50, dated in the Liste to IV/V, has been given various dates. Comfort follows the editors of the Yale papyri in preferring a date of c. 300 (Oates et al. 1967: 15-21).
Philip W. Comfort and David P. Barrett, eds. The Complete Text of the Earliest New Testament Manuscripts. Grand Rapids: Baker, 1999. Pp. 652 + 41 plates. ISBN 0-8010-2136-7. US $49.99.
1. This book is a presentation of transcriptions of all those mss, papyrus and parchment, which the editors deem to have been produced before the fourth century. It generated a certain amount of interest in the files of the TC-list. In order to avoid any prejudice, I avoided studying the correspondence until I had completed my own survey. I turned to it after my own examination, and shall deal with the principal issues at the end of this review. Sundry footnotes dealing with points from the list were added later.
2. The contents of the book include a list of the manuscripts in their canonical order, so that one may see which are extant in any given portion; an introduction with bibliography; transcriptions of 55 mss, each with an introduction; plates of 41 of them; and an index of names. While we are on the subject of names, I shall refer to this book by the first editor alone since, in spite of the two names on the title page, the Acknowledgements and the introductory material to each ms are generally written in the first person singular (e.g., 'My studies' [p. 494], 'My opinion' [p. 599]), although 'we' is also found in the first part of the introduction.
3. The volume describes itself as providing the 'complete text of the earliest New Testament manuscripts'. The first question is therefore one of selection. The mss included are
7, dated to III/IV in the Münster Liste (Aland 1994), is excluded, while 13, 18, 37, 72, 78 and 92, which are given the same dating, are included. Some of the mss that have been included are dated in the Liste to the fourth century: 17, 24, 35 (with a ?), 86. But a number of other possible fourth century candidates that have been excluded are 6, 8, 10, 62, 71, 81, 88 and 89. The inclusion of 17 and 24 seems to be on the grounds that datings by Grenfell and Hunt (Grenfell and Hunt, eds. 1898-) tend to be too late (more of this in a moment). 35 has had various dates between the third and seventh centuries given it, and here the third is preferred. There is no explanation given for the inclusion of 86. It would have been helpful to have learned, perhaps in the Introduction, why 6, etc., are not included.
4. 50, dated in the Liste to IV/V, has been given various dates. Comfort follows the editors of the Yale papyri in preferring a date of c. 300 (Oates et al. 1967: 15-21).
5. The first four of the majuscules included are datable to c. 300. The fifth was dated by Roberts to III (Roberts 1950:24-26), and more recently by K. Aland to V (Aland 1967:92). There is no tendency to push parchment mss dated IV in the Liste back to III/IV, comparable to that just noted for the papyri.
6. The precise number of separate papyri continues a matter of debate. Quite properly, Comfort treats 4/64/67 as a single entity. But not all identifications have been generally accepted. That 15 and 16 have the same scribe is a matter which has been debated. Comfort believes that they belong together. Unfortunately, he only says that 'a close study reveals their common identity' (p. 85). Without properly set out grounds for the view, one is not able to refute it properly. The plate of 16 is not terribly good, and, if the hand is the same as that of 15, then the two photographs must be to different scales, so one is not much the wiser from the information provided. This question will have to be looked at separately. There is a little more detail when we come to 49/65. I remain to be convinced, however, that these two leaves share a common scribe. Alpha is quite different in appearance and formation, and that alone should give one pause for thought. It is a pity that there is no beta in 65 for comparison with the small bowed form in 49. However, there is no room here for a proper comparison, which again will need to be undertaken elsewhere, so I content myself with declaring the matter sub iudice. The identity of 77 and 103 (P. Oxy. 64.4403), considered possible by the editors of this recent Oxyrhynchus volume (Handley et al., eds. 1997: 6), is affirmed as 'far more likely than not' (p. 599) by Comfort, who brings textual as well as palaeographical evidence into play.
7. It consists of some general comments on the papyri, an account of the editorial rules followed, a six page discourse on palaeography and the dating of mss, a bibliography of general works, a list of the nomina sacra and numerical values of letters, and a list of 'handwriting types' which is repeated from the palaeographical section. The appearance of a beginner's guide to dating mss is rather strange in what should be a technical work. It is pitched at a low level. So far as it goes, it is tolerably accurate, although it does suggest that attempts at early dating encounter 'immediate opposition because it is believed that the time lapse between the autograph and the copy is too short' (p. 17). It has also to be said that some attempts at an earlier dating owe more to apologetics than to palaeography. I make this statement generally, and not with reference to Comfort's work. In fact his rejection of the dates given to 46 by Kim (Kim 1988) and to 4/64/67 by Thiede (Thiede 1995), and his opinion of Hunger's dating of 66 at 150 (Hunger 1960) show him to be properly cautious.
8. The plates are fairly clear, though for some reason many seem rather dark. There is rarely a scale (photographic departments are not always very helpful in this regard).
9. The introduction provides brief helpful information under the headings 'Contents', 'Date', 'Provenance', 'Housing location', 'Bibliography', 'Physical features' and 'Textual character'. For most this is supplemented with a discussion, of varying length. Sometimes the palaeography is dealt with. The tendency here is for there to be a lack of detail, and sometimes bad argumentation. For the former I take the instance of 46, where in quite a full discussion we are told that the editors find similarities between it and a number of other mss, but with comments such as 'very similar', or 'many similarities' (p. 195). Without being told what these likenesses are, we have all the work to do again if we are to evaluate the decision. With regard to the argumentation, I return to the question of Grenfell and Hunt. Since Grenfell and Hunt thought that the codex form was only introduced in the fourth century, they sometimes date in that century a ms whose hand could (they observe) be dated earlier. But it does not follow that all mss which they dated to the fourth century must be older. This would only be the case if there were palaeographical grounds for an earlier dating. In such instances, one is justified in following up their suggestion and looking at a possible earlier dating, now that we know the codex form to have been adopted much earlier. Thus, on 24, we read that 'Grenfell and Hunt dated 24 to the earlier part of the fourth century. Since it was part of a codex, it could be slightly earlier' (p. 105). Some of their fourth-century datings are correct.
10. The manner and accuracy of the transcriptions is of course central to the success or failure of such a book. The manner is to provide transcriptions with restored material in square brackets. Sometimes large amounts of text which are missing are restored; for example, in the transcription of 28 the first fourteen lines of text on the recto and thirteen on the verso,1 which are missing, have been provided in full within square brackets. It is hard to see a justification for this. Although most transcriptions restore all the text missing on each line, in some only the partially extant words at either end are supplied. There is no explanation of this variation in presentation.
11. Conspicuously absent is the use of dots beneath a letter to indicate that it is uncertain, or of a dot beneath a space to indicate traces of ink. This is due, we are told, to 'publishing constraints' (p. 14). What these restraints might be, the present reviewer is at a loss to imagine. It takes no time at all to create a font with such markers, and the ability to show various degrees of uncertainty is essential if a book like this is to be helpful. As it is, we are given the impression of certainty in many places where certainty is unattainable.
12. We are given chapter and verse (the latter inserted into the text), but not line numbers. Chapter numbers are given at the beginning of a papyrus, or where a new one begins, but not at the beginning of each page.
13. How were the transcriptions produced? We are told that each
transcription accords with the editio princeps produced by those editors who first published the work. However, we often deviate from their transcription when we disagree with the editors on the reading of the text as we saw it - whether viewing the actual manuscript or the photographs [p. 14].Unfortunately, we are not told which mss the editors had actually examined, and for which their authority might thus count for more. The reference to inspection of 'several manuscripts' (p. 11) suggests that the answer is not very many. It has to be said that the first part of this description does not make perfect sense. It says 'The transcriptions follow the first editors, but frequently deviate from them'. Only one or the other half of this statement can be true.
14. The process of checking all the transcriptions would require a lot of work--quite often a visit to some pleasant but unfortunately distant library in order to peer at a fragment. Here we run up against the problem of reviewing this kind of book, where detailed knowledge of the mss is necessary if one is to offer an informed opinion. The problem is increased here, since one of the aims of reviewing in the electronic journal is to get the review out quickly. One needs to find an approach which does justice to the book while avoiding doing all the work over again. Fortunately, I am in a good position to produce a detailed analysis of at any rate a part of the materials. Having spent some years working on the papyri of John as editor of the IGNTP (Elliott and Parker, eds., 1995), I have been able to apply the detailed knowledge, frequently based on my own examination of the MS or on a detailed report of an inspection in answer to my questions, to the Johannine fragments. There are a good number for comparison: 5, 22, 28, 39, 45, 52, 66, 75, 80, 90 and 95. The IGNTP edition of the papyri contains transcriptions of 9 of these (all except 66 and 75), and collations of all 11.
15. 5 contains portions of chapters 1, 16 and 20. The IGNTP transcription was produced and corrected against the ms, which is in the British Library. There is one place where I think, on reflection, that we have all been wrong so far: on line 20 of the recto of the second fragment, 16.20, ed. pr.2 and Comfort read <grc>l[[o]]uphqh</grc>, that is <grc>lophqh</grc> corrected to <grc>luphqh</grc>; IGNTP reads <grc>louphqh</grc> without a correction. Looking again, there is evidence for the deletion of the omicron, but it also appears that the upsilon is a correction of an iota. So the correction is <grc>loip</grc>- to <grc>lup</grc>-, just as it is in line 22. Elsewhere also, Comfort always follows ed. pr. where it differs from IGNTP. This includes several tremata where there is none in the ms, and restoring <grc>petrou</grc> at 1.40 where there is too little room for it. Comfort differs from both ed. pr. and IGNTP at a further 5 places: bracketing the final letter of <grc>duo</grc> in 1.37; reading the omega of [<grc>w</grc>]<grc>ra</grc> in 1.39; reading a trema over upsilon in <grc>u?</grc>[<grc>mwn</grc> at 16.22; reading <grc>k_</grc> in [<grc>k_e_</grc> at 20.15; reading upsilon of <grc>ou</grc>]<grc>k</grc> at 20.24. Three of these are not absolutely beyond debate, but the third and fourth are definitely wrong.3
16. One finds more problems with 22. This is a rather difficult fragment of a roll. It is a ms on which my fellow-editor (W. J. Elliott) and I spent some time, and we are confident that we read everything that could be read. The text of Comfort is almost identical to that of the first editors, Grenfell and Hunt (P. Oxy. 10.1228), including a typographical error: at Fragment 1, line 1 (15.25) they and Comfort read <grc>sk</grc>]<grc>adalisqhte</grc>.4 The reconstruction should be <grc>skand</grc>]<grc>alisqhte</grc>, with delta supplied and the bracket after it. See the plate in the IGNTP volume. There are in all 19 differences between IGNTP and ed. pr. Comfort follows ed. pr. in all but two. At Fragment 2, line 6 they agree with IGNTP in placing trema rather than single dot over <grc>u</grc> in <grc>umwn</grc>. And at line 20 ed. pr. reads <grc>p_</grc>[<grc>r_s_</grc>, IGNTP <grc>p?_</grc>[<grc>r_s_</grc>, and Comfort <grc>p_r_s_</grc> [. This is probably a typo: it is quite clear from the plate that only the beginning of the superline and the left of the crosspiece of pi are visible. There is another error in the transcription of the nomen sacrum at line 3 of the first fragment (15.26), where <grc>p_s_</grc> is read instead of <grc>p_r_s_</grc>. The editors have not provided a reconstruction of the whole of the missing text (the lines were evidently rather long), but only of the partially extant words at the beginning and end of each partial line. This is justifiable, but the user deserves an explanation of what has been done.
17. Turning to 28, we find that the text is again identical with ed. pr., except that on line 21 of the verso it reads all of <grc>de</grc> at the end of the line, rather than <grc>d</grc>[<grc>e</grc>. In the other ten places where IGNTP disagrees with ed. pr., it follows the latter.
18. The next ms is 39, and here we find some places where Comfort differs from both ed. pr. and IGNTP: at line 24 of the verso, he reads <grc>martu</grc>[<grc>rwn</grc>, whereas ed. pr. and IGNTP read <grc>martur</grc>[<grc>wn</grc>. At line 3 of the recto ed. pr. and IGNTP read <grc>e</grc>[<grc>legon</grc>, whereas Comfort has [<grc>elegon</grc>. Otherwise it agrees with ed. pr. in the eight places where IGNTP differs.
19. 45 is a ms whose Johannine leaves I and my co-editor examined rather carefully over some days (that is, we worked with the ms). There are several problems with the edition of Kenyon (Kenyon 1933). One is that the punctuation includes medial and low points and the double point like our colon, all of which were represented by Kenyon as a high point. Another is that two fragments of F17 are in the wrong place, so that the plates are misleading. The first fragment was correctly placed by Kenyon in his transcription, but the second was not. Comfort follows Kenyon. This leads them astray on both sides of the leaf, at 11.32 and 54. In the recently edited fragment containing verses from chapters 4 and 5 (Skeat and McGing 1991), Comfort follows ed. pr. The same is true of F16v, except that there is a typo at line 26, where <grc>ekuklwsan</grc> should read <grc>ek</grc>[<grc>uklwsan</grc>. This is a serious mistake, since it implies that the ms contains the rest of the line (three complete words) where there is really only the beginning of the first of them. And at line 5 <grc>egw h</grc>[<grc>lqon</grc> should read <grc>egw d</grc>[<grc>e hlqon</grc> (there is a variant at stake here), and a point should be included in lines 5 and 7. The same pattern of following ed. pr., even where there are quite significant failings, is found elsewhere in the transcription. There is more visible in the first line of F16v than is given, and one letter of line 32 is also visible. Similarly, the first and last lines of both recto and verso of F17 contain legible letters which are ignored by ed. pr. and Comfort.
20. Coming to the oldest fragment, 52, it is surprising to find that the seven extant lines on each side have been supplemented with another 11 on each side (thus hypothetically filling the gap between recto and verso). This is a fruitless and potentially misleading exercise. The verso is identical with ed. pr. and IGNTP in what it transcribes, and agrees with IGNTP in assuming the omission of <grc>eis touto</grc> in line 2. But the recto (on which ed. pr. and IGNTP agree) contains two aberrations. The second omicron in <grc>i+oudaioi</grc> is not bracketed, even though there is a hole where it would be, and in the next line we find <grc>lo</grc>[<grc>gos</grc> instead of <grc>l</grc>[<grc>ogos</grc>. Comfort may have been misled by what appears to be a trace of ink on the edge of the fragment as it appears in the photograph. In fact it is not ink at all.
21. 80 is a tiny fragment, containing more hermeneia than Gospel. Note the false insertion of a space before tau in <grc>r</grc>[<grc>hma</grc>]<grc>t</grc>[<grc>a</grc> in line 2 of the verso. Otherwise the text is that of ed. pr.
22. 90 is a papyrus that is tricky in places. On the whole, the editors follow ed. pr. The only surprise is that in line 8 of the recto (18.37) they read <grc>gegennhma</grc>[<grc>i</grc>, where ed. pr. has <grc>ge?g?e?n?n?h?</grc>[<grc>mai</grc> and IGNTP reads <grc>gegenn</grc>[<grc>hmai</grc>. Apart from the obvious problem with the lack of dots, this seems overconfident in what it can restore--see the plate in IGNTP!
23. 95 is pretty straightforward to read. The main problem is what to read in the missing part of line 2 on the recto. There is the wrong amount of space for any known reading. Ed. pr. left a space. IGNTP read <grc>autw tw uiw</grc>. Comfort takes the way out of not restoring complete lines (again without explanation).
24. We come finally to the two most extensive witnesses, 66 and 75. With regard to 66, Comfort and a group of his students assisted in the IGNTP edition by providing the first check on the collation (for some reason this contribution is incorrectly described on p. 374 as 'proofreading' on a 'new transcription'). It would be quite inappropriate for this reviewer to make use of his knowledge of Comfort's contribution to a piece of teamwork, and it is hard to find a way of assessing the work which did not. So I shall concentrate on an area which was not reflected in that checking, and to which Comfort did not refer at the time, but which comes to the fore in the volume under review: his assessment of the corrections. The description of earlier work (pp. 374-376) is reasonably accurate, in that most scholars have assumed that all the corrections were the work of the first hand. However, one correction by another hand has always been known at 13.19, and IGNTP identified another at 14.22 (see page 6 of that volume). Comfort claims that both Colwell and Rhodes believed there to have been more than one hand at work (Colwell 1965, Rhodes 1967-1968). In fact, in spite of the ambiguity of Colwell's statement (Colwell 1965: 118), neither writer explicitly identifies the mass of corrections with a second hand. This, however, is what Comfort has done. He divides the corrections between c1 (the first hand), c2 (the official corrector, who was responsible for the pagination to 99 and 'made several substantive corrections'), and c3 (who did the rest of the pagination and 'made small corrections'). It is at c3 that I will look first.
25. The change in page numbers from 100 onwards looks good. The fact that the ink looks much blacker in the facsimile is not an infallible guide. But the identification with the corrector at 13.19, the larger of the two additions by this hand recognised by earlier work, is not convincing. The hand at 13.19 writes a good square hand, with alpha in three strokes, whereas in page <grc>r_a_</grc> it is in two, with a curved bow. Admittedly at page <grc>r_i_a_</grc> it looks more like the 13.19 correction.
26. Turning to the other corrections attributed by Comfort to c3, they all come after page <grc>r_</grc> (96 in ed. pr., leaf 50 verso in Comfort; there is a mistake in the pagination of the ms, hence the inconsistency). The addition of <grc>au</grc>[<grc>twn</grc>] at 15.25 does not look (so far as it is visible) different from the majority of corrections. All the others are on pages 99 and 100. That is, the IGNTP identification of the <grc>kai</grc> compendium at 14.22 is not accepted (in fact, Comfort attributes it to the first hand). In addition to the obvious 13.19 correction, he finds five more c3 corrections:
27. Turning to the c1 and c2 corrections, there are two questions. The first is whether the two can be differentiated. After that one has to compare the corrector(s) with the scribe. My first reaction was rather negative. I then tried to devise a way of running an independent test on the theory, and hit on the method of going through the corrections, trying to allocate them to c1 or c2 according to Comfort's criteria. If I could make the distinction, at least in the majority of cases, there was a possibility that there were two hands. But if I could not, then I would not be convinced. Readers of this review might want to make the same experiment for themselves. The results of the test, which I ran for the first four or so chapters, turned out to be mixed. The first quite certain identification of c2 in a larger addition was the insertion of <grc>erxomenos</grc> at 3.31. But I then found that the words added at 3.17 were also meant to be c2, and on reflection one understood why. In about two-thirds of the places, I found the corrector that Comfort did. The fact that the longer corrections are more easily recognisable as 'c2' might conceivably be explicable as a rather more mannered style in longer corrections.
28. There is, however, one further test which can be run very easily. In examining a hand, one should always compare pages that are far apart in order to see whether it changes or remains constant. For example, a scribe might start to hurry and write more artificially, or he might have begun in a script with which he was unfamiliar and as the copying progressed reverted in some features to one that he knew better. Since the facsimile of 66 is loose leafed, it is very easy to compare leaves from anywhere in the codex. If one takes pages 5 and 94 of the facsimile and places them beside each other, one is struck by several differences. The latter is written larger, with fewer lines to the page, a pronounced upward slope and an apparently more free style, with various hooks and loops that could be attributed to speed of writing. The curious thing is that in detail a number of letter forms are actually more formal. Alpha sometimes has something more like the three stroke form of a square hand than the looped bow which one finds earlier (e.g., its first occurrence in the last line of page 94). Mu is occasionally, or so at least it seems to me, with a less curved and more angled central part. One can at once see that at least most of the range of letters found in Comfort's first and second hands as illustrated on p. 377 is found in the first hand. An exception, so far as I can tell, is the vertical of rho without a slight thickening to the left at the bottom, which according to Comfort is a feature of c2. The point is that the difference of Comfort's c2 from the first hand is that it is more formal. The question is whether the letter forms are so far removed from everything else that one would have to conclude that it was the work of a different scribe. It has to be borne in mind that a competent scribe would have had a number of different styles at his disposal. Scribes did quite often use a different style for Auszeichnungschriften, and perhaps these extensive corrections should be placed in just such a category. Certainly, it is the larger additions in the margins which are written most neatly, and this could be due to a desire not only to mark them out as corrections, but also to write them as neatly and therefore unobtrusively as possible.
29. There are thus a number of remaining questions with this proposed division of the corrections, and room for further examination of the problem by competent palaeographers. To deal fully with this would expand this review beyond all conscience and delay it unacceptably. It should be added that, since the Bibliotheca Bodmeriana does not permit study of the ms, the matter is unlikely to be resolvable in the foreseeable future.
30. Finally, and very briefly, 75. The introduction here deals mostly with a discussion of the character of the text (surprisingly, without any reference to the foundational work of C. M. Martini [Martini 1966]). Since the IGNTP does not contain a transcription of this ms, there is no direct comparison to be made. But I note that at 7.18, where we found an evident problem in ed. pr. (which wants to insert [<grc>doca</grc>]<grc>n</grc> at the end of the line), Comfort agrees with IGNTP in omitting it. At 9.33 IGNTP reads <grc>o?utos</grc> against ed. pr., followed by Comfort, <grc>a?utos</grc>.
31. My comments have hitherto tried to be descriptive rather than to offer an opinion. In this last section, I shall try to draw some conclusions from my observations.
32. In general, the transcriptions which I have studied are far more dependent on the editiones principes than the statement from page 14 quoted above would have us believe. That this includes the typographical error by the first editor of 22 is an instance of agreement in error which establishes dependence. It is also quite strange that Comfort claims that the IGNTP edition was 'an excellent help in the production of the transcriptions' (p. 22). If he had either worked from the plates or the ms, or used the IGNTP volume, he would have avoided this error in 22, and would have produced a far more accurate transcription of 45. Of course, I am not expecting that Comfort agree totally with IGNTP. Some readings will always be so difficult as to leave room for more than one decision. But the reproduction of failings in the ed. pr. is no use to anybody. Beyond this, I found a sprinkling of slips which are too frequent for one to be satisfied with the accuracy. These are two flaws with the transcriptions. The third is the absence of dots for doubtful letters. One cannot take an edition lacking these as an advance in the discipline. We thus have one failing in method, another in accuracy, and a third in presentation.
33. The palaeographical parts of the book are too slight to be of much use, with the exception of the fact that it raises several questions which will now require resolution, including the possible identity of 49 and 65, and the correctors of 66. Whether these will prove a distraction or an advance remains to be seen.
34. The final judgement on a book of transcriptions will be of its accuracy. It sometimes requires extensive use of such work before one can form that final judgement. It may be the case that some of the transcriptions here are acceptable. But there are enough errors in the ones that I have examined for one to lack confidence in the others.
35. It is very hard for scholars to achieve perfection. Yet the work of some people working on their own is remarkable. One thinks, for example, of Tischendorf in his editions of manuscripts and his editions, or of Scrivener in his transcriptions and collations. One realises that they are two out of many, and one then thinks of other figures of the past, whom out of charity we may leave anonymous, whose accuracy left a good deal to be desired. But to refer to past achievements and failings is no use, for we live in a very different age. When one looks at the circumstances of Tischendorf and Scrivener, one realises that they did not achieve their work simply as magnificent individuals. The one had royal, the other ecclesiastical patronage to provide the conditions to complete their work. In our own world, accuracy is more likely to be achieved by a group than by an individual editor. The superb achievement of the Editio Critica Maior of the Münster Institut shows to what new heights an expert and well-equipped team can raise the editing of NT texts. The fact is that individuals cannot expect to match it, because today the funding situation for large undertakings requires team projects with secure external funding. Within such a structure, the work of every contributor (including the principal editors) is subject to the scrutiny of colleagues. With the circulation of electronic versions of a text, it is possible to eliminate errors without creating new ones (unlike traditional typewriting). The result of this should be a high standard of accuracy, achieved in an environment where co-workers learn from each other. A single individual can never hope to achieve the accuracy or the necessary skills without such a context of scrutiny and care. The thesis that I am propounding is that the problems with this book arise not because of shortcomings on the part of the editor, but because such an undertaking is doomed to failure by its very nature. There is nobody who could be expected to produce a sufficiently accurate book of this kind under these conditions. I know that I couldn't.
36. I turn briefly to one point raised in the TC-List discussion, where the production of a transcription was compared to the copying process of a Byzantine ms. The case is not similar. Most scribes were producing a current reading text, tending to eliminate non-Byzantine readings in the process. The modern scholar seeks to produce an exact copy of the ancient witness. Any error destroys the only purpose of the copy.
37. This leads to another question, a rather imponderable one: how many errors does it take to make a transcription worthless? The answer, scientifically, should be that even one is enough to do that. This is rather uncharitable, and one should probably say that one in twenty or thirty pages would be too many. I would certainly be pretty worried if anyone could point to half a dozen clear errors in the IGNTP transcriptions of John papyri, or even to four.5 It should be emphasised that an error can include a bad omission or insertion of a dot, a bracket one letter too many to the left or right, a misreading of a punctuation mark or a failure to record one, or a typo. It does not matter whether the data may be 'textually insignificant', something as apparently slight as a possible itacism or inclusion of a movable nu. The job is to record the text of the ms as accurately as possible. The fact that I have recorded errors in Comfort's transcriptions of all the witnesses also transcribed in IGNTP except for 95, covering about fifteen pages of his text, raises grave concerns. Even if one were to be charitable and treat many of his readings as justifiable decisions, there would still be a list of ten errors in these fifteen pages.6
38. A misunderstanding in the TC-List discussion should be corrected. The precise use of dots is not to indicate that a letter is damaged, but that it is uncertain. There are parts of some letters which could not be mistaken for anything else in a particular hand: perhaps half an omega, or the top of upsilon or even the top left stroke of it, or part of phi. Where one is dealing with a literary text rather than a document, one is sometimes able to record a letter quite confidently even where it is relatively incomplete. It is where the traces could be the remains of one of several possible letters that the dot should be employed. This is important where there is known textual variation in assessing the likelihood that a witness supports a particular reading.
39. There seemed to be a point of view expressed on the list, that the provision of accessible transcriptions is so far more useful for those without access to a specialist library than the widely scattered existing versions, that it outweighs any problems with the accuracy of the volume. This is, however, not the case. Certainly, putting all the information in one volume is a good aim. But, once again, without accuracy there is no gain. What is virtually the transcription of an editio princeps with a sprinkling of new errors in a reasonably cheap ($50) edition has the potential to do a great deal of harm. It is seriously to be hoped that those many subscribers to the TC-List who have a great interest in the discipline but are not living over the road from a research library will not use this volume just because it is cheap and accessible. There are several indications that they are too sensible to do so. One message (Clayton Stirling Bartholomew, 23 May, 19:57:41) raised the associated issue of the intended readership and suggested that for this volume it is an 'empty set'. Another message (Robert Waltz, 11 June, 11:14:05) indicated that he was not going to buy the book until he was satisfied that it was worth it.
40. The book's introduction asks for any comments that 'will help make the book better' (p. 15). It also stated that the collection will be expanded as more evidence becomes available. I hope that the editor will seriously consider using his energies in a better way. The reason is simple. To be of any value, the book should (1) include dots beneath uncertain letters, (2) be based on the mss, with far less dependence on earlier editors and (3) be much more accurately produced. To achieve these targets would require doing the work all over again. The very sad conclusion is that it would be better for the editor to throw his energies and abilities into some more useful contribution to NT textual criticism.
© TC: A Journal of Biblical Textual Criticism, 1999.
1Throughout this review I use the words recto and verso in their papyrological sense, the former referring to the side of the ms with horizontal, the latter to that with vertical fibres. 
2Ed. pr. is the standard abbreviation for editio princeps, the first edition of a text.
3I must exonerate Comfort, though, from being wrong in reading [<grc>qew</grc>]<grc>rei</grc> at 20.11. Maurice Robinson (e-mail of 23 May, 21:50:36) thinks it reads ]<grc>rai</grc>. But the epsilon is certain.
4I have used lunate sigma throughout this review, though Comfort uses medial and final forms.
5There is only one place where the need to emend the transcriptions will be recorded in a list of corrigenda to the papyri in the IGNTP majuscules volume (the correction is 5 at Jn 16.20, noted above). There will be four corrections to the restorations and notes. I am grateful to Dr. Comfort for drawing several to my attention.
6There is little to be added to the discussion over 'accurate passages' against errors, conducted between Jim West and Maurice Robinson. In reply to the former, one would have to agree that it would be incredibly hard to produce a transcription in which the majority of the words, or lines, or even verses were wrong. Not to produce at least the occasional page without errors (and one should certainly get the first page of 66 right, when one is fresh and there are no lacunae) would be heinous. But the problem is that once we know there are a lot of errors, then we will suspect an editor on every occasion, because there is no way of knowing where the errors will be.
Handley, E. W. et al. 1997. The Oxyrhynchus Papyri. Vol. 64. London: Egypt Exploration Fund.
Kenyon, Frederic G. 1933. The Chester Beatty Biblical Papyri, vol. II. London: E. Walker.
Kim, Y. K. 1988. "Palaeographical Dating of P46 to the later First Century." Biblica 69: 248-57.
Rhodes, E. F. 1967-1968. "The Corrections of Papyrus Bodmer II," New Testament Studies 14: 271-281.
Roberts, C. H. 1950. The Antinoopolis Papyri. Vol. 1. London: Egypt Exploration Society.
Thiede, C. P. 1995. "Papyrus Magdalen Greek 17 (Gregory-Aland P64: A re-appraisal." ZPE 105: 13-20.
D. C. Parker
Reader in New Testament Textual Criticism and Palaeography
Department of Theology
University of Birmingham