Non-Roman fonts used: SPTiberian, SPIonic
1. Although it is generally agreed that the Ketiv-Qere system was developed during the Masoretic period,2 the ultimate origins of the readings contained in the system are still not fully understood. Historically, attempts to explain the origins of the Ketiv-Qere readings have centered around two basic models. According to one model, both the Ketiv and the Qere represent variant readings which can be traced back to an ancient collation of manuscripts. According to the other model, readers introduced the Qere into the written text (the Ketiv) with the intention of correcting what they perceived to be an error. Both views have been held in some fashion from early on in the study of the Masorah, and both views still exist in modern times. In addition, several new approaches have emerged, most of which attempt in some way to combine features of the two traditional models. It will be suggested here that these two traditional models have not supplied an adequate framework for evaluating the origins of the Ketiv-Qere readings, and that a better approach can be established by focusing on the central questions which cut across both traditional positions.
2. The collation theory was ascribed to David Kimh@i3 in Jacob ben Chayyim's Introduction to the Rabbinic Bible. Ben Chayyim cited the following explanation as that of Kimh@i:
Ben Chayyim himself objected to this view, since he believed that both the Ketiv and the Qere represented ynysm h#ml hklh.4 Yet he not only preserves Kimh@i's collation theory, but he also ascribes a similar position to the grammarian R. Isaac b. Moses Ha-Levi (Ginsburg 1867: 42-43).5
[It appears that these marginal and textual readings originated because the sacred books were lost and scattered about during the Babylonian captivity, and the sages who were skilled in the Scriptures were dead. Whereupon the men of the Great Synagogue, who restored the law to its former state, found different readings in the books, and adopted those which the majority of copies had, because they, according to their opinion, exhibited the true readings. In some places they wrote down one word in the text but did not punctuate it, or noted it in the margin but omitted it from the text, whilst in other places they inserted one reading in the margin and another in the text] (Ginsburg 1867: 43-44).
Myrpsh wdb) hnw#)rh twlgb# ypl Nk w)cmn hl)h twlmh yk h)rnw
wryzxh# hlwdgh tsnk y#n) wtm )rqmh y(dwy Mymkxhw lw+l+ wl+l+nw
Mt(d ypl bwrh rx) Mhb wklhw Myrpsb tqwlxm w)cm hn#yl hrwth
Mynpbm wbtk )lw Cwxbm wbtk w) whwdqn )lw dx)h wbtk dwrybh l(
Mynpbm rx) Krdbw Cwxbm dx) Krdb wbtk Nkw
3. The idea that the Ketiv-Qere readings originated in a collation of manuscripts is also one of the key components in many modern theories. This perspective may be represented by H. M. Orlinsky:
It is our hypothesis that the Masoretes first selected the three best manuscripts of the Hebrew Bible available to them. Where the three manuscripts had no variant readings, no difficulty was experienced in vocalizing the text. But where the manuscripts differed, the Masoretes accepted the reading of the majority and vocalized it; that reading became the Qere. The reading of the minority was left unvocalized, and became the Kethib (Orlinsky 1960:187).The proposal of David Kimh@i on the one hand and that of Orlinsky on the other differ on the date to which they assign the collation process. Yet, they agree on the basic premise that the Qere is to be identified with a reading which previously existed in a manuscript and which was derived by a process of collation and the mechanical acceptance of the majority reading.
4. Others, however, have argued that the Qere originated as an oral correction to the written text, without manuscript support. Ben Chayyim described (and refuted) the opinion of Don Isaac Abravanel,6 who held a view on the origins of the Ketiv-Qere which essentially accords with this correction model. Abravanel's view, according to Ben Chayyim, was that:
Abravanel offered two possible explanations for these "irregular" (Myrz) words. Perhaps the prophetic writers intended to convey mysteries of the law through these anomalous expressions, so that Ezra and his associates merely supplied in the margin "corrected" forms according to common idiom, leaving the unusual but intentionally mysterious readings in the text. The Qere, in this case, would simply assist the reader in grasping the proper nature of the expression (Nw#lh (b+) and the straightforward, contextual sense (Nyn(h tw+y#p), leaving unaltered the Ketiv, which contained a deeper significance.
[Ezra the scribe and his associates found the books of the law entire and perfect, but before betaking themselves to make the vowel points, the accents and the division of verses, they examined the text, when they found words which, according to the genius of the language and the design of the narrative, appeared to them irregular] Ginsburg 1867: 45-46)
Mtwmtw Mtwml#b hrwth yrps w)cm wt(ysw rpwsh )rz( yk
)rqmb Nyy( Myqwsp ypwsw Mym(+hw dwqnh tw#(l rrw(th# Mdwqw
rwpsh tnwwkw Nw#lh (b+ ypk Myrz wyl) w)rn r#) Myrbdhw
5. As another possibility, Abravanel suggested that the anomalous words were due to "a lack of necessary grammatical precision or a deficiency in the knowledge of precise writing" (hbytkh qwdqd t(ydy rwcqb M) yw)rk qdqdm ytlb), which was "from the prophet like an error which proceeded from a prince" (+yl#h ynplm t)cwyh hgg#k )ybnhm). Ezra was therefore forced to explain these words "in accordance with the narrative context" (rwpys ypk), this explanation being the subject of the Qere which is indicated in the margin (Cwxbm M# r#) yrqh Nyn( )whw) (Ginsburg 1867: 46).
6. Thus, according to Abravanel, Ezra perceived an "irregularity" in the text and corrected it without consulting any manuscripts, relying primarily on the context. Of course, in neither case was this intended to be a text-critical intervention, since the correction was envisioned not as a restoration of the original text but as some kind of improvement on it.
7. More recently, some scholars have supported a correction theory which likens the Qere, in text-critical terms, to a conjectural emendation. For example, J. Weingreen describes the Ketiv-Qere system as having originated because the Masoretes offered corrections to the text wherever a miscopying was evident. As Weingreen explains, "This they did by drawing the attention of the reader to errors and directing that the corrections, which they supplied, were to be made orally" (Weingreen 1982:15-16). The corrections, according to Weingreen, were made only in reading, and not in the text itself, because the Masoretes wished to preserve unaltered the written tradition which they had received, even though they believed it to be in error. As with the view of Abravanel, the anomaly in the text is corrected essentially by instinct, without consulting a manuscript. For Weingreen, however, the anomaly was considered to be a copying error, so that the correction was intended to restore the original text.
8. While the collation model and the correction model have each survived independently into modern times, many scholars have combined elements of both in explaining the historical development of the Ketiv-Qere.7 Christian Ginsburg proposed a two-stage process by which the Masoretes assembled the Ketiv-Qere readings.
9. The first stage began with the registration of anomalous forms in the Pentateuch by the Sopherim in the Second Temple period. This record, which was transmitted unchanged by the Masoretes, became the basis for the Ketiv-Qere readings in the Pentateuch (Ginsburg 1897: 421-422). The method of the Sopherim involved correcting "clerical mistakes" (Ginsburg 1897: 313), such as haplography and dittography, and comparing parallel passages and different recensions.8 When the correction offered by the Sopherim was based solely on internal evidence, the lack of manuscript support prevented them from inserting their reading into the text (Ginsburg 1897: 309, 314). Thus, all of the Qere readings from this first stage represent conjectural emendations, manuscript corrections at this point having been added directly into the text.
10. The second layer of variants was generated by the Masoretes themselves. Since the Prophets and Hagiographa were not as carefully preserved as the Pentateuch, the textual tradition which was handed down to the Masoretes for these books contained variants from other standard codices. The Masoretes simply catalogued these variants using the Ketiv-Qere system, refusing at this late stage to interfere with the process of transmission (Ginsburg 1897: 422-423). Therefore, according to Ginsburg, the Ketiv-Qere readings in the Prophets and Hagiographa contain manuscript variants which were handed down from before the Masoretic period. Thus, Ginsburg's two-stage theory ascribes the Qere readings to (1) oral correction (in the first stage), and (2) manuscript collation without critical activity (in the second stage).
11. Another theory on the origins of the Ketiv-Qere which combines collation and correction into a historical framework was proposed by Robert Gordis. According to Gordis, the first and earliest Ketiv-Qere variants involved the Tetragrammaton and the substitution of euphemisms for indelicate expressions, since both are mentioned in the Talmud (Gordis 1971: 29-31).9 In these cases, the Qere guards the reader against blasphemy and obscenity. The second category of Ketiv-Qere variants was intended to safeguard the reader against ignorance, by guiding the reader to the correct vocalic interpretation of the consonantal text (Gordis 1971: 35-37).10 This second category predated the Masoretic vowel pointing system, which eventually rendered this kind of guide unnecessary. Gordis described the first two stages as "genuine Kethib-Qere" (Gordis 1971: 37), because in both cases the Qere simply admonishes the reader to vocalize something different than what is written in the text.
12. According to Gordis, the introduction of vowel pointing made the use of the Ketiv-Qere system unnecessary for guiding the reader to the correct vocalic interpretation. But when the need arose to find a way to register manuscript variants in the text, the already existing Ketiv-Qere system was put to use for this purpose (Gordis 1971: 440-441).11
13. Although this third type of Qere was the latest to be incorporated into the system, the variants which were transmitted were not new. The readings which eventually became the Qere were based on highly regarded standard codices which were kept in the temple library, along with the official "archetype" manuscript which Gordis believed to have been established by this time (Gordis 1971: 46-49). After these early manuscript variants had been incorporated into the Ketiv-Qere framework (third stage), their original status as textual variants was forgotten and they were absorbed into the already existing system, so that even this type of Qere became obligatory in reading (Gordis 1971: 53-54). Thus, in a fashion somewhat similar to Ginsburg, Gordis described the origins of the Ketiv-Qere readings in terms of historical stages, with the earlier stages consisting in oral corrections and the later stage representing manuscript variants.
14. Lastly, two new approaches have been suggested for explaining the origins of the Ketiv-Qere, one based on the inner-biblical relationships between particular Ketiv-Qere variants, and the other based on the specific meanings of the terms bytk and yrq.
15. The first new approach is actually a variation on the manuscript collation theory. It is based on the observation that many Qere readings in the text of Samuel-Kings are equivalent to the Ketiv in their synoptic passages in Chronicles. According to Gillis Gerleman, this indicates that many Ketiv-Qere readings can be traced back to ancient critical activity, reflecting two competing recensions of an earlier historical source (Gerleman 1948: 24). The written text of Samuel-Kings represents a critically revised edition, while Chronicles contains a long accepted, popular text (Gerleman 1948: 24). Even though the critical revision (Samuel-Kings) successfully restored many of the original readings, it was not completely able to replace the familiar text known to most people (Chronicles). Therefore, the many Qere readings in Samuel-Kings which correspond to the parallel Ketiv in Chronicles are simply concessions to the popular version (Gerleman 1948: 25).
16. A similar view was also expressed by Alexander Sperber, who suggested that the Talmud (or its source) "designated with bytk that recension of historic narrative, which is now included in the Former Prophets, while yrq was applied to the other recension, which Chronicles now exhibits" (Sperber 1942-1943: 303).
17. Sperber differed from Gerleman in that he did not consider one recension to be critically revised and another to be popular. They merely represent different recensions of the historic Annales. Furthermore, the terms bytk and yrq are of no help in determining the origin of the Ketiv-Qere readings, since these terms are probably nothing more than popular explanations of the original symbols, which had been used to identify the two recensions, but which were misunderstood by subsequent generations (Sperber 1942-1943: 303). Still, both Gerleman and Sperber agree that the Ketiv-Qere variants between Samuel-Kings and Chronicles contain the readings of different ancient recensions.12
18. The other new approach was suggested by James Barr, who used the Masoretic terminology bytk and yrq as the starting point for his theory. According to Barr, an independent reading tradition had already developed by the time an authoritative written text came to be accepted. Yet, since this text did not agree in every detail with the oral tradition, the Qere notes were added in order to protect the written text from corruption by the Qere (Barr 1981: 35-37). Barr is open to the possibility that Qere readings were taken from manuscripts, but he considers it better to understand the function of the Qere notes from the perspective of the liturgical reading tradition (Barr 1981: 36).
19. Barr's view has been followed by W. S. Morrow and E. G. Clarke, who argue from the Ketiv-Qere variants in the Aramaic portions of Daniel that the Qere represents a later form of Aramaic, while the Ketiv retains an older, conservative pronunciation (Morrow and Clarke 1986: 419).13 Thus, linguistic evolution partially explains how a reading tradition would develop which is different from the manuscript tradition.
20. Barr's contribution is that he attempts to look at the Ketiv-Qere phenomenon through the lens of the oral and written traditions which most likely formed the basis for the development of the system as it now exists. Barr's theory does not, however, explain how a reading tradition developed in the first place which differed in such a variety of ways throughout the whole canon from what came to be the authoritative written tradition.14 Although linguistic evolution would explain some of the phenomena (cf. Morrow and Clarke 1986), it would not explain all of them (e.g., euphemisms or the confusion of similar letters). If the oral readings never existed in manuscripts, they must at some point have been invented by readers. Barr's view rightly emphasizes the importance of the terminology, but he does not directly address the ultimate origins of the readings represented by the Qere. Barr is suggestive, however, in recommending an approach that could be reconciled with either of the traditional models.
21. In fact, the traditional models of correction and collation are not necessarily exclusive of one another. The idea of collation has usually been associated with manuscript evidence, and the correction theory has most often presumed strictly oral corrections; but these particular associations are not necessary. It is of course possible for manuscripts to be used mechanically to supply marginal variants, but it is also possible that manuscripts were used at points to correct perceived errors in the authoritative text. Likewise, although Qere readings could represent nothing but oral corrections to what is written, it is also possible that the Qere is actually an oral record whose purpose is not to correct the Ketiv at all, but to protect it (thus Barr 1981). In sum, the basic conception of the two traditional models as competing theories has not served to produce a clear understanding of the central issues relevant for describing the origins of the Ketiv-Qere.
22. It would perhaps be helpful to consider each of these theories independently by asking two basic questions: (1) What was the intended status of the Qere in relation to the Ketiv? (2) Does the Qere represent a manuscript reading? By approaching these questions separately, it is easier to see how they are interrelated with each other, not opposed to one another. It is suggested that these two separate but interrelated questions provide a firm starting point for evaluating the evidence for the origins of the Ketiv-Qere readings.
23. Considering the first question, the belief that the Qere represents an intended replacement of the written text is based on the observation that many Qere readings appear to simplify or standardize difficult or unusual passages. For example, it seems logical to suppose that where the Qere provides a euphemistic expression in place of an unpleasant word, the Qere was added in order to "improve" the Ketiv. Thus, in Isaiah 13:16:
|Isa 13:16||K hnFl;ga#$%ft@i Mhey#$'n:w%||Q hnFb;ka#$%ft@i Mhey#$'n:w%|
24. It also seems likely that the Qere is correcting the Ketiv in places where the Qere gives a standardized plene or defective rendering for an unusual Ketiv reading. This may be seen in Num 23:13 with the imperative form of K7lh:
|Num 23:13||K MwOqmf-l)e yti@)i )n%F-K1l;||Q MwOqmf-l)e yti@)i )n%F-hkfl;|
25. Further evidence that the Qere is intended to replace the Ketiv can be seen in cases where the Qere appears to modernize an archaic form, as in the numerous places where the archaic/ dialectical yt@i)a is "corrected" to t@;)a. Thus, in Jdg 17:2:
|Jdg 17:2||K tyli)f yt@i)aw:||Q tyli)f t@;)aw:|
26. The numerous examples of Ketiv-Qere variants which are similar to these have often led to the belief that the purpose of the Qere was to serve as a replacement for the Ketiv.15 If this were the case, whether or not manuscripts were involved, the purpose of the notes would be to subordinate the Ketiv to the Qere in the mind of the reader or listener.
27. Yet, other aspects of the Ketiv-Qere system have been problematic for this theory. First, there are Ketiv-Qere variants in some places which are reversed (i.e. given as Qere-Ketiv) in other places. This occurs with the spelling of MyrIw%s)j / MyrIys)j:
|Jdg 16:21||K MyrIysi)j||Q MyrIw%s)j|
|Gen 39:20||K yr"w%s)j||Q yr"ysi)j|
28. Second, there are places where a Qere note is given for a particular Ketiv reading which occurs elsewhere in the same book without a Qere note. For example:
|Gen 24:33||K lko)vle wynFpfl; M#&eyy,IwA||Q lko)vle wynFpfl; M#&aw%y,wA|
|Gen 50:26||K MyirFc;mib@; NwOr)fb@f M#&eyy,IwA||(Qere lacking)|
29. A third objection to the preferred status of the Qere was highlighted by Orlinsky, who pointed out that sometimes the Qere contains a reading which is actually more difficult than the Ketiv (Orlinsky 1960:186). An example is given from the first Ketiv-Qere in the Bible:
|Gen 8:17||K )c'who||Q )c'y:ha|
30. Yet, although these objections are genuinely problematic to the position that the Qere was meant to be preferred, in the majority of cases, the Qere readings do seem to "improve" the written text. Even Orlinsky suggested that it was the reading of the majority of manuscripts that was vocalized (Orlinsky 1960:190-191). Also Gerleman, who likewise supported the "collation" theory rather than the "correction" theory, nevertheless conceded that the popular readings contained in the Qere of Samuel-Kings represented practical preferences established by familiar use (Gerleman 1948: 26). Barr has rightly emphasized that, while some Ketiv-Qere variants appear unsystematic, there are enough regularities to suggest some tendencies in specific directions (Barr 1981: 34). Perhaps if many of these intended improvements made their way into popular manuscripts, and these manuscripts were instrumental in the development of the reading tradition, then the reason why some Qere readings were not applied consistently or exhaustively was that the manuscripts contained changes in some places but not in others. This popular tradition had certain tendencies, but because the Masoretes were merely following it, they did not produce a completely unified system. In general, though, we may suggest that the Qere tradition represents the recommended reading.
31. With regard to the second question, whether or not Qere readings are based on manuscripts, the fact that numerous Qere readings are found in ancient textual witnesses indicates that at least some Qere readings were based ultimately on manuscript evidence. The LXX frequently corresponds to the Qere rather than the Ketiv. Despite some problems of method pointed out by Orlinsky (Orlinsky 1940: 38-41), Gordis' study demonstrates sufficiently that the LXX corresponds to the Qere at least as often as to the Ketiv, if not more often (Gordis 1971: 60-65).16 A similar abundance of Qere readings can be seen in the Vulgate and the Peshitta (Gordis 1971: 66).17 The translation of Aquila also shows a preference for the Qere (Reider 1916-1917: 292). Furthermore, Qere readings have been found in biblical manuscripts from Qumran. In at least seven places, the Isaiah scroll contains a reading which agrees with the Qere against the Ketiv (Roberts 1951: 69).18 The existence of all these Qere readings--not in marginal notes, but in texts--suggests that some kind of manuscript evidence was a part of the tradition which underlies the Qere notations.
32. At the same time, there are difficulties with the idea that the Qere readings were taken directly from a collation of manuscripts. First, if the Qere represents simply a variant reading from another manuscript, what is the significance of the terminology bytk and yrq, and why was the Qere considered obligatory in reading?
33. Second, if the Qere represents a manuscript variant, then why were no more than two readings ever recorded? Gordis suggested that since the number of manuscripts consulted was not large, there were never more than two variants found for the same reading (Gordis 1971: 49). Yet, if the variants are as old as suggested by Gordis, this explanation is difficult to reconcile with the textual diversity which is now known to have existed for the Hebrew Bible in antiquity. In fact, Arie Rubenstein, comparing the text and corrections of the Isaiah scroll from Qumran, Ketiv-Qere readings from MT, and the synoptic passages of Isaiah and 2 Kings, showed that three variants sometimes existed for a single reading (Rubenstein 1959: 127-133). The manuscript collation theory cannot sufficiently explain why only one variant was recorded in the margin.
34. Third, there is no evidence in the pre-Masoretic transmission history of the Hebrew Bible to suggest that manuscript collation was ever used to create a critical apparatus. It does appear that at one time textual variants were inserted directly into texts--sometimes at the end of verses, sometimes near the atnah@, but most often side by side with their corresponding readings (Zimmermann 1943-1944: 459-474). But there is no evidence for variant readings being placed in the margin. Even the frequently cited report of the three Torah scrolls found in the Azarah is not directly relevant to the existence of a critical apparatus,19 since the reading of the minority was not recorded but discarded.20 Thus, the activity suggested by the collation theory is without precedent for the transmission of the Hebrew Bible in antiquity.
35. In sum, despite the evidence for Qere readings in ancient texts, there are significant reasons to doubt that the Ketiv-Qere system represents a direct process of consulting manuscripts and recording variants. If, on the other hand, the Qere notes were taken from an authoritative reading tradition, this would account for the bytk / yrq terminology as well as the fact that only one Qere is given for each Ketiv.
36. In answer to the two questions which were asked about the origins of the Ketiv-Qere, it is likely (1) that the Masoretes intended for the Qere to be preferred over the Ketiv, and (2) that manuscript readings do underlie some Qere variants, although it is not likely that they derive directly from a critical collation of manuscripts. Perhaps the immediate origin of the Ketiv-Qere system was the need to record both an authoritative written text and a separate reading tradition, but the ultimate source of the reading tradition was a popular manuscript recension. This would account for both the presence of Qere readings in ancient sources and the function which the Ketiv-Qere system seems to have performed during the Masoretic period.
© TC: A Journal of Biblical Textual Criticism, 2003.
1This paper originated as part of a seminar on the Masorah led by Professor David Weisberg at the Hebrew Union College-Jewish Institute of Religion. I am grateful to Prof. Weisberg for his encouragement, and his helpful comments have improved the paper throughout. An earlier draft of the paper was read at the Southeastern Regional SBL Meeting in 2003, and I wish to thank those who participated for their comments. Any errors which remain are of course my responsibility.
2There are no references to the writing of Ketiv-Qere notes in either Talmud or in Jerome. See Orlinsky 1960: 186-187. Phenomena similar to what is found in the Ketiv/Qere system are sometimes seen in classical Rabbinic literature, e.g., the exegetical technique . . . )l) . . . yrqt l), or the discussion of wl versus )l in bSota 31a.
3yxmq dwd (1160-1235).
4Cf. pages 50-51, 54-57, 69-70; cf. also bNed 37b.
5= "Ephodi" (fl. 1360-1412).
6l)nbrb) qxcy Nwd (1437-1508).
7Cf. Tov 2001: 62-63.
8Cf. Ginsburg's account of the Sopherim regarding the Nbytk )lw Nyyrq and the Nyyrq )lw Nbytk (309-318). Ginsburg later refers to the Sopherim as "authorised revisers and redactors of the text" (421).
9Cf. bPesah 50a and bMeg 25a.
10E.g., the Qere indicated w where the older masculine suffix h might be misunderstood as the feminine h.
11According to Gordis, the procedures for handling variant readings had been evolving from before the destruction of the Second Temple. Early on, variant readings were disregarded. Later, they were incorporated into the body of the text. Finally, they came to be added into the margin, being registered as if Qere readings (Gordis 1971: 41-44).
12Saul Lieberman also argued that the Ketiv-Qere system represented ancient recensions. The contemporaneous circulation and use of "corrected editions" and "popular editions" finds a parallel in the transmission history of Homer (Lieberman 1962: 21-27). Critical editions (diorqw&seij) of the Iliad and Odyssey were produced in Alexandria by Zenodotus of Ephesus in the third century BCE. The text of Homer seems to have been in a state of flux in the third century, with some standardization beginning in the middle of the second century (Pfeiffer 1968: 109-111). Aristarchus of Samothrace (c. 217-145 B.C.E.) may have produced his own complete editions of the Homeric texts, or he may simply have based his critical comments on the popular editions (koinai\ e0kdo/seij), cf. Pfeiffer 1968: 213-215.
13Cf. also Fassberg 1989. Morrow and Clarke modify Barr's suggestion by arguing that the purpose of the Ketiv-Qere system was to preserve both the written and the reading traditions from cross-contamination (Morrow and Clarke 1986: 421). But in another article, Morrow returns to Barr's contention that the sole purpose of the Ketiv-Qere was to preserve the Ketiv (Morrow 1992: 27).
14For example, it is doubtful that a reading tradition for Chronicles could have developed that was so well-known and customary that it required the writing of Qere notes in order to protect the Ketiv. Such an explanation would be more credible if Qere notes only existed for passages which were read liturgically.
15For more examples of these phenomena, cf. Yeiven 1980: 56-57.
16Although Orlinsky has given reason to doubt the precise accuracy of Gordis' figures, the larger point made by Gordis still stands (according to Gordis, the LXX actually prefers the Qere to the Ketiv by a margin of 320 to 213).
17Again, according to Gordis' figures, the Vulgate prefers the Qere 330 to 124, and the Peshitta prefers the Qere 328 to 122.
18Roberts identifies the following passages: 13:16; 49:5; 54:16; 55:13; 56:10; 65:4; 66:17.
19Cf. pTaan 4.2, 68a; SifreDeut 356 (ed. Finkelstein, p. 423); ARN, Text B (ed. Schechter, p. 129); Soferim 6.4.
20Cf. Lauterbach 1917-1918: 385-423; Talmon 1962: 14-27; Zeitlin 1966: 269-272. Zeitlin doubts the historicity of the account, arguing that there was only one scroll kept in the Azarah (cf. mKelim 15:6, mMoedQat 3:4, tKelim BMes 5:8). In any case, the general wording of the passages, Myn#h wmyyqw dx)h t) (Mymkx) wl+b, does not suggest any kind of critical apparatus.
Barr, James 1981. "A New Look at Kethibh-Qere." OTS 21: 19-37.
Fassberg, Steven E. 1989. "The Origin of the Ketib/Qere in the Aramaic Portions of Ezra and Daniel." VT: 39: 1-12.
Gerleman, Gillis 1948. Synoptic Studies in the Old Testament. Lund: C. W. K. Gleerup.
Ginsburg, Christian D. 1867. Jacob Ben Chajim Ibn Adonijah's Introduction to the Rabbinic Bible. 2d ed. London: Longmans, Green, Reader, and Dyer. [Originally in volume one (Pentateuch) of Ben Chajim's "Rabbinic Bible," 1524-25].
Ginsburg, Christian D. 1897. Introduction to the Massoretico-Critical Edition of the Hebrew Bible. London: Trinitarian Bible Society.
Gordis, Robert 1971. The Biblical Text in the Making: A Study of the Kethib-Qere. New York: KTAV.
Lauterbach, Jacob Z. 1917-1918. "The Three Books Found at the Temple in Jerusalem." JQR 8: 385-423.
Lieberman, Saul 1962. Hellenism in Jewish Palestine. New York: Jewish Theological Seminary.
Morrow, William S. 1992. "Kethib and Qere." ABD. Vol. 4: 24-30.
Morrow, William S., and Clarke, Ernest G. 1986. "The Ketib/Qere in the Aramaic Portions of Ezra and Daniel." VT 36: 406-422.
Orlinsky, H. M. 1940. "Problems of Kethib-Qere." JAOS 60: 30-45.
Orlinsky, H. M. 1960. The Origin of the Kethib-Qere System: A New Approach. In Congress Volume, 184-192. VTSup, no. 7.
Pfeiffer, Rudolf 1968. History of Classical Scholarship. Oxford: Clarendon Press.
Reider, Joseph 1916-1917. "Prolegomena to a Greek-Hebrew and Hebrew-Greek Index to Aquila." JQR 7: 287-364.
Roberts, B. J. 1951. The Old Testament Text and Versions. Cardiff: University of Wales Press.
Rubenstein, Arie 1959. "A Kethib-Qere Problem in the Light of the Isaiah Scroll." JSS 4: 127-33.
Sperber, Alexander 1942-1943. "Problems of the Masora." HUCA 17: 293-394.
Talmon, Shemaryahu 1962. "The Three Scrolls of the Law That Were Found in the Temple Court." Textus 2: 14-27.
Tov, Emanuel 2001. Textual Criticism of the Hebrew Bible. 2d ed. Minneapolis: Fortress Press.
Weingreen, J. 1982. Introduction to the Critical Study of the Text of the Hebrew Bible. Oxford: Clarendon Press.
Yeiven, Israel 1980. Introduction to the Tiberian Masorah. Translated and Edited by E. J. Revell. Masoretic Studies 5. Missoula, MT: Scholars Press.
Zeitlin, Solomon 1966. "Were There Three Torah-Scrolls in the Azarah?" JQR 56: 269-272.
Zimmermann, Frank 1943-1944. "The Perpetuation of Variants in the Masoretic Text." JQR 34: 459-474.