Tuesday, December 07, 2010

Hill, Who Chose the Gospels?

One of the books I picked up at SBL, and the one I chose to read on the plane home was C.E. Hill, Who Chose the Gospels? Probing the Great Gospel Conspiracy (Oxford: OUP, 2010). Chuck, who has written a load of important things about the history of the NT canon (some listed on his faculty web page) has written a readable, but not sensational, treatment of the history of the four-fold gospel canon (he has also written a short summary here).

In fact Chuck doesn’t think that anyone chose the gospels, he thinks (quoting and following Bruce Metzger) that they ‘imposed themselves as canonical upon the church’. The approach works backwards from an excellent and thorough demonstration that Irenaeus was not alone in accepting an exclusive four-fold gospel canon towards the end of the second century (Hippolytus, Tertullian, Origen, Dionysius, Cyprian, Victorinus, Marinus and Euplus [yes a new one to me too] held fundamentally similar views). In working back through the second century Hill argues that Justin Martyr and Papias both also had a four-fold gospel collection, and there is perhaps room for some disagreement here (although it should be noted that Hill has published a long article on the “new” Papias material he discerned in Eusebius, and for a fuller argument see Hill, ‘What Papias Said About John (and Luke): A “New” Papian Fragment’ JTS 49 (1998), 582-629). If you work backwards from the really clear evidence I think you are more inclined to think that somewhat marginal evidence is actually proof that the four-fold gospel was known in a period when working inductively from the available evidence we might not be able to say that.

Overall I think this is a really useful book. It is not sensationalist (despite the talk of conspiracy, Hill is generally in respectful dialogue with other scholars [with a hint of frustration that they haven’t considered all the relevant evidence]). Occasionally I found myself not quite so convinced of Hill’s position as he was, but had to admit he had arguments for his position. Certainly if you accept that Papias had a four-gospel canon tradition this would be tremendously significant for how we read the less clear evidence in the early part of the second century.

I found quite a few problems / errors / something more than just differences of opinion, especially in relation to manuscripts and such things. Somethings to consider for the second edition:

  • p8. Here Hill is discussing how many other gospels existed in the second century and gives a list of nine, noting ‘It is not unlikely that more Gospels might have circulated before 175. But if they once existed they have left no record, even in later lists of books to be avoided ...’ But there are later lists of non-canonical gospels which do provide some record of numerous named gospels. For example, there is a Samaritan list of 35 named non-canonical gospels (J. MacDonald & A.J.B. Higgins, ‘The Beginnings of Christianity according to the Samaritans’ NTS 18(1971)54-80, esp. pp. 66-69), and the Decretum Gelasianum also names a number of gospels and other books.
  • p. 13. In discussing the Oxyrhynchus Papyri Hill mentions 500,000 pieces, ‘only about a tenth of which have so far been published’. But in fact it is more like a hundredth, since the most recently published is P. Oxy 5071 (Parsons, cited by Hill in note 7 confirms this).
  • p. 20f. Here Hill is discussing the lack of early manuscripts of Mark. He argues that since church fathers knew Mark, and since some manuscripts could have included Mark alongside other gospels, ‘there is reason to believe that at this time Mark was more widely used in Christian churches than the statistics of papyrus discoveries would, by themselves, lead us to believe’. I think on the contrary that there is no reason to believe this (and I have argued this more fully in an essay on Mark in a forthcoming book edited by Hill) - it is rather a consistent phenomena that church fathers who doubtless knew Mark did not often quote Mark (for Clement of Alexandria for example Cossaert has 274 quotations from Matthew; 143 from Luke; 72 from John and 1 from Mark; for Didymus Ehrman has 155 passages from Matthew; 146 passages from John; 111 passages from Luke; and 10 passages from Mark; Brogan found only one quotation from Mark in Athanasius!).
  • p. 25. Hill writes: ‘A “book” was a scroll, or roll, a long sheet of papyrus or parchment rolled up with rods attached at each end to serve as handles.’ In Greek bookrolls on papyrus no rods are used (either in contemporary illustrations or in actual archaeological finds).
  • p. 30f. Hill follows Hurtado’s idea that use of a codex might correspond with identification of the text as scripture by the scribe; and further that public reading in church required a large codex. Since non-canonical gospels are often on rolls or small codices, they probably were not regarded as equivalent with the four canonical gospels. But in discussing the size of two non-canonical texts Hill slightly cooks the books. This is perhaps not a big problem, but suggests that perhaps the overall argument is driving how the evidence is perceived and presented. Firstly by taking P. Oxy 4009 (G.Peter?) as representing a miniature codex (‘a strong possibility’), whereas we don’t really know that for sure, and the editors also suggest that it could have been from a double columned codex. Secondly by describing P. Ryl 463 (G. Mary) as ‘a miniature’ when as reconstructed (9cm x 13.5 cm) it certainly goes beyond the normal categorisation of a miniature codex.
  • p. 72. Here Hill is comparing Clement of Alexandria’s use of non-canonical gospels with his use of the canonical ones, citing a monograph by Mutschler [whose first name is misspelt as Bernard, when it is actually Bernhard] on Irenaeus that he used Matt 757 times; Luke 402 times; John 331 times and Mark 182 times. Now I haven’t been able to check Mutschler’s book, nor his definition of “use”; but it is plain that the more recent monograph actually on the subject (The Text of the Gospels in Clement of Alexandria by Carl Cossaert) offers significantly different figures: 274 quotations from Matthew; 143 from Luke; 72 from John and 1 from Mark (and actually discusses Clement’s lack of knowledge of Mark).
  • p. 84. Here Hill is discussing the Akhmin codex of the Gospel of Peter. He writes: ‘It has often been reported that this codex was found in the grave of a monk ... This is part of a legend that has grown up around the discovery. We don’t know if the person in whose grave it was found was a monk or not, or what the person thought of the book. As far as we know, the gravedigger could have thrown it into the grace to get rid of it!’ Well, the reason it is often reported is not because of legend, but because the original publication of the manuscript, by one of the French archaeologists who excavated the Christian cemetery at Akhmim, stated that it came from the grave of a monk. One might disagree with this, or wonder whether the archaeologists had sufficient basis for making this identification (as van Minnen does in an article cited by Hill), but the conclusion is more well-grounded than Hill’s speculation. Another grave in the same cemetery contained a mathematical papyrus - the phenomena of people being buried with texts that had some relevance/value to them is fairly widely attested.
  • p. 118. Here Hill is discussing various aspects of Skeat’s view that some NT papyrus manuscripts were originally four-gospel codices. I was especially interested in this bit, and read it especially carefully and found a couple of problems. a) He notes that the scribe of P75 has some harmonisations and suggests that these ‘seem to indicate the scribe’s knowledge of Matthew and perhaps Mark’. This reference to Mark goes beyond the evidence, even of Comfort and Barrett (cited as evidence) who note the scribe’s knowledge of Matthew (as also picked up in Royse), since the possible reference to Mark is only in a direct parallel to a passage also in Matthew - if the scribe knows uniquely Matthean readings it doesn’t make much sense to propose he drew a reading from Mark. b) Hill states ‘if copied around 200, it is more likely than not that it [P75] had such a companion volume [containing Matthew and Mark], whether attached or separated.’ This is interesting, but I have no idea how this likelihood is measured. I would think this is extrapolating beyodn the available evidence. c) In discussing Skeat’s view of P4, 64 & 67 he states that in Skeat’s view ‘the codex contained at least three Gospels, and other features of the papyri indicated that this condex in fact originally contained four.’ Hill gives no indication of what these ‘other features’ are. I had another look at Skeat and can’t find any comment in this direction. d) Hill says that ‘Skeat’s conclusions have indeed been accepted by a number of other papyrologists’, but the footnote refers only to van Haelst, writing more than twenty years before Skeat and who does not accept Skeat’s view (although he does accept that P4 is probably from the same codex as P64 & P67). e) Hill notes that Skeat’s views have ‘not gone unchallenged’ (with footnote to Head and Charlesworth) but then says: ‘it seems agreed, however, that the books of Matthew and Luke represented in P4, 64, 67 were copied by the same scribe, whether bound together with Mark and John or not.’ This fails to note the major disagreement about whether we should even think of Matthew and Luke as bound together.
  • p. 119. Hill says that P75 ‘has sectional divisions which would make it easier to read aloud to a congregation’. This is clearly either wrong or rather exaggerated. I’m not saying that P75 couldn’t be read aloud to a congregation, but it has less help in this than just about any other NT manuscript. (On p. 121 he refers to ‘the apparent liturgical design of the papyri P75’ which I also found very questionable).


  1. Looks interesting Peter. I'm so glad to see former Cambridge PhD students making mistakes too...
    You guys were probably at SBL when prof Markus Vinzent (Kings College)gave a paper here in Cambridge on the Resurrection of Christ in the Second Century. Much of what he had to said (Marcion especially!)is relevant for you guys. Check my short report if you like: http://resurrectionhope.blogspot.com/2010/11/controversial-new-hypothesis-about.html

  2. Oops, Chuck had OUP send me a review copy for the blog - but you've done the job Pete! Guess I will follow in your footsteps again. Let's face it: You are always a-Head :-)

  3. Well I paid good money for my copy. You can review it too - I quite like it except for the points noted.

  4. I've added a slight up-date to the point about p. 8.

  5. p.30f. I hope you will reconsider the ‘cooking the books’ comment on my treatment of P. Oxy 4009 and P. Ryl 463. You say, ‘But in discussing the size of two non-canonical texts Hill slightly cooks the books. This is perhaps not a big problem, but suggests that perhaps the overall argument is driving how the evidence is perceived and presented. Firstly by taking P. Oxy 4009 (G.Peter?) as representing a miniature codex ('a strong possibility'), whereas we don't really know that for sure, and the editors also suggest that it could have been from a double columned codex. Secondly by describing P. Ryl 463 (G. Mary) as 'a miniature' when as reconstructed (9cm x 13.5 cm) it certainly goes beyond the normal categorisation of a miniature codex.’

    Roberts, in Manuscript, Society, and Belief, 10-11, says of miniatures, ‘measuring from 15 x 11 cm. down to 7 x 5 cm. they are far too small for public use’. Gamble, Books and Readers, 235-6, give a similar range. Your figure for P. Ryl 463’s measurement, 9 x 13.5, fits within this range. Turner Typology, lists a 9 x 13.5 codex in his Miniature Category. As to P. Oxy 4009, the editors do say, ‘If there was only one column to the page, we have a miniature codex, with a page width of (say) 7 cm … In that case, the original page height may have been no more than 10cm… but it remains possible that we have here one of the rare examples of a two-column papyrus codex…’ Given that two-column Christian codices before the fourth century are acknowledged by everybody to be rare, I think it is safe to say that there is a ‘strong possibility’ that P. Oxy. 4009 was a miniature. ‘Strong possibility’, in my experience, does not mean ‘for sure.’ I go on to say, ‘That is, while the scribe or commissioner of these small codices may or may not have considered these Gospels to be Scripture (we do not know), these particular copies were evidently not intended for reading in public but in private.’ The last part of the statement (qualified by an ‘evidently’) is certainly not infallible, but it is based on the judgments of Roberts, Gamble, and Hurtado.

  6. p.72. Thanks for the correction on the spelling of Mutschler’s first name. Mutschler says he is basing the numbers on the GCS indices. Why the huge discrepancy between Mutschler (GCS) and Cosaert on the number of references to Mark in Clement? There is probably a big difference between a ‘reference’ (Bezugnahmen), almost however defined (I think these numbers include allusions), and a quotation which is identifiable and analyzable in terms of determining nature of the underlying text, which is what Cosaert was looking for. What I find in Cosaert’s book is ‘Since Matthew and Luke share most of the material in Mark, there is little distinctly Markan material available for textual analysis. Thus, unless a church father specifically refers to a passage as originating from Mark, Markan references cannot usually be identified with certainty.” (234). This seems to mean that there may be many more, but they are indistinguishable from Matthew or Luke. Cosaert actually treats four places where Clement quotes Mark, and see his Appendix I, which gives a bunch more possibilities.

    p84. The gravedigger comment was meant to be tongue-in-cheek, not a serious proposal; bad humor perhaps, and perhaps not well founded in archaeological psychology, but I thought an unsettling statement might be appropriate in the face of so many confident pronouncements by scholars about how valuable or cherished the Gospel of Peter was to this ‘monk’. Perhaps the codex containing the GP, the ApocPet, part of 1 Enoch and the Martyrdom of Julian of Anazarbus was sacred to the person in whose grave it was found. And perhaps the mathematical papyrus found in the other grave you mentioned was also a sacred text.

  7. p. 118. a) About my comment that P75’s harmonizations (acc. to Comfort/Barrett and Royse) seem to reflect knowledge of Matthew and ‘perhaps Mark’. You say that ‘this reference to Mark goes beyond the evidence, even of Comfort and Barrett (cited as evidence) who note the scribe’s knowledge of Matthew’. But here is what they say (p. 504) ‘One of the clear indications of the scribe’s Christianity was his knowledge of the other Gospels, especially Matthew’ and then they list the three ‘subtle harmonizations’, including number 3, ‘John 6:5 to Matt. 14:15 and/or Mark 6:36’. These are their words. We are all free, of course, to disagree; that’s why I said ‘and perhaps Mark’, reflecting accurately, I believe, C&B’s statements. You say, ‘since the possible reference to Mark is only in a direct parallel to a passage also in Matthew - if the scribe knows uniquely Matthean readings it doesn't make much sense to propose he drew a reading from Mark.’ I think it makes sense if you introduce it with ‘perhaps’. Are you saying that if we know that a scribe knew Matthew, we must assume he didn’t know Mark? Did a scribe who harmonized to Matthew ever harmonize also to Mark? This is the only harmonization listed in John (as opposed to the two Matthean ones in Luke), so, perhaps it was made already in the scribe’s exemplar, and who is to say whether that earlier scribe did not know Mark (and perhaps was even ignorant of Matthew)?

    b) You challenge my statement that ‘if copied around 200, it is more likely than not that it [P75] had such a companion volume [containing Matthew and Mark], whether attached or separated’ as going beyond the evidence. This does go beyond the evidence supplied by P75 itself, but I was not limiting myself to that rather narrow slice of the evidence we have available for such judgments. We know that there was at least a conceptual four-Gospel canon among certain groups of Christians by ca. 200, and that canon consisted of Matthew, Mark, Luke, and John. In other words, we know there existed people who, if they had Luke and John, would also have wanted to have Matthew and Mark too. Do we know of persons or groups who accepted only Luke and John? Not likely a Marcionite or (acc. to Irenaeus) an Ebionite. I think this makes it ‘more likely than not’ that the owner(s) of P75 would have possessed Matthew and Mark as well. Not certain, but more likely than not. Perhaps it is going too far to imagine that they would have had these two other valued Gospels bound together, but on analogy with Luke-John, it may not be a serious violation.

  8. c) I did make a mistake here, and thanks for pointing it out. At the time of writing I overlooked the fact that Skeat actually did not offer evidence from the papyri themselves to move from (what he thought was) evidence of P4,64,67 being a three-Gospel codex to his conclusion that it was a four-Gospel codex. The ‘other features’ I failed to explain (for simplicity’s sake) were actually the features that made him conclude it had at least three. Those are what I had in mind, forgetting that he really had no reason for moving to ‘four’ except that he could not envision a three-Gospel codex. It remains that Skeat’s ‘conclusion’ (whether drawn from papyrological features or not) was that it was a four-Gospel codex.

    d) Here too you are right, I footnoted too quickly. Van Haest was carelessly cited by me as someone who must have been persuaded by Skeat (maybe he had been persuaded by Roberts). Graham Stanton was convinced by Skeat, but he was not papyrologist. How about Bruce Metzger and Bart Ehrman, The Text of the New Testament, 53? ‘The careful study by Skeat … has shown that they all derive from a single-quire codex that originally contained all four Gospels …’ Bagnall, Early Christian Books in Egypt, seems very sympathetic.

    e) I should not have said ‘whether bound together with Mark and John or not’ but ‘whether bound to each other and to Matthew and Mark or not’. A mere slip. My point in the sentence was simply that many (Roberts, Skeat, Charlesworth, Metzger-Ehrman, Bagnall?) have expressed themselves as in agreement with the common hand of the fragments.

  9. I see that some of my postings did not actually post. Just for the sake of completeness, I'll try again, even though this next one is out of order and should have come first:

    Thanks, Peter. I appreciate the review very much. And thanks for picking up on a number of points for me to re-examine if there is a second edition. I really value your sharp eye. You are absolutely right about the “tenth” of the 5000,000 Oxyrhynchus pieces comment and the contradiction with Parsons – an embarrassing slip on my part. May I give a response to the other points you brought up? (In several parts.)

    p8. About the later Samaritan and Gelasian lists of non-canonical Gospel. The question is, how many of them were written before C.E. 175? I already said on p.7 that gospels continued to be written after that date. I used the 175 figure and the number of Gospels written by that time because this is the date and these are the numbers Petersen used. But maybe I’m wrong. The next sentence contains the point: ‘But whether eight or eighty, this does not yet answer for us which Gospels, if any, were being used and valued by most Christians in the second century’, etc. If anybody wants to argue for more Gospels in existence by then, it is fine with me to do so. It won’t change the point I was making.

  10. p20f. On the use of Mark. I stand by my comment that Mark was more widely used than the statistics of papyrus discoveries would, by themselves, lead us to believe. I’m talking not only about actual quotations in surviving writings but use in other ways, such as liturgically (as I think Justin would lead us to believe). Mark is also mentioned by name by Papias, virtually by Justin, by Clement, and I think must be inferred from the Muratorian Fragment, besides of course Irenaeus. These already far outnumber the extant papyrus copies of Mark (namely, 0) that we have from before the end of the second century. It seems that Matthew was much more often preferred for citations of synoptic material, for whatever reason, but Mark held a place in the collection of four.

    p. 25. I should remove the reference to handles. But Turner, Greek Papyri. An Introduction, 7, says that the roll was vulnerable ‘even if protected by rollers (which have rarely survived)’. Adam Bülow-Jacobsen, ‘Writing Materials in the Ancient World’ in Bagnall, The Oxford Handbook of Papyrology, 19, speaking of the measures taken to protect the ends of a roll, says ‘At the beginning of the roll was the protokollon, an unwritten sheet, while at the end there was probably the o0mfalo/v or umbilicus, the wooden stick around which the papyrus was rolled, but even if no umbilicus was present, the end was protected inside the roll.’

  11. [This is the last one]

    p.119. You say that my reference to sectional divisions in P75 which would make it easier to read aloud to a congregation is ‘clearly either wrong or rather exaggerated’. Martin & Kasser, Papyrus Bodmer XIV-XV I, 14-16, detail the sectional divisions. Hurtado, Artifacts, 180-181, after mentioning P75 and its sectional divisions, says ‘To repeat a point for emphasis, this means that the early manuscripts in which these devices were deployed are artifacts of early Christian exegesis of these texts, and probably also reflect something of how these texts were read liturgically, by about 200’. See also his article in Evans and Zacharias, Jewish and Christian Scripture as Artifact and Canon (2009), page 78, and in that same volume Charlesworth’s article, in which he lists the sectional division system in P75 as one of the features indicative of its public/liturgical setting (citing Turner on the larger script of P75, making it easier to read aloud). Comfort &Barrett (503) say of P75 ‘The scribe even added a system of sectional divisions to aid any would-be-lector’. Of course, nobody knows for certain if P75 was read out liturgically, we only know that it has a number of scribal features which set it off from a number of other NT papyri, and many non-Christian papyri, which many experts believe would have been aids to public reading. And after all, it is not illegitimate or anachronistic or mere wishful thinking to imagine manuscripts of the Gospels such as P75 being read publicly in services of worship, because we know from literary sources that they were.

  12. Chuck, those two comments which did not post actually got stuck in the automatic spamfilter (which I just noticed). I think this is a risk with very long comments, so it is better to divided them in shorter pieces (Peter Head will appreciate that too, because we can easier break the 50-comments barrier). I don't know exactly how the spamfilter works, but apparently your comments came through on your second attempt.

    In fact, many real spamcomments get stuck in that filter, so on the whole it is good.

  13. Thanks, Tommy, for the explanation. Next time I'll make the comments shorter to get a higher comment count!

  14. Your comments are interesting. I have a basic knowledge of academic NT study, albeit somewhat dated, due to a seminary education (Covenant Theological Seminary, St. Louis, M.Div, 1980). I have never lost interest in the topic.

    As a high school teacher, I am somewhat saddened by the prevalence of variations on the Tubingen School's late-date view (born of the attempt to make things fit an Hegelian scheme) which I have found among educated people. I've been given Pagels' _Gnostic Gospels_ by those near and dear trying to "set me straight" and came away surprised at her reconstructions of early ecclesiastical history (reconstructions whose sources seem quite elusive to me). Hence, I'd like to follow this blog.