Abstract
Were there any so-called “variants” in the scrolls? The assumption of contact between any two scrolls has been inferred from corrective additions and deletions. It has usually been claimed that corrections were based on MT. I divide the scrolls into different groups because in each group there could have been a different incentive for correction towards an external source. The strength of the argument depends on this division. In an analysis of the evidence (see the Appendix) in each of the groups, a different correction pattern was detected, and often the corrections were revised towards a text different from MT. In all these cases, the corrections were not made towards external sources, but the source of the scrolls. Strengthened by theoretical arguments, we suggest that the scrolls contain no notations of variants, that is, elements that were copied as variants from other scrolls or were corrected according to external sources.
1 Background
The main topic at the heart of this investigation is not the textual relation between the Qumran scrolls, the quality of the texts, or the Hebrew textual traditions and their origins, but the comparison of scrolls in antiquity. Did the scribes and readers compare scroll A not only with its source, scroll B, but also with a scroll C and, if so, were practical conclusions drawn during such a procedure? Several scholars claim that the corrections found in many scrolls were based on MT and rarely also on the Vorlage of the LXX. For such comparisons, one needs suitable conditions, and they probably existed in some cases. Think about the long tables in locus 30, the so-called scriptorium, at Qumran: scribes may have looked up passages in other scrolls, compared details, and inserted changes. There may also have been similar environments elsewhere; this assumed procedure could also have taken place in the temple library if such a library ever existed.
The practical result of such a comparison could have been the notation of corrections in scroll A according to scroll B; indeed, there are hundreds of corrections in the scrolls, but were they inserted based on a scroll B?
Every Scripture scroll (in this study: biblical scroll, the traditional Hebrew Bible) had to be compared with at least one scroll, scroll B, which is the scroll from which it was copied. However, the question in which I am interested is whether the scroll was also compared with a third source, scroll C.
This description leads to the mentioning of the second axis of this paper, namely the existence of so-called “variants” in the scrolls. The definition of a variant in a Judean Desert scroll would be the notation of a detail by the original scribe, a later one, or a reader based on another scroll. The definition of a variant approaches that of a correction based on another source, but the two don’t overlap completely. The two definitions regard a similar phenomenon from different angles. In both cases, contact between two sources is assumed. If we ask ourselves whether there are any variants in the Dead Sea Scrolls, we mean to ask whether there was any contact between the base scrolls and external sources.
We talked about scroll A, the scribe’s scroll, and his source, scroll B, but we actually know little about these sources since most scrolls from antiquity have been lost. Nevertheless, in a few cases intelligent guesses have been made.
-
4QDanb may have been copied from 4QDana (thus Eugene Ulrich).1 The two scrolls agree closely with each other in an overlapping section. They have the same layout in eight lines (in which “the two Qumran manuscripts overlap in the Dan 8:1–5 passage, five lines actually end with exactly the same word”2).
-
4QTest (4Q175) 14–20 quotes from a text resembling 4QDeuth (and Deut-LXX) in the quotation from Deut 33:8–11 as shown by several unique agreements between the two scrolls3 as opposed to 4QDeuth’s disagreement with MT and SP (lines 14, 17, 18, 18, 20; see Table 1, cols. 3–6). Yet, this scroll was not the immediate source of 4QTest because of the differences between them in lines 16, 17, 17, 19, 20.4
-
4QTest lines 21–30 are based on 4QapocrJoshb, or vice versa.5
-
Several compositions may have been based on 1QIsaa or a similar scroll. It has been suggested that this scroll was the basis for quotes in 4QTanḥ., but I believe that that source was a scroll similar to 1QIsaa.6 This pertains also to possible quotations from this scroll in 1QHa.7
Table 1
4QTest and its possible sources
|
4QTest |
4QDeuth |
MT |
LXX Summ. |
MT Summ. |
|
|---|---|---|---|---|---|
|
4QTest 14 |
|
= |
8 om |
= |
≠ |
|
4QTest 16 |
|
|
9 |
≠ |
≠ |
|
4QTest 17 |
|
= |
9 |
= |
≠ |
|
4QTest 17 |
|
– |
9 |
= |
≠ |
|
4QTest 17 |
|
|
10 |
? |
≠ |
|
4QTest 18 |
|
|
10 |
= corr |
= corr |
|
4QTest 18 |
|
|
10 |
? |
= corr |
|
4QTest 19 |
|
|
11 |
? |
= corr |
|
4QTest 19 |
|
– |
11 |
= |
≠ |
|
4QTest 20 |
|
|
11 |
= |
≠ |
|
4QTest 20 |
|
– |
11 |
≠ |
|
1.1 Corrections according to External Sources?
In this study I am interested in scenarios in which a scribe or reader corrected scroll A in accord with a third scroll, scroll C, that possibly had much authority, and did not base his readings on his source, scroll B. The first step on this road is an examination of whether there is any evidence at all for such assumed corrections.
The correction procedure is part of the scribal world of the ancient scribes as we observe from some nonbiblical scrolls.
For example, 1QHa has occasionally been corrected by way of addition or deletion based on two earlier copies of H, 4QHc (4Q429) and 4QpapHf (4Q432):8
Table 2
Assumed correction of 1QHa according to parallel sources
|
4QHc (4Q429) 1 ii 1 |
1QHa XIII 17 |
|
4QHc (4Q429) 1 ii 5 [ |
1QHa XIII 20 |
|
4QHc (4Q429) 1 iv 2 |
1QHa XIII 32 original text that cannot be recognized was corrected to |
|
4QHc (4Q429) 1 iv 5 |
1QHa XIII 34 original text that cannot be recognized was corrected to |
|
4QpapHf (4Q432) 3 2 |
1QHa X 6 |
As in the case of 1QHa, Scripture scrolls may also have been corrected according to parallel sources. Indeed, from the beginning of scroll research, claims have been made that some scrolls have been corrected towards the proto-MT and the LXX.
The method to be followed in examining this thorny topic is as follows: the assumption of contact between any two scrolls is not inferred from shared readings,9 but from paratextual notations, namely:
-
Corrective additions, intralinear and marginal;
-
Deletions using a variety of systems (crossing out with a line, cancellation dots or strokes, use of parenthesis signs, box, or erasure).
In all these cases, the change in the scroll may have been executed on the basis of an external source (scroll C), but it could also have been inserted based on the original scribe’s source, scroll B, or his own insights. Further, the intervention could have been performed by a later scribe or reader.
Many of the corrections agree with MT. In these cases, the discussion has always been one-sided, that is, it has been claimed that the correction was based on MT.
For example, on the corrections in MasEzek (§ 2.1), Talmon notes: “Several supralinear emendations of scribal mistakes, presumably by the scribe himself, bring the text in line with MT.”10 Talmon made a similar remark on the corrections in MasLev (§ 2.1).11
Likewise, Barthélemy remarked on the supralinear corrections in MurXII: “These have the overall effect of bringing the text of this proto-Masoretic manuscript into conformity with what was to become the classical Tiberian 𝔐.”12 Barthélemy made the same remark about the nine supralinear corrections in 1QIsab: “Here, as in Mur, all the corrections have the effect of making 1QIsab conform to 𝔐.”13
Eugene Ulrich made a similar remark about the many corrections in 4QJera: “… despite the fact that the original scroll had been corrected early toward the proto-MT in many loci ….”14 Likewise, Armin Lange and Karin Finsterbusch described the corrections of 4QJera as corrections towards MT: “The second scribe of 4QJera executed his corrections also in an archaic script. These corrections turned 4QJera from a semi-Masoretic into a proto-Masoretic manuscript.”15
Similar judgments have been made regarding presumed corrections toward the LXX in 2QDeutc by Jozef T. Milik,16 and in 5QDeut by Jozef T. Milik, Armin Lange, and Natalio Fernández Marcos.17
The ease with which scholars decided that a scroll was corrected to MT may have resulted from the recording system of the content of the scrolls in the official publications, in which each reading of a scroll was compared with MT (Figure 1).
Figure 1
Recording system of the scrolls in DJD
Citation: Textus 33, 2 (2024) ; 10.1163/2589255X-bja10044
This annotation system is neutral, meant as a conventional system of recording details of the new source, but notes like “scroll = MT,” “scroll ≠ MT,” “scroll = LXX” were easily reinterpreted as closeness or approximation to MT. When meeting several notations of the type of “scroll = MT,” it is easy to make a second step, and to claim that the agreements between a scroll and MT derived from a revisional process equating the scroll with MT. Such a claim may be unsupported, but when the agreement with MT pertains to a supralinear correction it is given additional force.
The practices of the DJD series are not unusual. The apparatuses of the Göttingen editions of the LXX likewise indicate that certain inner-LXX variants agree with MT. This information is helpful since these variants may reflect revisional activity (Hexaplaric or Lucianic) towards MT, but the information may also be misleading when the agreement is coincidental (Figure 2).
Figure 2
Recording system of the Göttingen editions of the Septuagint
Citation: Textus 33, 2 (2024) ; 10.1163/2589255X-bja10044
2 Evidence
I now turn to the evidence for the assumed corrections in the scrolls, one by one, trying to determine whether there is any evidence in favor of the assumption of a revision toward MT or LXX or any other source. The alternative assumption would be that the scroll was corrected by the original or a later scribe or reader, especially when a mistake was made and was recognized during the copying or at a later stage.
I divide the scrolls into different groups because in each group of texts there could have been a different incentive for correction towards an external source according to the nature of the texts. The strength of my argument depends on this division. Indeed, in the so-called nonaligned scrolls, many corrections were made against MT, not in any specific direction, probably towards the scroll’s source.
| 2.1 |
Proto-MT scrolls |
| 2.2 |
MT-like scrolls |
| 2.3 |
A pre-Samaritan scroll |
| 2.4 |
Nonaligned scrolls |
2.1 Proto-MT Scrolls
Proto-MT scrolls are almost identical to the medieval MT. When they are not, in rare instances, the text is corrected, usually towards the text that is now included in MT. Almost all the corrections appear to be scribal errors. In each case, we need to ask ourselves whether the proto-MT text was attested at all in the environment where the scroll was copied, for example at Qumran or Masada. Although it is very hard to obtain such data, this is a necessary question since according to the best of our information, the proto-MT did not circulate everywhere in ancient Israel.
-
4QGenb (50–100 CE). With 358 partially preserved words, this scroll reflects only one variant in spelling (
למארת 1:15 [= vv. 14, 16 MT]; MT למאורת ), and none in content. The addition of the erroneously left out gimel in 1:16ה ג דלים reflects a scribal error. -
MasLevb (30 BCE–30 CE). The ninety-six fragmentary lines of MasLevb contain no variants, but there is one orthographical difference. Two apparent mistakes and one orthographic variant were corrected towards a text that is now MT:
10:17 +
הוא = MT11:32
יט מ א = MT (scribal error)11:35
וכ י רים = MT (orthographic variant). -
MasEzek (50–1 BCE). The 120 fragmentary lines of MasEzek contain eight orthographical variants (Ezek 35:12; 36:17, 18, 21, 26, 30; 37:6, 10) and three content variants (36:21; 37:7, 7).18 This scroll usually agrees with MT, and contains five corrections, almost all of presumed errors, one against MT:
35:13
דבר י כם , על י = MT36:22
בית against MT, probably an error (cf. v. 21)36:25
ט מ אותיכם = MT36:30
ותנ ו בת = MT37:4
הנ ב א = MT. -
MurXII (c. 115 CE). This long scroll, with 3,60519 more or less identifiable words differs from MT in only forty-one instances, twenty-three in spelling and eighteen in small content details. The closeness of MurXII to MT is visible in its unusual spellings.20 Ten corrective additions, above and below the line, revise MurXII towards the text now included in MT:21
Orthography (2):
Amos 9:8
השמ י ד = MTMic 7:4
מב ו כתם = MTSupralinear correction of scribal errors (7):
Joel 3:2
ההם ה = MTAmos 7:13
עוד = MTAmos 8:11
והשלחתי רעב בארץ = MT. The text makes no sense without the added words.Obad 2
בגוים = MTMic 4:7
את = MTNah 1:11
על = MTHag 2:3
אתו = MTElement omitted: Obad 13 vav of
ואל ‘apparently’ scratched out22 = MT; not visible on the plate.The number of corrections in this scroll (ten) is seemingly impressive,23 but when compared with other scrolls it is very minimal. In this long scroll we find an average of one scribal intervention in every fifty-four lines, as opposed to an average of one intervention in four to ten lines in many other scrolls.24 MurXII is therefore one of the carefully transmitted texts. Most corrections pertain to elements that were omitted erroneously by the first hand, with one possible erasure (Obad 13). On the other hand, the forty-one variants pertain to qualitative differences in content and orthography. Therefore, in this carefully written scroll, the ten corrections represent rare mistakes that must have been corrected according to the scroll’s Vorlage.25 It seems that quantitative differences were corrected when the omission was recognized, while qualitative variants, that is ‘differences,’ were left in the text. This difference in approach to quantitative and qualitative variants supports the view that this scroll was not revised according to an external source.
2.2 MT-Like Scrolls
MT-like26 scrolls differ more from the medieval MT than do the proto-MT scrolls (§ 2.1), but they are part of the Masoretic family.27 They also include some corrections in the direction of the text that is now MT.28 In my view, it is difficult to claim that these corrections were made in conformance with MT since most deviations from MT in these texts were not corrected to MT.29 I submit that all these instances are corrections of scribal mistakes made in order to align the text with the scroll’s source.
-
4QGen-Exoda, a relatively long scroll (125–100 BCE) with only a few corrections: two corrections of errors and one orthographic correction.
-
4QGenc (20–68 CE, close to MT-SP). In this scroll, there’s one large scribal omission (Gen 41:9–10), added above the line, without which the text makes no sense.
-
4QGeng (75–50 BCE). This scroll with 132 preserved words contains four variants in content and nine in orthography. It is closer to MT than to SP and the LXX. It also contains two corrections towards the text now included in MT:
1:14
אלהים between ויאמר and יהי (error).2:6–7 or 2:19
וי י צר -
4QDeutf (75–50 BCE). This scroll contains six corrections, probably two errors (8:4, 24:6), one phonetic difference (8:9), and three variants (8:8, 23:4, 27:4), four times towards a text that is now MT and twice in a different direction.
-
1QIsab (50–25 BCE). The base text of this long scroll was corrected only eight times towards a text that is now MT.30 These corrections involve letters or words added to the text and words deleted, probably by the scribe himself.31
Orthography (2):
41:7
ט ו ב = MT. The prima manu reading טב is not attested elsewhere in Scripture.66:24
ה א נשים = MT. The p.m. reading הנשים is probably a phonetic variant.Erroneously omitted letters (4):
55:13
ספד corrected to ס ר פד = MT. MT סרפד is a hapax, so that a mistake is likely.56:6
בני ה נכר = MT. The p.m. reading may well reflect a variant, occurring also in Isa 60:10; 61:5; 62:8. For the corrected form, see בן הנכר in 56:3.58:11
וכ מ וצא = MT. The p.m. reading is a clear error.61:1
ש לחני = MT. The p.m. reading is a mistake caused by the context that is replete with words starting with a lamed (no less than five consecutive words immediately following and one occurring two words before the present word).Erroneously omitted words (2):
55:10
כי = MT before כאשר .65:24
והיה = MT before טרם .The eight corrections (= MT) form a minute group as compared with a very large group of divergences from MT in this scroll that have not been corrected: 161 differences in orthography32 and 183 in textual details.33 It is incomprehensible that, with so many deviations from MT left in the scroll, only eight details would have been singled out for correction towards that text. Rather, these instances of scribal errors were corrected to the Vorlage of the scroll that happened to be close to the medieval text. Some of the instances listed above are clear examples of scribal error.34
-
4QIsab (68–1 BCE, close to both MT and the LXX). All ten scribal interventions in this scroll are supralinear, nine errors and one phonetic variant, all towards a text that is now MT.35
-
4QIsad (approximately 50 CE, close to both MT and the LXX). One scribal intervention in this scroll is supralinear and two corrections are indicated with dots.
-
4QJera (225–175 BCE). All but seven (12:4; 14:6; 17:18; 18:19; 18:23; 22:16) of the thirty scribal interventions in this scroll are supralinear, subdivided into four groups. The uncorrected text often forms a very unlikely text (8:12; 17:10, 18; 18:19, 23).
Orthography (1):
Jer 9:14
Linguistic corrections (7):
Two phonetic corrections (14:6 [dotted letter]; 17:17), five different forms (four supralinear corrections [9:11; 10:11; 17:21, 24]; one erased letter together with a supralinear letter [17:11]).
Additions of details (18):
A large addition of 7:30–8:3 between the lines, in the margin, and under the text (see below); smaller pluses (10) in 8:12; 13:5; 17:14, 16, 19, 19, 22; 18:23; 22:12, 14; erasures (3) together with supralinear additions in 17:18, 19, and 22:3; additions of single letters (4): 12:5; 17:10, 16, 21.
Erasures of errors (7):
12:4; 17:18, 18; 18:19, 23; 22:16.
The thirty scribal interventions of 4QJera in 130 partially preserved lines present an average of one intervention in every 4.3 lines. Together with 1QIsaa, this scroll is thus among the scrolls displaying the highest level of scribal intervention.36 This situation was created by careless copying visible in the large number of presumed scribal errors37 among these interventions.38 Beyond the corrected elements, the scroll differs from MT in nineteen details in orthography and in sixteen other details, which have not been corrected.39
The most striking feature in this scroll is the very long omission of a complete paragraph (7:30–8:3), after an open paragraph (7:29) and probably before an open paragraph (before 8:4, reconstructed).40 Another striking feature is represented by the many supralinear and linear corrections in col. XI (sixteen) displaying many errors as well as many erasures in a single column.
Almost all the mentioned corrections made in 4QJera agree with MT in a text that was already close to MT. However, the fact that three of the corrections revised the base text away from MT (9:11; 17:11; 22:3), together with the fact that most of the deviations from MT have not been corrected, suggests that the apparent corrections to MT revise the scribe’s errors, probably in agreement with his source. In the Appendix, the great majority of the corrections are indicated as ‘error.’ See further in the Appendix (Table 4), 4QGend, 4QpaleoExodm, 4QLevb, 4QIsaa.
2.3 A Pre-Samaritan Scroll
4QNumb (30 BCE–20 CE, pre-Samaritan, QSP) is analyzed separately since its text is closely related to SP and secondarily also to the LXX. This long scroll contains seven supralinear corrections, five errors, and two phonetic variants, all correcting towards the text that is now MT. The pre-Samaritan text deviates several hundreds of times from MT, so that the corrections should not be taken as revisions in that direction, especially since the text leans towards SP and the LXX.
2.4 Nonaligned Scrolls
So-called nonaligned scrolls contain a much larger number of variants than the texts in §§ 2.1–3. Since the corrections pertain only to a small minority of the variants (mainly scribal errors) it should not be claimed that they were based on MT. This can indeed be shown as many corrections were made in the direction of texts that differed from MT (Appendix, Table 6).
-
4QExodc (68–1 BCE). The scribe of this scroll was notoriously negligent with five supralinear additions, all corrections of erroneous omissions, including two long ones (8:8; 10:1–2; 10:9; 13:19; 17:13) towards a text that is now MT. In addition, this scribe made several mistakes listed by Judith E. Sanderson.41
-
4QDeuth (50–1 BCE) contains three corrections of errors.
-
4QSamc (100–175 BCE, QSP) contains nine supralinear corrections (four errors, three orthography variants, two linguistic variants). Most corrections correct against MT (6/9), so that this scroll was probably corrected towards its base text.
-
4QXIIc (approximately 75 BCE, probably QSP). This scroll contains twelve corrections: seven errors, five early variants.42 Most corrections were made against MT (8x).
-
4QXIIg (66–1 BCE, QSP). This scroll contains six supralinear corrections (three errors, one phonetic variant, and two early variants).43 Two corrections were made against MT (2/6).
-
4QCantb (25–1 BCE). In this scroll, which differs much from MT,44
עת , erroneously copied from 2:12b in v. 12a, was crossed out with a line, thus creating a text like MT. -
11QP sa (1–50 CE).45 The scribe of this scroll, which displays many deviations from MT, was very negligent, as shown by the many (twenty-six) supralinear additions and some deletions with cancellation dots: scribal errors,46 phonetic differences,47 spelling differences,48 but also involving two variants.49 Otherwise all corrections agree with MT.
Two Tetragrammata canceled with cancellation dots in Ps 138:1 (col. XXI 2) and Ps 145:1 (col. XVI 7) seemingly indicate corrections towards MT, but these instances should be viewed in the light of other data. Both Psalms 138 and 145 contain many Tetragrammata, making it easy to understand that the scribe was overzealous in using this word, which was subsequently erased (= MT). Besides, instead of MT
אֲרוֹמִמְךָ אֱלוֹהַי , the scroll’s writing of Ps 145:1 ארוממכה י̣֗ה̣֗ו̣֗ה̣֗ אלוהי was likely influenced by Ps 30:2 אֲרוֹמִמְךָ יְהוָה , and in 138:1 possibly by the frequent combination of the verb זמר and יְהוָה .5011QP sa contains so many variants vis-à-vis MT that it is not plausible that the scroll was corrected towards MT in the few instances mentioned above. One correction was made during the copying itself.51 This fact, together with two corrections towards a text different from MT (n. 50), makes it likely that the scroll was corrected towards its own Vorlage. This assumption is strengthened by the fact that the noncanonical segments of this scroll were corrected in the exact same way as the canonical segments.52
-
1QIsaa (125–100 BCE). The scroll for which most data is available is 1QIsaa. The character of this scroll doesn’t differ from the others, but it has a few special features.53 According to my counting, the scroll contains 245 instances of scribal intervention consisting of large and small supralinear elements and dotted letters. Of these, 137 agree with the medieval MT and 108 differ from it. Accordingly, clearly the corrections did not have the proto-MT in mind.
Some of the deviating additions pertain to corrections in language54 and orthography,55 while many reflect variant readings.56 The data, not included in the Appendix, are culled from DJD XXXII.
However, two riddles remain in which knowledge of the proto-MT seemingly may be presumed.
Of the twenty-one instances of cancellation dots in this scroll, nineteen agree with MT,57 e.g., 3:18 ·
י̣ה̣ו̣ה̣ ·with supralinear אדוני = MT אֲדֹנָי , while only two disagree.58 Three interventions crossed out with a line likewise agree with MT,59 while one instance goes against MT (7:22; MT כול כָּל ). The high degree of agreement of cancellation dots with MT differs from the supralinear additions that agree in only half of the instances with MT. This could indicate that the cancellation dots were inserted by a different individual from the one(s) who inserted the supralinear elements. Of course, the very nature of the techniques of deletion with dots and a line, as opposed to supralinear elements, makes them unique among the editorial techniques. The dots could have been inserted during the copying process by the original scribe,60 but also by a later scribe or a reader. In these cases, a revision according to the proto-MT is remotely possible, but it remains difficult to understand why these items would have been singled out for this purpose.Two dotted corrections in the Isaiah scroll present a second riddle:
Isa 36:4
MT
אִמְרוּ־נָא אֶל־חִזְקִיָּהוּ 1QIsaa
אמורו נא אל חזקיה מ֗ל֗ך֗ י֗ה֗ו֗ד֗ה֗ 2 Kgs 18:19
MT
אִמְרוּ־נָא אֶל־חִזְקִיָּהוּ Cf. Isa 37:10
MT
כּ֗ה תֹאמְרוּן אֶל־חִזְקִיָּהוּ מֶלֶךְ־יְהוּדָה When writing
מלך יהודה in the scroll in 36:4, the scribe may have been influenced by the longer text of Isa 37:10 (= 2 Kgs 19:10).61 That longer text included the formula ‘King of Judah’ after Hezekiah. Remarkably, the harmonizing plus was deleted with cancellation dots based on the scroll’s source (and hardly on proto-MT of Isaiah).Isa 36:7
MT
לִפְנֵי הַמִּזְבֵּחַ הַזֶּה תִּשְׁתַּחֲווּ 1QIsaa
לפני המזבח הזה תשתחוו ב֗י֗ר֗ו֗ש֗ל֗י֗ם֗ 2 Kgs 18:22
MT
= versions
לִפְנֵי הַמִּזְבֵּחַ הַזֶּה תִּשְׁתַּחֲווּ בִּירוּשָׁלִָם The originally long text of the scroll, with
בירושלים , may have referred to Jerusalem mentioned earlier in the verse, but remarkably ‘Jerusalem’ occurs also in the longer text of 2 Kgs 18:22 MT and versions. Here again, the dots on ב֗י֗ר֗ו֗ש֗ל֗י֗ם֗ may correct to the scroll’s source, but also to proto-MT of Isaiah.Summarizing the evidence of 1QIsaa, the many supralinear corrections in this scroll are clearly not based on proto-MT, while the scribal dots in this scroll pose a riddle. It is very unlikely that only these twenty-one words were chosen to be corrected to the proto-MT. In such an unlikely scenario, we need to think about the realia as well. In my view, this scroll was penned by a Qumran scribe62 and we have not found any proto-MT scrolls at Qumran. More on this in § 3.
-
4QTest does not contain a running biblical text, but a collection of three Scripture quotations, each of which is based on a different kind of text. Lines 14–20 quote from a text like 4QDeuth (and Deut-LXX) in the quotation from Deut 33:8–11 as shown by several unique agreements between the two scrolls as opposed to its disagreement with MT and SP. Further, in lines 18–19, the original text of 4QTest agrees three times with 4QDeuth, subsequently corrected to the readings included in MT and SP. The quotation is not identical to either 4QDeuth or the LXX but reflects a text that was close to these two sources. These corrections should not be understood as signs of an overall revision of this segment towards MT since 4QTest and MT-Deut are not close to each other (they differ in eight details). Nevertheless, a limited correction to MT SP is not impossible. Another possibility would be a correction towards a nonaligned text that shared some readings with MT, SP, LXX, while some are independent.63 For details, see Table 1.64
Summarizing §§ 2.1–4, the proto-MT texts (§ 2.1) were occasionally corrected towards MT; yet the few deviations from that text were usually not corrected (best example: MurXII). Most of the deviations of the MT-like scrolls (§ 2.2) from MT were not corrected (best examples: 1QIsab and 4QJera). The few corrections towards MT in the pre-Samaritan 4QNumb (§ 2.3) are a small minority in a scroll that contains hundreds of readings agreeing with the SP and the LXX against MT. The last group, that of the nonaligned scrolls (§ 2.4), differs from the others since many scrolls correct often towards a text different from MT. In my view, in all these cases, the corrections were not made towards the proto-MT or other scrolls, but to the source of the scrolls themselves. In § 3 we will address the theoretical background of a possible revision.
2.5 The LXX
Several scrolls contain individual readings agreeing with the LXX (especially 4QDeutq and 4QSama), but we focus here on possible corrections in Qumran scrolls towards the Hebrew source of the LXX. Why someone would insert such corrections based on the source of the LXX at all is another question, to be dealt with in § 3. Three of the following scrolls (nos. 2, 3, 4) have been mentioned in the scholarly discussion. In my view, while 5QDeut probably presents the most appealing data, in no case is the evidence convincing (Appendix, Table 7).
-
4QGenj (68–1 CE). Two pluses of this scroll, which may be described as MT-like, are corrected in the direction of the reading reflected in the LXX (41:16, 24). Nevertheless, revision in accord with the LXX is unlikely since, in other verses, the scroll differs fourteen times from the LXX.65
-
2QDeutc is described as follows by Maurice Baillet: “Le texte se rapproche de la LXX et de la Vulgate.”66 However, this fragment, of which a mere twelve fragmentary words have been preserved, does not show a close relation to any tradition.67
-
5QDeut. Milik’s contention that 5QDeut (chapters 7–9) has been revised four times according to a Hebrew text close to the Vorlage of the LXX would have been of special interest had the evidence been conclusive.68 Two of the corrections indeed agree with the LXX against MT.69 The third correction (8:19) is based on a reading that is, at best, dubious.70 The fourth addition (9:2) is probably irrelevant.71 Summarizing, the evidence of the three pluses of this scroll that agree with the LXX, though not completely certain, is remarkable. At the same time, the LXX disagrees ten times with 5QDeut, especially in its small additions (7:19, 19, 19; 8:15, 15, 16, 18, 18, 19, 23). The sum of this evidence does not favor the assumption that this text has been corrected towards a Hebrew text close to the LXX.72
-
4QJoshb. The single supralinear addition of
חטים at the end of Josh 2:15 equals the LXX (θερισμοῦ πυρῶν = קציר חטים , for which cf. Gen 30:14, Exod 34:21, 1 Sam 6:13, etc.), but elsewhere this scroll always disagrees with the LXX.
3 Theoretical Background of a Possible Approximation to Proto-MT or LXX
The evidence examined in §§ 1–2 concerning the source of the scribal corrections in the Judean Desert scrolls as being the MT or LXX did not yield positive results, while a few riddles remained. We were not able to confirm the oft-repeated suggestion that the corrections in the scrolls were based on the proto-MT or LXX. Faute de mieux, we suggest that the best option is to assume that the corrections were based on the sources of the various scrolls, and sometimes on the scribes’ or the readers’ intentions. This suggestion is strengthened by the fact that the text was corrected away from MT in several scrolls. The practical result of this analysis is that I have not found any factual basis in the data for the assumption of contact between scroll A and a scroll C.
We now want to approach this problem from another angle. We looked for evidence of possible corrections based on external sources, but is it at all imaginable that a scroll was corrected according to the proto-MT or the Vorlage of the LXX in the last three centuries BCE or the first two centuries CE? Bypassing the LXX, since there is no real evidence linking scrolls with that version, what does one have in mind when saying that a text was corrected to the proto-MT? Which physical copies represented the proto-MT in those early centuries, and where in ancient Israel were they available? By necessity, I turn to speculation regarding what is known about the textual situation in ancient Israel at the time when the mentioned scrolls were corrected. I first refer to the presumed dates of some of the scrolls for which I found most evidence regarding the presumed corrections, in chronological sequence:
4QJera (225–175 BCE) (MT-like)1QIsaa (125–100 BCE, QSP) (nonaligned)1QIsab (50–25 BCE) (MT-like)MasEzek (50–1 BCE) (proto-MT)11QP sa (1–50 CE, QSP) (nonaligned)MurXII (c. 115 CE) (proto-MT)
Several textual developments took place during the long period covered by these scrolls (225 BCE–115 CE), and we must be careful not to indulge in circular reasoning regarding textual theories. One of the issues mentioned at the beginning of this study was whether suitable conditions existed in antiquity for a comparison between scrolls. I think there were such conditions at Qumran and maybe also elsewhere if different scrolls were extant at the same location.
Yet, the correction of manuscripts according to proto-MT is by no means a proven reality for Hebrew sources. It is attested widely for Greek sources, but that’s a completely different story.73 A first glimpse of such a revision is visible in the kaige-Th revision of the OG of the Minor Prophets in the first century BCE, that is, before the destruction of the temple.74
If scrolls were revised at all according to proto-MT, we need to think about the milieu in which the mentioned scrolls could have been corrected to the proto-MT. For such a revision to have taken place, one needs to assume the availability of proto-MT scrolls at a certain location. Further, we need some background information about the origin of the scrolls found at Qumran since most of them probably were not copied there. Certainly, all the scrolls found at the other Judean Desert sites (Naḥal Ḥever, Naḥal Murabbaʿat) and at Masada were not copied there either. But we do know that the proto-MT scrolls were found at all the Judean Desert sites except for Qumran, at Masada, and presumably also in Jerusalem although we have no physical evidence of the presence of these scrolls.
This brings us to the matter of the presumed revision according to MT. Due to this kind of background information, we cannot state with certainty that all the scrolls discussed in §§ 2.1–5 were originally found at places where the proto-MT was easily available. In other words, I don’t know from where 4QJera and 1QIsab were imported to Qumran and I therefore don’t know whether the proto-MT was available in those localities. However, I do have a view about two central Qumran texts, 1QIsaa (125–100 BCE, QSP [Qumran scribal practice]) (nonaligned) and 11QP sa (1–50 CE, QSP) (nonaligned). Could these texts have been revised to the proto-MT? In my view, 11QP sa is a typical exponent of the Qumran environment, and to some extent also 1QIsaa.75 Now, according to my analysis, there probably were no proto-MT scrolls at Qumran.76 As a result, these texts could not have been revised according to the proto-MT. This argument is speculative, but it is additional to the argument that there is no evidence for any revision.
Correction to the Vorlage of the LXX would have taken place only in a few scrolls, but in my opinion such a suggestion is totally unacceptable. The LXX became an important source in the history of the biblical text only at a later stage when it was accepted by Christianity. I see no reason why its Hebrew source would have been so important that texts would have been revised towards its content. That suggestion is a misconception of scholarship.
The upshot of this analysis is that there is no theoretical support for the assumption that any scroll was corrected to either MT or LXX. I also don’t think that one should take refuge in the general argument that there was a movement towards stabilization of the Masoretic Text.77 Such a view has not been proven; I have argued against it,78 and even if it were correct, one still needs proof of detailed data.
The arguments adduced above argue against the assumption of revision toward MT. The creation of a master copy of the text of the Torah in the temple library (the so-called copy of the temple court) by a comparative eclectic process may be considered a parallel, but it really isn’t. That copy was not created by revision to MT, but possibly by an eclectic procedure, if the tradition is reliable at all.79
4 The Notion of Variants in the Dead Sea Scrolls: Conclusion
In the previous paragraphs we suggested that the Judean Desert scrolls were not compared to one another and that no details were copied from one manuscript to another by way of correction. Ours is a working hypothesis, to which we may have found two exceptions in 1QIsaa (§ 2.4). If this thesis is correct, by implication, the scrolls contain no notations of variants, that is, elements that were copied as variants from other scrolls or were corrected according to external sources.80
Appendix
Table 3
Corrections in proto-MT texts81
|
Reference |
Scroll |
Correction |
Relation to MT |
Presumed nature |
|---|---|---|---|---|
|
Gen 1:16 |
4QGenb |
|
equal |
error |
|
Lev 10:17 |
MasLevb |
|
equal |
error |
|
Lev 11:32 |
MasLevb |
|
equal |
error |
|
Lev 11:35 |
MasLevb |
|
equal |
ortho |
|
Ezek 35:13 |
MasEzek |
|
equal |
error |
|
Ezek 35:13 |
MasEzek |
|
equal |
error |
|
Ezek 36:22 |
MasEzek |
|
diff (om) |
error |
|
Ezek 36:25 |
MasEzek |
|
equal |
error |
|
Ezek 36:30 |
MasEzek |
|
equal |
ortho |
|
Ezek 37:4 |
MasEzek |
|
equal |
error |
|
Amos 9:8 |
MurXII |
|
equal |
ortho |
|
Mic 7:4 |
MurXII |
|
equal |
ortho |
|
Joel 3:2 |
MurXII |
|
equal |
error |
|
Amos 7:13 |
MurXII |
|
equal |
error |
|
Amos 8:11 |
MurXII |
|
equal |
error |
|
Obad 2 |
MurXII |
|
equal |
error |
|
Mic 4:7 |
MurXII |
|
equal |
error |
|
Nah 1:11 |
MurXII |
|
equal |
error |
|
Hag 2:3 |
MurXII |
|
equal |
error |
Table 4
MT-like scrolls
|
Reference |
Scroll |
Correction |
Relation to MT |
Presumed nature |
|---|---|---|---|---|
|
Gen 35:17 |
4QGen-Exoda |
|
equal |
error |
|
Exod 1:17 |
4QGen-Exoda |
|
equal |
ortho |
|
Exod 5:13 |
4QGen-Exoda |
|
equal |
error |
|
Gen 41:9–10 |
4QGenc |
|
equal |
error |
|
Gen 1:22 |
4QGend |
|
diff |
ortho |
|
Gen 1:14 |
4QGeng |
|
equal |
error |
|
Exod 17:13 |
4QpaleoExodm |
|
equal |
error |
|
Lev 13:58 |
4QLevb |
|
equal |
error |
|
Lev 14:45 |
4QLev-Numa |
|
equal |
error |
|
Deut 8:4 |
4QDeutf |
|
equal |
error |
|
Deut 8:8 |
4QDeutf |
|
equal |
var |
|
Deut 8:9 |
4QDeutf |
|
equal |
phon |
|
Deut 22:14 |
4QDeutf |
|
equal Q |
var |
|
Deut 24:6 |
4QDeutf |
|
equal |
error |
|
Deut 27:4 |
4QDeutf |
|
diff |
var |
|
Isa 41:7 |
1QIsab |
|
equal |
ortho |
|
Isa 55:10 |
1QIsab |
|
equal |
error |
|
Isa 55:13 |
1QIsab |
|
equal |
error |
|
Isa 56:6 |
1QIsab |
|
equal |
var |
|
Isa 58:11 |
1QIsab |
|
equal |
error |
|
Isa 61:1 |
1QIsab |
|
equal |
error |
|
Isa 65:24 |
1QIsab |
|
equal |
error |
|
Isa 66:24 |
1QIsab |
|
equal |
phon |
|
Isa 1:4 |
4QIsab |
|
equal |
error |
|
Isa 1:5 |
4QIsab |
|
equal |
error |
|
Isa 5:21 |
4QIsab |
|
equal |
phon |
|
Isa 5:25 |
4QIsab |
|
equal |
error |
|
Isa 5:25 |
4QIsab |
|
equal |
error |
|
Isa 17:14 |
4QIsab |
|
equal |
error |
|
Isa 18:7 |
4QIsab |
|
equal |
error |
|
Isa 40:4 |
4QIsab |
|
equal |
error |
|
Isa 40:26 |
4QIsab |
|
equal |
error |
|
Isa 51:16 |
4QIsab |
|
equal |
error |
|
Isa 48:21 |
4QIsad |
|
equal |
error |
|
Isa 49:1 |
4QIsad |
|
equal ( |
error |
|
Isa 49:4 |
4QIsad |
|
equal ( |
error |
|
Isa 6:7 |
4QIsaa |
|
equal |
error |
|
Jer 7:30–8:3 |
4QJera |
|
equal |
error |
|
Jer 8:12 |
4QJera |
|
equal |
error |
|
Jer 9:11 |
4QJera |
|
diff |
ling |
|
Jer 9:14 |
4QJera |
|
equal |
ortho |
|
Jer 10:11 |
4QJera |
|
equal |
error |
|
Jer 12:4 |
4QJera |
|
equal |
error |
|
Jer 12:5 |
4QJera |
|
equal |
error |
|
Jer 13:5 |
4QJera |
|
equal |
error |
|
Jer 14:6 |
4QJera |
|
equal ( |
phon |
|
Jer 17:10 |
4QJera |
|
equal |
error |
|
Jer 17:11 |
4QJera |
[ |
diff |
var |
|
Jer 17:14 |
4QJera |
|
equal |
error |
|
Jer 17:16 |
4QJera |
|
equal |
error |
|
Jer 17:16 |
4QJera |
|
equal |
error |
|
Jer 17:17 |
4QJera |
|
equal |
phon |
|
Jer 17:18 |
4QJera |
|
equal |
error |
|
Jer 17:18 |
4QJera |
{ |
equal |
error |
|
Jer 17:19 |
4QJera |
|
equal |
error |
|
Jer 17:19 |
4QJera |
|
equal |
error |
|
Jer 17:21 |
4QJera |
|
equal |
error? |
|
Jer 17:21 |
4QJera |
|
equal |
error |
|
Jer 17:22 |
4QJera |
|
equal |
error |
|
Jer 17:24 |
4QJera |
|
equal |
ling |
|
Jer 18:19 |
4QJera |
{ |
equal |
error |
|
Jer 18:23 |
4QJera |
|
equal |
error |
|
Jer 18:23 |
4QJera |
{( |
equal |
ditto |
|
Jer 22:3 |
4QJera |
|
different |
var |
|
Jer 22:12 |
4QJera |
|
equal |
error |
|
Jer 22:14 |
4QJera |
[ |
equal |
error |
|
Jer 22:16 |
4QJera |
} |
equal |
ditto |
Table 5
A pre-SP scroll
|
Reference |
Scroll |
Correction |
Relation to MT |
Presumed nature |
|---|---|---|---|---|
|
Num 19:1 |
4QNumb |
|
equal |
error |
|
Num 20:26 |
4QNumb |
|
equal |
phon |
|
Num 22:17 |
4QNumb |
|
equal |
error |
|
Num 22:33 |
4QNumb |
|
equal |
error |
|
Num 23:3 |
4QNumb |
|
equal |
error |
|
Num 24:6 |
4QNumb |
|
equal |
error |
|
Num 26:31 |
4QNumb |
|
equal |
phon |
Table 6
Nonaligned scrolls
|
Reference |
Scroll |
Correction |
Relation to MT |
Presumed nature |
|---|---|---|---|---|
|
Exod 8:8 |
4QExodc |
|
equal |
error |
|
Exod 10:1–2 |
4QExodc |
|
equal |
error |
|
Exod 10:9 |
4QExodc |
|
equal |
error |
|
Exod 13:19 |
4QExodc |
|
equal |
error |
|
Exod 17:13 |
4QExodc |
|
equal |
ortho |
|
Deut 1:1 |
4QDeuth |
|
equal |
error |
|
Deut 2:3 |
4QDeuth |
|
equal |
error |
|
Deut 4:34 |
4QDeuth |
|
equal ( |
error |
|
2 Sam 14:10 |
4QSamc |
|
diff ( |
ling |
|
2 Sam 14:11 |
4QSamc |
|
diff |
ortho |
|
2 Sam 14:22 |
4QSamc |
|
diff |
ortho |
|
2 Sam 14:30 |
4QSamc |
|
diff |
ortho |
|
2 Sam 14:32 |
4QSamc |
|
equal |
error |
|
2 Sam 14:32 |
4QSamc |
|
equal |
error |
|
2 Sam 15:1 |
4QSamc |
|
diff |
error |
|
2 Sam 15:8 |
4QSamc |
|
equal |
phon |
|
2 Sam 15:12 |
4QSamc |
|
diff |
var |
|
Hos 4:18 |
4QXIIc |
|
equal |
error |
|
Hos 13:10 |
4QXIIc |
|
diff |
var |
|
Hos 14:3 |
4QXIIc |
|
diff |
error |
|
Joel 1:17 |
4QXIIc |
|
diff ( |
error |
|
Joel 1:18 |
4QXIIc |
|
equal |
error |
|
Joel 2:11 |
4QXIIc |
|
diff ( |
error |
|
Joel 2:13 |
4QXIIc |
|
equal |
error |
|
Joel 2:19 |
4QXIIc |
|
diff (om) |
var |
|
Joel 4:9 |
4QXIIc |
|
diff |
var |
|
Joel 4:18 |
4QXIIc |
|
diff (om) |
var |
|
Joel 4:19 |
4QXIIc |
|
diff (om) |
error |
|
Zeph 2:15 |
4QXIIc |
{ |
equal |
var |
|
Hos 10:8 |
4QXIIg |
|
equal |
error |
|
Hos 11:10 |
4QXIIg |
|
equal |
error |
|
Hos 12:8 |
4QXIIg |
|
equal |
phon |
|
Joel 2:5 |
4QXIIg |
|
diff |
var |
|
Amos 1:14 |
4QXIIg |
|
diff |
var |
|
Amos 2:8 |
4QXIIg |
|
equal |
error |
|
Ps 104:24 |
11QP sa |
|
equal ( |
error |
|
Ps 104:32 |
11QP sa |
|
equal |
error |
|
Ps 105:6 |
11QP sa |
{{ |
equal |
error |
|
Ps 118:26 |
11QP sa |
{ |
equal |
error |
|
Ps 119:27 |
11QP sa |
|
equal |
error |
|
Ps 119:42 |
11QP sa |
|
equal |
phon |
|
Ps 119:140 |
11QP sa |
|
equal |
error |
|
Ps 119:154 |
11QP sa |
|
equal |
error |
|
Ps 119:155 |
11QP sa |
|
equal |
error |
|
Ps 119:175 |
11QP sa |
|
equal ( |
var |
|
Ps 121:2 |
11QP sa |
|
equal |
phon |
|
Ps 122:3 |
11QP sa |
|
equal |
error |
|
Ps 123:1 |
11QP sa |
|
equal |
error |
|
Ps 137:1 |
11QP sa |
|
equal |
error |
|
Ps 138:1 |
11QP sa |
|
diff (om) |
error |
|
Ps 139:17 |
11QP sa |
|
equal |
ortho |
|
Ps 141:10 |
11QP sa |
|
equal |
ortho |
|
Ps 143:3 |
11QP sa |
|
equal |
phon |
|
Ps 143:5 |
11QP sa |
|
equal ( |
error |
|
Ps 143:6 |
11QP sa |
{ |
equal ( |
error |
|
Ps 144:1 |
11QP sa |
|
equal |
error |
|
Ps 145:1 |
11QP sa |
|
diff (om) |
var |
|
Ps 145:3 |
11QP sa |
|
equal ( |
error |
|
Ps 145:6 |
11QP sa |
|
equal K ( |
var |
|
Ps 145:15 |
11QP sa |
|
equal |
ortho |
|
Ps 145:15 |
11QP sa |
|
equal |
error |
|
Cant 2:12 |
4QCantb |
|
equal |
error |
Table 7
LXX
|
Reference |
Scroll |
Correction |
Relation to MT |
Presumed nature |
|---|---|---|---|---|
|
Gen 41:16 |
4QGenj |
|
diff (om) |
var = LXX |
|
Gen 41:24 |
4QGenj |
|
diff (om) |
var = LXX |
|
Deut 7:15 |
5QDeut |
|
diff (om) |
var = LXX |
|
Deut 8:12 |
5QDeut |
|
diff (om) |
var = LXX |
|
Deut 8:19 |
5QDeut |
|
diff (om) |
var = LXX |
Eugene Ulrich, The Dead Sea Scrolls and the Origins of the Bible (Grand Rapids: Eerdmans; Leiden: Brill, 1999), 148–162 (162, n. 27).
Ulrich, The Dead Sea Scrolls, 162, n. 27.
Julie A. Duncan, “35. 4QDeuth,” in Eugene Ulrich and Frank M. Cross, Qumran Cave 4.IX: Deuteronomy, Joshua, Judges, Kings, DJD XIV (Oxford: Clarendon, 1995), 61–70 (68–70).
On the striking phenomenon of the agreements between the corrections in lines 18 and 19 and MT, see § 2.3.
Most scholars believe that 4QTest quoted from 4QapocrJoshb, but Hanan Eshel, “The Historical Background of the Pesher Interpreting Joshua’s Curse on the Rebuilder of Jericho,” RevQ 15 (1992): 409–420 believes that the quoting went in the opposite direction.
Flint assumes a direct connection, especially in 4QTanḥ. (4Q176), frgs. 8–11: Peter W. Flint, “The Interpretation of Scriptural Isaiah in the Qumran Scrolls: Quotations, Citations, Allusions, and the Form of the Scriptural Source Text,” in A Teacher for All Generations: Essays in Honor of James C. VanderKam, Vols. 1–2, ed. E.F. Mason et al., JSJSup 153/1 (Leiden: Brill, 2012), 1:388–406 (403). According to my count, the two agree four times in small details and ten times in spelling while, according to Flint, 4Q176 agrees in eight of its fourteen “alternative” readings with 1QIsaa. My tabulations are different, but among the variants of this scroll four remarkable readings stand out. In line 8 the two agree in a plus of the Tetragrammaton against MT 54:6, in line 10 they agree in the reading
For example, Isa 6:10
Thus already Malachi Martin, The Scribal Character of the Dead Sea Scrolls, 2 vols., Bibliothèque du Muséon 44, 45 (Louvain: Publications Universitaires et Institut Orientaliste, 1958), 2.494 in general terms and, in detail, Eileen Schuller, “Some Contributions of the Cave Four Manuscripts (4Q427–432) to the Study of the Hodayot,” DSD 8 (2001): 278–287 (285).
If a scroll shares idiosyncratic readings with MT, they do not count in determining correction according to a specific source because MT represents a large group of scrolls and not a single source. For example, the closeness of MurXII to MT is visible in such unusual spellings as
Shemaryahu Talmon in idem and Yigael Yadin, Masada VI: The Yigael Yadin Excavations 1963–1965, Final Reports, Hebrew Fragments from Masada (Jerusalem: Israel Exploration Society, 1999), 68.
Ibid., 47–48: “… the scribe may have adjusted an originally defective spelling
Barthélemy, Studies, 387.
Barthélemy, Studies, 405.
Eugene Ulrich, The Dead Sea Scrolls and the Developmental Composition of the Bible, VTSup 169 (Leiden: Brill, 2015), 145.
Armin Lange and Karin Finsterbusch, “The Hebrew Texts of the Bible and Their Value for Textual Criticism,” in The Comparative Textual Criticism of Religious Scriptures, THBSup 8 (Leiden: Brill, 2024), 79. In a similar vein, but less clearly, see Armin Lange, “7.2.1.2 4QJera (4Q70),” in Armin Lange and Emanuel Tov, Textual History of the Bible, The Hebrew Bible, Vol. 1B, Pentateuch, Former and Latter Prophets (Leiden: Brill, 2017), 514–515.
Maurice Baillet in idem et al., Les ‘petites grottes’ de Qumrân, DJD III (Oxford: Clarendon, 1962), 61: “Le texte se rapproche de la LXX et de la Vulgate.” See § 2.5.
Milik, ibid., 169–171; Armin Lange, “ ‘They Confirmed the Reading’ (y. Taʿan. 4.68a): The Textual Standardization of Jewish Scriptures in the Second Temple Period,” in From Qumran to Aleppo: A Discussion with Emanuel Tov about the Textual History of Jewish Scriptures in Honor of His 65th Birthday, ed. Armin Lange et al., FRLANT 230 (Göttingen: Vandenhoeck & Ruprecht, 2009), 29–80 (62); Natalio Fernández Marcos, “5QDt y los tipos textuales bíblicos,” in Biblia Exégesis y Cultura: Estudios en honor del Prof. D. José María Casciaro, ed. G. Aranda et al. (Pamplona: EUNSA, 1994), 119–125. For an analysis, see § 2.5 below.
Talmon, Masada VI, 68–72.
Lange, Handbuch, 346 counts 3,803 partially preserved words.
See n. 9 above.
Barthélemy, Studies, 387.
Thus Jozef T. Milik in Pièrre Benoit, O.P., Jozef T. Milik, and Ronald de Vaux, Les grottes de Murabbaʿât, DJD II (Oxford: Clarendon, 1961), 183.
There may be a few additional corrections in columns that are not clearly legible.
Emanuel Tov, “Approaches of Scribes to the Biblical Text in Ancient Israel,” in The Scribe in the Biblical World: A Bridge Between Scripts, Languages and Cultures, ed. Esther Eshel and Michael Langlois, BZAW 547 (Berlin: De Gruyter, 2023), 3–21, Table 4.
Jozef T. Milik, DJD II, 183 remarked that it is difficult to decide whether the corrections were made based on the scroll’s source or on a different manuscript.
For the term, see Tov, TCHB4, 63. Lange, “They Confirmed,” 54 names these scrolls “semi-Masoretic” (see also Lange, Handbuch, 16), while for Barthélemy, Studies, 383–409 they are “pre-Masoretic.”
Tov, Scribal Practices, 332–335 lists nineteen scrolls of this group.
Table 4 in the Appendix contains relevant examples from scrolls that contain multiple instances of corrections. Scrolls that contained one or two supralinear corrections are included in my database, but not in the Appendix: 11QpaleoLeva, 4QDeuti,m,n,k1,k2, 4QJoshb, 4QJudga, 4QSama, 1QSam, 4QEzeka,b, 4QXIIa,e, 4QP sb,d,e,g, 4QQoha, 4QDana.
All revisions are inconsistent, but to invoke inconsistency in this case would be exaggerated.
The evidence is presented by Dominique Barthélemy, Critique textuelle de l’Ancien Testament, OBO 50/3 (Fribourg: Éditions Universitaires; Göttingen: Vandenhoeck & Ruprecht, 1992), cii–cxvi, = idem, Studies, 389–409 (405). Barthélemy speaks about nine instances, but one instance is not an approximation to MT: In 55:7, a word was erased before
Thus Eugene Ulrich and Peter W. Flint, Qumran Cave 1.II, Parts 1–2: The Isaiah Scrolls, DJD XXXII (Oxford: Clarendon, 2010), II:200: “All the corrections appear to be penned by the original scribe; no later hands are detectable in the manuscript.”
Ulrich–Flint, DJD XXXII, 200–204.
Ibid., 208.
The level of scribal intervention in this scroll is that of an average biblical scroll. Based on a calculation of 564 fragmentarily preserved lines presented in DJD XXXII, we reckon with an average of one scribal intervention per seventy lines.
The evidence is listed by Eugene Ulrich et al., “56. 4QIsab,” in Qumran Cave 4.X: The Prophets, DJD XV (Oxford: Clarendon, 1997), 22.
See the comparative data in Tov, Scribal Practices, 331–335.
Emanuel Tov, “The Textual Base of the Corrections in the Biblical Texts Found at Qumran,” in The Dead Sea Scrolls: Forty Years of Research, ed. Devorah Dimant and Uriel Rappaport (Leiden: Brill; Jerusalem: Magnes and Yad Izhak Ben-Zvi, 1992), 299–314 (309–310).
It is difficult to determine who inserted the corrections, the original scribe or a later one, because most corrections are small and often crammed in between the lines. However, the very long addition in 7:30–8:3 was added by a scribe other than the original one. See my analysis in Ulrich et al., DJD XV, 152 and Kipp Davis, “Margins as Media.”
See the apparatus in DJD XV. Lange, Handbuch, 299 calculates seventy-four variants.
The handwriting of the long addition in the interlinear space, in the intercolumnar margin, and below the text reveals a different writing style from that of the main scribe (note the different shapes of the ʾaleph, bet, lamed, mem, final mem, ‛ayin, and shin). Likewise, in col. XI 6, the added
Judith E. Sanderson, “14. 4QExodc,” in Qumran Cave 4.VII: Genesis to Numbers, ed. Eugene Ulrich and Frank Moore Cross, DJD XII (Oxford: Clarendon, 1994 [repr. 1999]), 97–125.
The evidence is listed by Russell E. Fuller, “78. 4QXIIc,” in DJD XV, 237–251 (239).
The evidence is listed by Russell E. Fuller, “82. 4QXIIc,” in DJD XV, 271–318 (275).
In thirty-two details or groups of details as well as two large omissions of verses.
Based on the publications of James A. Sanders, The Psalms Scroll of Qumrân Cave 11 (11QP sa), DJD IV (Oxford: Clarendon, 1965) and Florentino García Martínez, Eibert J.C. Tigchelaar, and Adam S. van der Woude, “5. 11QPsalmsa, Fragments E, F,” in Qumran Cave 11.II: 11Q2–18, 11Q20–30, DJD XXIII (Oxford: Clarendon, 1998), 29–36.
Ps 104:24, 32; 105:6; 118:26; 119:27, 140, 154, 155; 122:3; 123:1; 137:1; 138:1; 143:5, 6; 145:3, 15. See also the list of erasures in Sanders, DJD IV, 14.
Ps 119:42; 121:2; 143:3.
Ps 139:17; 141:10; 145:6, 15.
Ps 119:175; 145:6.
Ps 9:12; 27:6; 98:5; 101:1.
Ps 105:6.
For example, cols. XIX 8; XXIV 12, 13; XXVII 2; XXVIII 3, 11.
The data were examined according to the interpretation of the edition of Ulrich–Flint, DJD XXXII, further aided by the insights of E. Yehezkel Kutscher, The Language and Linguistic Background of the Isaiah Scroll (1 Q Isa), STDJ 6 (Leiden: Brill, 1974), 519–547. Not included are additions combined with erasures (e.g., Isa 5:29) or instances in which the supralinear letters were added above the line because there was no room in the column at the end of a sheet, e.g., 3:9
E.g., 1:1
E.g., 1:3
E.g., 12:2
3:18*; 7:16; 8:7*; 10:17; 11:4; 12:6; 15:7; 19:5; 21:1; 23:8; 27:4; 34:12; 35:10; 36:4, 7; 37:27; 41:20; 48:4; 49:14*. The starred references record occurrences of divine names.
3:17*; 34:17.
2:4; 12:6; 21:1.
Certainly in 10:17
Also the continuation of this phrase in 1QIsaa is harmonized with 37:10
This scroll was well preserved in a jar and is quoted by the sectarian 1QHa, see n. 7 above. By the same token, in several cases, 4QTanḥ (4Q176) is close to 1QIsaa against MT, see n. 6 above. Furthermore, the scroll is replete with marginal notations that are also found in the sectarian 1QS and with letters in the Cryptic A script that probably have a sectarian background. For both, see Emanuel Tov, Scribal Practices and Approaches Reflected in the Texts Found in the Judean Desert, STDJ 54 (Leiden: Brill, 2004), 203–209.
Suggestion of the anonymous reader of this manuscript.
The base text
See the analysis of James R. Davila, “9. 4QGenj,” in DJD XII, 65–73.
Maurice Baillet in idem et al., Les ‘petites grottes’ de Qumrân, DJD III, 61.
This text, written in the Qumran scribal practice, agrees more with MT against the LXX than vice versa (vv. 9, 10).
Jozef T. Milik, DJD III, 169–171. See n. 17 above and my study “The Textual Base of the Corrections in the Biblical Texts Found in Qumran,” in The Dead Sea Scrolls: Forty Years of Research, ed. D. Dimant and U. Rappaport (Leiden: Brill; Jerusalem: Magnes and Yad Izhak Ben-Zvi, 1992), 299–314.
The addition of
According to the new readings of the Leon Levy Dead Sea Scrolls Digital Library, the aleph of the possible plus in 8:19 in 5QDeut [
Above “Anak,” Milik reads a yod (
On the other hand, Julie A. Duncan, “New Readings for the ‘Blessing of Moses’ from Qumran,” JBL 114 (1995): 273–290 (288) cautiously suggested proximity to the LXX.
See Emanuel Tov, “The Possible Revision of Hebrew Texts According to MT,” in From Scribal Error to Rewriting: How Ancient Texts Could and Could Not Be Changed, ed. A. Aejmelaeus, D. Longacre, and N. Mirotadze, De Septuaginta Investigationes 12 (Göttingen: Vandenhoeck & Ruprecht, 2020), 147–163.
The kaige-Th revision embedded in the Greek Minor Prophets Scroll from Naḥal Ḥever (8ḤevXIIgr). See Emanuel Tov with the collaboration of Robert A. Kraft, The Greek Minor Prophets Scroll from Naḥal Ḥever (8ḤevXIIgr) (The Seiyal Collection I), DJD VIII (Oxford: Clarendon, 1990).
For the Qumran origin of 1QIsaa, see n. 64. For the liturgical Qumran environment of 11QP sa, see Tov, TCHB4, 130, n. 42, with bibliography.
Tov, TCHB4, 37–40.
The proliferation of proto-MT and MT-like (for Lange: “semi-MT”) scrolls around the turn of the eras as presented by Armin Lange, “1.2 Ancient Hebrew-Aramaic Texts,” in THB, Vol. 1A, 82–166 (149) is impressive. However, (1) the chart does not distinguish between Qumran and the other sites, and this period happens to be the peak for the other sites; (2) the period of the peak of the MT-like scrolls at Qumran overlaps with that of the totality of the scrolls, and this fact makes me wonder about the strength of the argument relating only to the proto-MT and MT-like scrolls.
Tov, TCHB4, 374–378 (“The Myth of Textual Stabilization”).
However, the tradition is problematic as it is unclear how the master copy was created. Shemaryahu Talmon, Text and Canon of the Hebrew Bible (Winona Lake, IN: Eisenbrauns, 2010), 329–346 analyzes the different versions of this tradition. The talmudic tradition supposedly records the limiting of the differences between three specific texts by comparing their readings in each individual instance of disagreement. At the end of this comparison, presumably a new copy was created that contained the majority readings of these scrolls (the agreement of two sources against the third one). Although such an activity seems to be implied by the baraita quoted below, the procedures followed are unclear and thus the story itself is not trustworthy. It is unclear what the relation is between the three scrolls; seemingly there were three pairs of three scrolls. In the story of the three scrolls, the majority reading reflects MT.
Three scrolls of the Law were found in the temple court. These were the maʿon (“dwelling”) scroll, the zaʿaṭuṭê (“little ones”) scroll, and the hy’ scroll. In one of these scrolls they found written, “The eternal God is (your) dwelling place (
מָעוֹן maʿon)” (Deut 33:27). And in two of the scrolls it was written, “The eternal God is (your) dwelling place (מְעֹנָה mᵉʿonah = MT).” They adopted the reading found in the two and discarded the other. In one of them they found written, “He sent the little ones (zaʿaṭuṭê) of the sons of Israel” (Exod 24:5). And in two it was written, “He sent young men (naʿᵃrê = MT) of the sons of Israel.” They adopted the reading found in the two and discarded the other. In one of them they found written הוא hwʾ [MT: הִוא ] nine times, and in two, they found it written היא hyʾ eleven times. They adopted the reading found in the two and discarded the other (y. Taʿan. 4.68a; Sof. 6.4 probably has a better version: “In one הוא was written eleven times, and in two היא was written eleven times.”).
This study was read as a paper at the SBL Annual Meeting in San Antonio in November 2023.
Col. 3 lists the corrections. Thus in the first example,
