[[From A Multiform Heritage: Studies on Early Judaism and Christianity in Honor of Robert A. Kraft, ed. Benjamin G. Wright (Scholars Press Homage Series 24; Atlanta: Scholars Press 1999) 245-262]]


A Coptic Manuscript of the Gospel of John at the
University of
Pennsylvania Museum

Benjamin G. Wright
Lehigh University
James N. Hubler
Lebanon Valley College

Robert Kraft has gained acclaim for his wide range of scholarly interests and skills, and he has received international recognition for his pioneering work in developing electronic data bases and in utilizing computer technology in the study of ancient texts. In addition to being a scholar with an international reputation, his students know personally his commitment to training young scholars and including them in his research. The unraveling and identification of what began as a large clump of papyrus in the University of Pennsylvania Museum provide eloquent testimony to Bob’s wide array of scholarly skills.
     In 1979 Kraft began work on the lump of papyri fragments that bore the Museum number E16292. Although aware that they seemed to be hopelessly fused together, he succeeded in separating them and identifying them as a Sahidic Coptic Gospel of John. In 1988 he turned the project over to Ben Wright and Noel Hubler. As a celebration both of his 65th birthday and of his continuing and productive scholarly career we offer this publication of E16292 as a tribute to Bob’s work with papyri, computers and salvage, but most especially to the priority he places on mentoring his students.

Introduction to the Collection

     At the end of the 19th century, the University of Pennsylvania aggressively began acquiring papyri in various languages, the most numerous being Greek, Hieroglyphic, Hieratic, Demotic, Arabic and Coptic. Much of the current collection was brought from Egypt by W. Max Müller, who was [[246]] associated with the University Museum for over 30 years.1 Müller, however, did not uncover his manuscripts in the field, but he purchased them from antiquities dealers in Egypt. The Museum collection is indeed extensive in some areas, but unfortunately almost nothing, other than Egypt in general, is known of the provenance of much of the material.

1For a history of the papyri acquisitions see, John R. Abercrombie, “Egyptian Papyri,” Expedition (Winter 1978) 3-12.

     The Museum’s holdings were hardly touched from the time Müller made his purchases until 1966 when Kraft and some interested graduate students began to investigate what the Museum owned. In 1971 a program was begun to organize the Museum’s manuscript collection. Only minimal attention was focused on the Coptic, however, because of the relatively large amount of material in that language and its relatively poor state of preservation.

     At present the larger collections, Greek, Arabic, and Coptic, still require extensive work. As far as the extent of the Coptic holdings is concerned, in 1989, the last year that we worked on the entire collection, the Museum shelf list showed 131 acquisition numbers as Coptic (including some bilingual documents). Among these numbers are several “dump” envelopes containing numerous small fragments. No one still quite knows how many different documents comprise the collection.
As a result of the paucity of work done on the Coptic materials, most of the documents remain unidentified. During the several earlier expeditions into the Coptic collection, Kraft and one of his students identified eleven texts. These included a large Psalms manuscript (an announcement of which was published by Kraft),2 a fragment of the Acts of the Apostles (published by Kraft),3 some fragments of Psalms, several non-literary texts (one private letter was published by Leslie S. B. MacCoull),4 and the Gospel that is the focus of this article. In the course of our work in the collection, we identified small fragments of several New Testament books: Matthew, 2 Corinthians, Philippians, Colossians and Hebrews. Outside of the Bible, we have found parts of at least three folio pages from the life of an as yet unidentified saint, a short text that praises St. Stephen, as well as numerous personal letters. Still, a large number of texts remains to be studied.5 [[247]]

     2Robert A. Kraft, “An Unpublished Coptic/Sahidic Psalter Codex at the University Museum in Philadelphia: a Preliminary Report,” Biblical and Armenian Studies (ed. M. E. Stone; Jerusalem: St. James, 1976) 8149.
     3Robert A. Kraft, “A Sahidic Parchment Fragment of Acts 27.4-13 at the University Museum, Philadelphia (E 16690 Coptic 1),” JBL 94 (1975) 256-265.
     4Leslie S. B. MacCoull, “A Coptic Letter in the University Museum, Philadelphia,”
BASP 13(1976)15-16.
     5We worked on the entire Coptic collection for most of two summers, 1988 and 1989. In addition to extensive work on the Penn Gospel, we surveyed the entire collection, trying to identify, or at least to describe the contents of, as many documents as possible. The notes from [[247]] our work can be found on Kraft’s web page (URL http://ccat.sas.upenn.edu/rs/rak/kraft.html under “Papyrological and Manuscript Studies”).

The Penn Gospel

     In 1979, while making one of his periodic forays into the Coptic collection, Robert Kraft discovered, among the papyri acquired by Müller, a large lump of papyrus (Museum no. E16292) that had apparently been used in antiquity as stiffening for a book cover. The lump was part of a larger purchase made by Müller in Cairo from a certain Bernard Moritz, and like the other pieces acquired in this lot its provenance is unknown. Since the papyrus was inscribed, Kraft decided to try to separate the individual fragments from the larger mass—a task that took several months to accomplish. Kraft accomplished the separation by first ascertaining that the ink was not water soluble. After trying unsuccessfully to separate the fragments by leaving the lump in a very humid container (humidifier), Kraft took the step of submerging the entire lump in water. “With some gentle prodding at times” he waited for the individual fragments to float to the top. In all, there were approximately 58 fragments of various sizes stuck together in the original lump.

     While examining the fragments, Kraft found one that contained the name LAZAROS, and further examination confirmed that this particular piece contained John 11.1, “Now a certain man was ill, Lazaros of Bethany....” In the brief time he spent on the fragments, Kraft confirmed the Johannine text of several others, and our subsequent work has confirmed that all of the fragments separated from the lump belong to that Gospel.

     The text originally occupied approximately 40 folio pages of a larger book. Although no single full page is preserved, our reconstructions show that each page measured approximately 22.8 cm. x 19.4 cm. The text was written in two columns/page with approximately 32 lines/page and 13 letters/line.6 Of [[248]] John’s 21 chapters, the fragments preserve some part of all but 1, 2, and 12. It cannot as yet be determined in what type of book the Gospel was contained, but it does not seem to have been a New Testament, at least not in its modem form. The final verses of John (21.21-23,
25) are extant together with the Gospel’s concluding title lines, but the opposite side of this final Johannine fragment contains a text written in a hand different from that of the Gospel. It does not appear to be a long colophon. Unfortunately, we have not been able to identify the text, but using electronic search methods we have concluded that it is not from a book of the New Testament as published in Horner’s critical Sahidic text.  The possibility remains, of course, that it contains a variant New Testament text that departs significantly from that of the Sahidic New Testament that we have in electronic form, but given our inability to fmd any connections with Horner’s New Testament text that seems unlikely to us.

     6The 54 lines of the papyrus which are complete, or nearly complete, average 12.85 letters per line. To get an idea of how many lines existed per column, we examined fragments where three columns existed or could be reconstructed. Many fragments have columns which presumably were separated by two columns of text, because they were the first and last columns on a folio page. By counting the characters between fragments it was possible to estimate the number of lines between fragments. By adding the number of lines that most likely filled the gap to the number of lines on one fragment, it was possible to estimate the number of lines over three columns. Excluding fragment 3.6, whose column arrangement is unclear due to a double transposition (see further below), nine fragments were examined using this method. At the low end, fragment 4.2 had between 89 and 94 lines over three columns. At the high end, fragment 2.8 had 101 lines over three columns, while fragment 3.8 had between 98 and 102 lines over its three columns. Overall the fragments averaged 95.94 lines over three columns or 31.98 lines per column.
     Similar numbers are achieved if we look at consecutive columns. Fragments which come from consecutive columns on a folio page (usually from columns 2 and 3) average 31.36 lines per column. To arrive at this number we excluded Fragment 4.16, which, if it contained a text of John like Horner’s critical edition (G. W. Horner, ed., The Copric Version of the New Testament in the Southern Dialect, Otherwise Called Sahidic and Thebaic [Oxford, [[248]] 1911-1924]), would have had 45 to 47 lines per column. This number is 10 lines more than the next highest line count and may well reflect some sort of textual omission in the Penn Gospel.

     On paleographical grounds, the manuscript can be dated to about the 9th or 10th century.7 The scribe who copied it employed a full stop and an enlarged. space after some large sense breaks, often coinciding with verse breaks in modem editions. Super-literal strokes do appear occasionally, but with no evident consistency.

     7The palaeographic charts in Maria Craemer’s Koptische Paleographie (Wiesbaden O. Harrasowitz, 1964) were the primary basis for this conclusion. We have not been able as yet to check the Penn Gospel against Henri Hyvenat, Bybliothecae Pierpont Morgan codices coptici (Rome, 1922) in which a number of manuscripts from the 9th and 10th centuries an published in facsimile.

     At this stage of our study, we can only offer some preliminary and tentative remarks about the text-critical place of the Gospel in the Coptic tradition. We collated the text against Horner’s critical edition and against Herbert Thompson’s collation of two Chester Beatty manuscripts of John.8 Unfortunately, Horner’s edition is long out of date, but it remains all that there is. More recent publications of Coptic biblical texts have not yet been incorporated into a much needed new critical edition, and we have not had the opportunity to collate our text with all those that have appeared since Horner’s work. [[249]]

     8Sir Herbert Thompson, Coptic Version of the Acts of the Apostles and the PauIine Epistles in the Sahidic Dialect (Cambridge: Cambridge University Press, 1932). At the end of the volume is an appendix in which Thompson collates two Chester Beatty manuscripts of John against Horner. These two manuscripts, designated A and B by Thompson, are apparently Chester Beatty 813 and 814 as mentioned in a publication that we have not as yet seen: Johannes H. Quecke, Das Johanrsesevangelium saidisch. Text der Handschrift PPalau Rib lnv.-Nr. 183 mit den Varianten der Handschrften 813 und 814 der Chester Beatty Library und der Handschrift M 56 (Papyrologica Castroctaviana, Studia et Textus 11; Rome/Barcelona 1984).

     The text-critical situation in the Sahidic is complicated by the fact that many, if not most, of the witnesses used by Horner for John are fragmentary, as is the Penn Gospel. Thus, we can, at this juncture, only indicate manuscripts, or in the case of the Bohairic, traditions, listed by Horner to which our fragments seem to be related. Additionally a large number of the variants are orthographic and thus are of little text-critical value.

     Four Sahidic manuscripts show the closest relation to our John text: Horner’s Ms 69, a Vatican manuscript, Ms 86, also in the Vatican, and the two Chester Beatty manuscripts (A and B).9 Both of the Vatican manuscripts are fragmentary, but they have some important textual variants in common with our gospel. The Penn Gospel shares five substantial variants with Ms 69, involving: haplography, an object pronoun, a circumstantial and an alternative version of the end of John 7.47. Ms 69 only varies from Horner’s text in three places not in common with our fragments, and all three are not significant variants. Ms 86 also has the variant in 7.47 as well as two other substantive variants in common with E16292, and it does not vary from Horner independently of our gospel.

     9For the contents of 69 and 86, see Horner, Coptic Version, 3.351, 352. On the Chester Beatty manuscripts, see the collation in Thompson, Coptic Version, Appendix.

     The two Chester Beatty manuscripts each contain the entire gospel. For the portions contained in our fragments, the Penn Gospel is closest to Chester Beatty A, which has seven variants in common with our text, including a different word order for
5.32 and a different text for the end of 7.47. It also varies from Horner eight times independently, but four involve the omission or of a particle, the omission of the article and the omission of the direct object marker. Chester Beatty B agrees with our manuscript seven times, including 5.32 and 7.47, but it varies independently eleven times. Among these, however, are a number of variants entailing the use of a particle.

     The Penn Gospel has a number of orthographic proclivities, but grammatically it is unremarkable. Orthographically, it shows some signs of phonetic changes from earlier Coptic manuscripts. It tends to fill out consonants with overstrikes by prefixing
E. At John 11.19 and 21.25 the Penn Gospel adds an E in front of a consonant with overstrike and at 20.10 it has an EM where Horner has MM. By contrast, it consistently drops E in perfect relative forms (see 7.38 and 11.33; compare NSEHh for ENSEHh at 20.30).

     The Penn Gospel also demonstrates some changes in diphthongs, perhaps reflecting an assimilation of some diphthongs to simple vowels in later Sahidic. It uses, for example,
EI for I in 11.40 and at 20.20 it changes OU to O. Perhaps its most striking orthographic change is the spelling of sWPE as hW[PE] at 21.4. The spelling parallels the form of sWPE in the Amharic dialect. The Penn Gospel also displays unfamiliarity with Greek, as it transposes letters in Greek terms at 11.18 and 18.22. [[250]]

     Grammatically, the Penn Gospel is inconsistent in its use of circumstantials. Twice it uses initial forms where Horner has circumstantials, but once it has a circumstantial where Horner’s text has an initial. The Penn Gospel uses a first perfect at 5.6 and an imperfect at 18.22. In both places, Horner has circumstantials. Both of these variants are also found in Chester Beatty A and B. The Penn Gospel, however, uses a circumstantial where Horner has a first perfect at 11.13. Only Ms 69 shows the same circumstantial.
     The E16292 fragments also seem to have some relationship to the Bohairic translation of John. They agree with the Bohairic in a number of substantive variants. The relationship between Sahidic and Bohairic Coptic in this period, however, complicates any attempt to understand the textual situation of the Penn Gospel. Without going into detail, the question must be asked if E16292 provides an example of the Sahidic and the Bohairic translation being based on the same type of Greek text, or if it demonstrates a Sahidic text form developing towards the Bohairic, which by this time has become the standard liturgical text of Coptic Christianity. At this time we cannot say.10

     10For details on the various translations of the Coptic Bible, their status as independent translations of the Greek and the possible relationships between the translations in the different dialects, see Bruce M. Metzger, The Early Versions of the New Testament: Their Origin, Transmission, and Limitations (Oxford: Oxford University Press, 1977) 99-152.

     The Penn Gospel shows a remarkable double transposition on fragments 3.6, 1.2 and 4.9, which we discovered originated together on the same sheet of papyrus. They preserve parts of John 8.48-9.22. The fragments are inscribed over four columns in the following arrangement:

     recto: column 1:                             column 2:
        8.48-51(section 1)                         9.17-22(section 4)
    verso: column 3:                             column 4:
        9.12-16(section 3)                         8.53-55 (section 2)

     Although the four columns are on the same sheet, they evidently all did not appear on the same manuscript page.11 The Penn Gospel has about 2 cm between columns on the same page, but there are 5.5 cm between the columns on fragment 3.6. They, then, are the inside columns on two different pages of a single papyrus sheet that had been bound in a larger quire. The large area between columns was where the sheet had been folded.

         11For details on the development of the codex see E. G. Turner, The Typology of the Early Codex (Philadelphia: University of Pennsylvania Press, 1977).

     Thus, column 1 contains section 1, while its verso, column 4, continues with section 2. Still on the verso, column 3 has section 3 and its recto has section 4, column 2 of the fragment. If the quire were laid out to open from the right, section 1 should be to the right of section 4 and section 2 should be to the left of section 3 as follows:

column 1 = section 4 column 2 = section 1 [[251]]

column 3 = section 2 column 4 = section 3

But the order of the columns in the Penn Gospel is reversed:

column 1 = section 1 column 2 = section 4

column 3 = section 3 colunm 4 = section 2

     If the quire (and therefore the codex) opened from the left side of the page, then the arrangement would follow naturally. If the quire opened from the right, as one would normally expect, then column 4 would come first, followed by columns 1 and 2, and then column 3. The order of the text would then be:

John 8.53-55   8.48-51   9.17-22   9.12-16

A right-handed quire would require a double transposition. The scribe started with 8.53-55, then skipped back to 8.48, then continued on to 9.17, then jumped back to 9.12.

     Such a double transposition could be explained in the following way. The scribe copied from an exemplar which had slightly longer lines or columns than the Penn Gospel. The page (or pages) in the scribe’s exemplar had been torn and reinserted backwards. Thereby, section 2 preceded section 1 and section 4 preceded section 3 in the exemplar. Such an inversion of pages in the exemplar would explain the double transposition of passages in the Penn Gospel.

     Apparently, the columns in the exemplar contained more letters than the Penn Gospel. Between section 2 and section 3, where there would be two intervening columns of text missing, Horner has the equivalent of between 105 to 113 lines of text. Based on averages from other fragments, we would expect 96 lines over three columns, so 9 to 17 lines between the sections are missing. Presumably, the Penn Gospel did not have as many letters per column as the exemplar.

The Text of the Penn Gospel

     The text of the Penn Gospel is listed in chapter-verse order of the Gospel of John. First, we list each fragment number followed by its text.12 Following the fragment number, <-> indicates the direction of the papyrus striations for the first side. Next, we list column numbers if multiple columns are preserved. Then follows the text with chapter and verse numbers noted. Where a passage is preserved on multiple fragments, which are either joined or nearly so, the multiple fragments are listed at the head of the section. The first fragment is unmarked, while an equal sign and fragment number in the margin identify the beginning and end of subsequent fragments. If two fragments preserve text from the same line, braces [[252]] enclose the text from the second fragment. An empty line between passages represents a gap between fragments.

      12Kraft assigned fragment numbers when he separated the lumps of papyrus. The first number designates the lump from which the fragment came. The number after the decimal designates the layer in which the fragment lay in the lump. Since these numbers were assigned as the fragments were separated, they do not match the chapter-verse order of the book.

     A dot over a letter indicates that it is only partially preserved. A period in the line indicates that ink is visible but that the letter is illegible. A raised period in the line is a punctuation mark in the text. Decorations were also visible in the margins on fragment 3.5, at the beginning of column 2, John 7.28, and at the end of fragment 4.12, John 8.14. There are six lines of decoration at the end of the Gospel text followed by the title.


fragment # first side other side
4.16 John 03.03 John 03.09-10
2.11 John 04.39-41 John 05.05-07
2.12 John 05.32-33 John 06.01-02
3.3 + 4.15 John 06.04-08 + 07-08 John 06.22 + 24
1.4 John 06.43-44 John ????
2.10 John 07.06-07 John 07.12-13
3.5 + 1.3 John 07.22-28 John 07.32-38
4.10-12 John 07.47-50 John 08.12-14
2.9a John 08.31 John 08.37-38
3.6 + 1.2 + 4.9 John 08.48-09.15 + 09.16 John 08.53-09.20 +09.21-22
2.8 John 09.28-30 John 10.06-08
4.8 + 1.1 John 10.37-38 + 10.42-11.03 John 11.06-09 +11.13-15
3.7 John 11.18-21 + 11.26-29 John 11.32-33 + 11.39-41
3.8 John 13.29-31 John 14.08-????
2.4 John 15.25 + 16.04-05 John 16.????
4.6 John 16.25-26 John 16.30-32
2.3 John 17.15-18 John 17.22-23
4.2 John 18.21-23 John 18.36
2.1 John 19.18-20 John 19.32-34
4.5 John 20.04-05 John 20.10-11
2.2 John 20.20-21 + 20.25-26 John 20.30-31 + 21.03-04
4.3 John 21.20-23 + 21.25 new text = ??

Fragment 4.16 <-> = John 03.03 , 03.09-10

Fragment 2.11 <-> = John 04.39-41 , 05.05-07

Fragment 2.12 <-> = John 05.32-33, 06.01-02

Fragments 3.3 + 4.15 <-> = John 06.04-08, 06.22-24

Fragment 1.4 <-> = John 06.43-44, ????

Fragment 2.10 <-> = John 07.06-07, 07.12-13

Fragments 3.5 + 1.3 <-> = John 07.22-24, 07.28-38

Fragments 4.10-12 <-> = John 07.47-50, 08.12-14

Fragment 2.9a <||> = John 08.31, <-> 08.37

Fragments 3.6 + 1.2 + 4.9 <-> = John 08.48-09.16, 09.17-22

Fragment 2.8 <-> = John 09.28-30, 10.06-08

Fragment 4.8 <-> = John 10.37-38 + 10.42-11.03, 11.06-09, 11.13-15

Fragment 3.7 <||> = John 11.18-21 + 11.26-29, <-> 11.32-33 + 11.39-41

Fragment 3.8 <-> = John 13.29-31, 14.08-????

Fragment 2.4 <-> = John 15.25 + 16.04-05, ????

Fragment 4.6 <-> = John 16.25-26, 16.30-32

Fragment 2.3 <-?> = John 17.15-18, 17.22-23

Fragment 4.2<-?> = John 18.21-23, 18.36

Fragment 2.1 <||> = John 19.18-20, 19.32-34

Fragment 4.5 <-> = John 20.04-05, 20.10-11

Fragment 2.2 <||> = John 20.20-21 + 20.25-26, <-> 20.30-31 + 21.03-04

Fragment 4.3 <-> = John 21.20-23 + 21.25

Unplaced [needs checking]


2.5-7a-b and 9b [no 6?]

3.1-2, 4, 6, 10


Addendum by RAK 30ap2010

Wright and Hubler do not comment on the many instances of apparent "mirror writing" on the preserved fragments, presumably caused by ink blotting between the tightly compacted lumps, and/or the impact of any adhesive used to create the lumps. A closer examination of those instances may serve to identify further fragments or better readings, and also to confirm the order in which the already damaged fragments were lumped. Among the most obvious examples are frgs 2.12, 4.9, 1.3, 4.5, [among those fully processed by now], where the mirror writing is perpendicular to the main text. The back side of 2.12 <|>, for example, gives a clear mirror image of 2.11 <->; 4.10 <|> is mirrored on 4.9 <|>; 1.1 <|> has a clear mirror image on the margin, not yet identified. Both sides of 1.3 have mirror images -- on side one <-> is 1.2 and on the other side 1.4 (and clearly the page was fragmented before this phenomenon occurred). Virtually every page has such mirror images, which are more difficult to decipher unless in margin areas or perpendicular to the main preserved text.