Therefore, a simple letter-space count is sufficient and should provide reliable data for estimating the width of the column.11

2 Height

The best-case scenario for reconstructing column height is when the column at hand is attested as one complete fragment, or at least attested in physical joins of several fragments. Such joins may be obtained also from material reconstruction of the scroll.12 When no firm evidence for the column height is preserved, other clues should be sought. When a long parallel text exists, its division between consecutive columns could provide the number of lines per column. Fragments constituting parts of two consecutive columns, or two fragments from two consecutive columns may be used when filling these columns with the parallel text.13 A custom-made font should be used when reconstructing the partial columns in the layout of the target scroll. Such practice is rather cost-effective, especially when reconstructing large stretches of text, which would otherwise require drawing by hand or meticulous measurements and calculations. The potential error for estimating the number of lines in a column using such font is given in chapter 10.

Another possibility is to try and reconstruct the column height. Joining long and narrow fragments, preferably ones that hold remains of top or bottom margins, may complete the column height.14 Such a vertical sequence of narrow fragments can be expected mainly in papyrus scrolls, since papyrus tends to break in long vertical stripes. The case of the cryptic copy of the Rule of Congregation from cave 4 (4Q249a) is a good example. This highly fragmentary papyrus shows three fragments joined vertically to produce a length of 14 lines, the minimum column height (figure 22).15 This join was not sufficient to provide the exact number of lines since it did not contain top or bottom margins, yet it gives a minimum number, which proved significant when combined with other data.

Figure 22
Figure 22

Reconstruction of column II of 4Q249a pap cryptA Serekh haEdah. The joins between fragments 4Q249a 6+8+9 (at the center of the image) provide a minimum height of 14 lines. Font design: Nir Yenni

© IAA, LLDSSDL, Shai Halevi

Two parameters establish the column height: the number of lines per column and the height of the writing block measured in centimeters. While the latter is generally stable throughout the scroll, the number of lines may vary between adjacent sheets.16 In order to convert the number of lines into the height of the column, measured in centimeters, one has to establish also the space between lines. The distances between lines along the height of a column are not necessarily even; in the sheet containing 1QS columns 8–11, for example, the distances between lines increase in the bottom parts of columns.17 This phenomenon is consistent throughout the sheet and can thus serve as a significant indicator when a fragment from the same height in the sheet occurs. In scrolls containing rulings, the rulings were carried out for each sheet of parchment separately and are thus consistent for each individual sheet but may vary in another sheet.

3 Margins

For the sake of reconstruction, we distinguish the top and bottom margins from the intercolumnar margins. We further distinguish the latter from the intercolumnar margins spreading across the seams of two consecutive sheets.

The presence of top/bottom margins enables the secure location of a fragment on the vertical axis. The size of the bottom or top margins may vary throughout the scroll but is quite consistent within a sheet.18 It is therefore helpful to group together fragments according to the size of the top or bottom margin. Such grouping may lead to finding fragments that belong to the same sheet.

The width of intercolumnar margins may vary throughout the scroll, and there is no certain way to determine the width of a given margin. This factor is crucial for material reconstruction in the Stegemann Method. In fact, work according to this method begins by collecting all fragments that attest to margins and transitions between columns and sheets, placing them on the canvas, and fitting the text around them as anchors.

Estimating the width of intercolumnar margins uses the same markers as those used above for estimating the width of a column. Excluding cases of indentation,19 it is safe to say that all lines of a given column stand flush to the right. One fragment showing the beginning of a line is sufficient for indicating the right margins and the beginning of lines throughout the entire column. Left margins are harder to establish, because lines do not end at the same vertical line. Additional indicators such as vertical ruling, guiding dots, or scribal marks may provide further information and increase the level of certainty. The vertical ruling is the most accurate indication for the left edge of the column. When such a ruling is absent, we suggest seeking a sequence of three consecutive lines that end on a vertical line.

Intercolumnar margins that appear between sheets tend to be wider. The exact position of these transitions is important for a reliable reconstruction. It depends on the number of columns in the sheets. The mode number of columns per sheet in Qumran is between 3 to 4, but this number may rise up to seven columns per sheet (1QapGen, 1QpHab) and cases of one column per sheet are also known (4QDeutn, 4QDa).20 Within the fragments that contain margins, one should therefore look for stitching holes or for the actual thread at the edge of the fragment. Further, oblique indications for stitching may arise from guiding dots that appear in a vertical line at the beginning and ends of sheets and assist the scribe in the drawing of lines. If not the seam itself, one may sometimes see vertical abrasion of the leather caused by the press of the seam on the verso of an inner layer or on the recto of an outer layer, which designates the existence of stitching either at the end or the beginning of the adjacent column, depending on the way the scroll was rolled.21

Finally, the edges of a written sheet may be discerned by an unusual width of the last column. A relatively wide or narrow column in comparison to the other columns of the sheet may be a result of the scribe’s incapability to divide the sheet into even columns, due possibly to miscalculation.22 A similar phenomenon may be evident at the beginning of sheets as well.23

The column is the fundamental building block of the scroll. Its characteristics, i.e., the column width and height, the size of the margins, and the position of the stitching cord, are essential for a stable and valid reconstruction of the scroll. This chapter highlighted how this essential information can be retrieved and serve the material process. Having established the measurement for every column and sheet, the scholar may proceed to the next step in the material and digital reconstruction.


The digital canvas is in fact a modern application of the method presented by Steudel, “Assembling and Reconstructing,” 516–34. For a more recent digital reconstruction see Torleif Elgvin, “1QSamuel – A Pre-Canonical Shorter Recension of 2Samuel,” ZAW 132 (2020): 281–300. This reconstruction was carried out using fonts designed at the Haifa project for the DSS as is duly acknowledged. We thank Prof. Elgvin for sharing his article with us before publication.


Tov, Scribal Practices, 77–84.


Herbert, Reconstructing Biblical Dead Sea Scrolls, 21–26, conveys a method for calculating a “scribal margin policy” for each scroll, offering helpful hints for where the margin can be expected.


Tov, Scribal Practices, 82–83.


The mode number is more indicative in this case than the mean number because exceedingly short or long lines, such as lines ending with vacat, may dramatically affect the mean width and provide a false estimate.


Herbert, Reconstructing Biblical Dead Sea Scrolls, 5–26, 34–62.


See Herbert, Reconstructing Biblical Dead Sea Scrolls, 5–26. The six steps include: Assessing average width for every letter according to its occurrence in the scroll; employment of vertical dividers for a more accurate assessment of column width; calculation of “critical deviation” for overruling reconstructions that exceed a 5% deviation from the expected deviation; identification and calculation of margins when they are not preserved; developing a scribal margin policy to assess the location of the left margin; and finally, analyzing the columns according to their specificities.


When the specific habits of the scribe of one particular scroll are known, they should of course also be followed in the reconstruction, for example in scrolls that use indentation at the beginnings of paragraphs, as in 1QS. Elgvin’s reconstruction of 1QSamuel incorporates the indentations into his reconstruction in an effective example of using computer fonts and digital canvasses. Even more conspicuously, vacats seem to operate as thematic markers (rather than mere technical spaces) in the pesharim; see Bronson Brown-deVost, Commentary and Authority in Mesopotamia and Qumran. JAJSup 29 (Göttingen: Vandenhoeck & Ruprecht, 2019), 45–51, 232–33; Gregory H. Snyder, “Naughts and Crosses: Pesher Manuscripts and their Significance for Reading Practices at Qumran,” DSD 7 (2000): 26–48. In such scrolls, vacats should naturally be placed in the reconstruction accordingly.


Carol A. Newsom and Yigael Yadin, “The Masada Fragment of the Qumran Songs of the Sabbath Sacrifice,” IEJ 34.2–3 (1984): 77–88.


For example, Daniel Falk, Daily, Sabbath and Festival Prayers in the Dead Sea Scrolls, STDJ 27 (Leiden: Brill, 1998), 38, 60; Strugnell and Harrington, DJD XXXIV, 4.


Herbert, Reconstructing Biblical Dead Sea Scrolls, 60–62; Puech, “Édition et reconstruction des manuscrits,” 111. See also Brown-deVost, Commentary and Authority, 52, n. 139. Bronson Brown-deVost demonstrates that when the width of the line is known, the number of missing letters and words can be estimated quite accurately.


For an example see Eibert Tigchelaar, “הבא ביחד in 4QInstruction (4Q418 64+199+66 par 4Q417 1 i 17–19) and the Height of the Columns of 4Q418,” RevQ 18.4 (1998): 589–93, here 592–93.


Finding fragments from two consecutive columns may be achieved by searching among those fragments for areas of similarity in shape or damage patterns, which can be assigned to consecutive turns of the scroll, as explained in chapter 11.


Stegemann, “Methods for the Reconstruction,” 205.


Ben-Dov, Stökl Ben Ezra, and Gayer, “Reconstruction of a Single Copy,” 33–34.


In rare cases the difference can grow up to five lines, depending on the size of the sheet. See Tov, Scribal Practices, 93–95; Stegemann, “Methods for the Reconstruction,” 198. In 11QTa, for example, the number of lines per column ranges between 22 and 30 lines (Qimron, The Dead Sea Scrolls, 1.137), and in 1QIsaa the numbers vary between 28–32 lines.


See the experiment report in chapter 10.


For a detailed analysis of technical aspects of margins see Tov, Scribal Practices, 99–104.


See for example: 1QS in many instances, 4QEnc ar (4Q204) VI 9, 4QapocrDan ar (4Q246), ii 4, 1QSamuel; see Tov, Scribal Practices, 146.


For a detailed description see Tov, Scribal Practices, 80–82; Stegemann, “Methods for the Reconstruction,” 197–98 n. 70–77.


See for example 11QTa col. XLIV; Stegemann, “Methods for the Reconstruction,” 212, n44.


See for example the change of column width in the final column of fragment 4Q416 2 i–iv.


Tov, Scribal Practices, 83; Stegemann, “Methods for the Reconstruction,” 198.

  • Collapse
  • Expand