Chapter 16 Material and Digital Reconstruction of 4Q418a

In: Material and Digital Reconstruction of Fragmentary Dead Sea Scrolls
Eshbal Ratzon
Search for other papers by Eshbal Ratzon in
Current site
Google Scholar
Open Access

After editing each wad and fragment of 4Q418a it is time to present their original order together with the known text from parallel copies, based on the protocol presented in the methodological chapters of the present book (see chapter 12). The following reconstruction of the copy 4Q418a is not the final word in the reconstruction of Instruction, but rather only a skeleton, on which the reconstruction of other copies can rely (see chapter 17).

While the digital canvas suggests an exact placement of each fragment and a specific width for each column (figure 119), the accompanying comments give an additional range for these data. Providing a margin of error alongside the reconstruction is a methodology not previously used in the study of the DSS. We borrow it from the exact sciences and introduce it here as a proper way to present the uncertainty of any reconstruction. Establishing a margin of error requires quantifying all elements of the reconstruction using a numerical figure, a rather difficult and sometime counter-intuitive act. While most of the errors were calculated mathematically, some elements are impossible to express in numbers and could only be demonstrated verbally. Moreover, a certain subjectivity remains in the estimation of several components of the errors. Keeping that in mind, these reservations do not mean that the attempt to estimate errors should be abandoned as scholars should be as transparent as possible about their uncertainties. The data arising from the reconstruction of the other manuscripts of Instruction will allow us to narrow down this range, producing an increasingly accurate reconstruction.

1 The Reconstruction of 4Q418a: State of the Art

The main advancement in ordering the fragments and wads of 4Q418a was achieved by Eibert Tigchelaar.1 The fact that the fragments were preserved in wads indicates that deterioration took place while the scroll was still rolled, and that all fragments belonging to a respective wad were located on the same height in the scroll in consecutive turns. Since the fragments of each wad remained glued even after the rest of the scroll had vanished, the order of the fragments in each wad corresponds to their original order. Tigchelaar consulted the PAM images that documented the peeling process, achieving additional material and textual insights for the various fragments. He concluded that all extant wads were once part of a single stack of fragments, which originated from an area near the bottom margin. In addition, he successfully reconstructed the order of the wads as follows2:

Figure 99
Figure 99

Tigchelaar’s reconstruction of the order of fragments in 4Q418a

The beginning of the scroll stands at the right-hand side of the sequence, its end to the left. Capital letters indicate the particular wads, marked above the list of fragments that are included in that wad. In Tigchelaar’s notation, one question mark represents one potentially missing layer, while two question marks represent an unknown number of missing layers.3

According to Tigchelaar, the scroll was rolled in the normal way with the beginning of the composition to the outside. Therefore, the more a fragment is located in a lower layer when the wad is placed on a table, i.e., the more it lies to the outside of the scroll, the earlier it belongs in the composition.4 He also established the height of all columns as 36–37 lines.5 The reconstruction suggested hereby confirms this number, which is higher than the average of 20 lines per column in the DSS corpus.6 However, as demonstrated in Appendix 1, an average error of ±7% may be expected for the number of lines in a column, which amounts to 2.5 lines. Thus, the number of lines in this scroll may be between 34–39. Other copies of Instruction have fewer lines per column (21 in 4Q416, 28 in 4Q417 and 4Q418). Variation of the number of lines in copies of the same composition is frequent in other DSS.7 According to the current reconstruction, these 36–37 lines add up to 23.3–23.8 cm, and including the potential error of the height the range is between 22–25 cm (see Appendix 1). No upper margins were preserved, and the bottom margins that were preserved are broken. Thus, we have no way of measuring the full height of the sheets. Tov’s data shows that upper and bottom margins in Qumran are usually between 1.5 and 2 cm, but larger margins of up to 8 cm exist too.8 Thus, an educated guess would point to a sheet height of around 24–27 cm or slightly higher.

In addition, we were able to determine that the width of the columns of the scroll varies between 40 and 60 letters per line or between approximately 9 and 12.5 cm, with an average column of 10.7 cm and a standard deviation of 1.1 cm. This fits Davis’s note that most large scrolls, measuring over 30 cm in height, have columns 10–13 cm wide.9 According to the present reconstruction, the width of the intercolumnar margins within sheets is approx. 0.8 to 1.5 cm, with an average of approx. 1 cm and a standard deviation of 0.25 cm. The width of intercolumnar margins between sheets is 1.3 to 3 cm including the stitches, with an average of 2.3 cm and a standard deviation of approx. 0.8 cm.10 The number of columns from the beginning of the scroll to the last preserved fragment (frag. 1, which is the first layer of wad A) is 20, but crumbles of skin attached to the last fragment (named here fragment 0) testify to the existence of at least one more column. As mentioned in chapter 12, it is impossible to determine the length of the rest of the scroll with any accuracy. The results can be seen in figure 119.

In addition to establishing the above data, we were able to answer some of the questions left open by Tigchelaar:

  1. Only one column preceded the column attested in fragment 12.

  2. Fragment 20+21 belongs to the left of fragment 9, and is joined to another fragment that was originally in wad B on top of fragment 9, to which we assign the number 9b.

  3. Fragment 22 is in fact a wad, containing several new fragments layered underneath. Although they cannot be separated without damage, the examination of this wad sheds more light on the length of the scroll.

  4. Several more hitherto unknown fragments remain attached underneath other fragments of wads A, C, and D, and on top of wad B. Some of them are legible.

  5. Fragment 19 follows immediately after fragment 22.

  6. At least one layer is missing between wad A and wads D+C. It may be found underneath fragment 7, the last fragment of wad A.

  7. In addition, we offer some corrections to the number and order of fragments in wad A.

Incidentally, the chain of events that took part as we created this reconstruction is quite significant. After the present reconstruction had already been finalized, an examination of earlier images yielded two more fragments that were once attached on top of wad B. In an astonishing way, these new fragments constituted a perfect match to fragments 20+21 and 22c, according to the place assigned to them in the reconstruction. Not only did the boundaries of one fragment (9b) match the boundaries of fragment 20+21, but also the remains of a right margin on the second fragment (9a) fell into place exactly where our reconstruction predicted it will be (see chapter 15). This new find significantly buttresses our suggested reconstruction.

2 The Wads

We will now present our digital and material reconstruction of 4Q418a with the additions and improvements of the work of previous scholars. While the methodological chapters of this book discussed each topic separately to facilitate understanding, in reality the order of work is not as clear-cut. We therefore first present each wad and the information learned from it, including textual parallels to other copies, identification of additional layers, joins, margins, and other clues that helped the reconstruction. This presentation includes analysis related to the above methodological chapters: “Recreating Single Columns Based on Fragments and Parallels” (chapter 9) and “Damage Patterns” (chapter 11). After the presentation of the wads, we return to the order of the methodological discussion above. We begin with wads D and C, which include the anchor fragments. They are placed approximately in the middle of the reconstruction. We then proceed with wads E and B to their right, which include additional important information that validates the exact place of the fragments. We finally end with wad A to the left of the scroll, with its own information.

2.1 Wad D (fragments 15–19; PAM 41.891, 41.909, 41.973, 41.997, 43.687; IAA Plate 511, fragments 14–16)11

Wad D contains several textual overlaps with other copies, which makes it the key for the entire reconstruction. Its most important parallel is 4Q416 2 i–iv, the largest fragment of Instruction, preserving four consecutive columns. Due to its exceptional size and the large amount of nearly uninterrupted text, this fragment is the main anchor of the textual reconstruction, in relation to which other fragments are placed (see images: B-496201, B-499639, B-499640).12

Figure 100
Figure 100

The text of 4Q416 2, 4Q417 2, and 4Q418 7–10 cast in the layout of 4Q418a, Columns IXXII. The continuous text parallels fragments from both wads D and E. Arabic numerals designate fragment numbers; Roman numerals designate column numbers.

© IAA, LLDSSDL, Shai Halevi and Najib Anton Albina

Two fragments from wad D find parallels in 4Q416 2. Fragment 19 (layer 5) overlaps with 4Q416 2 ii 14–16 and 4Q417 2 ii 18–21. Strugnell and Harrington suggest a second overlap between frag. 18 (layer 4) and 4Q416 2 iv 3–7. Although column iv of 4Q416 2 is preserved in a fragmentary shape, and the overlap is limited, the text of both copies fits well, and there are no attested variants between them. The editors’ suggestion was accepted by both Tigchelaar and Qimron (both offering slightly different reconstructions of the missing text).13 If they are correct, it means that the upper layers (which came from the inner parts of the scroll), belong to a later part of the composition. This is the indication that 4Q418a was rolled in the normal way, with its end inside. The parallel text reveals the number of letters in the columns to which fragments 18 and 19 belong (40–50 letters per line, a rather large variation), and hence their width and the distance between layers. Below we will explain Tigchelaar’s use of this parallel for reconstructing the height of the column.

Another important parallel within wad D is that of frag. 15 (layer 1) to 4Q415 11 3–4 and 4Q418 167b 3–4 (see figure 101 below).14 This parallel fills 11 additional lines in the textual reconstruction, determines the width of column XIV of our reconstruction, and affects the width of other columns around it.

2.2 Wad C (fragments 13, 14, 14a; PAM 41.410, 41.972, 43.687; IAA Plate 511, fragment 13)15

Fragment 13 (layer 1) overlaps 4Q418 167a 5. Since frag. 15 (wad D, layer 1) was shown above to overlap with the same parallels, it results that frags. 13 and 15 are distant joins. Fragment 13 contains a bottom margin, hence both wads C and D belong to the bottom of the scroll. Consequently, frag. 14 (layer 2) should also be joined to the bottom of frag. 16 (wad D, layer 2).16 As explained in chapter 15, by checking the verso of fragment 14 and filtering the new IR image provided by the LLDSSDL, we were able to detect two additional layers (14a and 14b).17 14a should be joined to the bottom of frag. 17, while 14b should be joined to the bottom of frag. 18. Indeed, the text of 14b matches the text known from 4Q416 2 iv that should precede frag. 18.

Figure 101
Figure 101

Fragments 15+13, 16+14, 17+14a, and 18+14b, columns XIXIV

© IAA, LLDSSDL, Shai Halevi and Najib Anton Albina

2.3 Wad E (fragments 22, 22a, 22b, 22c; PAM 42.247, 43.687; IAA Plate 511, fragment 19)

Our examination of the images of fragment 22 discovered that more fragments are attached underneath it, constituting a fifth wad (E). This wad contains four layers. Unfortunately, according to the professional estimation of the conservation team of the Israel Antiquity Authority (IAA), it is impossible to peel the layers of wad E since it would completely ruin the upper layer and perhaps also the following ones. The peeling would have allowed us to read the text on each fragment, confirming the exact number of layers. At the present, a thorough examination of the images reveals a few letters on two additional layers as well as a left margin on the second layer of this wad, underneath frag. 22.18 The left margin served as an anchor for placing frag. 22a and for the entire reconstruction (see below).

Fragment 22 is similar in shape to wad D, indicating that they were once piled together. It also overlaps 4Q416 2 i 7–10 and 4Q417 2 i 12–17. As Tigchelaar correctly calculates based on these parallels, 32–33 lines of text are missing, both between frag. 18 and 19 and between frag. 19 and 22. Tigchelaar concludes that “this number together with the 4 lines of the fragments, suggests we are dealing with a manuscript with a column height of either ca. 36–37 lines or ca. 18 lines.” The first option pertains if there was one column between each of these respective fragments, while the second indicates two intervening columns. Eventually Tigchelaar prefers the former, because frags. 9, 10, and 11 (wad B, layers 1–3), coming from an outer part of the scroll, all contain right margins and are thus one column apart (see below).19 Our digital reconstruction supports Tigchelaar’s call. The attempt to digitally create a two-column gap between two layers generated a too-wide distance between the fragments. Tigchelaar was uncertain whether an additional layer was missing between frag. 19 and 22. After determining the height of the columns, our trial and error of the material reconstruction did not permit an additional layer.

2.4 Wad B (fragments 9–12; PAM 41.410, 41.965, 41.972, 41.997, 43.687; IAA Plate 511, fragments 5, 10–12)

4Q418a frag. 11 (layer 3) overlaps with 4Q417 1 i, which in its turn parallels 4Q418 43+44+45 i–ii (figure 103).20 This is a very important overlap, because 4Q417 1 is considered to be part of the prologue of the entire composition.21 The overlap places wad B at the beginning of the scroll.22 Since the fragments of wad B contain bottom margins, it is evident that wad B originates from the same horizontal height as wad D+C. The similarity in the shape of the boundaries and cracks between frag. 11 and frag. 22 shows that wad B and E were indeed part of the same pile, and that wad B should be placed before wad E. This idea was raised by Tigchelaar, who proposes that all fragments of 4Q418a come from the same pile. Although he thinks that the evidence from the similarity of the boundaries and cracks of wad B is inconclusive, we support this evidence by means of the digital presentation (figure 102).23

Even stronger support is achieved by two previously unkown fragments that are visible on top of the wad in PAM 41.410. We name them fragments 9a and 9b. These fragments can be joined to the bottom of fragments 22c and 20+21 respectively.

Figure 102
Figure 102

Black: borders and cracks of frag. 22; Red: borders and cracks of frag. 11

Figure 103
Figure 103

Wad B. The text of 4Q417 1 and 4Q418 43+44+45+108 presented in the layout of 4Q418a

© IAA, LLDSSDL, Najib Anton Albina

2.5 Wad A (fragments 1–8; PAM 41.973, 41.997, 43.687; Plate 511, fragments 1–4, 6–9)

As explained in chapter 15, the earliest image of wad A already shows it at the stage after its separation into two piles. We were able to confirm that they both originated from the same wad in the documented order, with the exception of frags. 7 and 8, whose order should be reversed. In addition, remains of the upper part of frag. 6 are still attached underneath frag. 5. Our reconstruction anticipated an additional layer between wad D and wad A. This layer may still exist underneath frag. 7 (which we believe was the last layer of wad A). The ink from the recto of frag. 7 bled through the skin, but parts of it seem to be covered by the other layer (figure 104). Moreover, two separate layers seem to be visible on the bottom left edge of the verso (figure 105).

Figure 104
Figure 104

Left: IR image of 4Q418a 7 recto; Right: IR image of 4Q418a 7 verso, flipped and enhanced. The same word is circled in white both on the recto and on the verso. The horns of the א and the roof of the ת and ה are visible on the verso image, but the bottom parts of these letters are covered by another layer.

© IAA, LLDSSDL, Shai Halevi
Figure 105
Figure 105

Multispectral image of 4Q418a 7 verso. The area in the red circle is enlarged in the right image, showing two separate layers one on top of the other.

© IAA, LLDSSDL, Shai Halevi

As Tigchelaar notes, frags. 6–8 (layers 6–8) are nearly identical in shape with frag. 15 (wad D, layer 1). This indicates that wad A and D had once been part of the same pile. Because of the similarity to the first layer of wad D, Tigchelaar places wad A after wad D.24

Fragment 3 overlaps 4Q423 5. This parallel adds two more lines to the textual reconstruction below frag. 3 (layer 3), and eight lines to the top of the next column. Fragment 7 (layer 7) overlaps 4Q415 6. This overlap is significant for the reconstruction of 4Q415, as 4Q415 11 contains also a parallel to 4Q418a 15.25

2.6 Fragment 20+21 (Plate 511, Frag. 20+22; PAM 41.375, 42.760, 43.687)

Fragment 20+21 is the only single-layered fragment in 4Q418a. Its earliest image in PAM 41.375 clearly shows one fragment, which, having been broken during the 1950s, received two different numbers.26 The text on this fragment does not overlap with any known text from elsewhere in Instruction. Its shape was the first indication for its location. The borders of fragment 20+21 resemble the cracks of fragment 22.27 They are also as narrow as fragments 10–12 from wad B. Hence, a location between wad B and wad E seems like the best option. The borders of the previously unknown fragment 9b physically join fragment 20+21, which also confirms its position.

3 Reconstruction of 4Q418a

After discussing each wad separately, it is time to return to the order of the reconstruction procedure as described in the first part of the present book.

Placing all the fragments on the canvas. All fragments were prepared according to the methodology described in chapters 3–9. This task includes applying filters, examining the verso of each fragment, choosing the best image, enhancing it if necessary, scaling, removing the background, digitally restoring the fragments, and reading them. When each fragment is ready, it can be placed on the canvas for the reconstruction of the entire scroll.

Creating a custom-made font. Due to the degraded shape of the fragments we were unable to choose suitable letters from the fragments of 4Q418a. Instead we used letters taken from 4Q418, whose paleography is nearly identical to that of 4Q418a. After creating the basic font, we measured all the spaces between words in the existing fragments of 4Q418a, calculated their average, and applied it to the space glyph in the custom font.

The text preserved on the fragments was then typed as discrete text layers, and placed in the foreground in order to be seen above the fragments in the electronic display. We tried to fit every letter to its corresponding letter in the image without interfering with the space between letters or letter width. We then collected pairs of letters whose kerning required adjustment. The adjustment of kerning was done specifically for 4Q418a.

Recreating single columns based on fragments and parallels. After choosing the font size to adapt the printed text to the letters seen on the fragment, we continued to write the rest of the text in the same font size. For the current reconstruction, we used Qimron’s edition for the text that was not preserved on 4Q418a. We did not include his reconstructions of the unpreserved text, and calculated these lacunae independently.28 Fragments 18, 19, 22, and 22a are the key fragments for this reconstruction. Fragments 18, 19, and 22 overlap the large fragment 4Q416 2 (combined with the text of 4Q417 2). Although we know from the parallel copy the number of letters in each line of these three fragments, no right or left margin is preserved on any of them. Thus, theoretically each fragment can be placed anywhere within its column (near the right margin, in the middle, or near the left margin). In practice, not every placement fits the alignment of the words as they were preserved on the fragment. Of note, frag. 19 was not easy to adapt. After ruling out several locations, the suggested reconstruction was the only one possible. For fragment 22a, however, we have no parallel text, but it does preserve a left margin, anchoring it at the end of column VIII. After all the trial and error, arranging the text of 4Q416 2 and 4Q417 2 in the layout and font of 4Q418a, reveals the following measurements:

  • Column XI, attested in fragment 18, holds 44–48 letters per line, corresponding to an average width of 95.4 mm, which gives a density of 0.46–0.53 letters per mm.

  • Column X, attested in fragment 19, holds 40–46 letters per line, corresponding to an average width of 88.6 mm, which gives a density of 0.45–0.51 letters per mm.

  • Column IX, attested in fragment 22, holds 54–64 letters per line, corresponding to an average width of 120.4 mm, which gives a density of 0.45–0.53 letters per mm.

The experiment described above (chapter 10) for the use of a font in the procedure of reconstructing the width of a column, showed an expected error of approximately ±4%, which corresponds to 3.8, 3.5, and 4.8 mm respectively. This error should thus be assumed for columns IXXI.

The height of the columns was established by adjusting the rest of the text of 4Q416 2 and 4Q417 2 into the layout of 4Q418a between the parallel fragments. The text fits into 36 lines between fragments 18 and 19, and 37 lines between fragments 19 and 22.29 This difference falls within the margin of error for a font-assisted reconstruction of column height (see chapter 10). Consequently, we cannot be certain if it reflects the irregularity of the scribe’s handwriting or a difference in the height of the two columns. If indeed there was such a difference, in most scrolls all columns begin at the same height, but sometimes a line is supplemented at the bottom of the column. We therefore added that additional line at the bottom of column IX under fragment 22.30

Damage patterns. The damage patterns in this scroll are in fact the wads, whose analysis was already recounted in great detail. To conclude, after establishing the order of wads and the order of fragments within each wad, we can now order the fragments of the entire scroll, as seen in figure 106:31

Figure 106
Figure 106

The suggested order of fragments in 4Q418a

Having established the order of the fragments, we can now proceed to calculate the distances between them.

4 Placing the Fragments on the Canvas Using the Stegemann Method

In order to ascertain the distance between these fragments, an additional datum is required: the width of the intercolumnar margins. As can be seen from other scrolls that preserve several consecutive columns, the width of columns and margins is not uniform throughout a scroll. Average margins are between 10–15 mm.32 This variety within the same scroll allows some liberty in the reconstruction within certain limits, while at the same time calling for methodological prudence.

Two fragments preserve a large portion of margins: fragment 9 and fragment 5. The former also contains traces of stitches between sheets, and its width is 6 mm from the end of the line to the stitches. Unfortunately, neither of these fragments contains text from both sides of the margin, and therefore we lack their full width. The margin in frag. 5 is nearly 25 mm wide, a very wide margin compared to other scrolls,33 but later in the reconstruction process this intercolumnar margin proved most probably to belong to a space between sheets, which explains its excessive width. Following the methodology of trial and error, we tested several possibilities for the width of the margins between the anchor columns of 4Q418a, many of which contradicted some of the data on the preserved fragments. Eventually, the chosen width was set according to the restrictions of each column, as will be detailed below.

We set frag. 22a on the left margin of its column, and frag. 19 around the middle part of its column, the only possible place for it. The margin between columns VIII and IX was thus set to be 15.5 mm, and that between columns IX and X to 22.3 mm. The latter is wider since it lies between frags. 22 and 19, which belong to two different wads (E and D respectively). In this point we make the plausible assumption that the wads were separated in the places where the stitches used to be. The stitching eased the tightness of the roll and prevented the layers from attaching to each other. If indeed frags. 22 and 19 belong to two different sheets, one can expect the margins between sheets to be larger on average. Thus, the distance between frags. 19 and 22a, which includes the width of columns IX and X and the intercolumnar margins to their right is 211.5±10 mm.34

In order to place the rest of the fragments, we used the Stegemann method to calculate the incremental growth of distances between consecutive layers. The increase is dependent on the thickness of the skin and the tightness of the rolling. While we were unable to measure the thickness of the fragments, our examination in the IAA lab showed that their skin is thin compared to other scrolls. Since there is meager information about the thickness of the scrolls in general, we relied on the little available information in addition to crude estimations. According to Stegemann, the thickness of the Temple Scroll, which is one of the thinnest known scrolls, is 0.16 mm. He calculates its increase from one layer to another as about 1 mm, a rather tightly-rolled scroll. He describes the general picture as:

The circumference of a layer may increase by rates varying from about 1 mm, as in the Temple Scroll, up to about 5 mm, as in one of the manuscripts of 4QAngelic Liturgy, partly published by J. Strugnell in 1959, or in 4Q504 (4QDibHama), published by M. Baillet in 1982. Most of the leather scrolls, however, have rates of increase of about 2 or 3 mm, as the quality of the leather was only of medium grade.35

The skin of 4Q418a is very thin. From the fact that the fragments were glued to each other in wads we conclude that the scroll was rolled tightly. We therefore used an increase rate of 1.5 mm. Based on Ratzon and Dershowitz’s examination of scrolls that were preserved comparatively intact, the error for the differences is 1 mm (see Appendix 3).36 Thus, we placed frag. 22 between frags. 19 and 22a: 105±5 mm away from frag. 19, and 106.5±5 mm from frag. 22a.37

Choosing the points of recurring damage from which to measure the distances can affect the measured value. We examined the PAM images to decide which point of a certain layer was placed on the layer beneath it, also based on the similar cracks of consecutive fragments. In the few cases where this was not possible, we chose a point on the borders of the fragment in a place that seemed most similar to the borders of the other fragment. While the Adobe InDesign measuring tool gives a very high precision, the human factor gives a measurement error of approximately 0.5 mm.

Since frags. 19–22a served as the anchor for this reconstruction, we placed all the fragments to the left of frag. 19 in decreasing distances, and the fragments to the right of frag. 22a in increasing distances. The distance of each fragment from frags. 19 or 22a to the left and right respectively (Sn) may vary in a certain range because of the potential error, as explained in Appendix 3, and will be demonstrated in the following excursus.

5 Excursus: Calculating the Error for the Distance of Each Fragment from the Anchor Fragments and from Its Consecutive Fragment

The error for the distance of each fragment is calculated by means of the following equation (Appendix 3):

The error for the distance between two consecutive fragments is:

In this expression, n is the number of fragments from frags. 19 or 22a to the left or right respectively.

a1 is the distance between frags. 19 and 22 for the computation of the fragments to the left (105 mm), and the distance between frags. 22 and 22a for the computation of the fragments to the right (106.5 mm).

Δa1 is the error for the above distances, which is 5 mm.

d is the difference (the increase or decrease) of the distances between consecutive fragments (1.5 mm).

Δd is the error for the difference (d), which is 1 mm.

Δn is the error for the number of layers. Within the same wad this error is zero, but when moving from wad to wad, we assume a potential error of 1 layer. Larger errors are less probable. Therefore, for fragments within wads D and E the error is 0 and for fragments from wad A the error is 1. As the position of wad B was confirmed by the join to wad E, the error for its fragments is also 0.

While it is possible to calculate up to several digits after the decimal point, this kind of precision is meaningless in the present case. The values should normally be rounded to 10% of the magnitude of the potential error. Nevertheless, as we set the fragments in an increasing/decreasing distance of 1.5 mm from each other, we keep that precision, in order to make our work clearer to the reader. We gave the possible range of each result in parenthesis in rounded numbers.

In the following tables we give the distances of each fragment from their consecutive fragment and from the anchor fragment (frag. 22) and the respective error:

Table 17
Table 17

Positions of fragments to the left

Table 18
Table 18

Positions of fragments to the right

As seen in Tables 17 and 18, the computed length of the right part of the scroll from fragment 22 to fragment 12 is nearly one meter, and from fragment 22 to the left of the scroll until fragment 1 the length is over 1.3 meters. Thus, the entire preserved scroll between fragment 1 and 12 is over 2 meters long. Taking into account the possible error, the length of the preserved scroll is between 2–2.5 meters. The original scroll was probably longer, as it definitely contained at least one more column at its beginning and end. The potential error of the positions for each fragment varies between 4.7–19.4%. It is larger when the number of missing layers is not exactly known. Nevertheless, this error is considerably lower than the huge errors measured by Ratzon and Dershowitz for the reconstruction of the length of a scroll using the Stegemann method.38 The reason for better certainty in this case is the knowledge we possess regarding the number of turns in the preserved part of the scroll, as each layer constitutes one turn. Moreover, not every position within the potential range is possible, since some of them may lead to contradiction with other data, such as the existence of parallels and intercolumnar margins. In addition, one cannot simply change the position of one single fragment. The fragments’ positions are co-dependent. This error is thus only a starting point, which will be reduced later in the process. Nevertheless, it may change the reconstruction to include one more or less column in every direction. It is important to be aware of these margins of error for future reconstructions of other copies of Instruction. The restrictions deduced from the data of other manuscripts are expected to further narrow down the range of possible reconstructions.

6 Columns and Margins

The positions of the fragments according to the table suggest the width of each column. Here the calculated measurement is affected by certain limitations found on the actual fragments, such as the existence of real (i.e., not calculated) margins and of a parallel text. The width of the columns and margins together with their potential errors and possible ranges are given in Table 19 and will be followed by detailed explanations. Note that the upper and lower limits of each value are not always equal, depending on the data preserved on each fragment. All numbers are rounded up to one millimeter.

6.1 Considerations Underlying the Data in Table 19

The following section explains the reasoning for the width of the columns and margins, both to the right and to the left of columns IXXI. The latter columns served as the anchor for the entire array of measurements, as explained above.

In the discussion below we explain the considerations for the width of every column. A separate paragraph will state the error that could be expected for that column and the factors that led to this calculation. We distinguish intercolumnar margins within sheets from those margins that stand between sheets. The average of each group is calculated separately. The standard deviation from that average is taken to be the margin of error for each group. For the width of the columns we had more information, and were hence able to give a better estimation for the error of their width. There is often more than one way to compute the margin of error, and some of our estimations might be judged differently by other scholars. The most important point to be taken from the below explanations is the order of magnitude of the errors and the kind of considerations accompanying them.

6.1.1 The Left Part of the Scroll

To the left of columns IXXI, the anchor of our reconstruction, some of the fragments of wad D and all the fragments of wads A and C are placed. Several restrictions exist on the reconstruction of the left part of the scroll: the preservation of intercolumnar margins on fragments 5 and 1, and the fact that frags. 13 and 15 are paralleled in 4Q415 and 4Q418. This parallel determines the length of the lines of column XIV.39

6.1.2 Columns XIIXIII

Since the preserved fragments 14, 14a, 16, and 17 do not provide any information on how to divide their columns, we set them to the same width as column XI, with equal margins to the margin between columns X and XI (figure 107). Evenly dividing the uninscribed length between fragments 15 and 18, which do have textual parallels to indicate their columns’ width, gives the same result.40 Support for this decision comes from the fact that it allowed an easy reconstruction of column XIV with the parallel text to frags. 13 and 15.

Figure 107
Figure 107
Columns XII and XIII
© IAA, LLDSSDL, Shai Halevi and Najib Anton AlbinaOnline version available at

The error for the columns’ width results from all the used data: the errors for the distance between fragments 15 and 18, the parts of columns XI and XIV contained within this distance, and the width of the three margins in that area (XIXII, XIIXIII, XIIIXIV).41

6.1.3 Column XIV

Fragments 13 and 15 stand at the bottom of column XIV, with their text paralleling 4Q415 11 and 4Q418 167a+b. The combination of all three copies dictates the length of the lines in this column. The potential error derives only from the use of the font for the reconstruction of the text (figure 108).

Figure 108
Figure 108
Column XIV
© IAA, LLDSSDL, Najib Anton AlbinaOnline version available at

6.1.4 Column XV

Col. XIV (frags. 13, 15) is the last column with fragments from wads C and D. The subsequent columns are represented by wad A, where frag. 7 represents the next known layer, whose text may overlap 4Q415 6, but it is not possible to infer a length of a line from these fragments. The width of column XV should thus be deduced by other means. We know that frag. 7 was not part of column XIV, because the text of the lower part of that column is known from parallel copies. We thus expect that the next two layers both belong to column XV. The problem is that the text of frags. 7 and 8 is hardly comprehensive. Their vocabulary is different, and while frag. 7 uses the second person singular, frag. 8 uses the third person plural. The only solution to the problem is that another layer is missing between wads A and D. In fact, this layer may be present in an additional fragment underneath frag. 7 which should be named 7a, as was shown above in the discussion on wad A. Theoretically, it is possible to add more than one fragment to the reconstruction. We examined the option that two fragments existed between wad A and wad D (fragments 7a and 7b), and it did not lead to a contradiction. However, we prefer to choose the minimalist option that requires the assumption of as fewest unpreserved fragments as possible. The fact that a larger number of fragments could not be ruled out means that the certainty of the reconstruction from this point onwards is reduced, and another column may have existed between fragments 15 and 7.

According to the current reconstruction, fragments 7 and 7a belong to column XV, while fragment 8 belongs to the subsequent column XVI. The inclusion of the two layers (7 and 7a) in the same column requires a very wide column in addition to a very wide intercolumnar margin between cols. XIV and XV (figure 109). The margin of 28.5 mm between these columns is similar to the wide margin between columns XVII and XVIII; both instances stand between sheets. In the lack of additional information, we must estimate the error for column XV statistically, based on the standard deviation of the width of the reconstructed columns in this scroll.

Figure 109
Figure 109
Column XV
© IAA, LLDSSDL, Najib Anton AlbinaOnline version available at

6.1.5 Columns XVI and XVII

Fragment 5 contains a wide left margin, which means – when calculated in the larger framework of trial and error – that it must belong to the same column as frag. 6. Since the first letters on frag. 6 line 2 constitute the end of a word, a space should be assigned for the rest of the word and another word when placing the fragment, keeping the beginning of all lines aligned. Whether or not another fragment existed between wad D and A, exactly two columns must be reconstructed between the end of fragment 7 and the left margin of fragment 5, because, as explained above, fragments 7 and 8 cannot belong to the same column, and fragments 5 and 6 must belong to the same column (figure 110).

The minimum and maximum width of column XVII will be calculated separately. The minimum is between the margin on fragment 5 and two words away to the right of fragment 6. The uncertainty comes mainly from the possible range of the distance between the two fragments. The maximum can almost reach to fragment 8, leaving enough space in the previous column for the completion of the broken words appearing on the fragment, and subtracting the minimum width of the intercolumnar margin.42 Note that the maximum width of the column exceeds any reconstructed column and assumes that column XVII is very narrow, thus the chances for a very wide column decrease with its growth.

We allocated the rest of the space between cols. XV and XVII to col. XVI and the margins around it. The exact division between them may change.43 Thus, similarly to the method that was used for calculating the error of columns XII and XIII, here too the error takes into account the errors derived from the other known lengths: the distances between fragments 5 and 6 and fragments 6 and 8, the width of columns XV and XVII contained within this distance, and the intercolumnar margins between columns LXV and LXVI and between columns XVI and XVII.44

Figure 110
Figure 110
Columns XVI and XVII (fragment 5a is pasted on top of fragment 6).
© IAA, LLDSSDL, Najib Anton AlbinaOnline version available at

6.1.6 Col. XVIII

In the lack of sufficient information for columns XVIIIXIX, we start the calculation in column XX where more information is available, and calculate backwards. The results are nonetheless presented sequentially. Column XVIII is very narrow, nearly the narrowest reconstructed column. Its maximum width is restricted by its right-hand intercolumnar space, showing on fragment 5 and by the reconstructed beginning of the parallel text to fragment 3 (see below).45 There is no information regarding its minimal width. We therefore set it according to the standard deviation of the width of columns in this scroll (figure 111).

6.1.7 Column XIX

The text preserved on fragment 3 overlaps 4Q423 5. Since only the fragmentary line 34 is preserved on both copies, it cannot be used to reconstruct the width of the lines. Nevertheless, since at this point of the scroll the fragments stand quite close to each other, the circumference of each turn being rather short, column XIX was limited from both sides by fragments 4 and 2, which determine its upper margin of error (figure 112).46 There is no way to calculate the lower margin of error, hence we use the standard deviation.

Figure 111
Figure 111
Column XVIII
© IAA, LLDSSDL, Najib Anton AlbinaOnline version available at
Figure 112
Figure 112
Column XIX
© IAA, LLDSSDL, Najib Anton AlbinaOnline version available at

6.1.8 Column XX

Column XX is the last extant column and is attested by frag. 1. This fragment merely contains traces of ink together with a left margin. The words on fragment 2 begin at the same vertical line, which may indicate the beginning of the column, but one more word is also possible. The size of such a possible word is approximately the same as the margin of error for the distance between fragments 1 and 2 (14 mm), and should be added to the calculation of the lower error.47 The upper margin of error is based on the uncertainty of the distance between fragments 1 and 3, and the fact that the text of fragment 3 is not part of the same column as fragments 1 and 2.48

Figure 113
Figure 113
Column XX
© IAA, LLDSSDL, Najib Anton AlbinaOnline version available at

6.1.9 The Right Part of the Scroll

Since the reconstruction of the entire scroll depends on the anchor columns IXXI, we calculate the measurements of columns IVIII backwards from column VIII to column I.

6.1.10 Column VIII

Fragment 22a is the second layer of wad E, but it is only partly visible as the bottom layer in the recto images of frag. 22. Inspection of these images shows the left end of a column on frag. 22a. A further layer of that wad, frag. 22b, is only visible on the verso images of frag. 22. A few letters are visible on new images of the verso provided to us by the IAA. However, most of the text of this layer is covered by a yet additional fourth layer (22c). Only in cases where 22c is broken, is the ink of 22b visible. The traces on 22b must be part of the same column of the letters visible on fragment 22a (figure 114). When placed in the reconstruction, the distance between the left end of the letters on fragment 22a and the right end of the letters on fragment 22b is 107.6 mm. The result is a rather wide column, albeit not the widest in 4Q418a (the widest is 128.2 mm in column XV). It is similar in width to column III, which is reconstructed based on a parallel text.

The only error for the width of this column is derived from the uncertainty of the distance between the fragments, but the placement of the fragments is supported by the join of fragments 9a and 9b with 22c and 20+21. The error is probably very low, and cannot be estimated.

Figure 114
Figure 114
Reconstructed borders of fragments 22a and b and the reconstructed distance between them.

6.1.11 Columns VIVII

Fragments 22c and 20+21 belong to two consecutive columns (figure 115). The joined frag. 20+21 shows the ends of several words, hence it is at least a few letters away from the left margin of column VI. From the right edge of column VIII to the approximate right edge of column VI we measure 252.5 mm. Since there is no additional information about the content of frag. 22c, we simply divide this width evenly between columns VI and VII.

After the reconstruction was performed, we discovered fragments 9a and 9b on top of wad B. The location of fragment 9a was calculated from the right according to its distance from fragment 9. It joins to fragment 22c. Luckily, the word preserved on fragment 9a begins exactly where we reconstructed the beginning of column VII, thus supporting the reconstruction, fixating the fragments of wad E to their place, and limiting the margin of error. Column VII can only be slightly narrower depending on the width of the margin between columns VII and VIII. Its maximum width depends also on the width of its right margin and the minimum width of column VI.49

Column VI includes fragment 20+21. The completion of broken words to its right demands at least a word or two to its right until the beginning of the column. Its minimum limit is affected by any addition to column VII and the margin between columns VI and VII. In the absence of more information regarding the possible addition to column VII, we set the minimum of column VI based on the standard deviation of column width in this scroll. Its maximum width is affected by the minimums of the margins around it and the minimum width of column V.50

In the proposed reconstruction, columns VI and VII are quite wide, whereas column V is very narrow. While the paucity of information created large errors that allowed even wider measurements for columns VI and VII at the expense of column V, we find these options to be less likely. The probability of the potential reconstructions will decrease with the increase of the disproportion between the columns.

6.1.12 Column V

The right margin preserved on fragment 9 determined the right end of column V (figure 116). It is narrower than its surrounding columns, but has approximately the same width as the known column XI. Fragment 9 preserves stitches. It is quite normal that the first and last columns in a sheet are narrower or wider than the rest of the columns in that sheet.

Figure 115
Figure 115
Columns VI and VII with fragments 22c+9a and 20+21+9b
© IAA, LLDSSDL, Najib Anton AlbinaOnline version available at
Figure 116
Figure 116
Column V with fragment 9
© IAA, LLDSSDL, Najib Anton AlbinaOnline version available at

The maximum width of column V depends on the error of the distance between fragments 9 and 20+21 and the minimum size of the margin between their columns.51 There is not enough information to set a minimum for this column. We thus use the standard deviation.

6.1.13 Column IV

Fragment 10 contains a right margin without stitches. The margin between columns IV and V contains stitches that connect two sheets, as fragment 9 indicates. The margin to the left of the stitches is 6.5 mm wide. We assumed that the margins on both sides of the stitches were equal (figure 117). Additionally, we equated the width of column IV to the known width of its preceding column (III). These assumptions are not always true, but in this case they fit well with the distances established by the overall reconstruction.

Figure 117
Figure 117
Columns III and IV
© IAA, LLDSSDL, Najib Anton AlbinaOnline version available at

The error for the column’s width derives only from the size of its left margin, whose upper margin of error was determined based on the standard deviation of the width of intercolumnar margins between sheets. The lower margin of error was set to be only 10 mm, as this intercolumnar margin is already the narrowest of its kind in this scroll.

6.1.14 Column III

Fragment 11 parallels 4Q417 1 i 21–24. Casting the text of 4Q417 in the layout of 4Q418a reveals that column III contains approximately 55–60 letters per line, reaching a width of about 105 mm, with an average intercolumnar margin of 10 mm. The size of this margin is close to the average in the DSS.

The error for the width of this column is derived from the error of using a font for the reconstruction. As the fragment contains a right margin, this error can only affect the margin between column III and IV. This outcome dovetails with the present reconstruction. Had the distance between frags. 10 and 11 turned out to be smaller, the text of column III would have intruded into column IV and overlapped frag. 10. Otherwise, had this distance turned out to be larger, the resulting margin would have been too wide to an unattested measure. Thus, the preservation of intercolumnar margins in several consecutive fragments and the parallel text from 4Q417 1 aggregate to constitute a crucial point for validating the entire reconstruction.

6.1.15 Column II

A small trace of ink on the right edge of line 3 of fragment 12 demonstrates that a space for at least one more word is missing to the right of the fragment before the right margin. This means that column II was comparatively wide (112.5 mm), but again not the widest one (figure 118). Within the current reconstruction, it is possible to narrow the margin between columns II and III, while widening column II and vice versa.

Regarding the error, there is not enough information to limit the highest potential width of the column, and we thus derive it from the standard deviation. The lowest possible width is based on the distance of fragment 12 from the preceding fragment 11. Since the distance between fragments 10 and 11 was confirmed by the join of fragments from wads B and E, the error for the distance to fragment 12 is only 1 mm. The most contributing factor to the lowest limit for the width of column II is based on the maximal possible width of the intercolumnar margin between columns II and III.

6.2 The Beginning of the Scroll

No fragment of 4Q418a overlaps the text of 4Q416 1, which is commonly considered to contain the beginning of Instruction (see chapter 14). The first fragment of 4Q418a, fragment 12, does not show any parallels in other copies either. Fragment 11 parallels 4Q417 1. In their unpublished reconstruction, Steudel and Lucassen suggested that 4Q417 1 was an alternative introduction to the composition instead of 4Q416 1, and that both introductions were joined in the copy 4Q418. The main basis for this proposal is the fact that 4Q417 1 contains stitches, and it is unlikely that only one column was included in the previous sheet. Tigchelaar adds some more literary arguments in favor of their proposal, but eventually concludes that it must remain a possible but unproved hypothesis.52 Here we only address the material consideration, and demonstrate that materially the opposite hypothesis – that both introductory sections were included in both 4Q417 and 4Q418a – is also possible.

Figure 118
Figure 118

Column II

© IAA, LLDSSDL, Najib Anton AlbinaOnline version available at

The main objection mentioned above is solved if we assume that more text was included before 4Q417 1, between it and the text of 4Q416 1. The first sheet of the scroll 4Q417 would then have contained more than one column. In addition, the verso of 4Q417 1 shows an imprint of the ink from a previous turn. This turn cannot have originated within this fragment despite its excessive width, and must have therefore originated in a previous sheet.53

Collating the evidence from 4Q417 and 4Q418a will present us with the size of the missing portion of text at the beginning of the composition before 4Q417 1. The procedure runs as follows:

We calculate the length of the text at the beginning of 4Q418a based on a comparison to 4Q417. This section parallels frag. 11 (the second preserved fragment of 4Q418a), and according to its content stems from the first few columns of the composition. We assume that the scribes of both 4Q417 and 4Q418a started copying Instruction at the top of the first column of their respective manuscripts, as usual in the DSS corpus. The preserved text of 4Q417 1 begins in the first line of the column, meaning that 4Q417 contained an additional number of whole columns from the beginning of the known text to the beginning of the scroll. This copy holds 28 lines per column. It thus lacks 28n lines, with n a positive integer (or, in other terms, a natural number). In 4Q418a, on the other hand, the known text begins in line 11 of what we now call column III. The number of lines in this copy is 36. Thus, 4Q418a contained an additional number of lines expressed by 36m + 10 lines, with m also being a natural number, but unlikely the same one as n. While the number of letters per line in 4Q417 1 i and 4Q418a 11 is approximately the same, we cannot assume that this is true for all the rest of the columns, thus allowing the number of missing lines in both manuscripts to be close, but not necessarily exactly the same. Thus:

36m + 10 ≈ 28n

Let us now run some iterations of this equation:

Table 20
Table 20

Comparison of missing text at the beginning of 4Q418a and 4Q417

If 4Q418a 12 would have stemmed from the first column of the manuscript, 4Q418a would lack 46 lines from the cast text of 4Q417 1 to the beginning of the scroll, while 4Q417 lacks either 28 or 56 lines. This configuration would mean that the beginning of the composition was of different lengths in these two copies. If, however, another column existed before frag. 12, 4Q418a would lack 82 lines, while 4Q417 would lack three columns, amounting to 84 lines. The difference of 2 lines is negligible compared to the margin of error, and may also result from unequal column width in both manuscripts. Three columns preceding that of 4Q418a 11 is also a possibility, but a less likely one, since the number of lines lacking in this case is 118, while in 4Q417 four lacking columns add to 112 lines. We thus choose to reconstruct one missing column before frag. 12, which equals three missing columns of 4Q417 before its first fragment. This is a normal size of a sheet, which shows that both introductory sections – 4Q416 1 and 4Q417 1 – could have been included in 4Q417.

No fragment was preserved from the first column of 4Q418a. Assuming that it belonged to the same sheet as columns IIIV, we set it to be of the same size as column II. Setting the margin between columns I and II is based on a similar estimate. We do not know where exactly to locate the text of the beginning of 4Q416 1 in the first column of 4Q418a. 4Q416 1 contains 18 out of 21 lines of the column. The first line of 4Q416 1 may have originally belonged to the first through the fourth line of its column, although the first is very unlikely, as its content does not seem like the opening line of the entire scroll. In the present reconstruction we began the text in line 3 of 4Q418a, but it may have been located one line higher or lower.54 It is impossible to estimate the margin of error for the width of the first column, hence we use the standard deviation.

6.3 Sheets

Typically in the DSS, sheets contain 3–4 columns, but exceptional sheets may reach 1–7 columns.55 The number of columns per sheet in the suggested reconstruction of 4Q418a is as follows:

Table 21
Table 21

Number of columns per sheet in 4Q418a

The sheets of 4Q418a are comparatively large, which is not surprising considering the height of the columns. The stitches between the first and second sheets were preserved on frag. 9. We assumed that there were stitches between wads D and E, which caused the roll to be looser and the layers to separate. Again we reconstructed stitches between wads D and A. Their existence also agrees with the large margin between cols. XIV and XV. Wad A is obviously too large to be contained in one sheet. The large margin preserved on frag. 5 may indicate the end of a sheet.

7 Conclusion

The digital and material methods for DSS reconstruction have been exemplified in this chapter. We have offered a reconstruction and visualization of 4Q418a. This digital reconstruction supports the suggestions of previous scholars, answers some of the remaining questions, and adds several new insights. We were able to identify previously unknown fragments and incorporate them in the general reconstruction; we rearranged some of the already known fragments; presented the text from parallel copies in the layout of 4Q418a; demonstrated the possibility for placing all the preserved fragments in 23 consecutive layers; and established the number and size of the missing columns from the beginning of the scroll. Despite the large theoretical margin of error, the control points in the left and right parts of the scroll allow a larger degree of certainty, buttressing the basic reconstruction. The length of the preserved part of the scroll is between 2–2.4 meters, including the beginning of the composition. Fragment 1, the last preserved fragment in the scroll, does not constitute the end of the scroll. At least one more layer existed, indicated by the crumbles of skin attached to it (which we named fragment 0). The distance between frags. 1 and 2 indicates that there was enough space for more turns, but we cannot estimate how many. The methods presented in part 1 of this book also allowed us to improve the reading of 4Q418a in quite a few cases, discussed thoroughly in chapter 15. As it turns out, 4Q418a was a long scroll with exceptionally high columns. Instruction must have been highly cherished by the people supporting the production and penning of this scroll. The margin of error added to each stage of the reconstruction will allow scholars to correctly use this information for future reconstructions of other copies of Instruction.

Figure 119
Figure 119
Figure 119

A visualization of columns IXX attested in fragments 1–22c

© IAA, LLDSSDL, Shai Halevi and Najib Anton AlbinaOnline version available at

Tigchelaar, Increase Learning, 126–31.


Tigchelaar, Increase Learning, 130.


Tigchelaar, Increase Learning, 130.


Tigchelaar, Increase Learning, 127.


Tigchelaar, Increase Learning, 130–31.


Tov, Scribal Practices, 84–91. Tov classifies the scrolls that contain such writing blocks as “very large.” This group of scrolls contains mainly scripture. Scrolls with the same number of lines per column are 4QProva, 11QPsa, and 4QGen-Exoda. Tov debates whether or not the size of the non-canonical very large scrolls indicates their authoritative status.


Tov, Scribal Practices, 95–98.


Tov, Scribal Practices, 94.


Kipp Davis, “High Quality Scrolls from the Post-Herodian Period,” in Elgvin, Davis, and Langlois, Gleanings from the Caves, 129–138, here 130, n. 3. Note that the scrolls discussed in his paper are slightly later than 4Q418a.


This is an average width of margins compared to other DSS, see Tov, Scribal Practices, 103.


Fragments number 15, 16, and 18 have deteriorated and are no longer represented on the IAA plate.


For a textual reconstruction of the composite text of this fragment including several 4Q417 and 4Q418 fragments see Qimron, Dead Sea Scrolls, 2.152–57.


Strugnell and Harrington, DJD XXXIV, 490–91; Tigchelaar, Increase Learning, 133; Qimron, Dead Sea Scrolls, 2.157.


For the overlaps between 4Q415 11, 4Q418 167a+b, and 4Q418a 15+13, See Strugnell and Harrington, DJD XXXIV, 488; Tigchelaar, Increase Learning, 136; Qimron, Dead Sea Scrolls, 2.160.


Fragment 13 is lost, its only attestation being PAM 41.410.


Strugnell and Harrington, DJD XXXIV, 488; Tigchelaar, Increase Learning, 135.


We thank Eibert Tigchelaar for pointing out to us additional letters on the verso of frag. 14. He also spotted two more letters at the bottom of the verso of frag. 16, belonging to frag. 17.


Ratzon, “New Data,” 25–38.


Tigchelaar, Increase Learning, 129–31.


Fragment 4Q418 45 ii does not find parallel in any of the other copies. Since its text carries on to the consecutive column, it must be reconstructed in column IV, although its exact position and layout depend on the unknown width of both columns 4Q418 45 ii and 4Q417 1 ii. The text of 4Q418 45 ii is not attested in Qimron’s edition.


4Q417 1 was originally designated fragment 2 by Strugnell (Strugnell and Harrington, DJD XXXIV, 151, 169, 192). Elgvin, based on material considerations, claims that the original designation was correct, and that 4Q417 1 should be placed after 4Q417 2. See Elgvin, “The Reconstruction of Sapiential Work A,” 568–69; Elgvin, “An Analysis of 4QInstruction,” 12–18. However, Steudel and Lucassen (quoted in DJD XXXIV, 19) arrived at the opposite conclusion also on material grounds. The conclusion of the DJD editors is primarily based on the contents of 4Q417 1, which is more suitable for an introduction to the composition. Further, its content is similar to 4Q416 1; the latter fragment contains a wide right margin, suggesting it is the first column of a scroll. See also Tigchelaar, Increase Learning, 155–59. Material support for placing 4Q417 1 at the beginning of the scroll will be published as part of the edition of 4Q417.


In addition, Tigchelaar, Increase Learning, 54, proposes that 4Q417 1 ii parallels 4Q418 103 i, also bearing implications for 4Q418a frag 11. However, this parallel is based on a joined fragment to 4Q418 103 i that we do not accept.


Tigchelaar, Increase Learning, 129–30.


Tigchelaar, Increase Learning, 129. In fact, frags. 6–8 are also similar to frag. 22. This similarity may theoretically indicate that wad A should be placed between wads B and D, but for frags. 6–8 to be in proximity to frag. 22, one should assume that frags. 6–8 were placed above frags. 1–5, in an inner turn of the scroll. This can be ruled out, because as shown in the previous chapter, part of frag. 6 was torn away and is still attached to the verso of frag. 5, thus proving the documented order.


Strugnell and Harrington (DJD XXXIV, 480–81) proposed that 4Q418 103 ii is an overlap for frag. 4 (layer 4). Their proposal was accepted by Tigchelaar, Increase Learning, 137 and Qimron, Dead Sea Scrolls, 2.169. However, their claim that “the text of line 2 coincides almost exactly with that of 4Q418 103 4” is more of an overstatement, see chapter 15. For the implications of the present reconstruction on the reconstruction of 4Q415 see Dayfani, “Material Reconstruction.”


According to DJD XXXIV, 492, it once belonged to a wad, but this is probably a mistake, as discussed above. See Tigchelaar, Increase Learning, 138–39.


Tigchelaar, Increase Learning, 129.


In some cases, the newly computed size of lacunae contradicts Qimron’s reconstructions, but these cases will be published separately.


The space between lines in frag. 18 is uneven and narrower than the space between lines in frag. 19. We have noticed that the space between lines in frag. 18 grows narrower towards the bottom of the fragment, thus, it is likely that the spaces in the lines above were wider. Perhaps frag. 18 has shrunk near its bottom.


At this point, the fine-tuning of the experiment in chapter 10 proved significant. Before adjusting the space and the kerning to 4Q418a, the text between frags. 22 and 19 required only 36 lines, ignoring the 37th line required in column X. The right number was achieved only due to the accurate kerning, showing the significance of such adjustments for the reconstruction.


Fragment 20+21 is joined to fragment 9a from wad B, but was not preserved as part of wad B.


Tov, Scribal Practices, 103.


Tov, Scribal Practices, 103.


The error is composed of the uncorrelated errors of the width of columns IX, X, and their intercolumnar margins. The error for the intercolumnar margins was separately calculated for margins within sheets and those between sheets, using the average and standard deviation of each type. It is thus calculated: .


Stegemann, “Methods for the Reconstruction,” 195.


Ratzon and Dershowitz, “The Length of a Scroll.”


The error for the distance between frag. 22 and frags. 19 and 22a depends on the errors for the distance between frags. 19 and 22a, and the error for the increase of the distances between layers (d). The latter, however, is negligible, and we were thus able to divide the error equally between the two distances.


Ratzon and Dershowitz, “The Length of a Scroll.”


Parallel texts exist also for fragments 7 and 3. However, in these cases there is no sufficient information to determine the width of the lines.







Again the upper and lower margins of error are not equal:









Tigchelaar, Increased Learning, 191–192, where he also summarizes Steudel and Lucassen’s suggestion.


The details of this proof will be provided elsewhere with the full material reconstruction of 4Q417.


The widths of the first columns of 4Q416 and 4Q418a are approximately the same, thus the number of missing lines at the top of these columns should also be the same.


Tov, Scribal Practices, 75–76.

  • Collapse
  • Expand