The complete or almost-complete scrolls known to scholars began deteriorating while still rolled, thus creating recurring damage patterns, as is evident in large scrolls like 1QS and 11QPsa. These patterns constitute significant benchmarks for the material reconstruction, i.e., for finding the relative position of the fragments within a scroll. In this chapter we present the core of the Stegemann Method together with updates from recent scholarship, adjustments to digital tools, and a critical view of the method’s precision.1 Establishing the sequence of fragments used to be carried out with images of fragments pasted on semi-transparent paper, which was subsequently rolled.2 Today we do the same work digitally, but the main principle has not changed. Similar digital work has been carried out by some scholars for a while, and here we seek to stabilize a pipeline for it.
In standardizing his method, Stegemann offered a list of steps for reconstructing scrolls. Some of his points, like collecting the data and reconstructing the columns, were discussed elsewhere in this book, and are in fact used by all editors of the scrolls whether or not they use unique points of the Stegemann Method. The heart of the method, which differentiates it from all other methods, lies in the assumption that the scroll was deposited in the cave while still rolled, and began deteriorating in that condition. Since certain areas of the scroll were damaged by the same causes (humidity, animal activity, stress from cords, strings or stones, seam imprints, etc.), these areas may present similar damage patterns. In many cases, sections of several turns of the scroll remained stuck together even when the rest of the scroll decomposed. Stegemann offers a way for finding the fragments stemming from the same area of the scroll, and arranging them according to their original order, relying on the fact that in scrolls the outer circumference is larger than that of the inner turns. Thus, fragments with larger distances between similar damage patterns stem from the outer turns of the scroll. However, arranging the fragments in the inside or outside of the scroll does not necessarily indicate their order in the composition, since while most scrolls were rolled with the beginning on the outside, some were left rolled with the beginning on the inside.
The natural processes of shrinkage and detachment of parts of a given fragment may cause irregularities in the measured increment of gaps between respective damage patterns. The increment may stop or even be reversed in the middle of the incremental series, due to shrinkage at one specific spot, thus rendering the Stegemann method inaccurate.3 It is therefore preferable to run the procedure on old PAM images – provided they are properly scaled – rather than on the scroll in its present condition or on the new IAA images. Ratzon and Dershowitz have shown, however, that even the use of the oldest images does not entirely resolve the problem.4 Best results are achieved if at least four or five consecutive damage points are preserved (NB: not reconstructed but physically preserved), in which case the effect of skin irregularities is minimized. Thus, in the case where one increment is negative and the other is positive, the prudent scholar would be aware of the discrepancy and would seek more data to normalize. In order to choose the right increment, it is important to examine all images and detect places where the scroll has shrunk or where cracks affect the distance between damage points, rendering one of the measurements less reliable.
We divide the discussion into three main topics: 1) the direction in which the scroll was rolled, i.e., was it rolled in the normal way with its beginning on the outside or the other way around?; 2) the application of the method to cases in which fragments were preserved in wads (stacked in piles), for example in 4Q418a discussed extensively in the present book; and 3) the more common case, in which the fragments were found scattered and a spatial reconstruction is sought using the recurring damage patterns. We will not discuss here the more classical methods for the joining of fragments that are surveyed by Stegemann himself, since they were not in use in the current manuscript of 4Q418a. Reading this chapter assumes prior reading of several previous chapters: Verso (chapter 6), Finding Wads (chapter 7), and Recreating Single Columns (chapter 9).
1 The Direction in which the Scroll Was Rolled
The normal way of rolling would be with the end of the scroll at its inner part, so that the reader will encounter the beginning of the scroll, or the handle sheet, at its outermost turn.5 For a reader finishing a scroll, re-rolling it from the end towards the beginning would require some effort, but users usually took the time to do so. In some scrolls, however, this effort was not carried out and the scroll is rolled with its beginning at the inner part.
The direction of rolling can be determined in scrolls that show a significant series of damage patterns. Thus, for example in this image of a section of the scroll 11QpaleoLeviticusa (11Q1):
One can clearly see the recurring patterns of deterioration at the bottom of the scroll. These patterns are small and close to each other at the left end of the scroll, whereas the damages grow larger and the gaps grow wider apart at its right end, closer to the beginning. This is a sign of the “normal” state of affairs. The same is true for 11Q5 (11QPsa):
In contrast, as the scroll 1QS was found rolled “backwards,” the damage patterns are closer at the beginning of the scroll and grow wider at its end.
1QS and 1QM are the only large scrolls preserved while rolled backwards.6 While this situation can easily be discerned in well-preserved scrolls, it is quite difficult to discover in fragmentary scrolls.
Tov added a second indication for identifying the direction in which the scroll was rolled. He suggested that the end of the scroll, which was also probably located at the innermost roll, was thus protected from decay by the outer layers.7 Hence, if only the end of a composition is extant, one may posit that the scroll was properly rolled, and vice versa. This rule of thumb is applicable, however, only in cases where the very first or very last pages survived; it is harder to implement when various fragments survived that are hard to place in a serial manner. Moreover, while Tov’s suggestion sounds reasonable, he did not test it against additional factors. The statistics in fact point against his conclusion. Based on his data, the beginnings of approximately 5.5% of the scrolls (51 manuscripts) were preserved, compared to only 3.1% (29 manuscripts) of the ends, implying that only a little less than half of the scrolls were rolled the normal way.8 However, the great majority of those scrolls that have been found as extant rolls (i.e., from caves 1 and 11) had clearly been rolled in the standard way. If they represent the entire corpus, their evidence counters Tov’s sample. The fact that more beginnings of scrolls were preserved than ends may be explained by other means. For example, it is possible that the fabric wrapping the scrolls protected the external layers, while the inner layers were more exposed to humidity and insects, penetrating through the exposed upper and bottom parts of the scroll and the space left in the middle.9 This may explain why most scrolls preserve neither the beginning nor the end. In addition, it might be worth not only inquiring about the first and last columns, but also running a more systematic log of the scrolls, examining ranges rather than specific points: which percentage of the fragments remained from the first region of the scroll, and which remained from its concluding part? This may lead to refined results. One may not overrule the possibility that some scrolls were left open in the middle, rolled from both sides or folded or in another condition; but such cases are hardly attested and difficult to prove.10
In some fragmentary scrolls, the fragments are preserved stacked in piles. In chapter 7 we explained how to identify such wads; here we focus on their use for ordering the fragments and reconstructing the scroll. Observing the wad with the text on top, fragments from external turns of the scroll are attached underneath fragments from the subsequent turn towards the inside. Thus, the order of the layers from the top to the bottom of the pile represents their original order from the inner turns of the scroll to its external turns.
Knowing the material order of the fragments in the original scroll does not necessarily indicate the order of the text in the original composition, however, as the scroll may have been rolled either way. Knowledge of parallel copies may solve this problem, as in the case of 4Q418a and other scrolls preserved in piles such as 4Q385a and 4Q82.11
As mentioned in chapter 7, although the peeling of wads was carried out with great care and documented both on the PAM images and in the editions, it was not devoid of problems. Some mistakes remained in the record, as we were able to detect, either with regard to switching the order of layers (for example in 4Q418a 7 and 8, see chapter 15), or – quite rarely – by reversing the order of the entire wad. It is therefore vital to re-examine the order of layers using close inspection of the PAM images. Here is the procedure for tracking down the wads and establishing their order, exemplified by the wads of 4Q418a:
The procedure begins with the oldest image in which the wad can still be seen before separation. A close examination of the earliest available image may shed light on the order of the wad, allowing a critical view of the scholar’s documentation. For example, in the case of wad A of 4Q418a we learned that the original order of two fragments had accidentally been reversed in later photographs. In addition, we located an early image for one of the wads, taken prior to the earliest documentation mentioned by Strugnell. This early image showed additional pieces of parchment, not accounted before (see chapter 15).
Whether or not an image of the unseparated wad is found among the PAM images, later PAM images may also add further documentation on the separation stage. This information is culled from the next image, in which the layers of each wad are placed sequentially on the plate. Numbers indicating the order of layers are usually recorded on pieces of paper in the photograph. Sometimes the layers were separated in several stages, with each of the stages documented on a separate PAM image. In such cases it is important to search for the earliest image of each layer.
After adapting the scaling of all images (chapter 4) and removing the background from all images (chapter 5), an image manipulation program is used to place the respective fragments as layers on top of the image of the original wad. The borders of the respective layers should be matched with the pertinent points on the original wad. In cases where the individual layers disintegrated while being separated, the individual pieces should be collected and virtually reassembled on the canvas.
Our treatment of wad A exemplifies the insights gained from this procedure. 4Q418a frag. 6 broke into many pieces while being separated from other layers, all recorded on PAM 41.997. We re-organized these pieces, placing them in their original position. Information on this original position was culled from the earlier images before separation, as seen, for example, in PAM 41.973.
As can be seen in the image (figure 37), the reassembled pieces of frag. 6 only cover about a half of the surface of the wad, while the upper half of the wad seen in the image comes from the next layer, not from frag. 6. That missing upper part of frag. 6 can now be found in a separate layer, where it is part of frag. 8 (figure 38). Surprisingly, that part is not found on frag. 7 – which is purportedly the next layer in the wad according to DJD – but rather on frag. 8. We are thus able to conclude that it was frag. 8, not frag. 7, that stood directly underneath frag. 6. This find contradicts the order recorded in the PAM image and reported by the editors.12
The same procedure carried out with regard to 4Q418a frags. 6–8 can be expected to confirm the documented order of the layers. In any event, the documentation should be examined critically by cross-checking it against other evidence.
3 Establishing the Order of Scattered Fragments
The discussion thus far involved cases in which the wads reached the editorial team while still attached, with information on the separation stage duly documented. However, in most cases the fragments appear separately, whether received by the editorial team in this state or being separated by them without proper documentation. In these cases, the physical shape of the fragments may indicate their original order. Fragments with similar damage patterns may have originated from the same vertical height in the scroll, possibly from close – though not necessarily consecutive – turns of the scroll. One should search for similarities in the external borders of each fragment, and for cracks and breaks on their surface. In addition, look for the information available on the verso that may have been derived from the recto, for example impressions of the seams from the previous or the next turn that are left on the scroll.13 Some damages may have been caused by a string holding the scroll tight.14 After collecting and recording such information, the editor can suggest the placement and configuration of the fragments. This is in principle an objective suggestion, based on the data, i.e., on the objective recurring patterns of deterioration. However, since the separate fragments continued to deteriorate even after their separation, suggesting a material reconstruction involves a certain measure of subjective judgement.
The subjectivity involves not only pointing out the decay patterns, but also – mainly – marking the exact point from which measurement should take place. The similar pattern spreads over quite a wide area, and the choice of the exact point from and to which the distances are measured is not at all obvious. Ratzon and Dershowitz demonstrate that this choice significantly influences the computations of the length of the scroll and their errors.15
The procedure for finding fragments with similar patterns of damage was proposed by Stegemann and Steudel:
Detect the most significant fragments with recurring contours or other damage patterns (figure 39). Fragments showing upper, lower, and intercolumnar margins are especially helpful.
Using an image manipulation software, mark the borders and every crack and hole in the fragment, while removing the image of the actual fragment.
Place the fragment outlines one on top of the other as digital layers in a canvas. Check whether at least three damage patterns correspond and therefore indicate a wad.
Repeating damage patterns may indicate that the fragments come from the same vertical location in the scroll. If they also come from the same sheet, space between lines may serve as further verification of the suggested correspondence, since in many scrolls the space between lines remains stable, at least in the same sheet (see chapter 9).
The order of the fragments with similar contours should now be established. The Stegemann method is based on the intuitive principle that the distances between damage points stemming from an external part of the scroll are larger than those stemming from its inner parts. The scholar should conceive an order of the similar fragments based on various textual and material observations. Of the former type, an order may be shown by a parallel text, or, in its absence, may be posited by the thematic sequence. Of the latter, an order may be conceived by means of the size of each point of the recurrent damage, for example a gradually growing size of a hole in the successive layers. The method offers rules for establishing the circumference of the scroll in each of its subsequent turns. If no textual evidence is available, and the fragments were not preserved in a wad, the certainty of the restoration is diminished. This suggested restoration may gain support, however, by means of checking that the material traits of the fragments fall into place according to the suggested order, and according to the known size of columns and margins. For example, one should establish that fragments showing an inter-column margin would indeed reproduce that margin in the exact spot according to the suggested order. This protocol in turn dictates the exact distance between the respective fragments, as described in detail in the next chapter.
Stegemann did not in fact invent the method, which had been in use by papyrologists beforehand. The method was passed orally and was not published in print as a formal method. Stegemann standardized it for the study of the DSS (Stegemann, “Methods for Reconstruction.”) For a recent summary of the method and its later developments see also Stökl Ben Ezra, Qumran, 53–55, and the introduction to this volume. For its earlier use in the field of the DSS, see Milik, “Dires de Moïse,” 91.
Stegemann, “Methods for Reconstruction”; Steudel, “Assembling.”
For this difficulty and others like it in determining incremental growth of damages, see Ratzon and Dershowitz, “The Length of a Scroll.”
Ratzon and Dershowitz, “The Length of a Scroll.”
See the discussion with additional background in Tov, Scribal Practices, 40.
Another example of a smaller scroll is 1Q22. See Milik, “Dire de Moïse,” 91.
Tov, Scribal Practices, 108–12.
Tov, Scribal Practices, 108–12. In addition, Elgvin claimed that if damage patterns indicate that a fragment reflects tiny turns, i.e., a circumference of the scroll of 2.5–4 cm, the fragment should be located at the innermost end of the scroll. See Elgvin, “1QSamuel.”
For a survey of the fabrics found in Qumran see G.M. Crowfoot, “The Linen Textile,” in Qumran Cave 1, DJD I, Dominique Barthélemy and Józef T. Milik (Oxford: Clarendon Press, 1955), 18–38; Mireille Bélis, “The Unpublished Textiles from the Qumran Caves,” in The Caves of Qumran: Proceedings of the International Conference, Lugano 2014, ed. Marcello Fidanzio, STDJ 118 (Leiden: Brill, 2016), 123–36; Naama Sukenik, “The Temple Scroll Wrapper from Cave 11,” in Elgvin, Davis, and Langlois, Gleanings from the Caves, 338–50.
Ariel Feldman suggested that the scroll 4QJosha was folded rather than rolled: Ariel Feldman, “Reconstructing 4QJosha (4Q47): The Contribution of Frag. 21,” [Hebrew] Meghillot 14 (2018/19): 3–12. The study of 4Q57 (Isaiahc) by Asaf Gayer shows that it may have been rolled from both sides.
For 4Q385a see Davis, The Cave 4 Apocryphon of Jeremiah, 73–84. A series of images depicting these patterns are reproduced passim. For 4Q82 see Russell Fuller, “The Twelve: 82. 4QXIIg,” in Qumran Cave 4, X: The Prophets, eds. Eugene Ulrich et al., DJD XV (Oxford: Clarendon Press, 1997), 271–318.
In chapter 15 we discuss the significance of this find for the reconstruction of the scroll.
Stegemann, “Methods for Reconstruction,” 195.
Stephen Pfann and Menahem Kister, “4QcrypA Words of the Maskil to All Sons of Dawn,” in Qumran Cave 4. Sapiential Texts, Part 1, ed. Torleif Elgvin et al., DJD XX (Oxford, Clarendon Press, 1996), 5–6.
See Ratzon and Dershowitz, “The Length of a Scroll.” Consider for example the series of holes in 4Q550 frag. 7 (B-363164). While they give a clear example of growing intervals of damage, the exact point from which to measure the growth is hard to determine.