HOME

The Date of the Magdalen Papyrus of Matthew (P. Magd. Gr. 17 = P64):
A Response to C.P. Thiede
[1]

By

Peter M. Head

Published in Tyndale Bulletin 46(1995)251-285

(reprinted here with minor alterations)

 

Summary

This article considers Carsten P. Thiede’s arguments concerning the date of P64 and suggests that he has both over-estimated the amount of stylistic similarity between P64 and several Palestinian Greek manuscripts and under-estimated the strength of the scholarly consensus of a date around AD 200. Comparable manuscripts are adduced and examined which lead to the conclusion that the later date is to be preferred.

I. Introduction

This article is a response to the arguments of Carsten Peter Thiede which were republished in the previous issue of Tyndale Bulletin.[2] The most significant and controversial of Thiede’s findings was that the Magdalen Papyrus of Matthew was written sometime before the end of the first century. This conclusion, and the palaeographical basis of Thiede’s arguments, will be the focus of our response. In particular we shall examine the manuscripts appealed to by Thiede as palaeographically similar to P64. After a discussion of the history of the study of this manuscript (section II, an area not covered by Thiede), we shall begin our analysis with the manuscript itself (including a plate) and a transcription which varies at a number of points from that of Thiede (section III). In Section IV, with the aid of further plates, we shall investigate several of the manuscripts appealed to by Thiede as early comparative material in order to assess his claim that significant manuscript discoveries require a radical reassessment of the date of P64. This will be followed by a similar investigation of the arguments which lead previous generations of scholars to accept a date of around AD 200 (Section V); and a brief discussion of various other features of the manuscript which relate to its dating (section VI). Our conclusion (section VII) will follow from these comparisons.

             A further comment by way of introduction and orientation may be appropriate. The steadily increasing numbers of New Testament manuscripts on papyrus and the confident allocation of dates by the various handbooks can obscure the fact that we have no absolutely secure dates for any NT manuscript on papyrus.[3] Indeed we have no dated manuscript of the NT until the Uspenski gospels of AD 835.[4] This is not particularly unusual, as literary documents were not customarily dated in antiquity (the first literary manuscript dated by the scribe is a text of Dioscorides from AD 512 now in Vienna).[5]

             While documentary papyri such as private letters or receipts often contain a date, there were fairly strict distinctions made between writing styles used for documentary purposes and those appropriate for literary works. This means that arriving at the date of a New Testament manuscript, which in this case, as in the vast majority of cases, is written in a literary rather than documentary hand, always needs a careful investigation of the evidence and a weighing of the arguments. In this context new arguments together with appeals to newly discovered datable parallels and a call to reassess old conclusions should be welcomed as an inevitable requirement of New Testament scholarship. While in most cases neither certainty nor precision can be attained it is normally possible to allocate an approximate date and to give a range within which the document can be dated with a reasonably high degree of probability.

II. The History of the Discussion

Even before the publication of his article in Zeitschrift für Papyrologie und Epigraphik, Thiede’s conclusions were given a high-publicity airing in The Times of London on Christmas Eve 1994.[6] In addition to highlighting the potentially radical implications of Thiede’s dating of the fragments, this article referred to the lack of attention given to the manuscript by scholars. In this section, we shall see that several scholars have paid attention to the question of the date of the manuscript by appealing to comparable, datable, manuscripts, which led to the consensus date of around AD 200 for the fragments. It will also be clear that Thiede has not refuted, or even given attention to, either the arguments of these scholars or the manuscripts to which they appealed.

             The modern history of the fragments began in 1901. They were purchased from a dealer in Luxor by Revd. Charles B. Huleatt (1863-1908), who identified them as portions of Matthew’s Gospel, and presented them to Magdalen College, Oxford, where Huleatt had been a demy (foundation scholar). No indication of the provenance of the fragments was given, although the small envelope in which Huleatt kept the manuscript contains the enigmatic comment ‘these found year after but evidently from same leaf’.[7] This suggests either that the fragments were found in successive years and Huleatt perhaps purchased them in two instalments, or that additional fragments (from the same leaf) have subsequently been lost.[8]   Huleatt is reported in the Librarian’s report of 1901 to have suggested a date in the third century. The librarian, H.A. Wilson, reported that A.S. Hunt favoured a fourth century date, and in view of Hunt’s status as a papyrologist, it was this date that prevailed.[9]

             An edition of the manuscript with plates was not published until 1953.[10] In addition to editing and transcribing the text,[11] Roberts classified the hand ‘as an early predecessor of the so-called “Biblical Uncial”‘ which began to emerge towards the end of the second century.[12] Hunt’s date was too late, he argued, partly because of his presumption (then common) that codex manuscripts demanded a date later than the third century and partly because of the scarcity of Biblical papyri in the early period. This identification of the style of writing provided the methodological key for dating the manuscript: once the general style has been identified or recognised the peculiarities of the individual manuscript can be compared with other examples of the style. If some of these can be dated, then it may be possible to construct a pattern of development into which the manuscript under scrutiny can be placed.[13]

             Roberts compared four other manuscripts (P. Berol. 7499; P. Oxy 843; P. Oxy 1620; P. Oxy 1819) and concluded with the support of H. I. Bell, T.C. Skeat and E. Turner in favour of a date ‘in the later second century’.[14] We shall investigate the manuscripts appealed to by Roberts as comparable with P. Magd. Gr. 17 in section V; for the moment we simply note that in terms of procedure, Roberts first identified the style and then by comparative palaeographical analysis attempted to ascertain an approximate date, after which he appealed to other scholars to confirm his research.[15]

             In 1956 R. Roca-Puig published another manuscript of Matthew (P. Barc. Inv. 1), two small fragments containing portions of Matthew 3.9, 15; 5.20-22, 25-28.[16] Roca-Puig noted various features of the manuscript,[17] identified the hand as a form of ‘biblical uncial’ and cited P. Berol. 7499 and P. Oxy 1179 for comparison, concluding that it should be dated no later than the third century.[18] It appears to have been Roberts who first suspected that the two manuscripts were part of the same codex; subsequent correspondence with Roca-Puig ‘confirmed this beyond a doubt’ and a date late in the second century was agreed.[19]

             Since then, the conclusion that the two sets of fragments are from the same manuscript has been an agreed (and assured?) consensus. Granted a date around AD 200, P64+67 contests with P77 (P. Oxy 2683) the honour of being the earliest manuscript witness to any of the synoptic gospels. Different studies have discussed the significance of its paragraph divisions,[20] its codicological format,[21] and its singular readings.[22] The issue that has aroused repeated exploration is the relationship of P64+67 with P4.[23] The similarities of script, size (c. 16 letters per line, 36 lines per column), structure (dual columned), provenance (P4 was also purchased in Luxor in 1891 following discovery in Coptos in 1880) and subject matter are doubtless significant.[24] Nevertheless dissimilarity of papyrus colour and fibre, letter formation and use of outdentation preclude the conclusion that all three manuscripts belonged to the same codex.[25] There is therefore nothing for us to gain from an extended discussion of P4.

             This history of research would be of little interest were it not for Thiede’s recent claims. The initial coverage in The Times advanced the claim that parallel hands from Pompeii, Herculaneum and Qumran (datable before AD 70 or 79) secured a similarly early date for P. Magd. Gr. 17 (=P64), a hundred years earlier than previously thought. The basis for this claim was, apparently, a re-identification of the style of the script:

In the course of four trips to Oxford, it became clear to him [i.e. Thiede] that the papyrus was written in a distinctive script common in the 1st century BC but petering out by the mid-1st century AD.[26]

Such an early date for this manuscript would obviously have important implications for the date of Matthew and the history of early Christianity. The use of nomina sacra suggested the existence of extremely developed Christology in the earliest period. The journalist’s claims were both hyperbolic and somewhat ill-informed: ‘it means that the New Testament is not a 2nd-century version of an oral tradition, but an eyewitness account’.[27]

             This coverage provoked predictable controversy, especially among those who recalled Thiede as an advocate for O’Callaghan’s claims that numerous portions of the Qumran scrolls from Cave 7 are NT texts from pre AD 70.[28] A series of letters to the editor and articles and correspondence in other newspapers and magazines fuelled the controversy, focusing particularly on the historical implications of Thiede’s argument for the dating of Matthew.[29] It was only after some of the initial hubbub had died down that Thiede’s scholarly article became available.

             In terms of method, Thiede follows a standard palaeographical approach (cf. what we said of Roberts above): identify the style of the script, then relate the manuscript under scrutiny to other (hopefully datable) examples of the same style. Since new manuscripts have been discovered since Roberts published P. Magd. Gr. 17, and especially since several of these were certainly written before AD 70 (Qumran) or 79 (Herculaneum), their value as datable comparative scripts is inestimable.

             Before we turn to these manuscripts themselves, we need to clarify Thiede’s claim about the date of P. Magd. Gr. 17. Thiede argued that since the main comparative script (from Nahal Hever) is generally dated around AD 50, the possibility of a date around the middle of the first century for P. Magd. Gr. 17 would be possible.[30] This date, which was picked up in much of the reporting of Thiede’s findings,[31] is not the date that actually emerges from his paper. Although Thiede is sympathetic to the possibility that the manuscript might have been written prior to AD 70,[32] he repeatedly refers more generally to ‘a first-century date’, or ‘prior to the turn of the century’, and in conclusion to ‘some time after the destruction of the Temple in Jerusalem’.[33] All of these phrases suggest a date between AD 70 and AD 100, and this is confirmed in the following comment:

I do not give a precise date, but suggest a date in the last third of the first century: The ‘starting point’ is the middle of the century; I allow for a variation of c. 20 years + / - and then opt for the later end, ‘soon after A.D. 70’.[34]

III. Assessing the Transcriptional Details

In order to assess Thiede’s palaeographical arguments we must first familiarise ourselves with the document itself, hence our plate of P. Magd. Gr. 17 (=P64, see plate one).[35] In his ‘Reappraisal’, Thiede claimed to improve on Roberts’ transcription of the manuscript in four places.[36] We offer our own transcription in order to familiarise the reader with the contents of the manuscript before we turn to the question of dating by comparison with other manuscripts. Several of Thiede’s ‘improvements’ are debatable and are addressed in the following notes.[37]

 

The left column; Frag 1, Verso = col. 1, ll. 24-28. Matthew 26.7f.;[38]
the right column: Frag 3, Verso = col. 2, ll. 16-20, Matthew 26.14f.[39]

katexeen ep]i? [thv] k?e?                tot]e? pore?[uqeiv eiv

falhv] autou anakei                twn i]b? legom[enov iou

menou i]dontev de oi                   dav ]i?skariw[thv prov

maqhtai ]hganakth                  tou]v? arxier[eiv eipen

                                                            ti qe]lete mo?[i dounai

 

Frag 2, Verso = col. 1, ln 34-36. Matthew 26.10 [40]

o iv eipen a]u?[toiv ti

kopouv par]exete [th

g]u?naiki ergon ga[r

 

The left column: Frag 3, Recto = col. 1, ln 15-19, Matthew 26.22f.;[41]
the right column: Frag 1, Recto = col. 2, ln 24-28, Matthew 26.31 [42]

legein au]tw m?[hti                 autoiv o iv p?a?n?[tev umeiv (?)

egw eimi] k?e? o? d?[e apo               skandalisqh[sesqe

kriqeiv ]e?ipen o e?[mb                  en emoi en t[h nukti

ayav me]t? e?m?ou [thn                 tauth g?e?g?[raptai gar

xeira en t]w? t?r?u?[bliw

 

Frag 2, Recto = col. 2, ln 34-36, Matthew 26.32f .[43]

p?r?o?a?c?[w umav eiv thn

galeglaian a[pokri

qeiv de o petrov? e?[ipen

IV. Assessing Thiede’s Palaeographical Analysis

Thiede suggested that the consensus ‘around A.D. 200’ was merely a convenient ‘dumping ground’ and that Roberts’ revision of Hunt’s suggested date from fourth century to late second century needs further revision in light of new manuscript finds. These manuscript finds come with archaeologically fixed final dates (termini ante quem).

             The first is the Greek Minor Prophet scroll from Nahal Hever (8HevXIIgr), dated by most scholars between 50 BC and AD 50 (see plate two).[44] Thiede noted: ‘the identity and near-identity of several letters is striking: alpha, epsilon…, iota, omicron, rho and nu are particularly close’.[45] We have reproduced as plate two the page to which Thiede refers in his discussion (columns B1-2 = Zc. 8.18-9.5).[46] This represents the work of one of the two scribes involved in producing the manuscript (this incidentally shows that two contemporary scribes from the same location working on the same text in a similar style can nevertheless have quite distinct scripts).[47]

             Close inspection provides little support for Thiede’s contention that two scripts are similar. Other than the obvious similarity of bilinear majuscule lettering, this script (8HevXIIgr Scribe B) has a more decorated appearance than P64; note especially the ornamentation of the letter-forms in blobs, hooks and half-serifs. In the size and spacing of the letters, the thinner pen-strokes, and the use of small spaces between words, 8HevXIIgr appears quite distinct from P64.[48] Even the letters which Thiede specifically highlights are actually quite distinct: the alphas of 8HevXIIgr have a generally horizontal cross-bar (P64: distinct angle from lower left to upper right); the epsilons have detached cross bars (no parallel in P64); the iotas have decoration and do not extend as far below the line as P64; the rho has a pronounced decorative blob on the end of its down-stroke (contrast P64); the nus also have pronounced decorative blobs at each join (unlike P64). Of the letters to which Thiede draws attention, only the omicrons are similar, and little significance can be drawn from this.

             Furthermore to take into account only similarities between letters and to ignore their differences is methodologically untenable. In many ways, when trying to draw comparisons between scripts, differences between letter-forms are of more importance than the odd similarity (after all there are only so many ways in which letters can be written). For differences, note the upsilon (8HevXIIgr: straight lines and 90 degree angle cf. P64 with curve), sigma (8HevXIIgr: extended horizontals), and mu (8HevXIIgr: outward pointing down-strokes).[49] The only conclusion which can be drawn from this evidence is that there is no significant relationship between the script and style of P. Magd. Gr. 17 and 8HevXIIgr. There is, then, nothing to be gained from a discussion of the date of 8HevXIIgr, and Thiede’s claim to have found significant new evidence may be unfounded.

             The second source to which Thiede appeals for comparison is pap4QLXXLevb (a papyrus manuscript of Leviticus in Greek from cave 4; see plate three).[50] This manuscript, according to Thiede, ‘shows several letters resembling Papyrus Magdalen Gr. 17, such as the alpha, the beta, etc.[51] It is not clear how many letters are covered by Thiede’s ‘etc.’! In fact, however, even those letters specified are not actually very similar: alpha has a horizontal half-serif at the lower end of its upright strokes and a horizontal cross-bar (contrast P64 as previously mentioned), while beta is not even fully attested on P64. In general, the style of pap4QLXXLevb is decorative with thin strokes and numerous hooks and (half-) serifs and no descenders below the bottom line (unlike P64). Some letters are very different from P64. For example, epsilon is very rounded, pi has a pronounced curve in its right upright, sigma is quite rounded, tau has an extended cross-bar, and upsilon has a lower hoop.[52]

             Thiede also appeals, in a general way, to parallels among the Herculaneum papyri and to 7Q61. According to Thiede 7Q61 ‘has the characteristic Eta with the horizontal stroke above the median, evident in Magdalen Gr. 17.’[53] But, notwithstanding the previously mentioned methodological problem of appealing to individual letters, the eta in this small fragment has to be reconstructed from only partial manuscript evidence and provides no evidence relevant to Thiede’s assertion.[54] Since numerous styles are exhibited among those Herculaneum papyri which have so far been unrolled,[55] there are bound to be some general similarities to almost anything.[56]      Nevertheless, while parallels certainly exist between scripts at Herculaneum and the rounded and decorated scripts of 8HevXIIgr and pap4QLXXLevb, close parallels to P64 are not apparent.[57] This rounded formal hand with decorative elements, which was labelled Zierstil after Schubart’s description, is known, as our discussion suggests, in manuscripts from Egypt, Herculaneum and the Judean desert, predominantly dated between 100 BC and AD 100.[58] Since the similarities with P64 cannot be upheld, we shall not unduly extend our discussion at this point.[59]

             Another manuscript appealed to by Thiede is 4QLXXLeva, a fragment of the Greek Old Testament from Qumran (see plate four).[60] Thiede argued that ‘the alpha, beta, epsilon, eta, iota, kappa, eta (sic) etc. are identical or near-identical to what we find in Magdalen Gr. 17’.[61] The initial impression is that this script is quite different from those we have examined up to this point, generally lacking decoration and ornamentation.[62] As will be obvious from the plate, the layout and general appearance is quite different from P64: more upright with narrower lettering and, as Thiede noted, a slight right-hand lean and a tendency to ligature.

             As regards the letters mentioned by Thiede, we ought to note that the manuscript is not consistent in its letter-forms, so a number of different alpha-shapes occur (with very narrow horizontal bar, without any observable cross-bar, and with upward sloping cross-bar). The upward sloping cross-bar most closely approximates the alphas in P64 but could hardly be regarded as identical (compare the first two alphas in the eighth line from the bottom: meta parrhsiav, or the first alpha in the fifth line from the bottom: tav entolav mou, which are quite distinct from the more consistent alphas in P64; note the angle and width of the cross-bar and length of right-hand down stroke). Since no beta is completely attested in P64, it is strange to find an appeal here, especially noting the proposed beta in P64 (the first letter of line 2 in Frag 3 verso) which has a very small lower circle, unlike those in 4QLXXLeva (see e.g. line 1, 9, 14 etc.). Only two deltas occur in 4QLXXLeva (line 13: muriadav and line 15: h diaqhkh), and they are narrower than those in P64. The epsilons of 4QLXXLeva are much more circular than those in P64. The etas and kappas are not particularly dissimilar, but the iotas do not extend below the line (as in P64). Other letters in 4QLXXLeva which are quite different from those of P64 include mu (with outward sloping sides), pi (with pronounced curve in right hand upright), rho (more curved, single stroke), and upsilon (squarer upper section). Such a list of significant differences precludes any stylistic identification of 4QLXXLeva with P64.

             In conclusion it has become apparent that Thiede’s attempt to reevaluate the dating of P64 on the basis of newly discovered and closely related manuscripts has not produced secure results. The method of comparison, with its focus on a few similar letters, is flawed; and even if that were not necessarily agreed its execution—the actual letters appealed to as identical—fails to provide compelling similarities. If our discussion has adequately represented Thiede’s argument, it is difficult to escape the conclusion that the actual palaeographical basis of Thiede’s argument is without substance.

V. Assessing the Consensus Palaeographical Analysis

Although Thiede’s analysis fails to convince, a negative conclusion on that count hardly justifies the consensus date of around AD 200 without further discussion. In this section we shall assess the arguments, largely ignored by Thiede, of Roberts and Roca-Puig, with a particular focus on the palaeographical aspects of those discussions (analysis of handwriting style and appeal to comparable, ideally datable, manuscripts). These scholars identified the style of P64 as an early example of the ‘Biblical Uncial’, a style that is epitomised by the later biblical codices of Vaticanus and Sinaiticus, and appealed for comparison to other documents conventionally dated in the second or third centuries AD.[63]

             The closest parallel which Roberts could find was P. Oxy 843, a manuscript comprising about half a roll of Plato’s Symposium (200B-223D; twenty-five relatively complete columns are preserved; see plate five).[64] The script is slightly slanting, bilinear, and lacking in ornamentation; the writing is thick/heavy and its general appearance is similar to P64, although more spacious and a somewhat more literary production.[65] In terms of individual letters, there are numerous distinct similarities (e.g. gamma, epsilon, kappa, nu, pi, tau, upsilon) and some differences (e.g. alpha, theta and omicron less rounded than in P64).

             Schubart suggested that this script represented a transitional position between the earlier strenger Stil (‘severe style’, or ‘formal mixed’ in Turner’s terminology) and the Bibelstil (‘biblical majuscule’) which became clearly distinguishable in the third century.[66] The narrower oval shapes of letters such as eta, theta, omicron, sigma (characteristic of the severe or formal mixed style[67]) in P. Oxy 843 distinguish it from P64, which has the more rounded fuller versions characteristic of the biblical majuscule.[68] We should note that there is no external evidence by which P. Oxy 843 can be dated;[69] this has obvious implications for trying to allocate a date to P64.

             Roberts appealed to two other manuscripts, the style of both of which might be regarded as somewhat similar. We have reproduced a plate of the first of these: P. Oxy 1620, a fragment of Thucydides (1.11-14; see plate six).[70] This manuscript contains a compact upright script of generally similar impression and a range of very similar letter-shapes (e.g. alpha, eta, kappa, lamda, nu, pi, tau, upsilon). The major differences include narrower forms of epsilon and sigma (cf. severe style) and a rounded, flatter omega. The presence of cursive marginalia offers an independent palaeographical confirmation of a date around AD 200 (‘late second or early third century’).[71] P. Oxy 1819, a collection of fragments from a roll of Homer’s Odyssey (containing x-xii),[72] consists of closely written lines of small upright uncials. Roberts noted that ‘a and u are similar, m different and the hand in general is lighter and freer’.[73]

             The overall impression of this hand is, however, distinctly different from P64, with a more tightly-packed and literary script (including accents, breathings and elision marks), squarer and occasionally decorated letter-formation, and specific differences in numerous letters (e.g. beta, delta, kappa, mu, pi).

             We shall not, therefore, take further account of this particular manuscript. On the other hand, scripts similar to P. Oxy 1620 and to P64 have consistently been dated to a period around AD 200. Examples include: P. Oxy 2256 (fragments of a commentary on Aeschylus) ascribed ‘to the later part of the second or the earlier part of the third century’;[74] and P. Oxy 2516 (fragments of Antimachus) assigned ‘to the second century’.[75]

             Notwithstanding the experience and expertise of the editors who have assigned these dates, we need to note carefully that, with the partial exception of P. Oxy 1620, there is no external evidence for the dating of the manuscripts surveyed up to this point.

             The final manuscript appealed to by Roberts does come with a certain terminus post quem. P. Oxy 405 is a fragment of Irenaeus’ Adversus Haereses which was originally, before the text was identified, thought to date to the later part of the second century or the early part of the third (see plate seven).[76] Since Irenaeus wrote this work in Lyons around AD 180, this may represent an early copy which made its way to Oxyrhynchus.[77] The general appearance of the script and some letter forms are, as Roberts suggested, quite similar to P64 and both exhibit the use of nomina sacra, but we should also note that P. Oxy 405 is more regular, rounded and literary in style.

             Further datable manuscripts emerged in Roca-Puig’s discussion of the P64 and P67. It is noteworthy that with a larger amount of text available, the similarities of P64+67 with other scripts exhibiting the biblical majuscule style emerged. In addition to several manuscripts previously mentioned, Roca-Puig appealed for comparison to two of the earliest datable examples of ‘biblical majuscule’. These constitute an important base for comparison, although they both represent more ambitious literary productions that P64, because they confirm the general dating already arrived at by Roberts.

             P. Oxy 661 (a portion of poetry customarily attributed to Callimachus) has writing on the verso ‘in a cursive hand which is not later than the beginning of the third century’.[78] This places the main text late in the second century (see plate eight).[79] Comparison with P64 reveals not only a generally similar appearance, especially in the closely packed writing in the right hand column, but also numerous similarly constructed letters (e.g. delta, eta, kappa, lamda, mu, nu, pi, rho). Other letters, such as epsilon, theta, omicron and sigma are more regular and rounded in P. Oxy 661, in keeping with a more stylish production generally.

             Roca-Puig also appealed to P. Dura 2, two small fragments of Appian found at Dura-Europos and definitely written before AD 256. These are too small to reproduce here, as they exhibit only a few letters in a rounded biblical majuscule.[80] Two further examples of approximately datable biblical majuscule might also be adduced, which go a long way to confirm that the biblical majuscule style was being used across a broad geographical range in the period around AD 200 (plus or minus fifty years). P. Ryl. 16, a fragment of a broad-margined edition of a comedy, the verso of which was reused in a letter from Syrus to Heroninus in AD 256. Such a sumptuous edition was probably not rapidly destroyed for scrap, hence a date around AD 200 is offered by the editor.[81]

             P. Oxy 2832 consists of two fragments from an oracle book of Astrampsychus, the verso of which was reused for a private letter in the latter half of the third century.[82] This suggests an early or mid third-century date for the original script which is somewhat similar to P64+67.

             In comparison with these datable examples, P64+67 appears somewhat less bookish, regular and literary. It is tempting to say that P64 is therefore earlier than these other texts, but the development of any style is not as straightforward as this, and the most that can really be concluded is that P64 stands within the general area of AD 200 and apparently attests a transition between older literary styles and the developing biblical majuscule.

             Numerous other examples of early biblical majuscule exist, several of which are broadly comparable with P64. We note P. Berol. 7499 (dated by Schubart to the third century; Roberts argued that our hand is less uniform and regular, and is ‘demonstrably earlier’);[83] P. Oxy 224 & P. Ryl 547 (‘Later second century’);[84] P. Oxy 1179 (‘Early third century’);[85] P. Oxy 2356 (‘late second century’);[86] P. Oxy 2364 (no date assigned by editor);[87] P. Oxy 2750 (‘around the latter part of the second century A.D)’.[88] Plates of these are readily accessible and the interested reader is invited to compare them with those provided in this article (such a reader will certainly find these texts to provide far closer parallels than those claimed by Thiede which we examined earlier). One example of this type of material, which exhibits numerous reasonably close parallels with P64, can be reproduced here: P. Oxy 2498, a small fragment of verse attributed to Hesiod (see plate nine).[89] The editor, E. Lobel, suggested ‘this appears to be a late second-century hand.’ Note both the general appearance (closely written, bilinear, with slight variation between wide and narrow strokes) and the letter-forms (especially alpha, beta, delta, kappa, mu, nu, pi, upsilon). The resemblance to P64 (and even more so with P67) is undoubtedly significant.

             In concluding this section, we note that many of the parallel scripts claimed by Roberts and Roca-Puig are indeed far closer to P64 than those recently claimed by Thiede. Comparison with datable material supports a date within fifty years of AD 200 and there seems little point in attempting any more precision than that in a document without detailed provenance.

VI. Other Features of P. Magd. Gr. 17

Several other features of the manuscript are also more consistent with a date within the range suggested above than with a date in the first century. In terms of the general shape of the codex, it is possible to calculate, from the extant remains which average around 16 letters per line (with some variation from 14 to 18/19), that the page must have two columns of around 36 lines (plus or minus 2).[90] A reconstruction of the page represented by P64 has 2,276 letters from the oti of 26.2 to the eipen of 26.33 (two sides of one leaf). This represents 2.6% of Matthew’s text and suggests a length for the whole gospel of 39 or 40 folios.[91]

             Two-column pages are relatively rare in papyrus codices (compared with parchment-vellum codices where they are common). Turner provides twenty examples from both secular and Christian literature ranging from the second to the seventh century, but predominantly from the second to the fourth centuries.[92] In addition in terms of overall page size Turner classifies P64+67 in his Group 9 (broadly square with a breadth of 16-13 cm.; the proportion of breadth to height of around 7:8, varying at times to as much as 2:3); he further argues that papyrus codices of this type offer the earliest examples (along with his group 8: height double the breadth) and probably represent ‘the earliest format of the papyrus codex’, with numerous representatives from the third century AD and some perhaps from the second century.[93]

             We might also note at this stage the presence of nomina sacra in P64 (no evidence of upper lines in P64, perhaps obscured or abraded) which places this clearly in the Christian manuscript tradition (cf. Ep. Barn. IX.8). Although Thiede is quite correct to point out that Roberts hypothesised that the use of such abbreviations may have originated in the Jerusalem church before AD 70, the fact remains that all the examples offered by Roberts are from the middle to late second century and Egyptian provenance.[94]

             A further important factor is the use in P64 of abbreviated symbols to represent numbers (frag. 3 verso line 2: ib for dwdeka). This is not found in either the Greek literary manuscript tradition or in Jewish manuscripts of the Greek Old Testament (where numbers were written in full), but it is characteristic of early Christian manuscripts from Egypt.[95]

             Both of these attributes of P64 prove a Christian scribal provenance for our manuscript. Since the manuscript was purchased in Luxor in 1901, it probably came from a centre of Christian activity in ancient Egypt outside of Alexandria. No such centre is known in the first century, but any number of possibilities exist from the second century; if not Luxor itself then Arsinoe, Qarara, Oxyrhynchus, Antinoopolis, or Coptos might all be candidates, since second-century Christian texts have been found in these places.[96]

             It might fairly be said that all of the features discussed in this section suit a date of around AD 200 far more comfortably than a first-century date. That is not to say that any of these features render an early date impossible; if an early date was sufficiently well-established, the manuscript would constitute evidence which would add immeasurably to our understanding of a whole range of subjects. But we have already seen that the arguments presented for the early date are unpersuasive; all of the evidence points to the later date.

VII. Conclusion

We agree with Thiede when he wrote ‘Caution is always the best approach in the dating of manuscripts’.[97] In this article an attempt has been made both to hear and to critically investigate his claims regarding the date of P. Magd. Gr. 17 = P64. Although we recognise the service that he has performed in facilitating a reexamination of methodological presuppositions, our verdict on his claims is a negative one. The very early manuscripts to which Thiede appealed for close parallels to P64 turned out to be not as close as the somewhat later ones which he had overlooked. Although there is no absolutely definite evidence by which P. Magd. Gr. 17 = P 64 can be dated with certainty, the available evidence points to a date around AD 200. To be on the safe side I would suggest plus or minus fifty years as the possible range.


HOME



[1]An earlier form of this paper was presented to the Early Christian and Jewish Studies Seminar, chaired by Dr. W. Horbury in Cambridge on May 5th 1995 whose members are thanked for their helpful discussion.  Thanks are also due to the Librarian, Magdalen College, Oxford, Dr. C. Y. Ferdinand, for allowing access to the manuscript on two occasions, and to Professor H. Maehler of London and Dr. R.A. Coles of Oxford for reading this paper.

[2]C.P. Thiede, ‘Papyrus Magdalen Greek 17 (Gregory-Aland P64): A Reappraisal’, TynB 46 (1995) 29-42 (the editors promised a response on 40). Thiede’s article was originally published in ZPE 105 (1995) 13-20.

[3]Note that K. Aland (ed.), Kurzgefasste Liste der griechischen Handschriften des Neuen Testaments (ANTT 1, 2nd ed.; Berlin: de Gruyter, 1994) takes the numbering up to P99.

[4]Gregory-Aland 461 (a minuscule); for brief discussion and a plate, see B.M. Metzger, Manuscripts of the Greek Bible: An Introduction to Greek Palaeography (Oxford: OUP, 1981) No. 26 (102).

[5]For a list of dated manuscripts, beginning with Vindob. med. gr. 1 (AD 512), see R. Devreesse, Introduction à l’étude des manuscrits grecs (Paris, 1954) 286-320.

[6]M. D’Ancona, ‘Eyewitness to Christ’, The Times 24/12/94 Weekend Section, 1 & 3.

[7]Magd. MS. 845.

[8]In the editio princeps (see below) Roberts suggested the latter, but the former seems more likely to this writer.

[9]Magd. MS. NL/8/6 (a copy of which is kept with MS 845). In a letter to Wilson dated Dec. 5th 1901, Huleatt enquires about the safe arrival of the fragments and draws attention to the use of i8b8 rather than dwdeka (Frag 3, Verso, line 2).

[10]Colin H. Roberts, ‘An Early Papyrus of the First Gospel’, HTR 46 (1953) 233-37.

[11]Roberts also noted the double column format of the codex (‘15-16 letters to a line and approximately 35-36 lines to a column’, 233), suggested that the codex would contain about 150 pages (234). and noted both the nomina sacra and the system of paragraph division (234).

[12]Roberts, ‘An Early Papyrus’, 235.

[13]For Roberts’ own articulation of the method, see C.H. Roberts, Greek Literary Hands 350 B.C.–A.D. 400 (Oxford: Clarendon, 1955) xi-xvi; cf. further, E.G. Turner, Greek Manuscripts of the Ancient World. Second Edition Revised and Enlarged (ed. P.J. Parsons; ICS Bull. Supp. 46; London: ICS, 1987) 18-23.

[14]Roberts, ‘An Early Papyrus’, 237. He also noted similarities with P. Oxy 405 (235-36).

[15]The manuscript was designated P64 by K. Aland; cf. his Kurzgefasste Liste.

[16]First published in booklet form: Ramon Roca-Puig, Un Papiro Griego del Evangelio de San Mateo (Barcelona, 1956). Also published in Ramon Roca-Puig, ‘P. Barc. Inv. N. 1 (Mt. III, 9, 15; V, 20-22, 25-28’, Studi in onore di Aristide Calderini e Roberto Paribeni (3 vols; Milan, 1956/57) vol. 2, 87-96 (cited here); R. Roca-Puig, ‘Un pergamino griego del Evangelio de San Mateo’, Emérita 27 (1959) 59-73. Further publications (after the identification of P. Barc. Inv. 1 with P. Magd. Gr. 17) include: R. Roca-Puig, ‘Nueva publicación del papiro numero uno de Barcelona’, Helmantica 37 (1961) 5-20; C.H. Roberts, ‘Complementary Note to the article of Prof. Roca-Puig’, Helmantica 37 (1961) 21-22; R. Roca-Puig, Un Papir Grec de l’Evangeli de Sant Mateu (Barcelona, 1962); this edition (cited here) incorporates C.H. Roberts, ‘Complementary Note’. Cf. also S. Bartina, ‘Another New Testament Papyrus (P67)’, CBQ 20 (1958) 290-91.

[17]R. Roca-Puig, ‘P. Barc. Inv. N. 1’, 87-91. These features include the physical shape of the manuscript: an average of 16 letters per line (from 14 to 20) with perhaps 38/39 lines per column which suggested either a rather odd-shaped narrow page with one column or a double column page between 18-20 cm. high and 12/13 cm. wide; the punctuation techniques and paragraphing, and the nomina sacra.

[18]R. Roca-Puig, ‘P. Barc. Inv. N. 1’, 91-93 (in this conclusion the physical form of the manuscript also played a part; cf. our discussion later). He also noted a few variants and published an edition of the text (p 93-96). This manuscript was given the number P67 by K. Aland; cf. his Kurzgefasste Liste .

[19]Roberts, ‘Complementary Note’; for Roca-Puig’s agreement, see Un Papir Grec, 50.

[20]E. Bammel, ‘P64(67) and the Last Supper’, JTS 24 (1973) 189. Bammel described P64 as ‘the oldest implicit commentary of the early church’, arguing that since the paragraphos occurred at Mt. 26.31, v.30 was linked with the preceding section, suggesting that the psalms mentioned therein were Hallel psalms sung at the end of the passover feast.

[21]E.G. Turner, The Typology of the Early Codex (University of Pennsylvania Press, 1977). Turner’s discussions of P64 and P67 relate to both its external dimensions (25) and its dual column internal arrangement (36). We shall return to some of these features later.

[22]P.M. Head, ‘Observations on Early Papyri of the Synoptic Gospels, especially on the “Scribal Habits”‘ Biblica 71 (1990) 240-47. In this case the only pure singular reading is the curious spelling error of galeglaian in Mt. 26.32; the text otherwise exhibits little variation from that represented in NA26=27.

[23]First raised in K. Aland, ‘Neue Testamentliche Papyri II’, NTS 12 (1965/66) 193-195 (following a suggestion by P. Weigandt); regarded as probable by J. van Haelst, Catalogue des Papyrus Littéraires Juifs et Chrétiens (Série Papyrologie 1; Paris: Sorbonne, 1976) 126; affirmed in C.H. Roberts, Manuscript, Society and Belief in Early Christian Egypt (London: BM/OUP, 1979) 13; cf. also C.H. Roberts & T.C. Skeat, The Birth of the Codex (London: BM/OUP, 1983) 40-41.

[24]For measurements, see Turner, Typology, 144 and 149. Otherwise see most recently P.W. Comfort, ‘Exploring the Common Identification of Three New Testament Manuscripts: P4, P64 and P67’, TynB 46 (1995) 43-54.

[25]For these points see C.P. Thiede, ‘Notes on P4 = Bibliothèque Nationale Paris, Supplementum Graece 1120/5’, TynB 46 (1995) 55-57. Scholars wishing to pursue the significance of the similarities between these manuscripts would be well advised to take account of the Philo manuscript (Paris B.N. Suppl. gr. 1120) in which P4 was found. In style and construction, if not in script, this is also very similar (2 columns, 35-41 lines per column, similar dimensions; see Turner, Typology, 113, #244). Some facsimiles can be found in Mémoires publiés par les membres de la Mission Archéologique Française au Caire (Cairo, 1893) no page numbers.

[26]The Times (24/12/94) 1.

[27]D’Ancona, ‘Eyewitness to Christ’, 1 and 3.

[28]Cf. C.P. Thiede, The Earliest Gospel Manuscript? The Qumran Fragment 7Q5 and its Significance for New Testament Studies (Exeter: Paternoster, 1992).

[29]Notable contributors include: G.N. Stanton (The Times 29/12/94 with three other letters) to which Thiede responded (The Times 2/1/95 with four other letters); and J.N. Birdsall (Church Times 6/1/95 cf. also, more briefly, The Times 13/1/95) to which Thiede responded (Church Times 13/1/95). Time Magazine’s generally well-balanced coverage (by R.N. Ostling in 23/1/9, 57) developed previous accusations that biblical scholars were not giving credence to evidence which would undercut popular theories (referring to H. Riesenfeld as fearful ‘that Bible experts would ignore the evidence because they are consumed by literary theories and shun philology and history’).

[30]Thiede, ‘Reappraisal’, 36-37.

[31]E.g. D’Ancona: ‘the mid-first century AD.’ (‘Eyewitness to Christ’, 1); G. Stanton: ‘a mid-first century dating’ (Letter to the Times, 29/12/94).

[32]Cf. Thiede, ‘Reappraisal’, 38: ‘a Christian codex fragment of the first century, perhaps (though not necessarily) predating A.D. 70’.

[33]Thiede, ‘Reappraisal’, 37, 40, 40 (respectively).

[34]Quoted from correspondence to the present writer (20th Jan. 1995). In a public lecture for the Hellenic Institute in London on Jan. 28th, Thiede referred repeatedly to ‘a first-century date’.

[35]In what follows we shall generally refer to this manuscript (P. Magd. Gr. 17) by the simpler designation: P64.

[36]Thiede, ‘Reappraisal’, 32-33; cf. his transcript on 41-42.

[37]The transcript reflects the layout of the fragments on the plate; we have placed the fragments in textual order and identified their location in the two columns of their (hypothetically reconstructed) page.

[38]First line: only lower end of letters visible but iota fits with extension below line. The nomen sacrum ke is plausible, but not guaranteed. Note: thn kefalhn is equally possible.

[39]Second line: I8B8 = dwdeka. Roberts suggests [o] after beta (cf. editions) but, as Thiede argued, there is insufficient space available. Third line: traces of the first iota are evident.

[40]First line: Thiede: ...au]t?[o]i?[v ti. But there is little trace of an iota. Second line: Roberts notes a trace of upsilon and rho. Third line: an upsilon is plausible on basis of the extension below line; there is no trace of rho at the end of line.

[41]As noted, in the first line Thiede reads ... au]t?w?n?[ mhti (with P45 et al.). But there is evidence of a punctuation mark immediately between omega and mu, the down-strokes are also wider than other nus, and the remains of an upper diagonal are visible. Hence we support Roberts’ reconstruction here. Second line: there is no line visible above nomen sacrum; the letters are very obscure. Fifth line: predominantly guess work here.

[42]First line: Thiede omits umeiv on stichometric grounds. But this is inconclusive: i) calculations about what was not written outside the actual fragment can hardly ever be secure, since scribal errors can take so many forms. ii) Line lengths in P64 vary from 14-18 (average 16), but Thiede's reconstruction would be twenty letters (P67 does range from 13-20). In addition we should note the out-dented alpha and the presence of four iotas in this line (most of the other long lines in both P64 and P67 have numerous iotas). Fourth line: traces are compatible with gammas but only the epsilon is relatively clear.

[43]First line: lower traces of letters are compatible with those given (although Roberts initially omitted the whole line). Second line: The spelling is clear (this was misprinted as geleglaian in Thiede, ‘Reappraisal’, 42) although quite unusual and perhaps purely a mechanical error (cross bar of gamma) or an optical error of a scribe.

[44]E. Tov, The Greek Minor Prophets Scroll from Nahal Hever (8HevXIIgr) (The Seiyâl Collection I) (Discoveries in the Judean Desert VIII; Oxford: Clarendon, 1990). For a discussion of date, see P.J. Parsons, ‘The Scripts and Their Date’, 19-26; cf. Thiede, ‘Reappraisal’, 35. The terminus ante quem is c. AD 135 (Parsons, 22).

[45]Thiede, ‘Reappraisal’, 36.

[46]The same column is reproduced in W.H. Schmidt, W. Thiel & R. Hanhart, Altes Testament (Grundkurz Theologie 1; Stuttgart et al.: W. Kohlhammer, 1989) 216 (cf. Thiede, ‘Reappraisal’, 35 n. 26) and plate XIX in Tov, The Greek Minor Prophets Scroll .

[47]For discussion of the two scribes see Tov, The Greek Minor Prophets Scroll 12-14 (Tov), and 19-22 (Parsons).

[48]Note also the more obvious different features: larger letters appear at beginnings of lines, the leather scrolls, the Jewish provenance evidenced by the tetragrammaton in archaic script, as well as its archaeological discovery in Nahal Hever.

[49]Thiede claims that differences between the scripts (w.r.t. etas and mus) are alleviated by the fact that ‘the second scribe of the Nahal Hever scroll provides the comparable eta and mu more than once’ (‘Reappraisal’, 36). This is unsatisfactory. One would not want to deny that occasional uncharacteristic letter-forms are similar (this could no doubt be found in thousands of manuscripts of widely divergent date and provenance), but the argument effectively grants that the characteristic letter-forms are different. There is therefore no strong basis in terms of a palaeographical comparison for the foundation of Thiede’s argument. For a plate of Scribe A, see E. Würthwein, The Text of the Old Testament: An Introduction to the Biblia Hebraica (ET; London: SCM, 1980) 181 (Plate 30): from col. 17 = Tov, The Greek Minor Prophets Scroll, plate XI.

[50]P.W. Skehan, E. Ulrich & J.E. Sanderson, Qumran Cave 4. IV: Palaeo-Hebrew and Greek Biblical Manuscripts (DJD IX; Oxford: Clarendon, 1992). NB. Plate XL (= Lev. 4 & 5 [partial]).

[51]Thiede, ‘Reappraisal’, 36.

[52]For these descriptions, cf. P.W. Parsons, ‘The Palaeography and Date of the Greek Manuscripts’, in Palaeo-Hebrew and Greek Biblical Manuscripts 7-13, 10.

[53]Thiede, ‘Reappraisal’, 36. For 7Q61, see M. Baillet, J.T. Milik & R. de Vaux, Les ‘Petites Grottes’ de Qumrân: Exploration de la falaise, Les grottes 2Q, 3Q, 5Q, 6Q, 7Q à 10Q, Le rouleau de cuivre (DJD III; Oxford: Clarendon, 1962; 2 parts) Part One, 145; Part Two, Plate XXX.

[54]There is, in any case, no ‘characteristic eta’ in P64, as a glance at the plate will show: the horizontal stroke can occur medially (e.g. Frag 1, recto, line 3), above the median (e.g. Frag 3, recto, line 3) or below the median (e.g. Frag 2, recto, lines 2 & 3; Frag 3, verso, line 4).

[55]Cavallo, the editor of the standard accessible collection of plates, identifies eighteen different stylistic groups! See G. Cavallo, Libri scritture scribe a Ercolano: Introduzione allo studio dei materiali greci (First Supplement to Cronache Ercolanesi volume 13; ed. G. Macchiaroli; Naples, 1983) 28-44.

[56]For a good general introduction see I. Gallo, Greek and Latin Papyrology (Classical Handbook 1; ET; London: Institute of Classical Studies, 1986) 36-45 (approximately 1,000 papyri have not yet been unrolled).

[57]Parsons, ‘The Scripts and Their Date’, in Tov, The Greek Minor Prophets Scroll, 24, appeals to P. Herc 182 (Libri scritture scribe a Ercolano:  pl. 18), P. Herc 1005 (Libri scritture scribe a Ercolano: pl. 22), P. Herc 1186 (Libri scritture scribe a Ercolano:  pl. 33), P. Herc 1471 (Libri scritture scribe a Ercolano:  pl. 36), P. Herc 1423 (Libri scritture scribe a Ercolano:  pl. 50) and P. Herc 1507 (Libri scritture scribe a Ercolano:  pl. 53), all of which are dated by Cavallo in the first century BC. Certainly none of these offer close parallels with P64, indeed in many respects the scripts, although ‘bilinear serifed hands’ (Parsons), are quite different from the finds from the Judean desert (as one might expect considering the Herculaneum papyri come from the library of an Epicurean philosopher named Philodemus who died c. 40 BC).

[58]W. Schubart, Griechische Palaeographie (Handbuch der Altertums-wissenschaft I.4; Munich: C.H. Beck, 1925) 111ff. Schubart argued that this type of hand passed out of use around AD 100 (112). More recently Turner has questioned whether the mere presence of serifs or decorated forms was sufficient to justify classifying a distinct decorated style within the broader category of ‘Formal round’ (noting the presence of serifs in hands as early as the fourth century BC and as late as AD 200-250), (Greek Manuscripts, 21; he refers further to G. Menci, ‘Scritture greche librarie con apici ornamentali (III a.C. - II d.C.)’, Scrittura e Civiltà 3 (1979) 23-53 and tav. i-x). Cf. also Thiede, Earliest Gospel Manuscript, 24-25 for an earlier discussion in connection with 7Q5. For an Egyptian example from the first century BC, see P. Fouad 266 (Deuteronomy) [Greek Manuscripts, No. 56, 96-97; also Würthwein, Text, 178-79, Pl. 29; or Metzger, Manuscripts of the Greek Bible, Pl. 3, 60-61]).

[59]Thiede’s appeal to archaeological termini ante quem for the Judean and Herculaneum examples is somewhat disingenuous. Since the books at Herculaneum belonged to Philodemus himself, and do not include any author later than his death (40 BC), many scholars take that date as the effective terminus ante quem for the vast majority of the Herculaneum papyri (cf. Turner, Greek Manuscripts, No. 78, 134; Parsons, ‘The Scripts and Their Date’, 24; Gallo, Greek and Latin Papyrology, 37; Cavallo, Libri scritture scribe a Ercolano: , 50-56). In relation to the Judean finds, although the dates of AD 70 or 135 provide helpful end-points, the dating of the manuscripts tends to fall much earlier, with considerable debate about whether they might be closer to 50 BC than AD 50.

[60]Published in Skehan et al., Palaeo-Hebrew and Greek Biblical Manuscripts, 161-165 and plate XXXVIII.

[61]Thiede, ‘Reappraisal’, 37.

[62]Most scholars have dated 4QLXXLeva sometime in the first century before Christ. So Parsons, ‘Paleography and Date’, 10: ‘the general impression is of a script earlier than that of the Greek Minor Prophets Scroll’ (i.e. 8HevXIIgr); he also cites Roberts in favour of a first century BC (or possibly second century BC) date.

[63]For example, as we noted in Section II, Roberts compared P64 with P. Berol. 7499, P. Oxy 843, 1620, 1819, and Roca-Puig appealed to P. Oxy 1179, POxy 661, P. Dura 2, P. Oxy 405.

[64]P. Cairo 41082, published in OxyPap V(1908) 243-92 and Plate VI (a portion of the same column, the final one, is presented in Schubart, Griechische Palaeographie, Abb. 88).

[65]The introduction by Grenfell and Hunt draws attention to punctuation (single high points, double dots to identify dialogue changes, marginal paragraphus, diaeresis and other lectional signs and corrections from a secondary hand), OxyPap V(1908) 243.

[66]Schubart, Griechische Palaeographie, 132.

[67]See Turner, Greek Manuscripts, 22. P. London inv. 733 (British Museum Bacchylides papyrus) is the best example or paradigm of this style (cf. Schubart, Griechische Palaeographie, 126-129; cf. Abb. 85 for plate). Cf. F.G. Kenyon, The Poems of Bacchylides. Facsimile of Papyrus DCCXXXIII in the British Museum (London: BM, 1897). Cf. also P. Oxy 1174 (Sophocles, Ichneutae = B.M. Pap. 2068, Greek Manuscripts, 34; OxyPap IX (1912) plate II) a late second century example (cursive marginalia confirm such a date) with numerous similarities to P64 (as also, by the same scribe, P. Oxy 1175; OxyPap IX (1912) plates III & IV); P. Oxy 2365 (third century).

[68]The main characteristics of biblical majuscule are: ‘a preference for geometric forms; letters can be fitted into squares (the only exceptions being I, R, F, Y, W); a contrast in thickness between compact vertical strokes, thin horizontal and ascending strokes, descending diagonals of medium thickness…; absence of decorative crowning dashes or ornamental hooks’ (G. Cavallo & H. Maehler, Greek Bookhands of the early Byzantine Period (AD 300–800) (ICS Bull. Suppl. 47; London: ICS, 1987) 34. This description is based on the detailed study of the development of this type of script by G. Cavallo, Ricerche sulla maiuscola biblica (Studia e testi di papirologia 2; Firenze: Le Monnier, 1967). He attempted to trace the development of this type of hand from its origins in the second century AD through its classic forms and into its degenerative stages. Of over 130 manuscripts surveyed in his book only thirteen are datable on other than palaeographical grounds. Cavallo’s technique is to order the datable material and then trace the development of the hand through the other examples.

[69]Grenfell and Hunt date it ‘from about the year 200 A.D.’ (OxyPap V (1908) 243); Roberts notes the presence of a flat omega, a third century characteristic (‘An Early Papyrus’, 235).

[70]Two columns from a roll now in the University of Melbourne, published in OxyPap XIII (1919) 190 and Plate VI.

[71]Cf. OxyPap XIII (1919) 189; having compared the notes to those in P. Oxy 1234, the editor wrote: ‘the main text may therefore well be ascribed to the reign of Commodus [i.e. 180-192] or even M. Aurelius [i.e. 161-180]’ (190).

[72]Johns Hopkins University Library. Noted, but not fully published in OxyPap XV (1922) 224-25; a facsimile of fragment 2 (2 columns representing Odyssey xi.244-283, 284-323) published in New Palaeographical Society Series II. Plate 76.

[73]Roberts, ‘Early Papyrus’, 235-36.

[74]OxyPap XX (1952) 29 (cf. plates V or VI, or Greek Manuscripts Plate 25).

[75]OxyPap XXX (1964) 20 (cf. plate IV, or Cavallo, Ricerche, plate 10).

[76]OxyPap III (1903) 10-11 and plate 1 (P. Oxy 405 = Cambridge University Library MS Add. 4413). Grenfell & Hunt edited it, as an unidentified theological work; the manuscript was subsequently identified as Irenaeus, Adv. Haer. iii.9 by J.A. Robinson in Athenaeum (Oct. 24, 1903), noted and re-edited in OxyPap IV (1904) 264-65.

[77]For a recent brief discussion of the date of the work, see D.J. Unger & J.J. Dillon, St. Irenaeus of Lyons Against the Heresies (Ancient Christian Writers No. 55; NY: Paulist, 1992) 3-4. Cf. Roberts’ later comment that the treatise must have reached Oxyrhynchus ‘not long after the ink was dry on the author’s manuscript’, C.H. Roberts, Manuscript, 53; see also his ‘Early Christianity in Egypt: Three Notes’, Journal of Egyptian Archaeology 40 (1952) 94.

[78]OxyPap IV (1904) 63.

[79]Cf. also Roberts, Greek Literary Hands, 16a.

[80]Cf. Roberts, Greek Literary Hands, 16b; C.B. Welles, R.O. Fink, J.F. Gilliam, The Excavations at Dura-Europos. Final Reports V. Part 1. The Parchments and the Papyri. (New Haven, 1959) 70-71. For a recent plate and discussion (following new conservation work), see R.G. Babcock & W.A. Johnson, ‘The Appian Papyrus from Dura-Europus (P. Dura 2)’, Bulletin of the American Society of Papyrologists 31 (1994) 85-88. This includes a convincing re-attribution of both fragments to the same manuscript of Appian’s Iberica.

[81]Catalogue of Greek Papyri in the John Rylands Library, Manchester (vol. 1; ed. A.S. Hunt; Manchester, 1911) 25-26 & plate 5 (cf. also Cavallo, Ricerche, plate 22). The verso is P. Ryl. 236, the letter refers to the third year, which is known from other letters of Syrus (e.g. P. Flor. 241-258) to be the third year of Gallienus and thus AD 255/6; see further Catalogue of Greek Papyri in the John Rylands Library, Manchester (vol. 2; eds. J. de M. Johnson, V. Martin & A.S. Hunt; Manchester, 1915) 385-86.

[82]OxyPap XXXVIII (1971) 30-31, and plate IV.

[83]P. Berol. 7499 (Homer, Illiad, VIII.306-324; cf. Berliner Klassikertexte V.1 [1907] 4). A plate appeared in Schubart, Griechische Palaeographie, Abb. 93.

[84]P. Oxy 224 [OxyPap II (1899) 114-116 no plate] = P. Flor 76 (Euripides, Phoenissae, lines 1017-1=43, 1064-1071). When published Grenfell and Hunt noted both that it was found with other documents from pre AD 300 and that ‘the evidence is at present all against assigning this style of uncial to an earlier date than the third century’. P. Ryl. 547 is another piece of the same role (Euripides, Phoenissae, lines 646-57); by 1938 when it was published sufficient early examples of the biblical majuscule had been found to enable the earlier date noted above, see Catalogue of Greek Papyri in the John Rylands Library, Manchester (vol. 3; ed. C.H. Roberts; Manchester, 1938) 195-95 and plate 9.

[85]OxyPapIX (1912) 186 and plate 1 (Apollonius of Rhodes, Argonautica, ii.101-110).

[86]OxyPap XXIII (1956) 5-7 and plate III (Archilochus, Elegiacs).

[87]OxyPap XXIII (1956) 30-38 and plate VII (Choral Lyric perhaps Bacchylides) (the editor refers to P. Berol. 16139 as containing part of the same fragment and P. Ashmole inv. 20 as written by the same copyist).

[88]OxyPap XXXVI (1970) 10; cf. plate III (part of a column of a roll of Xenophon, Cyropaedia, I.1).

[89]OxyPap XXVIII (1962) 70-71; cf. plate in Cavallo, Ricerche, 10a (attributed to Hesiod’s Catalogue or (after Pausanias) Mega/lai70Hoi=ai).

[90]In all of this P64 coheres extremely closely with P67 (line lengths from 13-20 letters), two columns necessary (by deduction), approximately 37 lines per column.

[91]A quick check of the editions of Westcott and Hort (no apparatus) confirms that this section represents 1.75 pages of a 70 page text (i.e. one fortieth); as does a check of NA27 (2.2 pages of an 87 page edition) (i.e. one fortieth). Thus I suggest a figure of around 40 folios for Matthew’s Gospel. Roberts (‘An Early Papyrus of the First Gospel’, 234) suggested 150 pages for Matthew. This is impossible to reconcile with his figures (as ours) of 15-16 letters per line, 35-36 lines per column and two columns per page. His later estimate of 90 pages (‘Complementary Note’, 63) is closer to the mark but still a little on the long side.

[92]Turner, Typology, 36, Table 3 (cf. also his comments on 35-36).

[93]See Typology, 21-22 for a list, and 25 for discussion. It is notable that both P4 (III) and Paris BN Suppl. gr. 1120. Philo (III) (Turner No. 244) also have dual column pages of similar size. Other Christian manuscripts of similar size and construction include: P. Baden iv.56. Exodus (II) (Turner No. OT24); P. Chest. Beatty VI. Numbers v.12 - Deut 1.20 etc. (II/III) (Turner No. OT36 NB. area of writing); P. Egerton 3. Origen? (III) (Turner No. 553).

[94]Roberts, Manuscript, 46.

[95]Roberts, Manuscript, 18-19, and Turner, Greek Manuscripts, 15. Roberts argues that the presence of such abbreviated symbols in later non-Egyptian manuscripts such as Vaticanus and Sinaiticus suggests that the early papyri represented a general practice among early Christian books. This may be the case but Roberts’ tendency to attribute such conventions to apostolic instruction from the Jerusalem church overlooks the fact that similar abbreviations were, in Turner’s words, ‘common in documentary papyri’ (i.e. from Egypt).

[96]Roberts, Manuscript, 6; cf. also his ‘Early Christianity in Egypt: Three Notes’, Journal of Egyptian Archaeology 40 (1952) 92-96. Quite likely the Catechetical School of Alexandria influenced most of the peculiar conventions found in extant Christian manuscripts. This may have established its influence by the middle of the second century and influenced the transmission of Christian texts throughout Egypt as well as in other parts of the Empire (so especially G. Zuntz, The Text of the Epistles: A Disquisition upon the Corpus Paulinum (Schwich Lecture, 1946; London: OUP, 1953) 271-76; Roberts, Manuscript, 24; cf. Eusebius, HE, V.10.1: the school existed before Pantaenus (c. 180 - 190?)).

[97]Thiede, Earliest Gospel Manuscript, 22.