Philip W. Comfort and David P. Barrett, eds. The Text of the Earliest New Testament Greek Manuscripts: A Corrected, Enlarged Edition of The Complete Text of the Earliest New Testament Manuscripts. Wheaton: Tyndale House, 2001. 607 pp. $44.99 list.

1. The title of this book recalls the sad tale of its ill-fated earlier incarnation. The initial volume had been published and released by Baker Book House in 1999 following a number of lengthy postponements. Yet that eagerly-anticipated publication was swiftly recalled by Baker due to various severe criticisms, most of which concerned the fact of too many errors within the published transcriptions of the ancient manuscripts--this being a virtual "kiss of death" for a publication of this nature. Baker therefore removed that entire press run from the market, apparently with no intent to publish a corrected edition or even an errata sheet. The editors (Comfort and Barrett, hereafter C/B) therefore began an extensive correction and improvement process regarding both the texts and format of those manuscripts which had appeared in the former edition. In the process they expanded the volume by adding in many newly discovered papyrus fragments.

2. As a matter of disclaimer, the present reviewer was among those who had noted and criticized certain of the transcriptional errors in the former edition (he and nine others are named in the current acknowledgments section as having offered "helpful suggestions for the second edition")¹. My own suggestion during that period of tumultuous discussion on the TC-List was that the work should be begun anew, and every transcriptional detail checked and verified by outside parties in an attempt to produce a volume which could be considered trustworthy by all who might use it. C/B on the other hand chose to rework and expand the material themselves, and in this light a reviewer such as myself might tend to approach the new work with a certain degree of skepticism. Whether such has a legitimate foundation can of course only be determined by examination of what is presented in the new volume, and that discussion will occur later in this review.

3. The first matters to be discussed, however, relate to the plan, organization, and methodology for presenting the text of "the earliest New Testament Greek manuscripts." There are plusses and minuses in these areas which readers will react to in differing proportions, and some of these deal with certain opinions stated by the editors or what might be perceived as factual errors. For my part, I am more concerned with the textual transcriptions and their accuracy than with what might be stated in the introductory matter, but even so, there are some items which should be noted.

The Plan

4. The plan of the volume is to present complete transcriptions of at least the extant text of all New Testament manuscripts which date before 300 CE, whether written on papyrus or vellum. In a basic sense this has certainly been done, since all manuscripts which are recognized by palaeographers to date from that era are definitely included herein. There are 60 papyri included², some combined as part of what C/B and sometimes others consider a single manuscript, i.e., 4/64/67, 15/16, 49/65, 77/103. Also, four early vellum uncials are included which happen to meet the date requirements (0162, 0171, 0189, 0220). One item (P. Antinoopolis 2.54) is apparently only an amulet and is not considered to be a New Testament manuscript by the M�nster Institut. The inclusion of a "non-item" of course raises an obvious question: why include this amulet, and (since it is included), why not include transcriptions of all Ostraca and Talismen containing New Testament texts, particularly since these have been relatively inaccessible since the time of Gregory?

5. On the other hand, C/B appear to apply their own palaeographical criteria in dating various manuscripts and tend to claim an earlier date for many manuscripts included in their volume than might be allowed by other palaeographers. For example, C/B include the following papyri which are dated by Nestle-Aland27 (hereafter NA27) as clearly fourth century (IV), which under normal standards should suggest something later than 300 CE: 17, 24, 35 (the date given a "?" by NA27), 86. More serious concern might be offered in regard to 50, which NA27 dates as IV/V. Surprisingly not included is 7 (III/IV [?] in NA27).

6. Yet, given the fact that there might be some dispute regarding palaeographical dating, one might then wonder why some other papyri which NA27 date as IV were not included in the C/B volume (e.g., 6, 8, 10, 25, 62, 71, 81, 88, 89). Certainly the volume is already quite long, and one might presume that the omission of these fragments was necessitated more by reason of space requirements; yet this is not a criterion for exclusion, but only the 300 CE date (adjusted as it might be due to different palaeographical considerations).

7. For my part, I would have preferred an edition which was not date-delimited, but which simply would have included all extant New Testament papyri, regardless of assumed date. I would urge Comfort and Barrett to continue their work in this regard by preparing a companion volume which would complete that task as well as include all remaining uncial fragments which date to the ninth century. In view of the fact that the mere age of a manuscript has no necessary relationship to the quality of the text which it contains, this is particularly the case when such later papyri and uncial fragments are used authoritatively within critical editions of the Greek New Testament. A continued two-volume compilation would indeed be helpful for New Testament textual researchers and would overcome dating disputes in the present volume caused by differing paleographical opinion.

The Organization

8. As for the organization of the volume, the concept is good, although not implemented in what might be considered an optimal manner. Each manuscript is introduced with a preface listing its Gregory-Aland number, its papyrological or catalog number, its contents, date, apparent provenance, current location, physical features, bibliographical information, and supposed textual character, followed by a longer descriptive narrative in larger type when warranted. While all these aspects reflect necessary information or at least noble interpretative aims, the reader will quickly notice that the point size of the type for extended discussions within the introductory section appears to be larger than that used for the Greek manuscript transcriptions themselves. This, along with certain aspects of the content of the prefaced material, leaves an impression that the editors' opinions regarding the manuscripts are of greater importance than the actual transcription of the manuscripts or the text which they contain. Increasing the point size of the Greek transcriptions and reducing the point size of introductory discussion would have been preferable.

9. While discussing fonts and point sizes, other desiderata come to mind: the Greek transcriptions are simply too crowded on the line. The character spacing definitely should have been increased as well as the point size (the format utilized in the Oxyrhynchus papyri volumes serves as a good model). Considering that on most pages of transcription there is wasted blank space ranging from 30-50%, one would think that at the very least the character spacing could have been increased without lengthening the book. Such an increase in character spacing would also help the reader notice the square brackets [ ] which indicate marginal or internal lacunae; as it stands, these brackets are often difficult to spot, and a casual reader might consider as text what is only reconstruction.

10. Within the discussion categories cited above, many readers will differ sharply with C/B's often dogmatic and at times erroneous opinions. It would have been far better simply to present a consensus statement covering a broader range of opinions than continually to counter those which are commonly held, particularly when new comments or opinions will not be well received or considered accurate. Although this review cannot pretend to offer a detailed examination of such matters, the following lengthy set of examples taken from the editors' introductory matter regarding 75 and 39 will illustrate the point:

Regarding Papyrus 75

11. When discussing 75, the editors provide a five-page introductory discussion, with several comments being simply egregious. On pp. 503-504, C/B take great pains to identify the scribe of 75 as a "Christian." Note the heavy comments on this point: "a professional, Christian scribe"; "the large typeface indicates that the manuscript was composed to be read aloud to a Christian congregation"; "we have a manuscript written by a Christian for other Christians"; "this work was read by Christians"; "the scribe ... may have produced it for an Alexandrian Christian community, a community that had come to expect textual fidelity"; "he is the best of all the early Christian scribes." Of course any of these comments may be correct, allowing for certain subjective value judgments, but it appears that the editors do "protest too much" on this point. While a scribe's "Christianity" may perhaps be presumed for most extant NT MSS, in most cases such probably cannot be proven, let alone the particular type of theological leaning which might exist among the various "Christian" groups--orthodox, heterodox, or heretical--and certainly not merely by viewing a biblical manuscript prepared by a scribe who may belong to any disparate group which might purport to be "Christian" in one form or another.

12. This problem is exemplified by two of C/B's further remarks:

13. "Various marginal comments in Coptic show that it was [later] read by a Coptic community of believers" (504). Here C/B either minimize or are unaware of the sharp doctrinal divisions between Coptic and Orthodox Christianity, a division so severe that an exclusionary breach resulted from a very early period. Merely labeling the scribe or his/her community "Christian" leaves a highly erroneous impression in this case.
14. "The scribe's Christianity shows in his abbreviations of the nomina sacra, as well as in his abbreviation of the word 'cross' (stauroj). These are telltale signs of a scribe who belonged to the Christian community" (503). I must presume that C/B are unaware of the fact that the nomina sacra are used regularly within the Nag Hammadi literature. Even in the Gospel of Thomas, one finds not only the nomina sacra being utilized (e.g., IS³ for ihsouj throughout), but also the identical abbreviation for stauroj as is found in 75 (Thomas, logion 55). The alternative would be that C/B perhaps consider the Nag Hammadi Gnostic groups "Christian"; but I should assume this as unlikely in the absence of statements to this effect.

15. On the other hand, at times C/B are more precise than other sources, but with an intermingling of subjective opinion. For example, in regard to 75 in John, NA27 lists the latter portion of the contents as Jn 12:3-13:10; 14:8-15:10, whereas C/B state specifically: Jn 12:3-13:1, 8-9; 14:8-29; 15:7-8. Yet immediately following such precision, a subjective value judgment is added regarding a particular reading (a closely-parallel statement appears in their description of 66): "The manuscript does not include the pericope of the adulteress (7:53-8:11), making it the second-earliest witness (next to 66) not to include this spurious passage."

16. Here, the clearly factual statement could have ended after "adulteress." Given the wide range of difference regarding the relative dating of 66 and 75 (reflected even in C/B's discussions: 66 could date from "the first half of the second century" [Hunger] to "the first half of the third century" [Turner]; 75 could date from possibly "A. D. 145-46" through "the early third century" [Martin]), it seems gratuitous to declare absolutely that 75 is necessarily "the second-earliest witness" to any reading in comparison to 66, particularly when the thrust of the present edition is to present the text and transcription of the documents as primary data, and not to quibble over relative dating.

17. The same misguided dogmatism can be noted for the last part of the quoted comment. Regardless of whether the pericope adulterae is spurious or not, the point is certainly not established merely by presenting the transcribed text of the two earliest witnesses. At best one can only say that they happen not to include that passage, and that such may point toward non-authenticity, although one might equally posit deliberate omission of a controversial passage (cf. Augustine) or lection-related major feast-day usage as alternative solutions.

18. Further theoretical biases can be seen in other comments. For example, most modern eclectic critics will reject as faulty and overly simplistic the editors' claim that "none of the early papyri are Byzantine because they antedate the Byzantine period" (27), since this appears to beg the question. C/B also virtually state that none of the early papyri are Alexandrian for the same reason (28), while at the same time they minimize any real existence, influence, or impact from the "western" text on the early Greek papyrus documents (27).

19. The matter of texttype development becomes a key issue in C/B's statements, but a detailed theoretical history from their standpoint is wanting, even though such should be requisite whenever radically new views are being presented. Only if one accepts their apparent view that the earliest witnesses are by reason of their date the most authoritative and least corrupt--regarding even fourth century uncials such as and B as relatively more "corrupt" than those early papyri-- can certain peculiarities in their statements be understood. Thus C/B can state contra Colwell and (I suppose) contra mundi that "the proto-Alexandrian manuscripts [the early papyri] are usually purer than the later ones" (28).

20. Regarding manuscripts like and B, C/B claim that these were "produced by scribes trained in Alexandrian scribal practices, the best of its kind in Greco-Roman times. Such scribes were schooled in producing well-crafted, accurate copies" (28). I suspect that, at least in the case of and perhaps B as well, some will disagree strongly with such an assessment. C/B further claim that the implementation of those Alexandrian scribal practices in manuscripts such as and B actually corrupted the "purer" proto-Alexandrian text found in the early papyri, since "these manuscripts [the proto-Alexandrian papyri] display the work of scribes who had the least creative interaction with the text--they stayed with their task of making good copies" (28).

21. Another problem related to these views is that at certain points in their introductory material C/B appear determined to produce a text-critical commentary, which is out of place in a volume of this nature. This is exemplified in the case of 75 by a discussion of various readings intended to prove that the scribe was not only a Christian, but was familiar with "the other Gospels, especially Matthew" (504). I will spare the reader the discussion of what are claimed to be "three subtle harmonizations" which supposedly prove this point (Lk 8:21 to Mt 12:46-50; Lk 10:24 to Mt 13:17; Jn 6:5 to Mt 14:15 and/or Mk 6:36), but which are never demonstrated by the editors and which in any case are subject to question. Suffice it to say that this aspect of the discussion is highly speculative, often elevating apparent unconscious alterations or scribal errors into matters of highly sophisticated and finely drawn recensional activity. Most scholars would consider such to lie beyond the normal activity or capacity of an individual scribe. I suspect that the "subtle harmonizations" exist more in the imagination of the editors than ever in the mind of the scribe of 75.

22. One will further question whether anything found in 75 Lk 11:31 will reveal "the scribe's national prejudices"; nor will one necessarily see how the reading there--basilis[sa] / sa egerqhsetai ... oti hlqen ek twn peratwn thj ghj with the note that the redundant sa was erased and replaced with notou (NA27) by the corrector--"allows for the possibility that the queen who visited Solomon came from Egypt, not Sheba" (504).

23. Much the same can be said regarding other almost fantastic claims regarding the scribe's presumed ability or intent. C/B cite a variety of readings, all presuming deliberate scribal alteration with the intent of "improving" the text. The average textual critic will in most cases fail to notice such subtleties and instead presume a more "normal" transcriptional error or unconscious alteration to have been the cause of such alterations. Cases in point include the following:

24. Lk 16:30, where NA27/Byz differs from 75 only in reading poreuqh where 75 reads egerqh; C/B claim that this subtle alteration demonstrates his "Christian prejudices," since in this variant "he chose Christian terminology over Hellenistic in his description of how one goes to the living from the dead" (505). The alternative (an unconscious alteration to egeirw as a common synonymous term for rising from the dead) is not given as an option, nor is any evidence provided to indicate that the use of poreuomai in such a context would have given this or any other scribe a problem due to supposed "Hellenistic terminology." In fact, had this been a concern, one would expect to see a large number of scribes opting for the "smoother," "easier," and "doctrinally preferable" egerqh instead of merely 75; since this did not happen, the presumption of C/B appears to be faulty. This can be more readily seen by considering the nearby internal parallel in 16:31 (which C/B do not mention): there the NA27/Byz wording is the clearly "Christian" form anasth, while 75 (also 579 pc) once more reads egerqh, this time for no necessary reason. Note that in 16:30, while 75 reads egerqh and NA27 reads poreuqh, 579 pc read anasth; all this points to a simple substitution of synonyms, which is a common unconscious scribal error that suggests no necessarily intentional alteration.
25. In Jn 4:41, according to C/B, the scribe of 75 presumably "exposes his male, Judeo-Christian prejudices ... by writing a pronoun that indicates that only the males in Samaria believed in Jesus" (504). Beyond the fact that neither C/B nor anyone else can tell that the scribe of 75 was necessarily male, let alone had gender-related prejudices which spilled over into deliberate textual alteration, the change in question (75 along with it-e and it-r1 reading pleion instead of pleiouj) appears to be merely an unconscious harmonization to pollw (erroneously perceived as a neuter) immediately preceding. C/B also fail to consider the cultural question regarding the validity of a woman's testimony in that era and the perceived necessity for male confirmation; note particularly in this regard the wording of Jn 4:28, "the woman left her waterpot and went away into the city, and says to the men (kai legei toij anqrwpoij).
26. In Lk 16:19, the "temptation" to give a name to the rich man, which the scribe of 75 "could not resist," resulted in Neuhj, which C/B suggest "could mean 'foolish,'" but without providing any evidence (505). While most textual critics would suggest a 75 corruption by homoioteleuton from "Nineveh" (Nineuhj) as found in Sahidic sources, C/B seem intent to demonstrate that the scribe of 75 was consistently an "original thinker."
27. In Jn 7:34 C/B claim that "the scribe of 75 changed the verb 'will seek' [zhthsete] to an imperative [z)hthsat(e] because he had a problem with Jesus predicting that the Jews would seek him, when in fact they didn't" (505). Once more, C/B seek deliberate alteration by an "editorial" scribe who is determined constantly to improve the text of his exemplar when in fact accidental error appears far more likely. This change could have been influenced by simple attraction to the alphas in pemyanta two words earlier, if not simply a pure error in writing a for e.

28. The determination of C/B to view the scribe of 75 as a theologically precise innovator who apparently makes no sensible alteration by error or chance seems to run quite counter to the findings of Colwell and Royse. There is an agenda involved here, apparently to demonstrate by the scribe's supposed acumen that he/she is "the best of all early Christian scribes," who, "when he did deviate from his exemplar, he did not go in the direction of simplifying the text ... rather, he elevated it" (504). One can only wonder at C/B's claims of innovation by the scribe of 75 when its reading clearly appears to be due to accidental causes:

29. Example 1: In Lk 9:34, C/B claim that "the scribe followed the logical implications of the text, making a lexical adjustment to indicate that it was the three disciples who entered into the cloud" (505). Yet when one compares the text of 75 (eiselqein) with NA27 (eiselqein autouj, B L 1241 [~ C 2542] pc) and with the majority reading (ekeinouj eiselqein, 45 A D W Q Y f1 f13 33 sy-h sa ), not only is the supposed fine point of "lexical adjustment" unapparent, but the singular reading of 75 can be seen to have more likely derived by homoioteleuton from the clearly early (due to 45) majority reading by a skip forward from e of ekeinouj to e of eiselqein.

30. Example 2: In Lk 14:8, C/B assert that "the scribe deleted any mention of a wedding festivity because the gathering that occasioned Jesus' parable ... was not a wedding celebration" (505). Certainly the text in question reads in NA27 tinoj eij gamouj mh, while 75 reads only tinoj mh, omitting eij gamouj. Yet while C/B continually prefer to see the scribe as a creative editor, once more most textual critics would likely see another blunder caused by homoioteleuton, the scribe simply having skipped from j in tinoj to j in gamouj, thus omitting the mention of the wedding purely by accident (so also it-b sa).

31. Example 3: In Lk 24:27 C/B suggest that the scribe of 75 "attempted to adjust the text for the sake of clarity ... to make it clear that Luke meant Jesus used all the Scriptures that would aid in presenting the messianic prophetic picture, not that all Scripture is messianic" (505). To this one should compare the actual readings involved. While NA27 reads diermhneusen autoij en pasaij taij grafaij ta peri eautou, 75 transposes to diermhneusen autoij ta peri eautou en pasaij taij grafaij. Rather than being a singular reading demonstrating some sort of subtle creativity (which itself does not really support the claims of C/B made above), it would be far more likely once more to see an accidental omission from the NA27 base text by homoioteleuton, skipping from -ij in autoij to -ij in grafaij, which omission was immediately caught by the scribe and corrected by adding the omitted portion to the end of the phrase (a quite common phenomenon as demonstrated by many otherwise singular transpositions in various manuscripts).

32. All in all, C/B appear to be intent on elevating the scribe of 75 into an editor when in most such cases the scribe simply blunders or makes an alteration apart from editorial intent.

Regarding Papyrus 39

33. In contrast to 75, C/B's discussion concerning 39 is minimal. Nevertheless, similar problems exist in relation to one claim made by the editors regarding that document. After the quite correct statement, "Grenfell and Hunt said 39 generally agrees with B" (which is true, given the existence of at least 3 significant sensible variants where 39 and B concur in the extant portion of this short nine-verse fragment), C/B then radically enhance the statement by continuing: "In fact it agrees verbatim with B and nearly so with 75" (147). But this latter claim goes too far, especially considering the fragmentary nature of 39, which is only a portion of a single leaf, half the text in the extant columns missing, containing on both sides only Jn 8:14-22. Aside from the fact that the sample is not extensive enough to provide a definitive textual characterization, its "verbatim" agreement with B exists only if one accepts C/B's reconstruction of the missing portions--and this raises a further point which affects a good portion of the book.

Excursus: Reconstruction of non-extant portions of manuscripts

34. The reader of C/B is warned in the Introduction (19) that:

35. We have attempted to reconstruct the beginning and ending of several manuscripts, wherever we could determine original margins. These reconstructions, indicated by opening and closing square brackets, are conjectural. Bracketed portions within the transcriptions represent letters or words most likely to have been in the original manuscript. The supplied letters and words often, but not always, accord with the text printed in the twenty-seventh edition of Nestle-Aland's Novum Testamentum Graece [emphasis added].

36. Certainly it is valid to attempt the reconstruction of missing portions of manuscripts where lacunae of various sorts exist. In places where no known variant readings exist and where the number of missing letters can be accurately calculated, there is no serious question regarding the text to be supplied for the missing portions. Where the number of missing letters cannot be calculated with precision or where known variant readings exist which could fit into the same amount of space-- particularly sensible variant readings--all reconstructed text becomes less certain. Strongly definitive conclusions thus should not be drawn when the missing portions of text could seriously affect such evaluations. This should particularly be noted whenever C/B choose to reconstruct lengthy portions otherwise missing from various manuscripts.

37. For example, in 46 2Cor, most leaves lack the bottom two or three lines; in each case C/B reconstruct that missing portion, generally by supplying the NA27 reading as the basis for reconstruction. Little or no serious consideration appears to be given to the possibility of variant readings in such reconstruction.

38. A case in point would be the reconstructed four-line portion at the end of 5, leaf 1 recto (Jn 1:40b-41): the reconstruction assumes prwton in Jn 1:41, following NA27 (with 66 75 2 A B Q Y 083 f1 f13 892 al lat sy-p sy-h Epiph), although the alternate prwtoj ( * L Ws ) would be equally possible, given that 5 shows "agreement ... with against B ... in critical passages" (73). In the same verse, one can only wonder what critical insight (beyond wanting to end the reconstructed leaf with a complete word) led the editors to reconstruct ermhneuomenon instead of the NA27/Byz meqermhneuomenon, unless a highly gratuitous assumption was made that 5 would erroneously duplicate the ermhneuomenon which appeared in Jn 1:38 preceding.

39. There are other cases, particularly at the end of certain fragments, where C/B insert a reconstruction of several missing lines, even though no one could ever tell what might have been read in missing portions which had no further continuation. 86 stands as a case in point: while it may be granted from comparison of the recto with the verso that the manuscript apparently contained 33 lines per page with a common line length and even allowing that C/B's reconstruction of the final 23 (!) lines on the recto might be approximately correct, since the reconstructed text lines up with the continuation on the verso there is absolutely no way to assume anything regarding the reconstructed final 22 lines on the verso, since this portion of text has no further link. Reconstruction in this and numerous other cases in C/B is an exercise in futility and offers nothing useful or substantial to the reader. Gratuitous reconstruction only takes up space which could have been used more profitably for the inclusion of additional textual data.

40. Again, 29 contains only Ac 26:7-8, 20, yet C/B provide a wholly gratuitous reconstruction of two full leaves in the non-extant portions (Ac 25:26 through 26:7a and 26:8b-20a). The line length in the reconstructed portions also differs dramatically, averaging 31-32 lines on the verso, but only 23-24 lines on the recto.

41. Even the extremely small fragment 52, comprising parts of only five verses spread over a mere seven lines on the recto and verso (Jn 18:31-33, 37-38) is expanded in reconstruction into two full pages (recto and verso) of 18 lines each, the latter of which extends from the true "end" of the manuscript in 18:38 through 19:3a! It would have been better simply to end the transcription of the manuscript wherever the continuous text ends and allow the evidence to stand where it at least remains substantial⁴.

Resumption of Papyrus 39 discussion

42. Once reconstruction has been made, the "conjectural" additions in C/B appear to assume the role of factual data. This is exemplified by the dogmatic claim that 39 is not merely in "general" agreement with Codex Vaticanus (Grenfell and Hunt), but is in "verbatim" agreement. While one might be able to make such a claim as regards the extant portion of 39, the NA27 supplied text in the reconstructed portions-- even when agreeing with B--does not prove identity with Vaticanus at such points. One can cite several instances which raise questions regarding C/B's dogmatism on this point:

43. In Jn 8:15, verso, lines 11-12, C/B offer the reconstruction sarka kr]inete egw / ou krin]w oudena, in this form agreeing with Vaticanus. NA27, however, note that 75 pc in this place happen to read sarka krinete egw de ou krinw oudena. 75 and B here differ, and the reading of 39 remains uncertain. The presence of de easily remains possible within the missing portion of 39: the three lines preceding have 14-14-15 letters, the line in question has 13, and the three lines following have 13-14-13 letters according to C/B's reconstruction. The 13-letter line without de in C/B could just as easily have been a 15-letter line if de had been present. If so, then 39 would no longer be "verbatim" with Vaticanus. That the presence or absence of de remains uncertain should have prevented any absolute claim regarding identity of text with Vaticanus.
44. The same situation occurs in Jn 8:16, verso, line 15, where C/B's reconstruction has 39 reading alhqinh e]stin o (continued on next line as oti). While this agrees with Vaticanus as well as with NA27 (75 B D L T W 33 892 1241 L-844 L-2211 pc) , one must consider the possibility of the alternate reading alhqhj estin (66 Q Y 0250 f1 f13 ), which fits within the same space and equally satisfies the line length pattern (the three lines preceding have 13-13-14, the line in question in C/B's reconstruction has 13, the three lines following have 12-12-11; the change from alhqinh to alhqhj would reduce the line in question to 12 letters). Once more, a variant reading exists with early support, and such could equally have stood within the reconstructed portion. This leaves C/B's claim of "verbatim" identity with Vaticanus even further in want.
45. In Jn 8:19, recto, lines 5-6, C/B's transcription of 39 reads PHR so[u apekriqh / IHS ou[te eme oida (supported by 66 75 B D L T Y 070 0250 f1 ). Yet also possible is the variant reading apekriqh o IHS (supported by N W Q f13 33 1241 al), which, if applied to the second line of C/B's reconstruction, would increase its length from 14 to 15, with the three lines preceding having 13-12-13 and the three lines following having 14-15-14 letters. Once more, the reconstructed portion of C/B remains uncertain regarding the presence or absence of the article, and a dogmatic assertion regarding verbatim identity with Vaticanus cannot be claimed.

46. This problem is further exacerbated when even the extant portions of 39 are examined. Not only in places where reconstructed text is in view does 39 have possible differences from Vaticanus, but in the existing portions several actual differences occur. Consider the following:

47. In Jn 8:14, 39 verso, line 9: the C/B reconstruction reads ercomai] h pou upa (the last continued on the next line to form upagw). A dot is placed under the h to indicate its fragmentary or uncertain nature. If 39 in fact reads h, then at this point it does agree with Vaticanus. The NA27 apparatus, however, makes no claim regarding the reading of 39 but recognizes two competing readings: h is supported by 66 75c B D K N T Y 070 0250 f1 700vid L-844 L-2211, while the alternative kai is read by 75* L W Q f13 . It seems disingenuous to differ silently from the non-mention of 39 in an established critical apparatus, particularly when the claim of "verbatim" agreement with B is being touted for that particular manuscript.
48. Jn 8:15, 39 verso, line 10, in C/B's reconstruction reads gw umeij] kata th(n), the -gw being the end of upagw from 8:14 in the line preceding, and the (n) used (by me) to indicate the presence of the nu-suspension bar in the C/B text. While most manuscripts indeed read as C/B indicate for 39, Codex Vaticanus at the same point reads umeij kata ta th(n). Although the B reading probably resulted from dittography, it nevertheless remains a distinct "sensible" reading. In any case, it is clear that at this point 39 does not agree "verbatim" with B.
49. Jn 8:15, 39 verso, line 11: C/B's reconstruction reads sarka kr]inete egw, whereas B reads the itacistic kreinete. While this is not a meaningful variant, it still is a case of 39 failing to agree "verbatim" with B. The same can be said for 8:15-16, verso, lines 12-13, where the two occurrences of krinw in 39 (krin]w in the first occurrence, k]rinw in the second) differ from Vaticanus' itacistic kreinw. Also not precisely identical are cases where 39 uses the nomina sacra PHR and PRA while Vaticanus consistently writes pathr and patera in pleno. There is also a minor discrepancy where the nu-suspension is present in Vaticanus but not present in 39, although this lies in the reconstructed portion of 39.

50. The lesson to be learned (and the warning to be issued) is that the reader of C/B should accept nothing in the introductory comments as automatically correct, but should constantly attempt to verify all claims made by the editors, and to do this either from personal examination of the manuscripts or their photographs, or by comparison with the editio princeps or other sources.

The Text and its Transcription

51. The most important element in a volume which intends to present "the text of the earliest Greek New Testament manuscripts" is of course the text itself, its manner of transcription and presentation, and its reconstruction in places where lacunae exist. All the problems related to editorial opinion and factual error in the introductory material can be overlooked if the primary intent of the reader is to obtain an accurate representation of the actual text contained in those early manuscripts. Of course, one simultaneously has to expect a high degree of accuracy in the representation of that text or else the primary value of the book is immediately lost, even if every detail of introductory material were totally accurate.

52. As noted at the beginning of this review, the initial Baker release of the first version of C/B was viewed by various critics as seriously flawed due to inaccuracy in its representation of the text of the various manuscripts. At that time, only bare mention was made of certain inaccuracies in the introductory material (it appears that most of the introductory material remains unchanged in the current edition). It thus is not surprising that such introductory inaccuracies as noted above remain in the present volume. On the other hand, all the textual errors which had been previously noted appear to have been repaired (cases where differences of opinion existed in regard to unclear letters were not necessarily altered by the editors, however).

53. The question now is whether serious textual errors remain in the transcriptions which were not noted previously by various critics. For this reason, I previously had suggested that the entire set of transcriptions should be reexamined, with every transcriptional detail verified--not only by the editors but by outside readers. The editors chose not to implement such a process, but instead repaired at least the known damage. One can only hope that they did in fact reexamine all transcriptions in order to present the text of these early manuscripts in the most accurate manner possible. Within the time-delimited context for this review, the best anyone can hope to do is to examine sample passages within several manuscripts and compare these both against the photographs of the original and the transcriptions given by the editio princeps, comparing the result against the C/B edition. The manuscripts and portions here utilized as samples include 45 (in part of Luke, John, and Acts) and 46 (in Galatians), along with smaller segments scattered among various manuscripts.

Definite errors

54. One thing can be said from the start: the number of plain and clear errors is thankfully few in comparison with what had appeared in the original edition. The multitude of errors previously noted have all been corrected, and it appears that the remaining text has at least been reread in an attempt to avoid the severe criticisms which plagued the earlier edition.

55. This does not mean that the present edition is wholly free from error, however. For example, in 45 at Jn 10:40 (leaf 15 recto, line 14), the word prwteron is an error for proteron, confirmed by examination of the photograph. The correct form already stood as such in the editio princeps, so the fault here clearly lies with C/B.

56. In 45 Lk 14:24, C/B state that leaf 15 recto, line 14, ends with keklhmenwn. Kenyon's original edition stated that the ending was keklhmenw(n), with the final nu indicated by a suspension bar. Examination of the photograph shows Kenyon to be correct and C/B in error. Note that this is not merely a matter of C/B writing in pleno what appears as a suspension abbreviation in the manuscript: their regular practice is always to show suspended nu by the normal bar symbol, as witnessed even on the same page of 45, six lines preceding (where eipe(n) has the suspension). There appear to be a number of cases where a suspended nu is not indicated in C/B; one can only attribute this to error on the part of the editors, but an error which thankfully does not alter the meaning of the text. Nevertheless, such an error may at times obscure the proximate cause for certain types of variant readings (particularly omission by homoioteleuton) when such precedes or follows a suspended versus a normal nu.

Correction of previous errors

57. On the other hand, one must equally acknowledge that C/B at times have corrected errors which had appeared in the editio princeps, and thus their text represents an improvement. For example, in 45 Lk 14:31, leaf 15 recto, line 6 from bottom, Kenyon had transcribed ekte]leisai, while C/B give the correct ekte]lesai. Also, in 45 Jn 11:48, leaf 17 verso, line 11, C/B correct Kenyon's erroneous airousin to arousin. These corrections happen to be mentioned in their margin, although not all differences between the editio princeps and the C/B transcriptions are so noted; nor are these always noted even when an error of the original edition is involved (this partial inclusion of differences between C/B and the editio princeps is a totally new feature which was not present in C/B's first edition).

58. Also in 45 Jn 10:36, Kenyon had reconstructed as "vid." o U[i/oj QU] eimi, even while citing the extant u as having a nomen sacrum bar over it (the usual practice of the scribe of 45 is to abbreviate uioj); C/B correctly interpret the reading as o U[S tou QU, (which also happens to be the majority reading, in contrast to omission of the article in D W).

Underdots and brackets

59. The use of brackets to indicate the boundaries of reconstructed text (as opposed to what is extant in a document) reflects a standard practice and was already implemented in the original edition of C/B. The current edition has responded to certain additional criticism of the original volume by adding sublinear dots under letters which are considered to be uncertain. The inclusion of this necessary standard practice thus reflects a welcome improvement and allows the extant text to be defined more precisely⁵.

60. The problem facing the more expert readers will still be that which had existed in the previous edition: C/B do not merely reflect a general consensus regarding which letters are extant or those which are borderline, but on the basis of their own examinations of the manuscripts have determined for themselves where the various brackets or underdots should go. In many cases, no other scholar has agreed with their estimate. While C/B's placement of dots and brackets certainly is correct in many cases, in far too many other places a reader who consults the original (or photographs thereof) seriously could question the accuracy of such placement of dots or brackets, particularly in view of the purported precision of textual representation in both extant and reconstructed portions.

61. For example, 46 Gal 1:6 according to Kenyon is reconstructed as oti outw t[axewj, with a dot under the t. C/B in contrast read oti outwj tax[ewj, with dots under tax, but not under the j. While the text as a whole is not affected, it is obvious that the transcription and reconstruction differ dramatically according to editorial opinion. Based upon the photographs, I cannot say that the final sigma of outwj is definitely present, let alone fragments of the letters tax following; yet C/B place no underdot in regard to the sigma when such apparently should have been done.

62. Similarly, in 46 Gal 2:19, leaf 82 verso, line 10 from bottom, the line in C/B begins nomou nomw, whereas Kenyon's transcription is n]omou nomw. The initial n of nomou is clearly fragmentary--only the barest mark survives--and even by C/B's own standards it should have been marked by at least an underdot. I can only presume this was an oversight (error?) on the part of C/B, since nothing approaching even a clearly recognizable portion of the full letter is visible. In many similar cases, C/B place no underdot in regard to fragmentary letters; I can only presume that when the reading is "morally certain," any fragment of a letter will count as complete and certain.

63. This brings up another matter of contention, reflecting my own perspective. C/B appear at times inclined to mark with an underdot items ranging from the barest fragment of a letter to a nearly complete letter. I would prefer to reserve the underdot for letters which are partial but still recognizable. Non-recognizable letters represented only by ink fragments should remain in the bracketed portion outside of the extant text. An underdot can be used there to show that "something" is visible, even though one cannot tell for certain what it might be. Although such a practice might not follow contemporary convention, it would allow a more accurate manner of presentation for transcriptional data⁶. As matters stand, I fear that those who read C/B will think the clearly-readable extant text to be more extensive than it actually is, and that C/B are far more inclusive than most critics regarding fragmentary points of ink which are not otherwise recognizable as letters. This of course explains why C/B's placement of brackets and underdots often differs from the editio princeps as well as from other publications. No notes are provided to show where such differences happen to occur (if notes were provided, they would run into the thousands).

Minor matters

64. There is a typo on p. 28, lines 4-3 from end: "Caeserean" (probably not caught due to its forced hyphenation). For the rest, the English text appears to have been well proofed throughout.

65. The monochrome photographs of the various manuscripts were far too dark in the original edition; the current edition has improved them, but more improvement still needs to be made.

66. Regarding a few matters of interpretation which are not solely tied to C/B, but which are reflected in other sources:

67. 9, 1Jn 4:16, verso, line 7 of transcription, is given both by the original editors (Grenfell and Hunt) and C/B as reading the nonsensical nomen sacrum o XQS, with a note stating "the scribe appears to have mistakenly written X before QS" (81). Yet the photo of that leaf on the facing page (80) seems clearly to read o XRS, which reflects a known abbreviation for the sensible reading o xristoj. While such may later have been altered with an intent to change it to o QS, this is not clearly evident from the monochrome photograph; yet if so, there may also have been an attempt to alter the bar above the original X into something representing a supralinear dot. Such certainly seems far more plausible than either Grenfell/Hunt's or C/B's supposition that the scribe simply wrote utter nonsense in nomen sacrum format.

68. 5, John 1:34, leaf 1 recto, line 7: C/B's reconstruction of outoj estin o eklekto]j tou QU is apparently touted as one of the "critical passages" where "the agreement of 5 with against B is evident" (73). Such reconstruction, however, presumes a line length of 29: the full line reads ti outoj estin o eklekto]j tou QU th e, the ti concluding oti from the previous line and the final e (with underdot) being the first part of epaurion continued on the following line. Yet the line lengths preceding and following on the same recto side have respectively 24-25-26-25-28-25 and 27-23-28-29-27-27-27-31. Thus, while the reading eklektoj is possible, it is not at all certain, especially when the alternative uioj (assumed to be written in pleno, as found among various early papyri) would reduce the 29 letter length to a more nearly "average" 25 letters.

Conclusion

69. Buy this book!

70. After reading the preceding litany of problems and errors, a prospective purchaser might consider the problems cited to be so insurmountable that nothing useful remains within its pages; but this is not the case. Anyone can ignore various theories, idiosyncratic views, and even outrageous claims made by the editors in their introductory material (being warned beforehand should suffice), so long as the basic extant text of the included manuscripts remains quite accurate.

71. Certainly, at least one substantial error in the textual transcription has been noted, as well as at least one case of a nu-suspension being rendered inaccurately as a normal nu. No doubt other problems remain which have not yet been discovered. All of C/B's readings, underdots, and reconstructions should still be verified by independent scholarly review, and some form of errata list as well as suggested corrections should be compiled to indicate where other scholars might differ from C/B (an internet file such as that regarding errors in Swanson's apparatus would be optimal for such a purpose). Yet as long as the reader is careful not to accept C/B's reconstructed text or underdotted letters as possessing the same weight of authority as the unquestioned extant text of the manuscripts involved, the base presentation of these data can indeed be used with profit by text-critical students and scholars alike. One might prefer something substantially more than this, but quite frankly, this is all that is available for most interested readers.

72. The book unquestionably covers a wide range of material and remains the best option for the average student or scholar as opposed to purchasing a number of expensive volumes, photocopying portions thereof, or seeking out from widely scattered sources the original editions of such early manuscripts. The C/B manner of presentation allows the reader to see the line-by-line arrangement of the various manuscripts, their degree of scribal accuracy, their scribal errors and attempted corrections, their use of nomina sacra in context, and even peripheral material such as page numbers, stichoi totals, and other items which are not always presented in other sources. Such a collection of data as is found in C/B is especially not available anywhere else at such a low price (Münster and IGNTP please take note!).

73. For practical purposes, I indeed find the C/B material to be a reasonable alternative for most potential purchasers. I do wish that the problematic areas noted in this review could have been treated in a different manner, particularly in such a way as to maximize the usefulness of the extant text of the manuscripts and minimize the role of the editors' theories, conjectures, and opinions. I still think the work needs a thorough verification by outside readers in order for even the extant text to be considered wholly trustworthy, even though it does appear that nearly the whole of that portion is now reliable. Much more work is required in regard to brackets and underdots before any common agreement can be reached, and I would urge Comfort and Barrett to consider seriously differing opinions on those matters.

74. Along similar lines, I would strongly urge Comfort and Barrett to consider the preparation of a companion volume--not limited merely to the period preceding 300 CE--which would include all remaining New Testament papyri, as well as (at least) the smaller uncial fragments up through the ninth century. Were such a companion volume also to include the Talismen and Ostraca previously catalogued by Gregory, this too would be a welcome bonus. It would also be helpful if an indication were given (no detailed apparatus required) of those places where a manuscript might deviate in sensible readings from both NA27 and or happen to agree with D (i. e., a lesser form of what appears in the Chester Beatty and Oxyrhynchus volumes).

75. If I might be so bold as to state my opinion regarding "what the text-critical community really needs": it is not more critical texts with limited apparatuses from which one cannot reconstruct the running text of the original manuscripts. We already have enough of those. Rather, the pressing need is for the raw data of manuscripts, either in publicly-accessible photofacsimiles or preferably in skilled transcriptions following a line-by-line format such as Tischendorf had initiated in his Monumenta Sacra Inedita volume and C/B have presented in their edition. Even complete collation data for specific manuscripts would be a preferable alternative. If text-critical researchers can be provided the manuscripts themselves in at least transcription or detailed collation and if this can be done in a low-cost format for full access and evaluation, there will be greater opportunity for text-critical progress in the future.

76. So my advice to the reader of this review is simple: buy Comfort and Barrett. At least in this volume everyone can have access to a "reasonably accurate" presentation of the basic data, most of which is not readily available elsewhere. Even while allowing that certain aspects of presentation, introduction and format could stand a good deal of improvement, the volume remains unique and should become part of the repertoire of all text-critical students and scholars.

Endnotes

¹ In light of the virtual non-release of the initial volume due to its withdrawal by Baker, as well as the subsequent improvement and expansion in the new volume (which itself bears a different title, dropping the word "complete"), it seems a misnomer when Comfort and Barrett refer to this new work as a "second edition" rather than what amounts to a new work being released for the first time.

² The number of papyri has been expanded from the original volume by adding the latest Oxyrhynchus finds through 115.

³ In this review, nomina sacra are indicated by capital letters rather than overbars.

⁴ In many cases, C/B chose not to reconstruct missing text, even when such indeed may have been possible (e.g. 45 in Mt 20-21, Mk 4-6, 7-9, Lk 6-7, Jn 4-5, Ac 4-7, etc.; cf.also 22, 24, etc.). Yet, given their propensity for "extra" reconstruction where totally unnecessary (as noted in the main text), one seriously must wonder why more "legitimate" reconstruction was not attempted in those cases where such would have offered more profit to the reader.

⁵ According to C/B's Preface (13), "typesetting constraints stipulated by the first publisher kept us from doing this" [i.e. inserting underdots]. If this is indeed the case, then Baker Book House clearly should be criticized for not permitting a proper scholarly format for the presentation of what had to be considered a highly technical work.

⁶ Editor's note: The "Note on the method of publication and abbreviations" found in recent volumes of The Oxyrhynchus Papyri (London: Egypt Exploration Society) provides a succinct summary of the Leiden editorial conventions used for transcribing papyri. A sublinear dot indicates that the editor feels real doubt concerning the identity of a letter so marked. Unfortunately, the degree and nature of doubt that warrants a dot varies from editor to editor (see Herbert C. Youtie, "Text and context in transcribing papyri", Greek, Roman, and Byzantine studies 7 (1966), 251-8). The following, however, is beyond doubt: if a letter is so fragmentary as to be altogether unidentifiable then it should be represented by a bare dot. Square brackets are reserved for letters of which no trace remains; a letter of which any trace remains should not be placed within square brackets. For example, a].q?rwp?.[j indicates that the alpha and sigma are lost, traces of the nu and omicron remain but the letters cannot be identified from their remnants, the theta and pi are tentatively identified, and that the rho and omega are beyond doubt.

Maurice A. Robinson
Professor of Greek and New Testament
Southeastern Baptist Theological Seminary
Wake Forest, North Carolina, USA.