Media Papyri: Examining Carsten Thiede's Rediscovered Fragments by Sigrid Peterson, PhD University of Pennsylvania May 14, 1995 (Revised September 1995) INTRODUCTION: METHOD AND METHODOLOGY METHOD There is so much inaccessible and unpublished papyrus in various corners of the world, almost entirely from Egypt, that the great urgency within the field of papyrology is to publish the pieces and thus make them available to the scholarly world. For this reason, as well as others, a difference exists between the kind of close scrutiny of methodology common to the study of the historical import of ancient literature, and the close scrutiny of a publication in the field of papyrology. The former focusses on the assessment of historical information, while the latter form of review is concerned with the accurate representation of the physical details of a piece of papyrus or ostrakhan with written words or symbols on it, a document. There are some standards, and there is a desirable descriptive rigor in treating a papyrus or group of papyri. Describing the physical characteristics of the papyri, both the writing and the material, is the conventional way to start. This involves noting the size and determining the side of the sheet which is being used, and whether there are two sides or one which contain writing. For example, if both sides of a sheet are used and the handwriting is the same on both sides, then the document is tentatively identified as a codex. An assessment of the type of document follows, making distinctions, as a rule, between documentary and literary, between business and personal, and between legal and commercial. The papyrologist makes an attempt to locate the place of origin, using all available clues. Such clues may be archaeological context, place references in the text, close comparability in style with documentary papyri whose text fixes the place, the way the material is handled, and other specific characteristics of the piece of papyrus being described. Those visible aspects of the papyrus which correspond to known formats of manuscripts are assessed; these include whether it is a two-column or single- column codex, in the case of P. Magdalen 17, the subject for analysis in this paper. A broad determination of the style of handwriting is the next step, a step which takes advantage of the characteristics already noted. This is a sort of decision tree, as in diagnostic determinations of all kinds. In the case of papyri, the effort is to determine provenance -- place and date for the origin of the physical manuscript. Many of the further points concerning paleographical dating of Greek literary papyri are to be found in succinct summary in Eric G. Turner, in GREEK MANUSCRIPTS OF THE ANCIENT WORLD: SECOND EDITION, Edited by Peter J. Parsons, Bulletin Supplement 46, 1987, London:University of London Institute of Classical Studies, 1987: To obtain a more precise result [in dating]. . . it will be necessary to find a dated or datable handwriting which the piece under examination resembles. . . .Confidence will be strongest when like is compared with like: a documentary hand with another documentary hand, skilful writing with skilful, fast writing with fast. Comparison of book hands with dated documentary hands will be less reliable. . . .[[20]]For book hands a period of 50 years is the least acceptable spread of time [to suggest, as they are long-lived]. A palaeographer familiar with the material will refuse assent to a precise year date allocated to a manuscript simply by comparison with other texts and by no other criterion. How are `resemblances' between handwritings to be judged? The first point of similarity to strike an observer will be in the forms of the letters, but taken in isolation this feature is too arbitrary to be trustworthy. . . .The forms of letters must, as W. Schubart pointed out, be considered in relation to the manner of writing which they help to constitute. The letter-forms chosen by the scribe are, of course, the most important means by which stylization is achieved (Turner, 1987: 19f). In the case of the fragments of P. Magdalen 17, these steps are described in the beginning of Colin Roberts's first publication of the fragments (HTR 46, 1953). A sound basis in method is crucial to the further exercise of critical thought. Thiede's January 1995 article in Zeitschrift fu%r Papyrologie und Epigraphik provides a casebook demonstration of mistaken methods which invalidate his inference about the date of the copy of the Gospel of Matthew represented by the P. Magdalen 17 fragments. First of all, his redating is opposed to a formidable foursome, that of Bell, Skeat, Turner, and Roberts, all of whom agreed with Roberts that these fragments should be redated from the third or fourth centuries to ca. 200 ce. As indicated again later, Thiede does not explain why the judgments of these eminent papyrologists is faulty. Instead, he allows the existence of "new evidence" to carry the weight of his proposed change in dating. However, he does not use it to dissassemble the methodical assessment of evidence in Roberts's first publication, nor does he establish a good case for the relevance of his new evidence to the methodical consideration of these fragments of papyrus. This review will concentrate on understanding these fragments of Matthew as assessed by Roberts, and then by Thiede. It is thus a third-hand approach to the actual data. Stuart Pickering (1995) has noted a more direct approach to redating the Matthew fragments, beginning with Colin Roberts's assessment of the handwriting style as "Biblical uncial" - now called Biblical majuscule. His initial suggestion for dating is "try second half of third century." His implicit method would be to assemble a comprehensive group of Biblical majuscule manuscripts for comparison purposes. These would be placed in a rough sequence according to the development of letters. The method I would suggest would be to assemble a very long sequence of examples of "book hands," whether literary or documentary, looking for several index letters which have a particular identifiable shape at the beginning of the series, and go through a small set of changes as the sequence develops. This is a way of establishing some end points in time, after which a letter never looks a certain way, and similarly a time before which the letter is never -- in this kind of hand -- written in a particular way. The series would have examples of both assigned and internally confirmed dates. It would be necessary to exclude information from Herculaneum; such papyri can be dated firmly as earlier than the eruption of Pompeii in 79 c.e., but their relationship to documentary and literary papyri in other parts of the Mediterranean is not clear. See further the introduction in Turner (1987). An informal trial of this approach showed that the letter epsilon is squared off before 200 bce and after 200 ce, and is rounded during the 400 years in between. Since the epsilon is squared off in most instances in the Matthew fragments, the date of 200 c.e. remains credible by this approach. This method has the advantage of being able to make the strongest "truth" statement possible, in the epistemological sense. It can be demonstrated by the syllogism: If it rains the streets get wet; At this moment the streets are NOT wet; Therefore at this moment it is not raining. While there may be other causes for wet streets, if wetness is absent then rain is absent. There is much art and ambiguity to paleography. However, if the unsystematic survey just described holds up to systematic data collection, one could express the finding as "If the time is between 200 bce and 200 ce, the epsilon is rounded. In these fragments of Matthew the epsilon is not rounded. Therefore the fragments were not written between 200 bce and 200 ce. Some additional questions would be asked of this data set, such as: Are there are any indications of a two-column format for the page? Do documentary papyri use abbreviation or suspension when they abbreviate names? How do literary papyri treat names? How many letters to a line? When does overlining appear and/or disappear? Is the codex REALLY confined to early Christian texts? These are important questions in general, subject to change in subtle ways with new information and new discoveries. It is a complicated and circular process; without the knowledge gained in this way, the recent effort of Thiede to redate Biblical papyri simply contradicts what is known in the field. It does not follow accepted standards of practice, and is therefore suspect. METHODOLOGY A sound basis in method is crucial to the further exercise of critical thought, called methodology, about the place these fragments of papyri should take in our understanding of the NT. Methodology is the effort to infer that a situation existed, and that the inferred situation has a valid foundation. Without a sound basis in established method, inference collapses. Much of what passes for methodology in NT studies consists of case-building, of assembling a number of bits of information from a number of sources, and then arguing that the assemblage provides the basis from which the historian can infer the situation. It is crucial, for this process to be credible, that the historian remain relatively distant from the desired outcome, and alert to any evidence that contradicts that outcome. It is also crucial to be aware of the biases inherent in the sources themselves. Papyrology would seem to offer the possibility of avoiding a lot of the work involved in critical consideration of the sources, and jumping straight into case-building for a speculatively-developed historical theory. In the case of the Matthew fragments, this basic work had already been done by Roberts (1953). Thiede needed only to understand the implications of Roberts's work before building his case for the existence of an early manuscript of Matthew in Egypt. Because Thiede's ZPE article makes a few necessary corrections, he gives the appearance of having mastered the earlier work and then having gone beyond it. However, Thiede erred in his assimilation of Roberts's work, or in reproducing it. The Matthew fragments come from a two-column codex. Thiede barely notes this detail. Had he noted it, according to current standards of practice, he would not have been able to claim that the Matthew fragments stem from an Egyptian copy of the Gospel written down around 70 CE. This is because the earlier codices, according to current understanding, are written in one column rather than two. Of course this could be questioned; Thiede doesn't seem to notice or care or recognize the point. Thus Thiede does not use the method (Roberts, 1953) upon which he relies as his basis for inference in the case of two-column codices. He does appropriate Roberts's suggestion that areas where the papyrus is broken contain nomina sacra. These are two- or three-letter representations of "sacred" names, usually marked by an overhead line, not seen on these fragments. There are various technical aspects, mentioned above as this article traced Thiede's arguments. To reiterate: if nomina sacra for the name of Jesus are present, and if the Thiede's downdating could possibly be true, then these fragments are evidence that some segment of the early Jesus movement thought that Jesus was in a category with God, Moses, and David, and like them his name deserved to receive special treatment as sacred. This group produced bound codices of gospels which reflected their belief by using the nomina sacra for the name of Jesus, the argument would go. Since the evidence for the practice can be identified particularly in this supposedly early version of the Gospel of Matthew, the Gospel of Matthew must have been written within the lifetime of some of the disciples, making it an eyewitness account (more or less), and overturning most modern scholarship on the development of the synoptic gospels. Thiede wisely does not claim any of this in his January 1995 article in ZPE. Thiede's inferences are based on the absence of something that is evident in pictures of the fragmentary Matthew codex, namely that it consists of two columns, and on the presence of something that is not apparent in the photographs of the codex, namely the nomina sacra. Methodologically speaking, reconstructions of missing text cannot prove that words existed. Inference from reconstructed words is not valid. They cannot be used as evidence to answer a question of inference, only a question of description. Where several documents have similar phraseology, such as legal and commercial documents, one document can be reconstructed in terms of another. The reconstruction can be used to characterize the contents of the document without reference to the others from which it was reconstructed. However, they are not a basis for specifics of any sort. That is, an attempt to base an argument on reconstructions is an attempt to argue from uncertain evidence. Thiede's research is flawed in its conception and its presentation, and his findings therefore have no basis. His is a very limited contribution--he has established that what was once P. Magdalen 18 is now P. Magdalen 17. The sections which follow form a more technical review of Thiede's article. TECHNICAL REVIEW The following report has taken shape as the result of Dierdre Good's nudging, requests on the electronic discussion list called ioudaios-l for some (further) discussion of Carsten P. Thiede's reassessment of three fragments of a manuscript of Matthew. The fragments Thiede discusses are all from P. Magdalen Greek 17 (reclassified from P. Magdalen Greek 18), and are designated as {P}64 in the list of codices in the Nestle-Aland 26 Greek-Latin New Testament. There it is dated "ca. 200," in accordance with Colin Roberts's publication and redating of the fragment, to be found in Harvard Theological Review 46, 1953, pp. 233-7, plate [HTR]. Thiede's article is called "Papyrus Magdalen Greek 17 (Gregory- Aland {P}64 ): A Reappraisal," and appears in Vol. 105 of Zeitschrift fr Papyrologie und Epigraphik, pp. 13-20, and Plate IX.(ZPE) I. FULL ARGUMENT. As I understand the varied news accounts, Thiede called a press conference in December of 1994 to announce his forthcoming publication of a first century c.e. fragment of the Gospel of Matthew in a German journal of papyrology. The fragments were a) newly redated by paleography to the first century c.e., around 70 c.e.; b) contained a "stichometrically-plausible" instance of the nomen sacrum -- -- in fr. 1, recto, line 1, Mt 26.31; and c) therefore, first century followers of Jesus thought of him as divine, as bearing a name requiring special treatment in gospel accounts. II. ZPE ARGUMENT In his article in ZPE, Thiede does not address the implications of his redating and reconstructions of the P. Magdalen Gr. 17 fragments. There is no argument or discussion in the ZPE article of point c) above. Such a claim DOES seem to have been made by Thiede in his press conference, in some fashion. In turn, the media have omitted any critical distinctions and said things like, "new papyrus fragment shows that followers of Jesus knew he was divine." Someone associated with ZPE responded to mention of the flap on the papy-l list by noting that Thiede had put together some material that deserved to be aired. This is indeed the case. In the article Thiede confined himself to the following points: * The fragments comprising {P}64, formerly known as P. Magdalen Gr. 18, and so listed in Van Haelst's Catalogue, must be renumbered as Gr. 17 instead. Thiede's description of the error is not clear, but perhaps relates to his request to view Gr. 18, which turned out to be a tiny unrelated scrap. The Magdalen College Library now gives {P}64 the number Gr.17. * {P}64 and {P}67 from Barcelona (P.Barc. inv. 1) are part of the same manuscript, but this manuscript should not be linked with fragments of Luke known from {P}4. Pickering thinks the association should not be abandoned so quickly. {P}4 is also known as P. Paris Bibl. Nat. Suppl. Gr. 1120. To quote C. H. Roberts, "There can in my opinion be no doubt that all these fragments come from the same codex which was reused as packing for the binding of the late third century codex of Philo." [Manuscript, Society and Belief in Early Christian Egypt, Oxford University Press, 1979, p. 13.] * Nestle-Aland has mislabeled the contents of the P.Magdalen Gr. 17 fragments, which Roberts labeled correctly. It is a matter of Mt 26.31 appearing on a different fragment (fr. 1, recto) from Mt 26.32-33 (fr. 2, recto). Verso listings are correct. * Four variant readings, most of which Stuart Pickering has discussed more adequately than Thiede. However, the scribal error of GALEGLAIAN for GALEILAIAN, `Galilean,' of Roberts (1953) is reported by Thiede to have been misprinted in Roberts (1962) [`Complementary Note,' in R. Roca-Puig, Un Papiro Griego del Evangelio de San Mateo, Barcelona\2/1962, 59-60.] This transcription reads, as Thiede reports, GALIGLAIAN, to which Roberts added a note "`v.33, vel GALEILAIAN." Thiede transcribed GALEGLAIAN (op. cit., p. 20). In discussing the article with Robert Kraft, he mentioned that it was apparent that the papyrus does *not* have a gamma before the lambda, but rather an iota. Close attention to the photos of Roberts (1953) indicates that what has been taken as a crossbar to a gamma is probably a flaw in the papyrus, and only the vertical line (of an iota) should be read. * Thiede gives the history of dating the fragment, starting with its acquisition by the Rev. Charles B. Huleatt at Luxor in 1901. Huleatt suggested third century; a librarian reported that A. S. Hunt thought the fourth c. was more likely. Hunt, together with Grenfell, assigned manuscripts which came from codices to third century or later. Roberts (1953) dared to question this, and reassigned {P}64 to ca. 200, based on paleography. Roberts (1953) announced that he had obtained the agreement on the dating of Bell, Skeat, and Turner, major names in paleography of Greek manuscripts. Unaccountably, Thiede does not say why these notables were incorrect in their collective paleographical judgment as to the date of {P}64. * Thiede omits to note that {P}64 is clearly in two columns; he obscures this in his transcription, though the accompanying plate is similar to Roberts (1953) in presentation. Roberts (1953) in contrast notes the two-column format, and clearly labels his transcription according to columns. * Thiede argues that new papyri, published since Roberts (1953) allow the consideration of an earlier date. He mentions the Greek Minor Prophets Scroll, now called 8HevXIIgr, published in DJD VIII, Ed. by E. Tov, 1990, paleographically dated by P. Parsons. Thiede also mentions texts from Herculaneum until 79 c.e. -- the eruption of Vesuvius) and a recent publication (Kim, Biblica, 1988) that lowers the date of Bodmer-Chester Beatty papyrus II ({P} 46) from ca. 200 to ca. 100 c.e. He then adduces likenesses of individual letters to these early papyri from various parts of the Mediterranean. As Stuart Pickering indicated, most of this work is unsound in its reliance solely on individual letter forms. I would add that Thiede sees resemblances between serifed letters from serif-style mss and 'plain' letters from {P}64, where the overall style is also lacking in serifs or other ornamentation. * A sound investigation of the possibility of redating an individual ms would assemble a group of related ms without regard to their date, and then attempt to place the specific ms within a series of mss. This is a method which has led to good results with the paleographical dating of the Hebrew-Aramaic mss of the Dead Sea. Where there are few examples, as with the Greek mss of the Dead Sea, precise paleographical comparisons cannot be made, and dating is very hazardous. This is the case with the Greek Minor Prophets Scroll (8HevXIIgr). To use this ms as a basis for dating another ms, as Thiede has done, is to compound the unreliability of paleographical dating. In contrast to the method I have sketched, Thiede appears to have proceeded by assembling materials which *might* be datable to the first century, and then found individual letter forms from the Matthew fragments which are not unlike letter forms in his samples chosen only by their date. Such a method as Thiede's does not have what scientists call "face validity." There is no reason to think that the investigator has been striving for objectivity, when the methodology is so closely related to the results obtained. * While the initial methodological error of A. S. Hunt with respect to dating {P}64 - the Matthew fragments - occurred because he and Grenfell believed that codices did not appear until the third century (Roberts, 1953:234), codicological information is important and relevant. No one disputes that these fragments come from a codex. Eric G. Turner's investigation of the codex in The Typology of the Early Codex , University of Pennsylvania Press, 1977, sets a lower bound for codex development as the second century c.e. (p. 4), based on the dating of "Christian" materials, with Greek literary codices becoming prevalent a century later. Thiede does not call for, nor address, the implications of his findings for codicology. Should Turner's dates for codices be lowered? Thiede does not say. * Thiede concludes his argumentation with a discussion of nomina sacra. He argues that Kim's lowered dating of {P}46, which has clear nomina sacra , supports Roberts's speculation that nomina sacra were used in the first century c.e. Roberts did not redate his list of early papyri to support his contention, however. [Source is C. H. Roberts, Manuscript, Society and Belief in Earliest Christian Europe, Oxford Univ. Press, 1979.] This material is included because Roberts (1953) and Thiede (1995) both reconstruct nomina sacra] in unclear or missing portions ofthe fragments of {P}64. Whereas Roberts (1953) is methodologically within bounds in reconstructing nomina sacra for a manuscript of ca. 200 c.e., because a fair amount of other evidence exists to support the practice, Thiede (1995) is methodologically less secure in reconstructing nomina sacra for a date around 70 c.e., since he relies on their plain existence only in Kim's redating of {P}46 (Bodmer II). Both the media presentation of I above, and Thiede's article in ZPE depend on the perception of nomina sacra in the text of these fragments of Matthew 26, and specifically on the nomen sacrum IS for IHSOUS. III ASSESSING THIEDE'S ARGUMENT Thiede contributes greater precision to the specification of {P}64, as P. Magdalen Gr. 17, rather than Gr. 18. He notes Roberts's (1962) changed reading of GALEGLAIAN to GALIGLAIAN, or perhaps GALEILAIAN, which is helpful, as the source is not widely available. However, he still reads GALEGLAIAN in his own transcription. While he notes the relationship with the fragments in Barcelona, he did not obtain photos and include them in his argument. He also provides no reason for dropping {P}4 - the Paris Luke fragment which Roberts (1979) assigned to the same ms. His redating on paleographical grounds is seriously flawed in four ways. First, he does not indicate how four great paleographers could all concur on a lowered redating of the Matthew fragments to a date ca. 200 and still be in error. Second, he compares letters in these fragments from Egypt [Luxor is purchase place, hand compares with {P}4, from Philo codex binding] with material from Herculaneum in Italy (that may be from ca. 40 b.c.e. on provenance grounds, with a terminus ad quem of 79 c.e.) and from Qumran in The Land, and from elsewhere in the wilderness of the Dead Sea (Naxal Xever). Third, he compares individual letters without an appreciation of the characteristics of their formation or the hands of which they are a part. Fourth, his assembly of mss for comparisons is not a coherent set, and was apparently chosen primarily as a group of mss which COULD be dated in the first century c.e., regardless of their other features. Thiede does not recognize that a two-column codex such as {P} 64 --Magdalen Gr. 17 -- has no similarly-constructed examples with which to be compared. He also does not recognize the need to provide some explanation for the appearance of a two-column codex at least a century earlier than all other examples of two-column codices. See Turner, op. cit. Finally, Thiede (1995) and Roberts (1953) both transcribed the fragments as though they contained nomina sacra , and as though the use of nomina sacra was not restricted to KURIOS, KURIE, or QEOS, QEOU, but rather extended to abbreviations of IHSOUS. However, and I must state this emphatically, there is NO VISIBLE SUPPORT for reconstructing nomina sacra of IS or IH. That is to say, almost no ink-papyrus combination exists for the areas where these have been indicated. In working out the stichometry, using the available text of Matthew 26 in the relevant verses, I was able to supply alternative lines in every case where Thiede proposed abbreviation or suspension (use of first and last letters), except for the proposed use of letters instead of a word to signify the number 12. There, I agree, the stichometry (line length) is such that IB (Greek letters standing for 12) must be read. This was also Roberts's (1953) transcription. Specifically, in the case of Fr. 2, verso (Mt 26.10), Thiede reconstructs a first line as [oISeipenau]t[o]i[sti] -- which gives a 16-letter stich. There are at least two problems with this reconstruction. First, the column is missing both beginning letters and ending letters. Second, there are no letters on papyrus for this line. At most, there are two dots, which might be the bottoms of letters, and if they are the bottoms of letters, those letters just might be the indicated t and i of Thiede's line 1. In the case of Fr. 3, recto (Mt 26.22-23) both Thiede and Roberts reconstruct a line with KE, for KURIE of "Is it I, Lord." Thiede shows [imei]KEod[eapokri] for a 15-letter stich. That there is a line of text here in the papyrus is apparent. What it might contain is not at all clear. The only clear line follows, with both beginning and end of the stich missing. The possibilities for reconstruction are numerous; Thiede's line is not supported by the miscellaneous ink in various spots on the line. In the case of Fr. 1, recto (Mt 26.31) many might argue that the name IHSOUS *must* be suspended, using IH, or abbreviated, using IS, in order for the line lengths to come out right. I would point out that we have a line clearly beginning autoiso . . . . and a following line that is 16 letters long, (Thiede counted 17) consisting of one word, skandalisqhsesqe , with the words following in the text appearing on the line below. The text we now have suggests that the first line would read autoisoiesouspanteshumeis for an impossible 25 letters. Thiede suggested autoiso[ISpantes] at 15 letters. I suggest that autoiso[iesouspantes at 19 letters is possible. This possibility exists because the word autois extends into the margin by one letter, and the next five letters occupy the space taken by only four in the following line. This would mean that a line of 19 letters would come out no longer than a line of 16 or 17 letters, yet could still contain the name IHSOUS written out. Something has to be done to fit the first line into the column. That it has to be done using an abbreviation or suspension of IHSOUS is not automatically the case. It is a plausible solution, however, for a manuscript considered in relationship with other two-column codices and other manuscripts containing nomina sacra , which Thiede does not do. IV SUMMATION Thiede's 1995 article suggests a lowered date for {P}64 -- P. Magdalen Gr.17 -- by arguments which are methodologically unsound. His further argument that there are nomina sacra used in place of IHSOUS and KURIE is an extremely flimsy one. These fragments of papyrus do not witness directly to the reconstructions with recognizable inked letters on physical papyrus. The layout of visible letters in one case supports Thiede's (and Roberts's) observation that the text contains Greek letters which represent the numeral 12, rather than the Greek word for 12. In the other cases, other plausible reconstructions of the lines are also possible. In the absence of more data, such as the Barcelona fragments might provide, these fragments do not provide any firm evidence for the existence of nomina sacra in either Roberts's date of ca. 200, or Thiede's 1st century dating.