1 Introduction: The Historical Context of Manuscript Production and Scribal Practice at Qumran
The scrolls studied in this book were, for the most part, carefully-planned objects, created and used within the social and historical contexts of Second Temple period Judaism, a context bound up closely with the broader Hellenistic milieu of the eastern Mediterranean region. The Aramaic scrolls from the Qumran caves were presumably used at the site of Qumran, though some of them likely came to the settlement from outside, having been crafted and written before the first century BCE.1 The extent to which the scrolls studied here were manufactured at or around Qumran remains a question without a clear answer, though significant progress is now being made on this front due to new scientific research focused on determining the provenance of their skin and ink through chemical and other analyses.2 Preliminary studies have suggested that at least some of the scrolls were processed and written in the vicinity of the Dead Sea, local to Qumran.3 Wherever our scrolls were made, written, and used, it is obvious that an extended process of planning and production preceded someone reading or listening to them. This process involved a substantial investment of time and resources.
There is some evidence that the monetary costs of manuscript production were very high in ancient Mediterranean cultures, a factor tied to the cultural prestige of scrolls (also called bookrolls), the knowledge they contained, and the specialized nature of writing and reading more generally.4 A number of Roman-period authors commented on the high value placed on literary manuscripts or libraries perceived to contain precious knowledge, most of which would have been papyrus bookrolls.5 In the first century BCE, Cicero wrote to a friend that his personal library was “worth a considerable sum” (multorum nummorum).6 More than a century later, Pliny the Younger wrote that his uncle, Pliny the Elder, could have sold the notebooks (commentarios) for his Natural Histories in Spain for 400,000 sesterces.7 We are told that there were considerably fewer notebooks at that time (aliquanto pauciores erant) than the 160 eventually inherited by Pliny the Younger. If we estimate that Pliny the Elder possessed around 100 rolls at the time of his potential sale, we would be left with a cost of approximately 4,000 sesterces per “notebook.” The average wages of an ordinary Roman legionary soldier in the first century CE was roughly 1,000 sesterces, and so the entire collection would have far exceeded an average person’s lifetime wages, and even a single bookroll would cost several years’ worth of wages. Owning such an object clearly involved a serious financial investment. Another signal of the high value placed on written manuscripts was that they were occasionally listed among valuables seized as the spoils of war. For example, Plutarch reported that Aemilius Paulus took as war booty the bookroll collection of Perseus, king of Macedon, in 168 BCE.8 These sources admittedly reflect social locations quite different than those from which the Qumran scrolls originated, but they give us a sense of the high social and financial value attached to written scrolls during the Hellenistic and Roman periods, at least by the educated, literate classes.9
Anatomy of a scroll (1Q20 [apGen] cols. 20–22)
The costs of producing manuscripts were largely labor-based, first to manufacture the papyrus or skin writing surface, which required specialized tools and training, and then to employ extensively-educated scribes to lay out the blank manuscript (usually in columns) and to write the text. We can no longer determine the extent to which these labor costs may have been mitigated by the social contexts in which the scrolls kept by those living at Qumran were produced, since we cannot accurately calculate how much of the process was done by members of the Essene sect at or around the settlement, or elsewhere. We simply do not yet know – and we probably never will know with certainty – where the scrolls from the caves around Qumran were produced, though the scientific advances in determining provenance mentioned above hold out promise for advancing our knowledge somewhat. Still, whether a scroll was produced in Jerusalem, at Qumran, or elsewhere in Greco-Roman Palestine, we may safely assume that it was a costly endeavor for the individual or community creating it. In all of these places, an important potential difference compared with the Greco-Roman social contexts discussed above is that whoever produced the Qumran scrolls may not have had to pay the associated fees charged by professional craftspeople and scribes, who did their work to make a living.
With this general context in mind, in what follows I provide a synthetic overview of the physical and scribal features of the Aramaic scrolls from the Qumran caves, building on the individual scroll profiles. Along the way, I will discuss aspects of the manuscript production process, and for this reason the chapter is laid out in sections that follow this process step-by-step. I will begin with the materials used to create the blank sheets of skin or papyrus and end with the writing, correction, and repair of the scrolls included in the profiles. The two central goals of this chapter are: 1.) to provide an overall sense of the physical and scribal features of the Aramaic Qumran scrolls, and 2.) to facilitate comparison with other ancient manuscripts, most immediately with the other scrolls from the Qumran caves, but also with corpora farther afield, such as those from Samaria, Elephantine, and Oxyrhynchus.10
Parts of the column (11Q10 [Job] cols. 37–38)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Najib Anton AlbinaBoth the scroll profiles and the following discussion in this chapter rely on an array of terms for the physical features of a scroll and its columns. I therefore provide above two illustrations detailing these features and terms, one for the scroll as a whole and the other for its columns.11
2 Manuscript Features
2.1 Materials
Tov described the manufacturing and copying processes of the Qumran scrolls in general (Hebrew, Aramaic, and Greek), which may be viewed against the wider backdrop of scroll production in the Mediterranean Basin during the Hellenistic and Roman periods.12 An initial decision was that of the material to be used for a scroll, although in some (perhaps most) cases this choice was likely dictated by geographic location and the availability of resources. The two materials used for the scrolls included in the profiles are animal skin and papyrus.
2.1.1 Skin
Nearly all of the Aramaic Qumran scrolls are written on skin prepared from domesticated quadrupeds commonly found in Second Temple period Palestine.13 Judging by the pioneering DNA-based study of Anava et al., it seems that the majority of scrolls in the environs of Qumran were made from the hides of sheep (Ovis aries), though two of the scrolls they tested were made of cow (Bos taurus) hide.14 This study confirms the earlier observations of Poole and Reed, that the Qumran scrolls derived primarily from sheep and/or goats, and more rarely from cows or other quadrupeds.15 The use of skin, as opposed to papyrus, for creating most of the scrolls kept at or near Qumran is best understood against the broader cultural backdrop of skin manuscript usage by government-affiliated Persian and eastern Hellenistic scribal chanceries. From a technological perspective, letters and documents such as those found in the Arshama archive from the late fifth century BCE, first published by Driver, and the very similar fourth-century Bactrian archive published by Naveh and Shaked, reflect a manufacturing and scribal process strikingly similar to those found in our corpus.16
Papyrus fragment (6Q23 [papWords of Michael] frag. 1)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Najib Anton AlbinaOnce the hide had been removed from the animal, a standard process of several basic steps was followed in order to prepare it for cutting, the laying out of columns, and writing.17 The first step was to dehair (or depilate) the unsplit hide, which was typically done by soaking it in water with natural, enzyme-inducing agents that helped to loosen the hair from the skin.18 Salt, flour, other vegetable-based materials, urine, and dung were commonly used for this purpose, with the lime mixtures so common in later dehairing processes not yet in use for our manuscripts.19 This process also played a role in cleaning the hide and loosening its fiber structure for easier manipulation. The hair was removed by scraping once it had been soaked and loosened, after which the dehaired skin would be stretched, dried, and worked with a rock or other implements.20 The final stage of preparation was to dress both sides of the skin superficially (not by soaking or penetration) with a gallic acid tanning solution, which facilitated the permanence of ink when applied, served as a final means of cleaning, and generally made the surface of the skin more attractive and finished-looking. Investigation has shown that, for the Qumran scrolls, a dressing made from gallic vegetable tannins such as the gall apples of acacia trees was the norm.21 Beginning with Roland de Vaux, some have found evidence of an industrial tanning center at ‘Ein Feshkha, potentially related to the settlement at Qumran and thereby with the Dead Sea Scrolls, though this use of the site is still debated and deserving of further study.22
Skin and papyrus manuscripts arranged by palaeographic date (youngest to oldest)
When the skin was fully dehaired, stretched, dried, smoothed, tanned, and cured, it was ready for “sheets” to be cut from it for making scrolls. The large majority of these sheets were rectangular in shape, with the sheet size(s) being determined by the type and size of animal. Judging by the construction of Qumran scrolls with multiple sheets, and the regular differences in size from scroll to scroll, it is reasonable to conclude that sheets were often cut with a desired literary work or type of scroll in mind. Letter-based numbering is present at the top, righthand corner of each sheet of 1Q20 (apGen), and perhaps also on the single fragment of 4Q529 (Words of Michael), both written in a hand different than the main text. These letters suggest that the sheets were planned and prepared with the idea of a definite sheet sequence. At the same time, the letters used for the sheets in this scroll represent high numbers, making it very unlikely that the scroll started with the letter aleph. Considering these factors together, we may posit a scenario in which a large batch of numbered sheets (at least 20) were manufactured for future use, from which a scribe could then draw as needed for a specific literary work.
For scrolls that were large in height, such as 1Q20 (apGen) and 4Q202 (Enb), it seems probable that a small quadruped like a sheep or goat would have provided a single sheet, while the same hide might have supplied two or more sheets for scrolls of a smaller scale, such as 4Q535 (Birth of Noahb). While rectangular sheets intended for a scroll were the norm, exceptions do occur. In such cases, irregularly shaped pieces of skin that were left over after the rectangular sheets had been cut from the hide may have been put to use. A probable example of this among the Aramaic manuscripts is 4Q339 (List of False Prophets), which was written on a relatively small “card” of skin that was folded rather than rolled. It has also been suggested that 4Q242 (PrNab) may have been written on a single, small piece of skin.23 Comparable examples among the Hebrew manuscripts include 4Q340 (Netinim), 4Q341 (Exercitium Calimi C), and 4Q175 (Testimonia). We might also consider the nearly square sheet used for 1QIsaa cols. 26–27. It is notable that many of these small, irregularly-sized manuscripts are apparently scribal exercises, scholarly resource texts, or copies evidently intended for personal study or travel.
The last stage of the scroll construction process was to sew the individual sheets together, which was generally done with vegetable-based threads made of flax or similar materials. It is often assumed that this sewing was done after the scribe had laid out the columns and written the text, on which see below.24 This order of events – laying out the columns, writing the text, and only then sewing the sheets together – might be implied, for instance, by the sheet numbers on 1Q20 (apGen), providing easy reference to the order in which the sheets should be sewn together. However, it is clear that some scribes wrote their text after the sheets had been sewn into a larger scroll, as in the case of 4Q213 (Levia; see image below), on which the writing crosses over the sewn seam joining two sheets.
Sheet number פ (1Q20 [apGen] col. 5; marked by box)
The relative quality of the finished skins used for the Qumran manuscripts clearly varied, though this factor is impossible to assess accurately unless a scroll is examined in person (which was not possible for this study). The severely degraded state of most skins also inhibits a true appreciation of their original quality.
2.1.2 Papyrus
It is often claimed that papyrus had to be imported to Greco-Roman Palestine from Egypt.25 While this certainly could have been the case, there is good evidence that the papyrus plant (Cyperus papyrus) grew at multiple swampy locations in Palestine during the Hellenistic and Roman periods, into the early 20th century. This was most certainly the case in the Hula Valley and Lake Agmon, in the Galilee region, but also in the Sharon Plain and elsewhere.26 An overlooked detail that bears on this question is Josephus’s mention of Aristobulus’s successful battle against the Nabateans at a location named Papyron (
The pith of the plant’s harvested stalk was split into thin strips, which were made into individual sheets (
Sewing between sheets, with writing across the seam (4Q213 [Levia] frag. 1)
Manuscript repair by stitching (4Q547 [Visions of Amrame] frag. 5)
Of the eighty-nine literary scrolls considered in the profiles, only five (6%) of them are made of papyrus.32 They are:
-
6Q8 (papGiants)
-
6Q23 (papWords of Michael)
-
4Q196 (papToba)
-
4Q558 (papVisionb)
-
4Q559 (papBiblical Chronology)
Tov observed that approximately 13% of the literary scrolls from Qumran were written on papyrus, meaning that the percentage of Aramaic scrolls written on papyrus is significantly lower than is the norm for the Qumran corpus (Hebrew, Aramaic, and Greek) more generally.33 It is striking that two of our Aramaic manuscripts came from Cave 6, a cave that produced relatively few scrolls, but had an unusually high proportion of texts written on papyrus.34 Three of the Aramaic papyrus scrolls are copies of texts also found on skin scrolls: the Book of Giants, the Words of Michael, and Tobit. The two remaining texts, not identified in any other copies, are generically akin to other texts written on skin scrolls. Consequently, it would seem that a text’s genre was not a determinative factor in choosing the material on which it was written. It is generally believed that papyrus was considered a cheaper, less durable material for manuscripts than skin in antiquity, a view already implied in Pliny’s hyperbolic account of Eumenes’s invention of prepared skins (membrana,
In comparison with the Aramaic scrolls from Qumran on skin, the papyrus scrolls tend to be generous in terms of their margin sizes, line spacing, and letter size, which may reflect less concern with economy than for skin scrolls, due to the lower expense and prestige of papyrus. For the skin scrolls, there is a general (though not absolute) correlation between the quality and size of a scroll and the formality of its script and other scribal features; smaller scrolls – often written on lower quality skin – are more likely to have relatively informal, erratic scripts, with higher admixtures of rounded cursive elements. All of the papyrus manuscripts are marked by erratic letter size and formation, and by informal, cursive script features. 4Q559 (papBibChronology), 6Q8 (papGiants), and 6Q23 (papWords of Michael) are among the most erratic, semi-cursive scripts of the entire Qumran Aramaic literary corpus, with the scribes of both 6Q8 (papGiants) and 6Q23 (papWords of Michael) using highly idiosyncratic writing features (the use of a cursive final aleph and a cursive mem, respectively). The scripts of 4Q196 (papToba) and 4Q558 (papVisionb) are noticeably more squared, upright, and formal than those of the three other scrolls, but even these two (or possibly three) scribes mixed rounded, cursive features into their writing in a way not typical for skin scrolls of very good or excellent quality. It also seems that vacats were used less frequently in Qumran papyrus scrolls than in those written on skin, though the fragmentary state of the evidence leaves some question on this point. The only column of a papyrus scroll for which we have a good sense of its overall size is that preserved on 4Q196 (papToba) 2, which appears to be approximately 17.5 cm in height and 16 cm in width. The height of this column is around the average for those preserved on skin scrolls, but its width is the greatest of any of the Qumran Aramaic scrolls, again suggesting a somewhat different scribal approach for papyrus scrolls than for skin. The overall picture gained from the above assessment is unlikely to change appreciably were we to add other papyrus scrolls of possible relevance, such as 6Q7 (papDaniel), Puech’s 4Q558a (papUnidentified), or the other fragmentary papyrus manuscripts listed in Chapter 1.
These indicators that the Aramaic papyrus scrolls kept at Qumran were generally made to lower quality standards than skin scrolls from the same corpus is corroborated by a comparison with the classical standards of Hellenistic- and Roman-period literary bookrolls, described in great detail by Johnson.37 While some features of the Qumran scrolls occasionally meet the benchmarks of professionally-written literary bookrolls from places like Oxyrhynchus (e.g., letter height and line spacing), it is readily apparent that the spacing and scripts of the Qumran scrolls are significantly more erratic than roughly contemporary Greek literary bookrolls from Egypt. From the little evidence available, it seems that the Qumran scrolls also tended to be smaller in height and had different column ratios (much wider columns relative to height) than the hundreds of bookrolls studied by Johnson. Of course, it should be borne in mind that the two literary corpora were produced in different geographic and social milieux, although there is reason to believe that the professional guild standards clearly in effect at Oxyrhynchus were known and used widely throughout the Greek and Roman empires. It seems plausible, therefore, to assume that they were also known to Jewish scribes in Greco-Roman Palestine, an assumption supported by the multiple parallels adduced by Tov between the Qumran scrolls and wider Greco-Roman scribal practices.38 At the very least we can say that, if the artisans and scribes responsible for creating the Qumran Aramaic papyrus scrolls knew of the professional guild standards evident in literary bookrolls like those at Oxyrhynchus, they did not adhere closely to those standards. If we were to look at Qumran for scrolls that did follow the high professional standards of Hellenistic- and Roman period literary bookroll production, we would find them in the “Very good” and, especially, “Excellent” quality categories used in the profiles. At Qumran, all such Aramaic scrolls were written on skin, with representative examples being 1Q20 (apGen), 4Q203 (EnGiantsa)/4Q204 (Enc), 4Q205 (End), 4Q209 (Enastrb), 4Q554a (NJb), and 4Q544 (Visions of Amramb).39 This correspondence strengthens the notion that, in Greco-Roman Palestine, esteemed literary texts were very often written on skin scrolls because of the material’s implied high quality, prestige, and perhaps durability compared with papyrus. Cheaper, less durable papyrus, by contrast, was reserved almost exclusively for: a.) mundane legal, business, and personal documents such as deeds, loans, letters, inventories, and lists; and b.) lower-quality copies of the same types of literary texts also copied on skin. However, it important to note that a low-quality copy might also have been written on skin of sub-par quality or irregular size, as in the case of 4Q201 (Ena), 4Q212 (Eng), 4Q339 (List of False Prophets), and 4Q542 (TQahat)/4Q547 (Visions of Amrame). The situation was very different in Egypt, where papyrus was used for the full spectrum of textual needs, with a bookroll’s contents and intended use bearing on a manuscript’s quality, size, format, and scribal execution.
Pulling together the threads of the preceding discussion, it is evident that papyrus was chosen for scrolls of low to moderate quality. What such lower quality may tell us about the intended uses of these scrolls is a matter of educated speculation, and our answers will depend, to a considerable extent, on the social scenarios that we reconstruct for those who wrote and used them. The fact that four of the five scrolls treated here contain scribal corrections suggests that they were compared to other copies and were considered valuable enough to update and correct. Wise argued that such low quality manuscripts were “personal” or “private” copies as opposed to scrolls that circulated in the “book trade,” though Johnson has challenged and complicated this binary in important ways.40 If we imagine the scrolls from the Qumran caves being written for an ethno-religious group like the one that lived at Qumran – or even a wider spectrum of interested Jews living in Jerusalem or elsewhere – then lower quality manuscripts like those written on papyrus (or a low-quality skin scroll like 4Q542 [TQahat]/4Q547 [Visions of Amrame]) could be explained from the viewpoints of production, intended use, or perhaps a combination of both factors. On the production side, these scrolls might be the work of scribes in training or otherwise possessing less scribal skill, writing low-quality “practice texts” that might still be kept as serviceable copies. Papyrus may also have been used for texts still in the process of composition, resulting in something like a “draft copy” or a “working copy.”41 Depending on the social situation, there may simply have been a need for lower cost or investment of labor. In terms of intended use, there are several reasons why lower-quality, more expendable copies of some texts would have been desirable. These reasons may have included personal use or use among a small group not requiring an expensive, high-quality copy, portability with less concern about damage, and wider distribution facilitated by lower expense and investment of labor. In the end, all we can say with confidence is that the Aramaic scrolls from Qumran were produced along a spectrum of quality indicators, and that the papyrus copies fell towards the lower-quality end of this spectrum. The reasons for such lower-quality copies may have included concerns over expense, availability of materials, considerations of literary content, degree of scribal training, and intended use (e.g., personal study, portability, or further textual or compositional work).
2.2 Scroll Dimensions
The largest possible dimensions of skin scrolls were limited by the size of the animal used and conventions of usability.42 As discussed above, the sizes of papyrus scrolls do not differ substantially from those made of skin, suggesting that manuscripts of both materials adhered to a widespread cultural aesthetic for literary texts. The charts above aim to give a snapshot of the known dimensions of the scrolls included in the profiles.
Manuscripts with height and length substantially preserved
Manuscripts with only height substantially preserved
Based on the information above, we can observe that the height of scrolls in our corpus range between approximately 7 cm at the small end of the spectrum and 30 cm at the large end, with a mean height of almost 16 cm and a median height of 14.4 cm. There is no clearly observable correlation between the period of production and manuscript height, nor does there seem to be a sustained connection between height and the quality of scribal execution. Scrolls that are of very high scribal quality are both 14 cm (11Q10 [Job]; very good–excellent) and 31 cm tall (1Q20 [apGen]; excellent), the former being a century or less younger than the latter. At the same time, scrolls that are relatively small in height include those with both fair (4Q569 [Proverbs]) and very good (4Q246 [apocrDan]) levels of production and scribal execution. It is possible – even reasonable – that the length of a literary work impacted the size of the scroll on which it was planned to be written, though it is now impossible to assess such a correspondence in the large majority of cases. Still, it should be noted that two manuscripts of the Visions of Amram (4Q545 and 4Q547) written in the same century vary quite significantly in height and scribal execution, while a fairly long and expertly written scroll like 11Q10 (Job) is of only medium height (14 cm). Although the sample size is very small, it appears that there was a well-established range of potential sizes for scrolls containing a literary work (ca. 7–30 cm in height), but that within that range individual scribes – no doubt working at different times and in diverse locations – had considerable freedom as to the size of scroll they produced.
Unfortunately, there is little reliable evidence regarding the widths (i.e., lengths) of the scrolls comprising our corpus. It can be estimated that 11Q10 (Job) was approximately 7 m long, and 4Q112 (Dana) around 3 m. 1Q20 (apGen) must have been at least 3 m long, and was very likely much longer than that. By contrast, it is difficult to imagine a manuscript less wide than 4Q339 (List of False Prophets), at a mere 7 cm or so. Based on the data of the entire Qumran corpus, including the Hebrew texts, it seems that the outside limit of width for most scrolls was in the vicinity of 8–10 m, though most scrolls were considerably narrower than this, in the 2–6 m range.43
Examples of large and small scroll heights (to approximate relative scale)
2.3 Laying Out the Scroll: Ruling, Columns, Margins, and Spacing
Once the sheets of skin had been prepared and cut, a scribe very often added dry-ruled guidelines to facilitate the subsequent stage of writing, using a sharp instrument (perhaps made of bone, as Tov suggests) and a straight edge to score the skin and leave permanent, slightly-darkened lines on its surface. This was consistently done on the wool or hair side of the skin (the recto), since it received the ink better than the flesh (verso) side.44 In at least the case of 4Q115 (Dand), it has been suggested that “very diluted ink” was used to make some of the guidelines.45 A sheet that was fully ruled had vertical guidelines demarcating both the right and the left sides of its columns, and evenly-spaced horizontal lines to guide writing stretching the length of the sheet (i.e., covering multiple columns), similar to a modern piece of lined paper. Rarely, as in the cases of 1Q72 (Danb) and 4Q115 (Dand), two vertical guidelines were used on the right side of a column, a practice that, within the broader Qumran corpus, most often occurs at the right edge of a sheet.46 Guide dots were sometimes placed at the right and left edges of a sheet to indicate where the straightedge was to be placed for marking the horizontal script guidelines.47 While the majority of our skin scrolls were fully ruled, on occasion we find one that was only partially ruled, or has no dry-ruling at all.48 At other times, it is very difficult to tell whether or not a sheet was dry-ruled due to the poor state of its preservation and, in some cases, very lightly-ruled lines. Sheets or scrolls of papyrus were not in need of ruling and layout marks in the way that skin ones were, since the horizontal grain of the papyrus strips were almost always on the recto (writing side) of the sheet, and so provided a natural means to regulate script consistency and spacing. The following tables and figures summarize the data on use of scribal guidelines and guide dots.
Vertical ruling on right side of column only where left margin is extant
Vertical ruling on right side of column where left margin is not extant
Ruled script lines
Manuscripts with marginal dots for marking script guidelines
The only Aramaic literary scroll with part of its end preserved is 4Q245 (psDanc), which in frag. 2 has at least one partially or fully blank column.49 The narrowness of this column suggests that it likely fell at the end of a sheet, and therefore the manuscript.50 An uninscribed column at the end of the composition accords well with the general picture from the Qumran scrolls documented by Tov.51 Technically, we also possess the end of 4Q339 (List of False Prophets), though because it is written on a single card of skin, its “end” amounts to little more than the standard margin of a text column.
We possess slightly more evidence from the beginnings of our scrolls.52 4Q242 (PrNab), 4Q571 (Words of Michaela) and 4Q545 (Visions of Amramc) preserve evidence of an area of skin left blank before the first inscribed column, on the same sheet. For 4Q242 (PrNab) this blank area is not much bigger than an especially wide intercolumnar margin, though it should be borne in mind that we do not know the nature of this manuscript, and that some have suggested it was written on a single sheet of skin rather than a full scroll.53 The blank area on 4Q571 (Words of Michaela) is somewhat larger, close to the size of a full column for what is preserved. A different practice was used for 4Q529 (Words of Michael) and 4Q543 (Visions of Amrama), the first inscribed columns of which are preceded by a standard intercolumnar margin at the beginning of a sheet and a sewn seam. This seam strongly suggests that what is often called a “handle sheet” – a blank sheet of skin to which a wooden stick may, in some cases, have been attached – was sewn to the beginning of the scroll. Blank sheets or portions of sheets at the beginnings and ends of scrolls served the purpose of protecting its outer parts in storage, and allowed for handling by readers without potentially smudging or otherwise damaging the text itself through ongoing use.
Guide dots (4Q545 [Visions of Amramc] frag. 3)
Part of the end of a scroll (4Q245 [psDanc] frag. 2)
Ruling the sheet resulted in a series of blank columns surrounded on all sides by margins, a basic pattern that was followed even when sheets were not ruled. As has often been observed, the technology of writing in columns allowed for ease of reading in a scroll format, since only a small area of the scroll (one to a few columns) needed to be unrolled at any time for continuous reading. The sizes of complete or easily-reconstructed columns from our corpus vary considerably. In twelve of the eighteen examples included in the table below (67%), the column is taller than it is wide (a height to width ratio greater than 1), and in four cases (22%) the height is nearly or more than double the width. By contrast, a handful of scrolls (e.g., 4Q318 [Zodiology and Brontology] and 4Q547 [Visions of Amrame]/4Q542 [TQahat]) have some columns that are close to twice as wide as they are tall. Column heights range between the extremes of just over 4 cm (4Q535 [Birth of Noahc]) and nearly 27 cm (1Q20 [apGen] and 4Q202 [Enb]); however, the significant majority of scrolls have columns of ca. 10–16 cm. As for width, the outer limits are 6 cm (4Q561 [Physiognomy/Horoscope]) and 17 cm (4Q210 [Enastrc]), with a large group falling into the 8–14 cm range. Consequently, we can speak of a corpus-wide norm for columns of 10–16 cm in height and 8–14 cm in width, with occasional exceptions. These sizes are generally in keeping with the broader Qumran corpus, as discussed by Tov, with Aramaic scrolls from the corpus studied here falling into his small, medium, large, and very large writing block parameters.54
The partial beginning of a scroll (4Q571 [Words of Michaela])
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Najib Anton AlbinaExamples of column width (to approximate relative scale): Top image: 4Q542 (TQahat) frag. 1 (15.5 cm); Bottom image: 4Q554 (NJa) frag. 2 (6–7.5 cm)
Within the column, the number of lines and average letters per line depend on the tightness of spacing, the size of the script, and other factors like the use of vacats. The outer limits of lines per column are set by 4Q561 (Physiognomy/Horoscope), at 5 or 6 lines, and 4Q209 (Enastrb), at around 40, with a fairly even distribution on a bell curve between those extremes and a mean of nearly 19 lines per column. The average number of letters per line (not counting spaces between words) cluster mostly between 30 and 50 per line, as seen in the following chart. Significant outliers include 4Q339 (List of False Prophets; 15 letters), 4Q561 (Physiognomy/Horoscope; 23.5 letters), 5Q15 (NJ; 81 letters), and 4Q210 (Enastrc; 81.5 letters). Column size, lines per column, and letters per line do not seem correlated to the scribal quality of a manuscript in any straightforward way.
The visual compactness of the column was impacted by a combination of the line spacing, or leading, and the size of the script. The significant majority of manuscripts cluster in the 6.5–7.5 mm range for leading, as documented in the chart below. As already noted above, there is a recognizable tendency for papyrus scrolls to be spaced more generously than those on skin, despite what otherwise seem to be indicators of lower quality.
As a general rule, the lower margin of a scroll tends to be slightly larger than its upper margin, as already observed by Tov and others. This seems to have been part of a basic aesthetic of scroll (and, later, codex) layout, as is also true for Greek manuscripts from sites such as Oxyhrynchus.55 However, there are a number of scrolls that do not appear to follow this rule and instead have upper and lower margins of roughly equal size, such as 4Q318 (Zodiology and Brontology), 4Q544 (Visions of Amramb), 4Q547 (Visions of Amrame)/4Q542 (TQahat), 4Q550 (Jews at the Persian Court), and perhaps 4Q535 (Birth of Noahb).56
Intercolumnar margins range from erratic and very small (a few mm), as in parts of 4Q212 (Eng), to highly regulated and large (1.5–2 cm), as in 4Q554 (NJa) and 4Q554a (NJb). As a language written from right to left, the right side of the column (i.e., the left edge of the intercolumnar margin) always follows a straight vertical line, though in some scrolls this principle was followed more faithfully than in others. However, at the left side of the column, where the lines of writing end, there is much more variation. In some scrolls – again, 4Q212 (Eng) is a good example – the scribe varied greatly in where he would end lines, leaving the left side of the column with a jagged, unkempt appearance. The opposite approach was taken by the scribes of 4Q203 (EnGiantsa)/4Q204 (Enc) and 4Q552 (Four Kingdomsa), who were so keen to keep a neat, even column and intercolumn that they sometimes left larger than usual spaces between the last two words of a line in order to “justify” the margin. The same practice is adopted by the scribe of 4Q554 (NJa), though less rigorously than in 4Q203(EnGiantsa)/4Q204 (Enc) or 4Q552 (Four Kingdomsa).
Intercolumnar margins were of two types: 1.) Those that occur were two sheets meet, and so span the sewn seam connecting the sheets, and 2.) those that were in the midst of a sheet and so have no seam between them. The large majority of intercolumnar margins in a typical scroll would have been of the latter type, and tended to be of greater or lesser regularity in size depending on the individual proclivities of the scribe who either laid out (e.g., ruled) or wrote it. Understandably, those scrolls that were ruled with vertical lines on both sides of the column tend to exhibit more regularity in the size of intercolumnar margins. In general, more consistently-sized, even, and generous margins seem to be one indicator of a higher-quality scroll. Intercolumnar margins spanning a seam between sheets show more variety. Sometimes, margins similar in size to those in the midst of a sheet were also left at both of a sheet’s ends, resulting in an intercolumn between sheets wider than the norm for the scroll. This appears to have been the case in 1Q20 (apGen), 4Q544 (Visions of Amramb), and 4Q554a (NJb), though in the last two examples we have only one side of the margin preserved. At other times, smaller blank spaces were left at the ends of a sheet, resulting in intercolumns more similar in size to (perhaps even narrower than) those in the midst of a sheet (see, e.g., 4Q553 [Four Kingdomsb] 1 and 2, 4Q212 [Eng] 2, and the several seams of 4Q213–213b [Levia–c]).
Example of a wide column and an exceptionally large number of letters per line (5Q15 [NJ] frag. 1)
Surveying the data above, we may posit a correlation among the empty space left on a scroll by way of margins, the formality of its script, and the extent to which vacats are used. In general, scrolls with small or erratic margins have a strong association with informal scripts, and typically contain few or no vacats. Some good examples of this type of scroll are 4Q201 (Ena), 4Q212 (Eng), 4Q213–213b (Levia–c), 4Q541 (apocrLevib?), 4Q547 (Visions of Amrame)/4Q542 (TQahat), 4Q550 (Jews at the Persian Court), 4Q561 (Physiognomy/Horoscope), and several of the papyrus copies. By contrast, scrolls with large margins tend to be written in neater, more formal scripts with regular use of vacats. These same scrolls are very often fully ruled, and so have an overall aesthetic of open, well-regulated spacing. The banner example here is 1Q20 (apGen), but also meriting inclusion are 1Q71 (Dana), 4Q113 (Danb), 4Q115 (Dand), 4Q246 (apocrDan), 4Q203(EnGiantsa)/4Q204 (Enc), 4Q209 (Enastrb), 4Q552 (Four Kindomsa), 4Q554 (NJa), 4Q554a (NJb), and 11Q10 (Job). While most of these scrolls are of a fairly large format, there are notable exceptions (e.g., 4Q246 [apocrDan] and 11Q10 [Job]). The two typological groupings just outlined can be said to represent the ends of the manuscript quality spectrum for our scrolls, between which many of the remaining cases may be placed. With respect to the general (though not absolute) correlation between formality of script and other manuscript features – notably a scroll’s size and format – see now the corroborating observations of Longacre (“Style”) for the Dead Sea Psalms scrolls.
Line spacing examples (to approximate relative scale)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photos: Shai HaleviExample of a fairly typical upper to lower margin ratio (4Q246 [apocrDan] frag. 1)
Small and large top margins (to approximate relative scale)
Intercolumnar margins (to approximate relative scale)
Intercolumnar margin size (mm; without seam)
Intercolumnar margin (4Q530 [EnGiantsb] frag. 2)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Najib Anton AlbinaIntercolumnar margin with seam (4Q553 [Four Kingdomsb] frag. 8)
3 Scribal Habits
3.1 Writing the Text
The arduous process of preparing the sheets of skin and, in most cases, laying them out in columns with dry-ruling, culminated in a scribe writing the chosen text. The ability to write required extensive training, as reflected in written materials throughout the ancient Mediterranean Basin. A number of such writings were found at Qumran and other Judean Desert locations, where scribes apparently trained to write the alphabet (abecedaries) and lists of words (exercitium calami).57 Such texts were often written on inexpensive material like pottery sherds or left over scraps of skin or papyrus. Scholars have proposed that some longer texts, poorly written by apparently unskilled scribes, may also have been scribal learning exercises.58 Of the Aramaic scrolls studied here, Milik speculated that 4Q201 (Ena) was, perhaps, a “school-exercise,” and that 4Q551 (Narrative) was written by an “apprentice scribe,” though such opinions are impossible to verify.59
3.1.1 Ink and Related Writing Implements
In preparation for writing a scroll, scribes in antiquity needed to prepare their own inks and pens.60 The scribal profession was well-established in Hellenistic and Roman Mediterranean societies, and there is clear evidence that some scribes owned and used their own kits in plying their trade. These may have included pens, ink pellets, and an inkwell. If it was the writing scribe who laid out a skin scroll, tools for that purpose – a pointed object and a straight edge the length of a typical sheet (ca. 0.5–1 m) – were also required. Finally, for skin scrolls a needle and flax or animal sinew thread were needed to sew the sheets together. In the communal context of Qumran or the institutional domain of the Jerusalem temple it is impossible to determine the extent to which such tools of the trade may have been shared by groups of scribes, but they had to be readily available.
Scribes in antiquity mixed their own inks as needed, though it is difficult to ascertain a scribe’s possible role at earlier stages of the ink manufacturing process. The production of ink was a discrete area of artisanship in the Hellenistic- and Roman-period Mediterranean Basin. Combining the few texts that address the topic and scholarly inference from archaeological objects such as scribes’ writing kits, it is clear that there were a variety of methods for making ink.61 The simplest and presumably most widespread method was to mix a carbon-based soot such as lampblack or other burnt plant products (e.g., resin or pine wood) with a binding agent – typically gum Arabic, a naturally-occurring sap from species of the acacia tree family, but possibly also bone glue or other substances – and a suspension liquid such as water, oil, or vinegar.62 Pliny the Elder recorded that soot could also be scraped from furnaces at workshops made specifically for manufacturing ink and paint pigments.63 Ink production was an area of continual experimentation and, already during the Hellenistic and Roman periods, some ink recipes included metallic compounds rich in iron, copper, or lead. Such compounds could derive either from natural mineral sources like vitriols, minium, ochre, and hematite, or from the by-products of processed metals.64 These metallic additives may have contributed to coloration and adhesion, but they are known to have served as drying agents for the ink. Once these ingredients were mixed into a paste, the ink was dried and made into pellets or cakes for easy transport or storage until needed by a scribe.
A number of studies have confirmed that the inks used for the Qumran scrolls were primarily carbon or soot based. However, some of the ink studied by Nir-El and Broshi – in particular that of the Genesis Apocryphon (1Q20) – has unusually high levels of copper and lead compared with the other scrolls tested.65 The Genesis Apocryphon’s ink also has the peculiar characteristic of delaminating the skin in some places, in extreme cases “eating away” the skin completely to leave negative, ghost letters where the writing once was. The same phenomenon occurs among the Qumran Aramaic scrolls on 4Q115 (Dand), as well as on several Hebrew scrolls.66 Nir-El and Broshi surmised, based on the prior study of Haran, that the metallic compounds present in the Genesis Apocryphon’s ink must have come from its storage in a metal inkwell, since metallic inks had not yet been invented during the early Roman period.67 However, Pliny’s account makes clear that metallic compounds were intentionally added to carbon inks already in the first century CE, and likely well before that time. As Christiansen has shown, carbon inks and metallic inks were not the only two options, since what he called “mixed inks” – carbon-based recipes with metallic additives – were used during the period when the Qumran scrolls were written.68 Consequently, it is unnecessary to resort to a bronze or copper inkwell in explaining inks from the Qumran caves with metallic compounds. It seems more likely that lead- or copper-based additives were simply part of the recipes for some, but not all, of the inks used. However the high levels of copper and lead came to be present in some of the inks at Qumran, Cross assumed that they were responsible for the delamination of the skin on the scrolls exhibiting that feature.69 Nir-El and Broshi were more circumspect, admitting that the deterioration caused by the ink of the Genesis Apocryphon and similar scrolls may be attributed to multiple factors, including the reaction of binding constituents in the ink (e.g., vegetable gum or animal size) with environmental changes, or the lead and copper present in the ink.70
A mordant water, such as that suffused with the tannins of harvested oak galls or other tannin-rich plants, helped inks bond to the writing surface, and there is chemical evidence that such ink was used for at least some of the scrolls tested by Nir-El and Broshi.71 This water could have been added either during the initial stage of making the ink paste, or when water was added to the dried ink at the time of writing. Chemical tests on the Qumran scrolls suggest that sweet water was used at both stages, not water from the Dead Sea.
There is clear, compelling evidence that Jewish scribes during the Herodian and Early Roman periods used both ceramic and metal inkwells for mixing and holding their inks, though we possess no proof that the same was true for the preceding Hellenistic period.72 These inkwells were of the type used more widely in the Roman world, as seen by comparison of the inkwells found by de Vaux at Qumran with those depicted on the frescoes at Pompeii, and in various excavations around the Roman world.73 De Vaux announced three inkwells (two ceramic and one bronze) found during his original excavations, to which we may add a fourth discovered at the site by Steckoll and a fifth that was found at the nearby site of ‘Ain Feshkha but, according to the neutron activation analysis of Gunneweg and Balla, was produced at Qumran.74 In the same study, Gunneweg and Balla identified the unmistakable fragment of a sixth ceramic inkwell among the unpublished pottery collection from Qumran, though analysis showed that it had not been produced at the site.75 Finally, in 2001–2 Magen and Peleg discovered a seventh inkwell in an ancient dump to the east of Qumran, which Gunneweg tested in 2007 and determined was produced at Qumran.76 Several additional inkwells (both ceramic and metal) have been claimed to originate from Qumran, but their connection to the site cannot be verified.77 Of course, all of these inkwells pertain only to scrolls that may have been written at Qumran during the Roman period. Nevertheless, they suggest that inkwells of the type found at Qumran were used more widely by Jewish scribes during the Second Temple period.
The pens used to write our manuscripts were presumably made of appropriately-sized reeds (ca. 1 cm in width), which were sharpened to a triangular tip and split at the nib for better ink flow.78 Reeds of the sort needed grew at many places in Hellenistic- and Roman-period Palestine (and still today), including the Dead Sea region. Being made of perishable material and having little monetary value, no such pens have survived from Qumran or elsewhere in ancient Palestine, though examples from Roman Egypt and fresco images from Pompeii fit well the above description.
3.1.2 Scripts
Although each scribe had individual preferences and writing habits, the Qumran scrolls show clearly that well-established norms existed for writing that speak to the existence of guilds or shared communities of practice. Corpora such as the Elephantine papyri reveal that the role of scribe was often cultivated within a family setting, with scribal skills being handed on from father to son. In other social situations, such as that of Qumran, we might imagine non-familial apprenticeships and training arrangements. In both cases, it may be supposed that small scribal circles intersected with others in an interwoven network, comprising a wider scribal tradition across the Mediterranean Basin in antiquity. While we witness a basic continuity in the Aramaic square scripts used almost exclusively in the Aramaic Qumran scrolls, it is typically assumed that styles and aesthetic preferences shifted slowly over time, allowing for the typological, diachronic classification of scripts. For the scrolls studied here, the relevant periods commonly adopted in the scholarly literature are the Hasmonean (second–mid-first centuries BCE) and the Herodian (mid-first century BCE–70 CE).
Each script from among the Qumran scrolls may be placed somewhere on the related spectrums of formality – informality, and uniformity – variability. Classificatory terminology for Jewish scripts of the Second Temple period has varied, but the most commonly used taxonomy was heavily influenced by F.M. Cross, who proposed the four basic categories of formal, semi-formal, semi-cursive, and cursive.79 As Longacre recently observed, the scripts included by Cross and others under some of these headings vary significantly, limiting their usefulness.80 Rather than working with Cross’s formal-cursive system, some scholars have preferred to use a more intuitive formal-informal spectrum.81 Longacre has now proposed a more sophisticated taxonomy, advocating for the following script types: 1.) Ornate rectilinear, 2.) Ornate curvilinear, 3.) Simple rectilinear, 4.) Simple curvilinear, 5.) Semi-cursive, 6.) Cursive, 7.) and Extreme cursive. The first four categories belong to what Longacre and others call the “Square” scripts, with Semi-cursive being transitional to the “Cursive” scripts (the latter not being found among the manuscripts included in the profiles). Any of these script types can be written with varied “Levels of execution”: Calligraphic, Common, and Current.82 Longacre’s nuanced treatment has pushed the discussion of ancient Jewish scripts forward in important ways, and his insights will have to be accounted for in future palaeographic work. However, because his system has not yet been integrated into palaeographic assessments of most of the Qumran scrolls, in the profiles and in what follows I continue to use the system(s) developed by Cross and followed to various extents by Milik, Puech, and others.
I assume that scripts characterized by formality, squareness, ornateness, and high levels of uniformity are one indicator of a high-quality scroll, compared with more informal, rounded, simple, and varied (i.e., messy) scripts. As Longacre noted, formal scripts with high levels of execution demand considerable skill and attention from a scribe, and are more suitable to public uses.83 This observation is supported by a clear correlation between literary texts that were presumably held in high esteem and formal, square scripts written with high levels of execution, even if there are a number of exceptions for which we must account. By contrast, scripts with less formal characteristics, written with lower levels of execution, are very often used for mundane documents belonging to the private sphere (e.g., business documents, deeds, contracts, personal letters, and receipts). The latter class of scripts are obviously written with less care, the priority being function rather than aesthetic.
The scripts of the Aramaic scrolls from the Qumran caves range from formal to semi-cursive, for both the Hasmonean and Herodian periods, with formal Herodian period scripts showing an increasing penchant for calligraphic flourishes (see, e.g., 11Q10 [Job] and 1Q71 [Dana]). In a number of cases, we find a single literary composition written in scripts of notably different quality: the Enochic Book of Watchers (compare 4Q204 [Enc] and 4Q201 [Ena]) and Book of Giants (compare 4Q203 [EnGiantsa] and 4Q530 [EnGiantsb]), the Words of Michael (compare 4Q529 [Words of Michael] and 6Q23 [papWords of Michael]), the Visions of Amram (compare 4Q543 [Visions of Amrama] and 4Q547 [Visions of Amrame]), Tobit (compare 4Q197 [Tobb] and 4Q198 [Tobc]), and Four Kingdoms (compare 4Q552 [Four Kingdomsa] and 4Q553 [Four Kingdomsb]). If we allow as evidence the several copies of Daniel now containing only Hebrew portions of the book, that work should also be included in this list (compare 1Q71 [Dana] and 6Q7 [papDan]). As noted above, higher quality scripts are very often coordinated with wider margins, more generous use of vacats, and greater uniformity in spacing, making for a scroll with an orderly aesthetic that included more open space than scrolls of lesser quality, allowing for easier reading. Representative examples of the main classifications of script type found in the profiles are provided above.
Formal Hasmonean script (4Q543 [Visions of Amrama])
Semi-formal Hasmonean script (4Q542 [TQahat])
Semi-cursive Hasmonean script (4Q201 [Ena])
Semi-formal Herodian script (4Q536 [Birth of Noahc])
Semi-cursive Herodian script (6Q8 [papGiants])
Cursive scripts from the Judean Desert, as drawn by F.M. Cross
Letter height (to approximate relative scale)
Roughly 60% of the palaeographic dates are assigned to the first cent. BCE, making this far and away the century during which most of our copies are likely to have been made. The significant majority of scripts have a letter height (based on an average for medial letters only) between 2.5 and 3.5 mm, with a clear preference for slightly larger script sizes in papyrus scrolls compared with those of skin.
The social and geographic locations in which the Qumran scrolls were written has been a topic of vigorous study and debate, with an important facet of the discussion addressing possible cases of a single scribe copying multiple scrolls.84 Such cases have potential implications for the connections between different scroll caves, links between the caves and the site of Qumran, and other questions about the communities who created the scrolls kept near Qumran. I have gathered below cases where a scribe who wrote at least one Aramaic scroll is alleged to have written other scrolls as well. While this remains an area in need of further study, a very small number of cases in which a single scribe wrote multiple scrolls can be identified with a high degree of certainty.85
3.1.2.1 11Q18 (NJ) and at Least Nine Hebrew Scrolls
The highly-respected epigrapher Ada Yardeni proposed that over fifty manuscripts at Qumran and Masada were written by the same, Herodian-period scribe, including several of our Aramaic manuscripts.86 The Aramaic texts in her list are: 1Q32 (NJ?), 2Q24 (NJ), 3Q14 4 (Tob?) and 6, 4Q203 (EnGiantsa), 4Q531 (EnGiantsc), and 11Q18 (NJ). From an additional list of thirty-six manuscripts that Yardeni claimed were “perhaps also copied by this scribe,” the only Aramaic fragment belongs to a papyrus copy of Tobit in the collection of Norwegian antiquities collector Martin Schøyen, which an international group of scholars has now identified as likely to be a modern forgery.87 Yardeni placed heavy emphasis on what she called the scribe’s “peculiar lamed, with the ‘pressed’ and curved lower part.”88 Yet, she also allowed for “certain differences between groups of manuscripts” and development of the scribe’s writing style over time.89
Having examined images of many of the manuscripts discussed by Yardeni, particularly those written in Aramaic, I have difficulty accepting that all of them were written by one and the same scribe. This doubt is strengthened by the yet-unpublished research of Gemma Hayes, who is working as part of The Hands That Wrote the Bible project at the University of Groningen.90 Hayes’ research is based on a combination of computer learning models and traditional palaeographic analysis, and she has narrowed Yardeni’s long list of scrolls to only eight, which she argued can be assigned to an individual scribe with a high degree of probability based on letter formation and spelling practices. Her eight scrolls are: 4Q161 (pesher Isaiaha), 4Q166 (pesher Hoseaa), 4Q171 (pesher Psalms), 4Q397 (MMTd), 4Q439 (Lament by a Leader), 4Q215 (TNaphtali), 4Q474 (Text Concerning Rachel and Joseph), and 11Q18 (NJ). To this list we may confidently add 4Q175 (Renewed Earth), as observed by Eibert Tigchelaar.91 The scribe of these nine scrolls likely worked from around the mid-first century BCE to the early first century CE, based both on the traditional palaeographic dates of Cross and Yardeni and the digital writer identification methods of Hayes and her colleagues in Groningen, which are corroborated by a new Carbon-14 date for 4Q161 (pesher Isaiaha). This Herodian-period scribe was highly-trained and very adept, writing in an elegant script classified by Cross and those after him as “round semiformal.”92
The scripts of the eight scrolls identified by Hayes and 4Q175 (Renewed Earth), as a subset of Yardeni’s much larger group, are virtually identical. Considering the similar spelling conventions and other scribal preferences found in these nine scrolls, it is very likely that they were penned by the same scribe. Among the implications of this finding, I highlight only that the same scribe who wrote 11Q18 (NJ) also copied some of the most distinctive Hebrew sectarian works: the pesharim and MMT. This concrete scribal connection between the Aramaic and sectarian literatures attests to the active, ongoing interest in earlier Aramaic writings among those belonging to the Essene group responsible for collecting and curating the Qumran library.
3.1.2.2 4Q113 (Danb), 1Q11 (Psb), 4Q57 (Isac), and 11Q14 (Sefer ha-Milhamah)
In 2008, Eugene Ulrich identified a scribe who wrote in a hand “among the most careful, stately, and elegant seen in the Qumran collection,” as responsible for writing 1Q11 (Psb), 4Q57 (Isac), and 11Q14 (Sefer ha-Milhamah).93 His argument was based on general features of the scribe’s style – a notably angled stance, letter size and spacing, use of palaeo-Hebrew script for the Tetragrammaton – and more detailed comparison of letter formation. As Ulrich and others have observed, scripts can be very similar and still be the work of different scribes, since scribes writing formal scripts, especially, seem to have worked to a common standard. Nevertheless, in this case there are good grounds to believe that a single scribe wrote these scrolls.94 Several years after Ulrich’s article, Sidnie White Crawford related that Ulrich had added 4Q113 (Danb) to the list of scrolls attributed to this scribe.95 The identification was cautiously endorsed by Tigchelaar and we await the judgement of Hayes and the Groningen team.96 If we assume that the scribe of 4Q113 (Danb) also wrote the other scrolls identified by Ulrich, then we have another case in which a scribe copied both a partially Aramaic composition and a Hebrew sectarian one. Of course, we must bear in mind the somewhat ambiguous situation of Daniel as a bilingual work in its final form.
3.1.2.3 4Q207 (Enf) and 4Q214a–b (Levie–f)
According to Milik, the same scribe wrote both 4Q207 (Enf) and the manuscript he called 4Q214 TestLevib, designated by Stone and Greenfield as 4Q214a–b (Levie–f; see discussion below). Despite the small amount of text preserved on 4Q207 (Enf), the scripts in these manuscripts are exceptionally coherent, and the scribal connection between them looks solid.97 For further discussion of the complexities involved, see the profiles for 4Q207 (Enf) and 4Q214a (Levie).
The following three cases are of a different nature than those above, since in each case it has been argued that what some have designated different scrolls of the same or closely related works are, in fact, part of the same scroll written by a single scribe.
3.1.2.4 4Q213 (Levia), 4Q213a (Levib), 4Q213b (Levic), and 4Q214 (Levid)
Milik considered what would eventually be labelled by Stone and Greenfield as 4Q213 (Levia), 4Q213a (Levib), 4Q213b (Levic), and 4Q214 (Levid) to be a single manuscript, which Milik designated 4Q213 TestLevia.98 Since the editio princeps of Stone and Greenfield in DJD 22, their division of the fragments into four scrolls has generally prevailed, but Hanneke Van der Schoor argued convincingly in a recent article that Milik’s previous conclusion was correct.99 Van der Schoor’s argument rests largely on the legitimate point that informal scribal hands, such as that of the scribe discussed here, could vary appreciably even within the same fragment, something for which Stone and Greenfield did not adequately account. Despite this expected variation, the script of the fragments discussed by Van der Schoor is remarkably coherent. For further discussion of this scribe’s style, which is quite messy and irregular, see the profile for 4Q213 (Levia).
3.1.2.5 4Q214a (Levie) and 4Q214b (Levif)
A similar situation obtains for 4Q214a (Levie) and 4Q214b (Levif) as for the Aramaic Levi copies discussed above. Milik had originally posited a single scribe and manuscript, designated 4Q214 TestLevib, which was later split into two copies by Stone and Greenfield: 4Q214a (Levie) and 4Q214b (Levif).100 As with Milik’s 4Q213 (TestLevia), Van der Schoor has claimed that Stone and Greenfield’s division into two scrolls was unwarranted, and that the fragments included by them under 4Q214a (Levie) and 4Q214b (Levif) are best assigned to a single scroll, written by one scribe.101 This scribe wrote in an upright, neat script, likely during the Hasmonean period.
3.1.2.6 4Q542 (TQahat) and 4Q547 (Visions of Amrame)
Émile Puech suggested that these two manuscripts were written by the same scribe “à quelques années de distance.”102 The identification is, in my opinion, very convincing based on the striking similarity of the scripts and other scribal factors in these two scrolls. In fact, I have argued elsewhere that that 4Q542 (TQahat) and 4Q547 (Visions of Amrame) belong to one and the same scroll.103 Puech gives no explanation for his opinion that the two scrolls were written several years apart, and I see no firm basis for making this assertion given the expected variation in less formal hands. This scribe wrote in a relatively informal, irregular semi-formal style, and Puech dated his scribal activity to the early Hasmonean period.
3.1.3 Spacing and Vacats
Spacing conventions varied from scribe to scribe, but fell within a well-established range of practice. For columns that had been dry-ruled (i.e., in most cases), line spacing was already determined and needed only to be followed by the scribe. When there was no ruling on a scroll the scribe had more freedom to determine the spacing, both in the leading (where horizontal script lines were not present) and in the length of lines (where vertical column lines were not present, in whole or in part). In such cases, scribes who wrote in a more formal, consistent hand tended to have more regular spacing, resulting in a neater overall appearance for the scroll. Scribes who wrote in an informal, untidy script often had more erratic spacing practices as well, as we see, for example, in 4Q542 (TQahat)/4Q547 (Visions of Amrame).
Word spacing varies, from what appears almost to be scripta continua – with no more space between words than is typically left between letters within a word (1 mm or less) – to spaces of approximately 5 mm. Both of these extremes are very rare, with the large majority of scribes leaving an average of 1.5–2 mm between words. As was observed for line spacing and line length above, there is a recognizable connection between less formal scripts and more erratic word spacing practices, while scribes who wrote in more formal, regulated scripts tended to have more regular spacing between words. In a few cases, we find scribes who sought to keep a neat left side of the column, “justifying” the left margin by leaving a larger than usual space between the final words of a line when called for (see Figure 18, above). In our corpus, this occurs in 4Q203(EnGiantsa)/4Q204 (Enc), 4Q552 (Four Kingdomsa), and 4Q554 (NJa), the same practice being found in a number of Hebrew Qumran scrolls.104
Vacats provided a way for scribes to indicate what they considered to be breaks or pauses between small or large “sense units” of a text, and in this way vacats assisted with the mental work of reading a scroll’s contents. I consider the use of vacats to be one likely indicator of a manuscript’s quality. Considered together with a scroll’s formality and evenness of script, neatness, margin size, number of corrections, and other aspects of spacing, the presence of very few or no vacats (where we have enough text to determine this) generally indicates a scroll of relatively low quality, while many vacats – especially vacats of varying sizes, reaching to half a line or greater – generally indicates a scroll of relatively high quality. In most cases, the factors just listed coincide with one another to provide a robust overall sense of a manuscript’s quality. In some of the highest-quality scrolls, such as 4Q203(EnGiantsa)/4Q204 (Enc), 4Q531 (EnGiantsc), 1Q20 (apGen), and 11Q10 (Job), we find variation in the size of vacats in order to signal lesser or greater breaks in the flow of the text, with up to a full line or more to signal major pauses. In doing so, a scribe prioritized an easy, aesthetically pleasing reading experience over the preservation of precious scroll materials.
A column with small, medium, and large vacats (1Q20 [apGen] 22)
A column with no vacats (4Q212 [Eng] 1iii–v)
3.1.4 Other Scribal Peculiarities
In addition to the expected writing of a scroll in an Aramaic script, scribes would, on rare occasion, use practices that were unusual judged against the wider backdrop of the Aramaic scrolls studied in this book.
3.1.4.1 Interchanging Medial and Final Letter Forms
Scribes sometimes used medial letter forms in final position, or (much less often) vice versa. The reason for this practice is not always clear, but in some cases it may indicate a poorly-trained or careless scribe. In a considerable number of examples, such variation aligns with poorly-executed, informal scripts. Medial and final letter forms are occasionally exchanged in 4Q201 (Ena) and 4Q539 (TJoseph). 4Q541 (apocrLevib?) and 4Q213 (Levia) have medial pe in final position (these letters are unusually large in 4Q213), 4Q546 (Visions of Amramd) and 4Q560 (Magic Booklet) do the same thing with medial kaph, and the scribe of 4Q202 (Enb) used the medial forms of both letters at the ends of words. 4Q242 (PrNab) and 4Q553a (Four Kingdomsc) have medial mem in final position, and 11Q18 (NJ) does the same for tsade. 4Q208 (Enastra) has only medial letter forms at the ends of words. The scribe of 4Q542 (TQahat), however, takes the prize as most erratic in this respect, consistently using the medial kaph and tsade in final position, and less consistently placing final mem and nun in medial position.
3.1.4.2 Cursive Letter Forms
On occasion, we find that a scribe used cursive letters known from outside of the Qumran Aramaic corpus, but not common within it. A cursive form that can hardly be called rare among our scrolls is the looped tav, which is used in a significant minority of scrolls, often in combination with the more formal, square tav. Far less common is the cursive mem, used by the scribes of 4Q212 (Eng), 4Q558 (papVisionb), and 6Q23 (papWords of Michael). Unique in the corpus is the cursive aleph written by the scribe of 6Q8 (papGiants), which is accompanied by the equally rare practice of distinguishing between medial and final forms of the same letter.
3.1.4.3 Number Symbols
Our scrolls attest to scribes both writing out numbers in full and using shorthand numeric symbols, the latter derived from Hieratic but common to Aramaic literature of the Persian to Roman periods.105 Scribes who employed this system wrote 4Q318 (Zodiology and Brontology), 4Q540 (apocrLevia?), 4Q554 (NJa), 4Q558 (papVisionb), and 4Q559 (Biblical Chronology). Among these, the scribe of 4Q318 uniquely distinguished between medial and final forms of the symbol for number one.
Palaeo-Hebrew divine name (4Q243 [psDana] 1.2)
3.1.4.4 Palaeo-Hebrew Script for a Divine Name
In a single case – 4Q243 (psDana) 1.2 – we find that a scribe used palaeo-Hebrew script to write the divine name “your God” (
3.1.4.5 Tetrapuncta for a Divine Name
Another unique scribal practice among the Aramaic Qumran scrolls is the repeated use of tetrapuncta – four successive dots or strokes of ink – by the scribe of 4Q196 (papToba) as a substitute for a divine name. For some years it was assumed that the four dots in 4Q196 represented the four letters of the Tetragrammaton, since that specific substitution is well-known from Hebrew scrolls at Qumran.107 However, further research has shown that the dots in our scroll represent instead the name
3.1.4.6 Dicolon Symbol to Indicate a Break between Sense-Units
The scribe of 4Q156 (Lev?) repeatedly used two vertically-oriented dots – sometimes called a dicolon in the literature – to mark a minor sense-division in the text. This practice is unique to 4Q156 (Lev?) among the Aramaic scrolls from Qumran, though it is also used in 4Q364 (Reworked Pent B) and in Greek texts contemporaneous with the Qumran scrolls. For further details on use of the dicolon, see the profile for 4Q156 (Lev?).
3.2 Correction, Supplementation, and Secondary Use of Completed Scrolls
The stories of our valuable scrolls were only beginning once they had been written by their original scribes, and among the manuscripts studied here we find many signs of ongoing use, correction, and repair. Several distinct practices were employed for marking deletions and other paratextual or secondary additions. These are presented in groups below, according to the type of scribal practice.
3.2.1 Correction through Deletion
Scribes regularly made mistakes in their work, though we see strikingly different rates of such errors depending on the individual scribe. In general, there is a clear correspondence between less formal, more erratic scripts and a higher frequency of scribal mistakes, both factors contributing (often together) to what I take to be a lower quality copy. Our scribes had a few ways in which such mistakes could be corrected. In situations where a letter or word had been written accidentally or incorrectly – as decided either by the original scribe or a later reader – the mistake could be either marked so that later readers were aware not to read, pronounce, or copy it, or scraped from the skin with a sharp instrument, and so erased. In a few cases (e.g., 4Q213a [Levib] and 4Q548 [Visions of Amramf]) we find both types, signifying either distinct stages of correction or mixed usage by a single correcting scribe.
Partially extant tetrapuncta (4Q196 [papToba] 18.15)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Shai HaleviDicolon (two dots) to signal minor sense-division (4Q156 [Lev?] 1.6)
3.2.1.1 Erasure by Scraping
The most common way of deleting a letter or word in our corpus written by mistake was to scrape the skin’s surface with a sharp instrument. It is often impossible to tell with certainty if this was done by the original scribe or a later corrector, but occasionally it seems that the original scribe realized his mistake while writing and erased the letter(s) in question once they were dry (e.g., if he stopped writing mid-word). The practice of erasure by scraping is attested in around a dozen of the scrolls left to us: 4Q115 (Dand), 4Q202 (Enb), 4Q209 (Enastrb), 4Q210 (Enastrc), 4Q212 (Eng; uncertain), 4Q213a (Levib; uncertain), 4Q529 (Words of Michael), 4Q531 (EnGiantsc), 4Q541 (apocrLevib?; uncertain), 4Q542 (TQahat), 4Q548 (Visions of Amramf), 4Q551 (Narrative), 4Q554 (NJa; uncertain), 11Q10 (Job), and 11Q18 (NJ).
Erasure of one letter and at least two words by scraping (11Q10 [Job] 21ii.5–6)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Najib Anton Albina3.2.1.2 Deletion by a Strikethrough Line
The second most popular way of marking the deletion of a letter or word was to use a secondary line of ink through the cancelled portion of text. A single letter is typically struck through with a vertical line, as in 4Q196 (papToba) and 4Q530 (EnGiantsb). However, a horizontal line seems to strike a single letter in 4Q531 (EnGiantsc), while the scribe of 4Q213a (Levib) used lines in both directions. Horizontal lines mark the deletion of more than one letter in 4Q212 (Eng), 4Q530 (EnGiantsb), 4Q533 (EnGiantse), and 4Q554 (NJa).
Deletion of a letter with a vertical line (4Q530 [EnGiantsb] 7i.1)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Najib Anton AlbinaDeletion of a word with a horizontal line (4Q533 [EnGiantse] 3.2)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Najib Anton Albina3.2.1.3 Deletion by Cancelation Dots
A third, relatively rare way of signaling a cancelled letter was to mark it with one or more ink dots. A single dot above the intended letter was employed by the scribes of 4Q208 (Enastra), 4Q112 (Dana), and 4Q548 (Visions of Amramf), while the scribes of 1Q20 (apGen) and apparently 4Q542 (TQahat)/4Q547 (Visions of Amrame) used dots both above and below a letter. For the many ink dots in 4Q213 (Levia) – which are probably accidental – see the profile for that scroll.
Deletion with cancellation dot above the letter (4Q208 [Enastra] 16.4 and 18.2)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Najib Anton AlbinaDeletion with cancellation dots above and below a letter (1Q20 [apGen] 5.9)
3.2.2 Correction through Supplementation
In some cases, the original scribe of a scroll or a later, correcting scribe considered a letter, word, or phrase to be missing from the text as first written. These situations were typically corrected by supplementation of the original text with letters or words placed directly above the spot in need of emendation, though there were other methods, too.
3.2.2.1 Supralinear Additions
Additions of a single letter above the spot in need of correction is quite common among the Qumran Aramaic scrolls, as it is in the Qumran corpus more generally. A majority of the scrolls studied here have small corrections of this sort; they were very common. For most supralinear corrections it is difficult to determine with certainty whether they were done by the original scribe at or near the time of initially writing the text, or at a later time by a different scribe. In some cases, however, a scribal hand is sufficiently distinctive relative to the original hand that this determination can be made. In 11Q10 (Job), for example, the original scribe seems very likely to have made such corrections. Scrolls in which the correcting hand appears to be different than the original one include 1Q23 (EnGiantsa), 4Q246 (apocrDan), 4Q542 (TQahat), and 4Q550 (Jews at the Persian Court). Sometimes a repeated correction suggests a secondary corrector, as we find for the word
Supralinear addition of a letter (4Q531 [EnGiantsc] 17.1)
Scribal insertion mark below supralinear word (4Q196 [papToba] 6.8)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Najib Anton AlbinaSupralinear addition of multiple words (4Q202 [Enb] 1ii.2)
3.2.2.2 Sublinear Additions
The placement of a supplemental letter under the text to be corrected does occur, but much less frequently than those above the text. Such sublinear additions are found in 4Q559 (papBiblical Chronology) and, apparently, 4Q246 (apocrDan). The letter in the latter scroll – seemingly a large, cursive mem written in a hand different than the main text – is the subject of some debate.
Sublinear letter added (4Q559 [papBiblical Chronology] 4.5)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Najib Anton Albina3.2.2.3 Vertical Insertion of Text
Supplementation of the main, horizontally-oriented text with an addition that runs vertically along a column is known from a number of Hebrew Qumran scrolls, most famously 1QIsaa. We find only one such addition among the Aramaic scrolls, in 4Q542 (TQahat).
Vertical insertion of text (4Q542 [TQahat] 3)
3.2.2.4 Overwriting/Conversion of a Letter
Another way of correcting an existing letter or word was to write over it in an attempt to change the reading, a practice that is fairly widespread in our corpus and is mainly limited to scrolls written in informal, untidy scripts. At times, this practice is combined with the erasure of letters by scraping, as in 4Q212 (Eng). This type of correction is sometimes difficult to discern, but overwriting one or more letters in a word does or may occur in 4Q196 (papToba), 4Q208 (Enastra), 4Q210 (Enastrc), 4Q212 (Eng; entire word overwritten), 4Q530 (EnGiantsb), 4Q541 (apocrLevib?), 4Q542 (TQahat), and 4Q560 (Magic Booklet).
Tav written over another letter (4Q208 [Enastra] 5.2)
Courtesy of The Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority. Photo: Najib Anton Albina3.2.3 Other Paratextual Letters and Symbols
Some paratextual scribal marks have been mentioned above, such as the letters at the top, righthand corners of sheets on 1Q20 (apGen) and perhaps 4Q529 (Words of Michael), or the large, cursive sublinear mem (less likely, bet) on 4Q246 (apocrDan). To this we should add the large, oblique mem written in a hand other than the main text on the bottom margin of 4Q546 (Visions of Amramd) 9, the purpose of which is unclear.
Several times in the Aramaic scrolls studied here we find scribal symbols in the intercolumnar margins of scrolls evidently related to reading practices, called by Tov and others paragraphos symbols drawing on the study of Greek manuscript.110 Indeed, signs very much like those found in the Qumran scrolls are used in Greek manuscripts from the Hellenistic and Roman periods, and one can reasonably posit influence from Greek scribal practices on Hebrew and Aramaic ones.111
A straight, horizontal paragraphos line is used in the intercolumnar margins of 4Q532 (EnGiantsd) and 4Q542 (TQahat). The narrative context of the symbol in the former scroll is unclear, though Puech (DJD 31:103) suggested that it may be preceded by a vacat. In the latter scroll, however, the symbol clearly corresponds with a narrative pause, and so presumably aided readers in finding the break between two sections of the work. A somewhat differently shaped symbol of the so-called “fishhook” type occurs twice in 4Q213 (Levia)/4Q213a (Levib), and in each case we can see that the symbol corresponds to the beginning of a new narrative unit. It seems reasonable to assume that these marks were made after the text was initially written, an assumption supported in 4Q532 (EnGiantsd) and 4Q213 (Levia)/4Q213a (Levib) by the fact that the symbols are written in noticeable lighter ink than the main texts.
Marginal paragraphos symbol (4Q532 [EnGiantsd] 1ii.7)
Marginal “fishhook” paragraphos symbols (4Q213 [Levia] 1ii.11 and 4Q213a [Levib] 2.10–11)
Finally, the X mark at the blank beginning of a line on 4Q556a (Prophecyb) should be mentioned. The context of the symbol is now lost, but the fact that the X is evidently placed in a vacat suggests that it served to mark a matter of importance for readers of the text. This suggestion is supported by the several times a similar symbol occurs in the Hebrew Qumran scrolls (e.g., 1QIsaa XXVI.9, XXXV.10; 4Q177 [Catena A] 12–13ii.9, 29.2; and 4Q417 [Instructionc] 4.1). Assuming that this is correct, we find three types of symbol used to mark literary units for readers in the Aramaic scrolls from the Qumran caves: a straight horizontal line, a hooked line, and an X mark. All three marks are also well-attested in the corpus of Hebrew scrolls.
3.2.4 Opisthograph: 4Q201 (Ena)/4Q338 (Genealogical List?)
A single opisthograph has been identified among the Aramaic Qumran scrolls, with 4Q201 (Ena) being reused to write 4Q338 (Genealogical List?) on the verso (i.e., flesh) side of the skin. The relationship between the two texts, if there is any, can no longer be ascertained, though Milik and Tov entertained the possibility of a connection of both works with the biblical patriarchs.112 For further discussion, see the profile for 4Q201 (Ena).
4 Concluding Observations
4.1 Historical Context and Transmission
Viewed as an entire corpus, the Aramaic fragments studied in this book represent nearly ninety scrolls written over a period of more than two centuries, from roughly the late third or early second century BCE (4Q208 [Enastra]) to the mid-first century CE (11Q10 [Job]). The scrolls represented are overwhelmingly written on skin, though a small percentage are made of papyrus. In several cases, the same literary work is found on scrolls made of both materials: the Book of Giants, the Words of Michael, Tobit, and Daniel if we take into account 6Q7 (papDan), now containing only parts of the last Hebrew chapters of that book. The only clear case in which we are not dealing with a scroll is 4Q339 (List of False Prophets), for which the distinctiveness of the physical medium (a rectangular “card” of skin) matches that of its literary contents (a list). Indeed, to class this text among the Aramaic literature at Qumran is open to debate, since only the list’s heading is written in that language.
There is compelling evidence that some Aramaic scrolls in our corpus were written by the same scribe, while others were written by a scribe also responsible for one or more of the preserved Hebrew scrolls. It stands to reason that in these instances – especially when we find an Aramaic scroll written by a scribe who also wrote Hebrew sectarian literature – the likelihood of that scroll being produced at or around Qumran increases significantly. Other practices supporting a connection with scribes from the Essene group(s) responsible for the sectarian literature (at Qumran or elsewhere) are the use of palaeo-Hebrew script and tetrapuncta to write the name
Florentino García Martínez observed that there are no differences in the scribal practices between the Aramaic and Hebrew groups of scrolls from the caves around Qumran.113 One could, of course, find minor scribal details that occur in one group and not the other, but taking a wide view of the entire Qumran corpus García Martínez is essentially correct: the media, production techniques, scripts, and other scribal practices are remarkable consistent across the Aramaic and Hebrew scrolls. All of these scrolls were demonstrably part of one and the same Jewish scribal culture, which was itself well-integrated with the interconnected scribal guilds of the eastern Mediterranean Basin during the Second Temple period. If the scholarly consensus view that much of the Jewish Aramaic literature attested in our scrolls was composed during the Hellenistic period is correct – and there is every reason to believe that it is – then we can assume that the very large majority of our scrolls, if not all of them, are copies of earlier exemplars.
As seen especially in the preceding chapter on language, the scribes writing our scrolls felt comfortable introducing a limited amount of change during the copying and correction processes. Such change was typically restricted to several aspects of orthography and morphology, though occasionally it extended to more extensive modifications in phrasing. There is little evidence in our corpus to support the idea of scribes making major alterations to a composition during the process of transmission, though we must admit the absence of sufficient textual overlaps among multiple copies to make bold claims on this front.
The fact that many of the Aramaic compositions stored and used at Qumran were copied into the Herodian period proves their enduring appeal to at least some Jewish communities over several centuries. This appeal obviously included those associated with the site of Qumran, where the Aramaic literature was kept and copied. However, with the exception of Daniel, which begins and ends with sections written in Hebrew, the rabbinic circles that eventually rose to social prominence in later Roman-period Palestine do not seem to have carried forward most of this literature. Despite this apparent decline in interest, we do find some meager evidence of ongoing Egyptian Jewish transmission in the Cairo Geniza, with the Aramaic Levi Document and a Hebrew translation of Tobit, alongside the Christian preservation of Tobit in Greek and Latin translations, an Enochic collection eventually culminating in Ethiopic 1 Enoch, and some earlier Aramaic didactic literature (the Aramaic Levi Document, and perhaps other, now-lost works of a similar ilk) reworked into the Greek Testaments of the Twelve Patriarchs.
4.2 Indicators of Manuscript Quality
It was suggested above that a variety of factors coalesce to give an indication of the quality of a manuscript in antiquity, and that some of these factors may be weighted more heavily than others. One helpful way to think about these factors is to imagine a spectrum of quality, which in our case would range from scrolls deemed “excellent” on one end to those considered “fair” on the other (the hypothetical category “poor” being reserved for documentary and other texts of lower quality than any of our scrolls). We may then propose exemplar scrolls at the two ends of the spectrum, against which other scrolls may be held up for comparison.
At the “excellent” end of the quality spectrum we have scrolls like 4Q209 (Enastrb) and 1Q20 (apGen), written in very neat, formal scripts on high-quality skin that is generally free of defects.114 On such scrolls we find generous margins, even ruling and spacing with regular vacats that range in size (up to and sometimes exceeding a full line), and few scribal mistakes or corrections beyond the occasional addition or deletion of single letters. All of these factors speak to an intentionally high level of care and professionalism, and of a shared regimen of specialized training resembling that outlined by Johnson for the bookrolls of Oxyrhynchus.115 The very best scrolls are at times large in format, approaching 25–3o cm in height, such that they align with what Tov called de luxe editions drawing on terminology used in the study of Greek bookrolls.116 Scroll size, however, does not seem to be the most salient indicator of overall quality, since we find a few larger scrolls that do not exhibit the characteristics listed above, along with several smaller scrolls that do. As the scroll height increases, the typical dimensions of the writing block tend to change correspondingly, with taller scrolls having higher ratios of height to width. Considering the data of this chapter, there is some reason to believe that the size of a scroll was often more closely related to the length of the text(s) it was planned to contain than to its intended quality. The social functions of high-quality scrolls are not entirely clear, but their production obviously required an elevated investment of money, skill, and time. Such an investment is consistent with more public-facing or formal uses for a scroll, and speaks to the high social value placed on the text being written and the knowledge it conveyed.
Excellent original quality manuscript (4Q209 [Enastrb] 23)
At the “fair” end of the spectrum, by contrast, we have scrolls like 4Q212 (Eng) and 4Q542 (TQahat)/4Q547 (VisAmrame), written in informal scripts that tend towards cursive features and betray noticeably less attention to the consistency and tidiness of letters or the evenness of ink coverage (sometimes requiring re-inking). Spacing between letters, words, and lines is more erratic and crowded, giving the overall impression of less open space on the scroll. Margins tend to be small and uneven, mistakes or corrections are considerably more frequent than in high-quality scrolls, and vacats are either not used or are small when present.117 All or some of these features contribute to a scroll that is more arduous to read. In some cases, such as that of 4Q542 (TQahat)/4Q547 (VisAmrame), the skin is also of an obviously lower grade, marked by imperfections and repairs already when the scroll was first written. We must be cautious of too confidently linking the scribal features just listed with assumptions about why these poorer-quality scrolls were made or how they were used. It was suggested above that the investment required for a top-quality scroll may reflect the high social value placed on its contents, plausibly signaling a certain public or symbolic significance. However, it is more problematic to claim that lower levels of scribal execution and material fineness in a scroll reflects a proportionately low regard for its literary contents by whoever commissioned or made it. One reason for caution on this front is that we have several literary works with copies occupying both ends of the quality spectrum. It is, of course, possible – perhaps even likely – that the various copies of a literary work were written in disparate geographic and social locations, and that these locations entailed differing views of the work being copied; we simply lack the evidence to determine the nature or extent of such differences. On the other hand, suggestions that low-quality scrolls denote intended uses that were less formal and public – such as “private,” “individual,” “study,” or “working” copies – have some logical appeal.118 In this case, a scroll’s quality is tied more closely to its intended social location and function than to an estimation of its contents.
Good original quality manuscript (4Q213a [Levib] 1, 2)
Fair original manuscript quality (4Q542 [TQahat] 1)
Papyrus scrolls were evidently made according to somewhat different quality standards than those on skin. While scripts and scribal features such as vacats and corrections on papyrus scrolls tend to correspond to those found at the lower end of the quality spectrum for skin scrolls, the factors of scroll size, margin size, and openness of line and word spacing often resemble higher-quality skin scrolls. We cannot be sure of the factors involved in these differences, but it is reasonable to infer that they are best explained by the relative cheapness of papyrus compared to skin. This cheapness would have allowed scribes to dispense with some of the frugalities exercised when copying low-quality skin scrolls, while using levels of care and scribal precision falling well short of those on the best skin scrolls.
The discussion above is grounded in observable differences among the scrolls available to us, but it is important to admit the extent to which it is both tentative and heuristic. There are many factors involved in the production of our scrolls of which we cannot, at present, be sure. We might take 4Q201 (Ena) as an example, which according to Milik “seems to have been made from a very old copy, dating from the third century at least,” and “does not fit very well into the scribal traditions of the Jewish copyists of Judaea or even Egypt; the scribe would perhaps be dependent upon the Aramaic scripts and the scribal customs of Northern Syria or Mesopotamia.”119 Beyer, in similar fashion, sees reason to place the copyist in the northern Transjordan.120 Puech and Drawnel wisely dismissed such speculation as unverifiable and, therefore, unconvincing, Drawnel suggesting that “[w]ithout indicating the place of the composition of the text found in 4Q201, the script shows several affinities with the semicursive ‘Idumaean’ bookhand.”121 Still, Milik’s and Beyer’s theories raise the important question of how differing geographic locations, times of copying, and distinctive scribal microcultures or communities of practice may have influenced how a scroll was prepared, written, and corrected. A scroll written by a scribe in northern Palestine during the late third century BCE was presumably governed by somewhat different conventions and expectations than one copied at or around Qumran in the early first century CE, though it is now largely beyond our ability to judge accurately in which ways, and to what extent, those conventions and expectations differed. While we await new insights from pathbreaking technologies for assessing the dates and origins of the skins and inks used to write the scrolls, and for discerning the scribal hands represented in them, we must glean what we can from this rich source of Jewish scribal culture in antiquity.
Current scholarly consensus holds that the site of Qumran was settled sometime during the early first century BCE, while a portion of the scrolls kept in the caves around Qumran predate that time. A cogent defense of the first-century BCE settlement of Qumran can be found in Mizzi and Magness, “Qumran.” For a recent appraisal of the manuscript evidence against the archaeological backdrop of the site see White Crawford, Scribes, 141, 317–20. Emanuel Tov (Scribal Practices, 5) observed that, “It appears that many, if not most, of the literary texts found in the Judean Desert had been copied elsewhere in Israel. Therefore, the contents and scribal practices reflected in them represent not only the persons who passed through, lived, and wrote in the Judean Desert, but to an even greater extent the culture and scribes of Palestine as a whole.” Tov admitted that this opinion is difficult to verify at present, though it does have logical appeal due to the great number of scrolls kept at the site, evidently written by hundreds of scribes (Scribal Practices, 14–15).
The number of studies of this sort are expanding rapidly. The following articles are a representative sample, and provide a more extensive bibliography of the existing research. On the ink of the scrolls, see Nir-el and Broshi, “Red Ink”; Nir-el and Broshi, “Black Ink.” On the manuscript materials, see Wolff et al., “Provenance”; Rabin and Hahn, “Characterization”; Schuetz et al., “Temple Scroll”; Anava et al. “Genetic.”
See Rabin, “Archaeometry.”
On the social value ascribed to bookrolls in Greco-Roman antiquity see Johnson, “Reading,” 612–15.
See, for example, Johnson, “Libraries,” 359–61.
For the text and translation see Cicero, Letters, 2.276–77 (the Loeb letter number is 212 [XIII.77]).
Pliny the Younger, Letters, 178–79.
The books were the only items kept personally by the king, bestowed on his sons, “who were devoted to learning.” Plutarch, Lives, 430–31 (Aemilius Paulus 28.6). See further Houston, Libraries, 34–7.
On the potential differences between reading communities and reading events, see Johnson, “Reading.”
On the Oxyrhynchus papyrus bookrolls, see the excellent study of Johnson, Bookrolls, whose work influenced considerably my own approach in this chapter.
Unless otherwise noted, none of the scroll images used in this chapter are to actual scale. They are for the purposes of illustrating manuscript features and scribal practices only.
Tov, Scribal Practices. Cf. Johnson, Bookrolls.
I use the term “(prepared) skin” here in an effort to avoid the confusion in terminology often found in the literature, at the risk of sacrificing elegance for utility. One finds the terms leather, parchment, or vellum used of the scrolls by various scholars, though each of these terms can, in some contexts, imply specific methods of preparation that do not clearly apply to the Qumran scrolls. For further discussion and bibliography see Tov, Scribal Practices, 34–35, who prefers the term “leather.”
Anava et al., “Genetic,” 1220. None of the scrolls tested belong to the Aramaic manuscripts studied here. It should be added that Woodward et al., “Parchment,” 228, found that 11Q19 (Temple Scrolla) is made from goat and ibex skins. See also Parry et al., “Advances,” 505–6.
Poole and Reed, “Preparation,” 17.
Driver, Documents; Naveh and Shaked, Bactria. The major innovation represented in the Qumran scrolls is the stitching together of multiple skin sheets to form a longer scroll, an innovation possibly adapted from longstanding papyrus bookroll technologies.
The process is helpfully described and explored by Poole and Reed, “Preparation.” See also Tov, Scribal Practices, 33–35; Reed, “Tannery”; and Bond, Trade, 112–14, who cites a number of the ancient sources on the process.
As observed already by Sukenik (Scrolls, 25), and later confirmed by Poole and Reed (“Preparation”), the Qumran scrolls were almost always written on unsplit skin of the Gevil type, though Tov (Scribal Practices, 35) noted the important exception of 11Q19 (Temple Scrolla). This copy of the Temple Scroll is physically unique among the Qumran scrolls corpus, on which see now Schuetz et al., “Temple Scroll.”
Poole and Reed, “Preparation,” 12.
See the description of Bar-Ilan, “Writing,” 996.
Again, 11Q19 (Temple Scrolla) is an exception, on which see Schuetz et al., “Temple Scroll.”
Reed, “Tannery”; Magness, Archaeology, 215–16.
Tov, Scribal Practices, 36.
See, e.g., Tov, Scribal Practices, 37.
E.g., White Crawford, Scribes, 156.
Bein and Horowitz, “Papyrus”; Zohary, Plant Life; Zohary, Plants, 137. See also the comments of Pliny the Elder, Natural History, 13.73, who notes that Cyperus papyrus grew as far north as Syria.
Josephus, War 1.130; Josephus, Antiquities 14.33.
Kraeling (“Place Names,” 201, n. 14) states with reference to Jericho that “the papyrus plant, from which the name is derived, is found there, of course, but it is also found in many other places.”
The classical description of the process is the much-discussed account of Pliny the Elder, Natural History, 13.74–83. For discussion and bibliography, see Johnson, Bookrolls, 85–91, 141–43.
For a helpful, comparative overview of the evidence at Oxyrhynchus, see Johnson, Bookrolls, 100–41. Johnson (128) observed a tendency for earlier, Ptolemaic-period bookrolls to have wider columns relative to their height than later, Roman-period ones.
See Tov, Scribal Practices, 44.
6Q7 (papDaniel), a papyrus copy of at least part of Daniel, might be considered as part of this discussion, since it is plausible that the scroll once contained the Aramaic portions of the book in addition to the currently extant Hebrew portions of Dan 8, 10, and 11.
Tov, Scribal Practices, 44–45. Precisely the opposite obtains for Greek texts, which are overwhelmingly written on papyrus scrolls.
Tov, Scribal Practices, 47.
Pliny the Elder, Natural History, 13.69–70. See also Tov, Scribal Practices, 32–33; Wise, Thunder, 127–28. On the “cheapness” of papyrus for book-production in antiquity, see especially Skeat, “Papyrus.” The same view seems to be reflected at two places in the New Testament: In the Apostle Paul’s request near the end of 2 Timothy, that Timothy take to him “the books, and above all the parchments” (
Wise, Thunder, 127–34.
Johnson, Bookrolls.
Tov, Scribal Practices, 273–74.
Even closer in style to the bookrolls studied by Johnson are the Twelve Minor Prophets scrolls on skin, in Greek from Nahal Hever (8Hev1) and in Hebrew from Wadi Murabba‘at (Mur88), a topic that merits further exploration.
Wise, Thunder, 127–28; Johnson, Bookrolls, 157–60.
This might help to explain why there are so few papyrus copies of “biblical” books at Qumran, as noted by Tov (Scribal Practices, 47). Such books would have been among the most settled in terms of their textual state and authoritative status by the period during which the Qumran scrolls were written, while more recently composed texts were more open to revisions of various kinds.
For a recent reflection in the limits of usability for a scroll, and the typical outer dimensions of scrolls in antiquity (though not addressing directly the Aramaic scrolls from Qumran), see Carr, “Materiality,” 599–604.
See Tov, Scribal Practices, 76–77; Carr, “Materiality,” 602–4. 1QIsaa is 7.34 m long, and 11QTa (11Q19) 8.75 m long. Tov (76) lists three possible extremely long exceptions: 4QRPa–e (22.5–27.5 m), 4QJerc (16.3–17.6 m), and 1Q20 (apGen) (more than 11.83 m), though the lengths of 4QRPa–e and 1Q20 (apGen) especially are open to question based on the required, extensive reconstruction. As I have argued in the profile of 1Q20 (apGen), it is doubtful that the Morgenstern’s proposed reconstruction of that scroll (in “New Clue”), espoused by Tov, is correct. The discussion of Johnson (Bookrolls, 143–52) is of some pertinence to our discussion, though it must be remembered that he is speaking of papyrus scrolls, a thinner substance than prepared skin. Consequently, the diameter of rolled scrolls that he discussed (150) would have to be increased if applied to the skin scrolls from Qumran.
This can be seen by the fact that that 11Q19 (Temple Scrolla), which was unusually written on the flesh side of the skin, was subject to special surface preparations. See Schuetz et al., “Temple Scroll.”
Pfann, “Preliminary Edition,” 39. In my opinion, Pfann’s conclusion deserves further scrutiny. It appears to be possibly correct for the horizontal text guidelines, but not the vertical ones.
See Tov, Scribal Practices, 59–60.
For an extensive discussion of the Qumran evidence more broadly, see Tov, Scribal Practices, 62–68.
Some of the more certain examples of partially-ruled or unruled skin scrolls are 4Q198 (Tobc), 4Q201 (Ena), 4Q212 (Eng), 4Q213–213b (Levia–c), 4Q214 (Levid), 4Q339 (List of False Prophets), 4Q530 (EnGiantsb), 4Q540 (apocrLevia?), 4Q541 (apocrLevib?), 4Q547 (Visions of Amrame)/4Q542 (TQahat), and 4Q557 (Visiona).
The end of 1Q20 (apGen) is almost fully preserved as it was presumably stored in Cave 1 in antiquity, but it is evident that the scroll was cut following a seam between two sheets (between cols. 22 and 23) such that the original end of the scroll is no longer extant. The reason for cutting the scroll in antiquity is not clear.
Based on the best photographs, it appears that not all of the horizontal script lines continue past the last intercolumnar margin, suggesting that there was one partially or fully uninscribed (but fully dry-ruled) final column. This was followed by an area of unruled skin.
Tov, Scribal Practices, 111–12, 115–18.
For the wider context at Qumran, see Tov, Scribal Practices, 110–15. Tov’s suggestion (111) that 4Q534 (Birth of Noaha) preserves the beginning of the scroll seems to me unlikely. While it is true that 4Q201 1 may preserve the beginning of the manuscript (corresponding to 1 En. 1:1–5; see Tov, Scribal Practices, 110), there are no physical remains of the area preceding the fragment.
Tov, Scribal Practices, 36. The blank area on 4Q545 (Visions of Amramc) is approximately the same size, but is not well-preserved and may once have been larger.
Tov, Scribal Practices, 82–90.
Tov (Scribal Practices, 99) noted that the same is true for later rabbinic instructions on the writing of scriptural scrolls (b. Menah. 30a; y. Meg. 1.71d and Sof. 2.5), with approximately a 2:3 ratio for top to bottom margin size. For Oxyhrynchus, note the important qualifications offered by Johnson, Bookrolls, 130–41.
Tov’s observations (Scribal Practices, 101) about 1Q20 (apGen) are not quite accurate. It is true that the top and bottom margins are of similar sizes, but the bottom seems to have been slightly larger than the top (ca. 2 mm) in most cases where we can measure. This scroll is on the borderline of being included in the list above.
For a list of the texts see Tov, Scribal Practices, 13–14.
Tov, Scribal Practices, 14.
Milik, BE, 141; Milik, “Daniel,” 355. It seems preferable simply to say that these scrolls are of relatively low quality without making assumptions about the social settings in which they were written, given the very little information we possess about scribes in Jewish antiquity.
For helpful visual and textual examples of the types of inks, pens, and inkwells discussed below, see Willi, Roman Writing.
In general, see Christiansen, “Manufacture.” With reference to the Qumran scrolls specifically, see Plenderleith in DJD 1:39–40; Steckoll, “Inks”; Nir-El and Broshi, “Black Ink”; Rabin et al. “Characterization,” 129–32. On the several weaknesses of the study done on ink from an inkwell allegedly from Qumran by Rasmussen et al. (“Constituents”), see Rabin, “Analysis.”
Christiansen et al., “Composition,” 27825. On the possible use of bone glue as a binder for at least some of the Qumran inks, see Murphy et al., “Degradation,” 95. See also, however, the comments of Rabin et al. “Origin,” 100, on the ink of 1QHodayota.
Pliny the Elder, Natural History, 35.41–43, discussed in Christiansen, “Manufacture,” 172–75.
Christiansen, “Manufacture.”
Nir-El and Broshi, “Black Ink.”
Tov, Scribal Practices, 53–54.
Nir-El and Broshi, “Black Ink,” 162, citing Haran, “Workmanship.”
Christiansen, “Manufacture.” See also Rabin, “Historic Inks.”
DJD 12:133. He believed that the delamination was “presumably because of some residual acid in the ink from its storage in a metal inkwell,” noting that “the phenomenon is not unusual among the Qumran manuscripts.”
Nir-El and Broshi, “Black Ink,” 164–66.
Nir-El and Broshi, “Black Ink.”
Ceramic and metal inkwells are known to have been made and used during the Hellenistic period (see, e.g., Sjökvist, “Inkstands”), so that the use of inkwells by Jewish scribes preceding the Roman period is entirely plausible. Prior to the Hellenistic period scribes used wooden ink pallets suitable for use with rush brush pens. The switch to inkwells accompanied the change to reed pens, following Greek scribal practices. For discussion and helpful images, see Longacre, “Script,” 12–21.
Examples of ceramic inkwells similar to those from Qumran have been found in excavations at Meiron in the Galilee, the Burnt House in Jerusalem, and most recently (2020) Gush Etzion.
de Vaux, Archaeology, 29–30; Steckoll, “Notes,” 35; Steckoll, “Inkwell”; Gunneweg and Balla, “Neutron Activation,” 13, 32. See also Goranson, “Qumran.”
Gunneweg and Balla, “Neutron Activation,” 13, 32.
The inkwell is published in Magen and Peleg, Qumran, 20–21.
See Goranson, “Qumran,” 111; Gunneweg and Balla, “Neutron Activation,” 32.
As opposed to the earlier rush brush pens. On this change, which preceded the period during which our scrolls were written, see Longacre, “Script.”
Cross, “Development”; Cross, “Palaeography.” See also the overview of Longacre, “Formality,” 102–110.
Longacre, “Formality”; Longacre, “Style.”
E.g., Van der Schoor, “Variation.”
Longacre, “Formality”; Longacre, “Style.”
Longacre, “Formality”; Longacre, “Style.”
On this phenomenon at Qumran more broadly, see Tov, Scribal Practices, 23–24.
John Strugnell was of the opinion that 4Q542 (TQahat) and 4Q53 (Samc) were written by the same Hasmonean-period scribe (recorded in Bonani et al., “Radio Carbon,” 28). No palaeographic analysis was offered in support of this view, but close examination of the two manuscripts show that Strugnell is incorrect. There are resemblances between some letters, such as he and dalet, but others (e.g., lamed, mem [both medial and final], and tet) show beyond doubt that these manuscripts are the work of different scribes. Józef Milik wrote (BE, 273) that 4Q209 (Enastrb) “is written in the same beautiful Herodian script as 1QIsab, 1QM, 1QGenAp, and the original hand of 1QH.” These scripts are indeed very similar, but it is likely that Milik intended only to characterize them as scripts of a similar style, not as written by the same scribe. Of the scrolls that he lists, 4Q209 (Enastrb) and 1QM are remarkably alike, and deserve further consideration as having been written by a single scribe.
Yardeni, “Scribe.”
Yardeni, “Scribe,” 289–90. The manuscript number in the Schøyen collection is 5234. On the claims of forgery, see Davis et al., “Dubious,” 220–21.
Yardeni, “Scribe,” 287. On p. 293 she states that “[l]amed is the most characteristic letter of this scribe.”
Yardeni, “Scribe,” 287.
Hayes’ most official presentation of her research was in a public, online lecture titled “Digital Palaeography and the Scribes of the Dead Sea Scrolls.” The lecture was recorded, and is available online at: https://www.youtube.com/watch?v=zSxlKNi2lyY (accessed 15 June, 2021).
In Humbert and Fidanzio, Khirbet Qumrân, 258. This is despite the absence of this scroll from Hayes’ list, as acknowledged by Tigchelaar (n. 79).
Cross, “Development,” 173; Yardeni, “Scribe,” 287. This is also the style description adopted by Hayes.
Ulrich, “Identification.”
Both Ulrich (“Identification,” 205) and Tigchelaar (Humbert and Fidanzio, Khirbet Qumrân, 258) note some differences in letter size and thickness of the ink in 1Q11 (Psb) and 11Q14 (Sefer ha-Milhamah), but at least the second feature is easily accounted for by the use of a different pen.
White Crawford, “Collection,” 124, n. 57. White Crawford, Scribes, 162, n. 152.
Humbert and Fidanzio, Khirbet Qumrân, 258.
Milik, BE, 5, 244.
Milik, “Fragment,” 95.
Van der Schoor, “Variation.” See also Drawnel, “Milik,” 113–14.
Milik, “Fragment,” 95.
Van der Schoor, “Variation.” See also Drawnel, “Milik,” 113–15.
DJD 31:377.
Machiela, “Testament of Qahat.” See also the profiles for 4Q542 (TQahat) and 4Q547 (Visions of Amrame).
A list is provided in Tov, Scribal Practices, 106–7.
Tov, Scribal Practices, 212–13; Longacre, “Script,” 39–40.
Tov, Scribal Practices, 243.
It is clear, for example, that Tov (Scribal Practices, 218–19) assumed the tetrapuncta of 4Q196 to represent the Tetragrammaton. In this assumption he followed Fitzmyer (e.g., at DJD 19:30).
See Machiela, “Tetragrammaton.”
Tov (Scribal Practices, 218–19) draws a close connection between the use of tetrapuncta and what he calls the Qumran Scribal Practice, associated by him with the sect living at Qumran.
Tov gives a broader introduction to this practice across the Qumran scrolls in Scribal Practices, 180–84.
These and other practices can be set alongside a growing set of Greek influences in the scribal realm, such as those discussed by Longacre, “Script.”
Milik, BE, 139; Tov, Scribal Practices, 71.
García Martínez, “Scribal Practices.”
These formal scripts could be either curvilinear or rectilinear, tending more towards the latter over time, and beginning to exhibit a square module and ornamental strokes over the course of the Herodian period. See further Longacre, “Script,” 30–39; Longacre, “Style.”
Johnson, Bookrolls, 157–60.
Tov, Scribal Habits, 125–29. See also Johnson, Bookrolls, 155–60; Longacre, “Style,” 10–12.
In her study of the Aramaic Levi Document, Van der Schoor (“Variation,” 200–201) noted that variability of script often corresponds with a messiness in general layout.
On the inherent ambiguity of terms like these, devoid of more fulsome description, see Johnson, Bookrolls, 158–60. It should be noted, however, that the social situation described by Johnson likely differs significantly from that behind the Qumran library. Van der Schoor, “Variation,” 201, suggested that such copies may not have been “an official copy meant to be read by others, but possibly a non-final copy in the process of transmission.” Of course, there is no way to verify a statement of this sort.
Milik, BE, 140–41.
Beyer, ATTM1, 227.
Puech, “Notes,” 649; Drawnel, ABE, 70–71.