Hostname: page-component-74656f9884-skzmx Total loading time: 0 Render date: 2026-11-06T12:31:17.208Z Has data issue: false hasContentIssue false
  • English
  • Français

The Date and Content of P. Antinoopolis 12 (0232)

Published online by Cambridge University Press:  07 March 2012

Michael J. Kruger
Michael J. Kruger
Affiliation:
Reformed Theological Seminary, 2101 Carmel Rd, Charlotte, NC 28226, USA. email: mkruger@rts.edu
Rights & Permissions [Opens in a new window]

Abstract

P. Antinoopolis 12 (Aland 0232), a miniature codex containing remnants of 2 John, was considered by C. H. Roberts to be our earliest copy of this short epistle (third century) and also evidence of an early Johannine corpus. However, upon closer examination, both of these claims appear to be problematic. This article will argue that P.Ant. 12 is most likely a fifth-century codex (not third) and that it was far too small to have carried the entire Johannine corpus. Although there is no combination of books that fits perfectly into this codex, the most plausible suggestion seems to be that P.Ant. 12 originally held the book of Hebrews and the Catholic Epistles.

Keywords

P.Ant. 122 Johnminiature codicesCatholic Epistlesthe book of Hebrews

Information

Type
Articles
Information
New Testament Studies , Volume 58 , Issue 2 , April 2012 , pp. 254 - 271
Copyright
Copyright © Cambridge University Press 2012

In 1950, C. H. Roberts published the original edition of one of our earliest fragments of 2 John, P. Antinoopolis 12 (Gregory-Aland 0232)—henceforth P.Ant. 12.Footnote 1 This parchment leaf, discovered at Antinoopolis by J. de M. Johnson in the winter of 1913–14, measures just 8.8 × 7.3 cm and contains 30 well-preserved lines covering 2 John 1–9 (see Figures 1 and 2 below).Footnote 2 Not only did Roberts offer quite an early date for P.Ant. 12—‘not much later than the middle of the [third] century’Footnote 3 —but he also suggested that the portion of the codex prior to 2 John originally contained a number of other Johannine writings, namely the Gospel of John, Revelation, and 1 John.Footnote 4 This latter suggestion is (largely) based upon the fact that P.Ant. 12 contains page numbers 164 and 165 in the top margin of the recto and verso. If Roberts's analysis is correct, then P.Ant. 12 is not only our earliest copy of 2 John,Footnote 5 but also constitutes our earliest known corpus of Johannine writings.Footnote 6

Figure 1.

P.Ant. 12 Recto. Figure reproduced (in print only) courtesy of Imaging Papyri Project, Oxford and Egypt Exploration Society. All rights reserved

Figure 2.

P.Ant. 12 Verso. Figure reproduced (in print only) courtesy of Imaging Papyri Project, Oxford and Egypt Exploration Society. All rights reserved

The problem, however, is that both of Roberts's claims—regarding both the date and content of P.Ant. 12—run into serious obstacles upon closer examination. While a third-century date has been affirmed by J. van Haelst,Footnote 7 it has been challenged by K. Aland as well as G. Cavallo and H. Maehler who seem to go in the opposite direction to Roberts, placing the manuscript in the fifth or sixth century.Footnote 8 Obviously, such a disparity in dates (nearly three centuries!) warrants a closer examination of this manuscript. As for the suggestion that P.Ant. 12 formed an early Johannine corpus, it will be argued below that such an idea runs into insurmountable spacing problems. Remarkably, these problems have gone largely unnoticed until recent times. All of these issues, combined with the general neglect of P.Ant. 12 over the last half-century, suggest that a reassessment of the date and content of this manuscript is long overdue. The purpose of this article is to offer some preliminary steps in that direction.

1. The Date of P.Ant. 12

In order to establish a more definitive date for P.Ant. 12, we shall consider three factors: (1) size; (2) scribal hand; and (3) other scribal features (nomina sacra, punctuation, etc.).

1.1. The Size of the Codex

In most discussion of P.Ant. 12, the small size of the codex (8.8 × 9.9 cmFootnote 9) has played too little a role in establishing its probable date.Footnote 10 Although the average height of early Christian codices often exceeded 20 cmFootnote 11 —as can be seen in some of our most significant NT manuscripts, P45Footnote 12 (20.4 × 25.4 cmFootnote 13), P52Footnote 14 (18 × 21.3 cm), and P75 (13 × 26 cm)—small codices were not rare in early Christian communities.Footnote 15 Such ‘miniature’ codices (defined by Turner as less than 10 cm wide) were most likely designed for private use and could contain a surprising number of pages.Footnote 16 They were often quite elegant and provided convenient and portable access to various forms of Christian literature.Footnote 17 Biblical books were not uncommon in this tiny format; e.g. Pap. G. 39212Footnote 18 (0217) John; Pap. G. 3072Footnote 19 (0223) 2 Corinthians; PSI 251Footnote 20 (0176) Galatians; P.Oxy. 1080Footnote 21 (0169) Revelation; PSI 5Footnote 22 (0173) James. But, extra-biblical literature was also common; e.g. the Shepherd of Hermas,Footnote 23Acts of Peter,Footnote 24Acts of Paul and Thecla,Footnote 25 an apocryphal gospel,Footnote 26Protevangelium of James,Footnote 27Didache,Footnote 28 the Apocalypse of Peter,Footnote 29 the Life of Mani,Footnote 30Bel and the Dragon,Footnote 31 the Gospel of Mary,Footnote 32 VI Ezra,Footnote 33 and Tobit.Footnote 34

Most noteworthy for our purposes here is that the miniature codex did not become a regular part of Christian literary culture until the fourth century or later. Although the advent of miniature parchment codices in secular literature can be traced back to the time of Martial where classical authors (e.g. Homer, Virgil, Cicero) were available in pugillaribus membranis for the private use of the literate upper class,Footnote 35 this innovation did not appear to meet with much success and in the later years of Martial's publishing there are no more references to it. However, in the fourth century and later the Christian use of the miniature codex became quite widespread—so much so that Roberts originally hypothesized that ‘the miniature codex would seem to be a Christian invention’.Footnote 36 When we look at miniature codices of only NT texts and other Christian literature provided by the recent catalog of Thomas Kraus, the statistics are quite compelling.Footnote 37 Of 29 such codices, only two are dated to the third century, P.Ryl. 463 and P.Ant. 54, but the latter is not really a codex at all, but an amulet.Footnote 38 There are two other codices that date to the third/fourth century, but one of these is also an amulet (P.Oxy. 2684).Footnote 39 All the rest of these codices, 25 of 29, are definitively fourth century or later (with many in the fifth and sixth century).Footnote 40 It is also interesting to note that 23 of these 29 codices are on parchmentFootnote 41 —a material that did not become common in Christian manuscripts until the fourth century and later.Footnote 42 Perhaps it is not surprising, then, that the other miniature parchment codex discovered at Antinoopolis, the Acts of Paul and Thecla (P.Ant. 13), was dated to the fourth century.Footnote 43

Given that P.Ant. 12 is a miniature codex (8.8 × 9.9 cm) constructed from parchment, Roberts's mid-third-century date would be decidedly out of sync with the trends observed here. While such an early date is surely not impossible, it appears to be quite unlikely—unless there are other factors that compel us to accept it. We now consider what some of those factors might be.

1.2. Scribal Hand

The script of P.Ant. 12 is rounded and upright, with a smooth and flowing style that Roberts describes as a hand of ‘elegance’.Footnote 44 The ϵ, ο, σ, and ω are quite large and circular, creating a broad similarity to the classical ‘biblical majuscule’ of Sinaiticus and Vaticanus that is difficult to miss.Footnote 45 However, unlike the biblical majuscule, the hand of P.Ant. 12 has a more curved execution, often creating loops in the α, υ, μ, and even the ι. Not only are the ρ and the ι often extended well below the line (esp. verso ll. 5 and 8), but left-to-right oblique strokes are often prolonged and end in a flourish (esp. end of ll. 1, 3, 4, 7, 9, 10 of the recto). This flowing nature of the script is most aptly seen in the curved horizontal stroke over the nomina sacra that often ends in a ‘roundel’ or ink blob (ll. 9, 10, and 15 of the recto). Cavallo observes that this mix of large, rounded letters, along with a curved, flowing execution, are the classic marks of ‘Alexandrian majuscule’Footnote 46 which was rare prior to the fourth century and at its peak in the fifth.Footnote 47

The large, rounded ο and ω of P.Ant. 12 stand in contrast to the much smaller ο and ω (often raised above the line) that were typical in the third and early fourth centuries.Footnote 48 The small ο is evident in a number of third-century texts (e.g. P.Oxy. 847, P45, P75, P.Oxy. 1016, P.Oxy. 223, P.Oxy. 2341, and P.Oxy. 1015), and the small ω is evident in a number of third- and early fourth-century texts (e.g. P.Oxy. 1016, P.Oxy. 840, P.Oxy 847, P.Oxy. 856, P.Herm.Rees. 5, P.Oxy. 1015, P.Oxy. 2656, P.Oxy. 2458, B.M. Pap. 126, P.Herm. Rees. 5, P.Lond. inv. 2852, P. Chester Beatty XI, P.Oxy. 1352).Footnote 49 In addition, the large, rounded nature of the ϵ, θ, and σ in P.Ant. 12 stands in contrast to the later versions of Alexandrian majuscule, typically sixth century and beyond,Footnote 50 which are marked by a very clear narrowing of these same letters.Footnote 51 Note particularly P.Grenf. II.112, P.Oxy. 1820, and P.Oxy 2258 (all sixth-century versions of Alexandrian majuscule) which have quite narrow versions of ϵ, θ, and σ (and even ο).Footnote 52 This combination of factors suggests that the extreme ends of the dating range for P.Ant.12 (third century and sixth century) are the less likely options, making a fifth-century date, or perhaps late fourth century, the strongest possibility.

In Roberts's original dating of P.Ant. 12, he leaned on comparisons with the Chester Beatty Daniel and Esther which are dated to the third century.Footnote 53 However, while there are certainly similarities between these scribal hands, there are also notable differences. Take, for instance, folio 80 (recto) of the Chester Beatty Daniel. What immediately stands out is the much smaller ο and σ (e.g. ll. 2, 3, and 6), and the smaller ω often raised above the line (e.g. ll. 7, 9, and 11)—both features that are common in the third century but lacking in P.Ant. 12.Footnote 54 Moreover, it lacks the flowing style of P.Ant. 12, including the loops in the υ and the long extension of ι and ρ below the line.Footnote 55 In terms of comparisons, the Chester Beatty Daniel fits better with P.Bodmer II (P66), the Gospel of John dated c. 200Footnote 56—a text which Cavallo, not surprisingly, considers an early precursor to Alexandrian majuscule.Footnote 57

A likely fifth-century date for P.Ant. 12 is confirmed when it is compared to PSI 1.1 (P35), a fifth/sixth-century fragment of Matthew which Cavallo and Maehler consider to be an even more ‘fully-fledged “Alexandrian Majuscule”’.Footnote 58 There we see again the flowing script, often with loops and extensions, combined with the full-size, rounded ο, σ, ω, and ϵ. The ϵ frequently has the upper curve almost touching the oft-extended middle horizontal stroke. The formation of the ν in both manuscripts is also strikingly similar; the letter is quite broad and the oblique stroke is not only extended, but nearly horizontal. In addition, the υ is almost identical with its looped stem and curved arms, and both begin the left-to-right diagonal stroke of the δ with a slight hook or downward curve. The α in both documents is formed in a single motion, resulting in an upper loop and a lower loop—the latter often left incomplete or half-closed.Footnote 59 See Table 1 for examples of the similarities.

Table 1.

Images of P.Ant. 12 courtesy of Imaging Papyri Project, Oxford and Egypt Exploration Society. All rights reserved. Images of PSI 1.1 used by permission of Biblioteca Medicea Laurenziana, Florence

In addition to P35, there are other comparanda that provide support for a fifth-century date for P.Ant. 12. For instance, P.Heid.inv. G 5148, a fifth-century papyrus codex of Genesis, is described by Rodney Ast as having a script that bears ‘some resemblance to the developed stage of the “Alexandrian majuscule”’.Footnote 60 In this manuscript we see (again) the fully rounded versions of ο, σ, and ϵ, as well as broad versions of μ and ν. Moreover, the δ exhibits the same ‘hook’ at the start of the left-to-right diagonal stroke. PSI 2.138, a fourth/fifth-century papyrus codex of Homer, shares many of these same characteristics and also forms the α in a similar fashion as the hand of P.Ant. 12, creating an upper loop but often leaving the lower loop incomplete.Footnote 61 P.Amh. 2.192, a sixth-century papyrus codex of Deuteronomy, also bears a great deal of similarity to P.Ant. 12 but is likely a later, more developed version of Alexandrian Majuscule.Footnote 62 It, too, has the looped υ, and rounded ο, ω, ϵ, and σ. However, it lacks the flourishes and extensions so common in P.Ant. 12, such as the roundels at the end of the υ, the lengthened horizontal line of the ϵ, and letters that extend below the line. Its more squared, formal style suggests a time period slightly later than P.Ant. 12.Footnote 63 See Table 2 for comparisons.

Table 2.

Images of P.Amh. 2.192 used by permission of The Pierpont Morgan Library, New York

1.3. Other Scribal Features

In addition to size and scribal hand, there are other inscriptional features that can help date P.Ant. 12. Particularly noteworthy is the enlarged ϵ, preceded by a small space, which marks the beginning of a new section in line 11 (recto). The κ at the very beginning of line 1 is also enlarged despite the fact that it is the middle of the sentence. Such enlarged letters were relatively infrequent in Roman or Ptolemaic literary papyri,Footnote 64 but were not unusual in documentary papyri.Footnote 65 Although enlarged initial letters are found in some of our earliest Christian texts—P. Egerton 2 (II century), Chester Beatty Numbers/Deuteronomy (III century), Chester Beatty Ezekiel (III century)—the practice did not become abundant or pronounced until the fourth century or later.Footnote 66 More vivid examples of this practice include: Chester Beatty Melito (IV century), P.Berlin inv. 6747 (IV century), P.Oxy. 840 (IV century), P.Oxy 1351 (IV century), P.Oxy 2068 (IV century), Codex Alexandrinus (V century), Codex Bezae (V century), and P.Oxy. 1169 (VI century). Of course, this factor does not rule out a third century date for P.Ant. 12, but it does, once again, make that date much less likely.

Another consideration that led Roberts to an early date was the use of the nomina sacra in P.Ant. 12.Footnote 67 In particular, he appealed to the ‘unorthodox’Footnote 68 form of παρς for the genitive πατρος. However, is an odd form of the nomina sacra able to provide a reliable guide to dating? Although πατηρ is contracted as early as the second century (Chester Beatty Genesis), it was not consistently contracted throughout the first five centuries of early Christian manuscripts. Indeed, of the 133 manuscripts in which the word occurs during this time frame, 80 of them have it only written out in full.Footnote 69 Moreover, there are other odd contractions of πατηρ in later manuscripts; e.g. the fifth-century Freer codex (W) uses the unusual προς for the genitive and the fifth-century PSI 7.759 uses the very rare πρα followed by an apostrophe. In the end, it seems this particular instance of the nomina sacra is too idiosyncratic to provide much assistance to our quest for a solid date.Footnote 70

In sum, all of these considerations—size, material, scribal hand, enlarged letter—suggest that the most probably date for P.Ant. 12 is not third century, as Roberts suggested, but the late fourth or early fifth century.

2. The Content of P.Ant. 12

One of the most intriguing features of P.Ant. 12 is the page numbers in the top margins of the recto and verso, 164 and 165. These page numbers not only demonstrate that this small codex must have originally held a number of other writings, but they also allow us to make an educated guess about what other writings might have fitted within this space. As is now well known, Roberts originally suggested that ‘the codex held a corpus of Johannine writings, with the Gospel, Revelation, and 1 John all preceding II John’.Footnote 71 Since Roberts's original pronouncement, a number of scholars have (understandably) appealed to P.Ant. 12 as one of our earliest pieces of evidence for a Johannine corpus.Footnote 72 The problem, however, is that this hypothesis was built upon Roberts's own mistaken observation that P.Ant. 12 contained ‘400 words to the page’Footnote 73 when in fact it only contains around 400 letters to the page.Footnote 74 The latter would not even provide enough space for the Gospel of John, not to mention Revelation and 1 John.Footnote 75 In light of this reality, it is appropriate that we now consider alternative possibilities about what might have been originally included in P.Ant. 12.

To begin, we must first provide some fresh calculations about the amount of space available prior to our leaf of 2 John. Even though the bottom of the folio is missing, we can calculate the number of letters on the recto with a fair degree of accuracy because we know which words are at the beginning of the following page. The total comes to 405. Unfortunately, we cannot be as certain about the number of letters on the verso because we do not have the page that follows it. Even though the amount of letters per page can change throughout a codex due to a variety of factors,Footnote 76 we have little choice but to use 405 as our working ‘average’. Using this amount, we can calculate that the number of letters preceding 2 John is approximately 66,000 (405 × 163). This space would definitively rule out larger books like Acts (98,075) and the Gospel of John (73,932).

In terms of what might have filled this space, we can only offer conjectures. After all, it is always possible that P.Ant. 12 was part of a ‘composite’Footnote 77 codex like P72 with a combination of (seemingly) unrelated works—Nativity of Mary, 3 Corinthians, Jude, Melito's Homily on the Passover, 1 and 2 Peter, etc.—which were drawn from a variety of different codices and patched together, even retaining the old page numbers in the top margin.Footnote 78 However, while such possibilities always exist, it is reasonable for any hypothesis about P.Ant. 12 to be based on what Christians in the fourth/fifth century might have normally or regularly placed alongside 2 John. Even a quick look at the development of canonical collection units during this time period shows that 2 John found a secure home not only with the other Johannine letters but also with the broader corpus of the ‘Catholic Epistles’.Footnote 79 While David Trobish would argue that the Catholic Epistles were a unit by the middle of the second century,Footnote 80 David Nienhuis has argued that Origen and Eusebius show that this was not a stable unit until the late third and early fourth century.Footnote 81 Regardless of whether one sides with Trobisch or Nienhuis, by the time P.Ant. 12 was produced, it is reasonable to think that 2 John would have been regularly circulating within this relatively well-established corpus.Footnote 82

If so, then there is the question of which of the Catholic Epistles came before 2 John and which came after. A number of the Eastern church fathers (Eusebius, Cyril of Jerusalem, Athanasius, Epiphanius), as well as the major fourth/fifth-century uncials (א, A, B), placed the epistles in the order of James–Peter–John–Jude.Footnote 83 Other canonical lists (particularly from the West) reveal that the order of the Catholic Epistles was not so uniform; e.g. Augustine (Peter–John–Jude–James), Rufinus (Peter–James–Jude–John), Council of Carthage (Peter–John–James–Jude), Codex Claromontanus (Peter–James–John–Jude), and Innocent I (John–Peter–Jude–James). Depending on which option one takes, there are a number of different possibilities for how much space the Catholic Epistles might occupy prior to 2 John. But, even if we followed Rufinus's order and assumed that all the Catholic Epistles preceded 2 John (except, of course, 3 John), this would still not fill up the available space. The letter amounts are as follows:Footnote 84 1 Peter (9,321), 2 Peter (6,222), James (9,102), Jude (2,641), and 1 John (9,793), for a total of 37,079 letters.Footnote 85 This would still leave room for 28,917 letters.Footnote 86 What could possibly fit into that remaining space?

One could certainly hypothesize about various Pauline epistles that would make up this difference (the whole corpus could not fit), or perhaps even the Gospel of Mark (if the order of the Catholic Epistles were rearranged). While such possibilities surely could not be ruled out definitively, there is little reason to think a codex like this would include such a random sampling of NT books. A stronger possibility is the book of Revelation. Given that it typically stands next to the Catholic Epistles in most canonical lists (as well as extant codices), it seems more likely that we might find all these books together in the same manuscript. While Revelation normally is found at the end of a codex, perhaps we could hypothesize that it was moved earlier in the codex to stand alongside the Johannine letters—as was possibly done in Codex Bezae.Footnote 87 The problem with this scenario, however, is that Revelation and 1 John would only fill up about 57,000 of the 66,000 letters preceding 2 John, still leaving a gap of about 9,000 letters. In order to fill that gap, we would have to imagine that another one of the Catholic Epistles (maybe James?) came first in this codex, followed by Revelation, the Johannine letters, and then the remainder of the Catholic Epistles. Such a bizarre arrangement—with Revelation in the middle of the Catholic corpus—seems difficult to imagine.Footnote 88

Thus, as far as canonical books are concerned, we turn to the final option for filling this extra space: the book of Hebrews. Although some of our earliest codices place Hebrews within the corpus of Paul's letters (P46, א, A, B), other patristic testimony from this time period indicates that its location within the NT canon was not so stable.Footnote 89 In the Western church, where Pauline authorship was doubted most often, the ‘traditional sequence’Footnote 90 of Hebrews was after Philemon at the end of the Pauline corpus (and often immediately prior to the Catholic Epistles). This was the position for Hebrews offered by Augustine,Footnote 91 the synods of Hippo and Carthage, Amphilochius of Iconium, Jerome's Latin Vulgate, as well as the fifth-century uncial 048Footnote 92 and the sixth-century codex Claromontanus (D).Footnote 93 Given this historical context, it is not difficult to imagine that doubts about the Pauline authorship of Hebrews might have led some to place the book within a codex of the Catholic Epistles. If that were done, it would be natural to put Hebrews in the first position in the codex. What is particularly attractive about this suggestion is that it solves the complex spacing problems. The book of Hebrews contains 27,134 letters, making it a remarkably good fit into the remaining space of 28,917 letters (again, assuming all the Catholic Epistles preceded 2 John). On this hypothesis, the space preceding 2 John would have begun with the book of Hebrews and ended with 1 John. The space between Hebrews and 1 John would have been filled with 1–2 Peter, James, and Jude, though their exact order is unclear. And, of course, 3 John would naturally come after 2 John. While this solution nicely solves the spacing problems, its weakness is that it assumes the Catholic Epistles end with the Johannine letters—a relatively uncommon order (although, as noted above, Rufinus testifies to this orderFootnote 94).

In the end, it appears that none of the various combinations we have examined here is entirely satisfying. There does not seem to be a book (or set of books) that is a ‘natural’ fit into this space. Such challenges remind us that hypothetical reconstructions of codices are always an uncertain affair. Moreover, the fact that no combination of canonical books provides a neat solution raises the possibility (again) that P.Ant. 12 may very well have been a ‘composite’ codex filled with a mix of apocryphal and canonical materials. Nevertheless, when one takes all the issues into consideration, including the broader historical context of what books might naturally have circulated together, the suggestion that P.Ant. 12 was originally a codex of Hebrews and the Catholic Epistles still seems to be the most plausible option available to us.

As for what books might have come after the Johannine letters, that is even more difficult to determine than what might have come before. If P.Ant. 12 was a single-quire codex then it would be nearly 170 pages—approaching the upper limit of what it could comfortably hold (especially given its small size).Footnote 95 However, as Turner notes, ‘No example of a single-quire codex of parchment has yet been identified’.Footnote 96 Thus, it is reasonable to assume that P.Ant. 12 was likely a multi-quire codex, particularly given its fourth/fifth-century date. We have other examples of miniature parchment codices that are multi-quire, such as the fifth-century Mani Codex that had a total of 192 pages.Footnote 97 Even so, it seems unlikely that P.Ant. 12 could have reasonably held much more than Hebrews and the Catholic Epistles. The most natural candidate to end a codex like this would be the book of Revelation, but this would add approximately 116 pages, for a total of about 286 pages. That would create a codex around 5–6 cm thick—an unlikely scenario for a codex that is only 8.8 cm wide.Footnote 98

3. Conclusion

The purpose of this article has been to offer a reappraisal of both the date and content of P.Ant. 12. While Roberts originally suggested a date in the middle of the third century, and Aland a date as late as the sixth century, we have seen that a number of factors indicate that a fifth-century (or perhaps late fourth-century) date is more appropriate. Given that P.Ant. 12 is a miniature parchment codex, a date in the middle of the third century would be surprisingly early. Moreover, we have argued that the scribal hand—with its fluid style and large, rounded ο and ϵ—best fits with the ‘Alexandrian majuscule’ that reached its peak in the fifth century. The sixth century is unlikely because by that time Alexandrian majuscule often had versions of ο, ϵ, and θ that were decidedly more narrow. Such a conclusion finds further confirmation in the substantially enlarged letters at the beginning of new sections and new lines.

In regard to the content of P.Ant. 12, we can only be certain of what did not precede 2 John. Despite the oft-repeated claims that this codex was our earliest Johannine corpus, there is inadequate space to include the Gospel of John, Revelation, and 1 John. While hypothetical reconstructions are always limited by the available data, we have argued that the most reasonable combination of books (though not without its weaknesses) is Hebrews followed by the Catholic Epistles. If so, then P.Ant. 12 might be evidence that, in some circles, the book of Hebrews was more closely associated with the Catholic letter collection than the Pauline one. As such, this codex might be an additional data point in the already complex discussions about the place of the book of Hebrews in the canon of the NT.

References

1 Roberts, C. H., The Antinoopolis Papyri, Part I (London: Egypt Exploration Society, 1950) 24–6Google Scholar. A more recent (but very brief) treatment can be found in Comfort, Philip Wesley and Barrett, David P., The Complete Text of the Earliest New Testament Manuscripts (Grand Rapids: Baker, 1999) 648–9Google Scholar.

2 For an overview of the Antinoopolis excavations, see Johnson, J. de M., ‘Antinoë and its Papyri’, JEA 1 (1914) 168–81Google Scholar.

3 Roberts, The Antinoopolis Papyri, 24.

4 Roberts, The Antinoopolis Papyri, 24–5.

5 Apart from P.Ant. 12, the earliest copies of 2 John would be found in the fourth-century Sinaiticus (א) and Vaticanus (B).

6 For an analysis of the possibility of such a corpus, see Hill, C. E., The Johannine Corpus in the Early Church (Oxford: Oxford University, 2004) 449–64CrossRefGoogle Scholar. Hill also notes that Codex Bezae (D) is a Johannine corpus of sorts if one assumes that the missing pages would have included Revelation and 1–2 John. Further discussion of Bezae's contents can be found in Parker, D. C., Codex Bezae: An Early Christian Manuscript and its Text (Cambridge: Cambridge University, 1992)CrossRefGoogle Scholar; and Chapman, J., ‘The Original Contents of Codex Bezae’, The Expositor 12 (1905) 46–53Google Scholar.

7 van Haelst, J., Catalogue des Papyrus Littéraires Juifs et Chrétiens (Paris: Publications de la Sorbonne, 1976) 195 (#555)Google Scholar.

8 Aland, Kurt et al., Kurzgefasste Liste der griechischen Handschriften des Neuen Testaments (Berlin: W. de Gruyter, 2nd ed. 1994) 38CrossRefGoogle Scholar. While Aland offers a fifth-century date in the Liste, he offers a fifth/sixth-century date in Aland, K. and Aland, B., The Text of the New Testament: An Introduction to the Critical Editions and to the Theory and Practice of Modern Textual Criticism (Grand Rapids: Eerdmans, 1989) 126Google Scholar. While Aland and Aland offer no basis for the date they give, a fifth-century date is offered on the basis of palaeographical observations by Cavallo, G. and Maehler, H., Greek Bookhands of the Early Byzantine Period, A.D. 300–800 (London: Institute of Classical Studies, 1987) 22Google Scholar.

9 This is the reconstructed size suggested by Aland et al., Liste, 38.

10 Barker, Don, ‘How Long and How Old is the Codex of which P.Oxy. 1353 is a Leaf?’, Jewish and Christian Scripture as Artifact and Canon (ed. Evans, Craig A. and Zacharias, H. Daniel; London: T&T Clark International, 2009) 198Google Scholar, argues for an early date for P.Ant. 12 on the basis that small size was ‘a common format in early codices’. However, he is drawing on a statement from Roberts, The Antinoopolis Papyri, 24, where Roberts does not indicate what he means by ‘early’.

11 Turner, E. G., The Typology of the Early Codex (Philadelphia: University of Pennsylvania, 1977) 14–22CrossRefGoogle Scholar; Hurtado, L. W., The Earliest Christian Artifacts: Manuscripts and Christian Origins (Grand Rapids: Eerdmans, 2006) 162–3Google Scholar.

12 A helpful analysis of P45 can be found in Skeat, T. C., ‘A Codicological Analysis of the Chester Beatty Papyrus Codex of the Gospels and Acts (P45)’, Hermathena 155 (1993) 27–43Google Scholar.

13 Dimensions will list breadth first and then height in accordance with Turner's methodology.

14 Although P52 is a tiny fragment its overall dimensions can be estimated with a fair degree of accuracy. For fuller discussion of this fragment, see Roberts, C. H., ‘An Unpublished Fragment of the Fourth Gospel in the John Rylands Library’, BJRL 20 (1936) 45–55Google Scholar. In regard to the varying views about the date of this fragment, see Nongbri, Brent, ‘The Use and Abuse of P52: Papyrological Pitfalls in the Dating of the Fourth Gospel’, HTR 98 (2005) 23–48CrossRefGoogle Scholar.

15 C.H. Roberts, Manuscript, Society and Belief in Early Christian Egypt (London: British Academy, 1979) 10-11. For more on miniature codices, see Kruger, M. J., ‘P.Oxy. 840: Amulet or Miniature Codex?’, JTS 53 (2002) 81–94CrossRefGoogle Scholar; and Kraus, T. J., ‘P.Oxy. V 840—Amulett oder Miniaturkodex? Grundsätzliche und ergänzende Anmerkungen zu zwei Termini’, ZAC 8 (2004) 485–97CrossRefGoogle Scholar; ET in Kraus, Thomas J., Ad fontes: Original Manuscripts and their Significance for Studying Early Christianity (Leiden: Brill, 2007) 47–67CrossRefGoogle Scholar.

16 The Mani Codex is the smallest known miniature codex and is about the size of a matchbox (3.5 × 4.5 cm), yet still contains 192 pages. For more discussion see Henrichs, A. and Koenen, L., ‘Ein griechischer Mani-Codex (P.Colon. inv. nr. 4780)’, ZPE 5 (1970) 97–216Google Scholar. Other miniature codices also contained an impressive number of pages. The Acts of Peter, P.Oxy. 849 (early fourth century), contains the page numbers 167 and 168 in the top margin.

17 Roberts, Manuscript, 11.

18 9 × 13 cm.

19 8.5 × 12 cm.

20 7 × 12 cm.

21 7.7 × 9.3 cm.

22 6 × 7.5 cm.

23 P.Oxy. 1783 (V.H. 659), 9.3 × 12 cm. The abbreviation ‘V.H.’ refers to the catalog of van Haelst mentioned above.

24 P.Oxy. 849 (V.H. 603), 9 × 9 cm.

25 P.Ant. 1.13 (V.H. 610), 7.2 × 9 cm; and P.Oxy. 1.6 (V.H. 609), 6.7 × 7.3 cm.

26 P.Oxy. 840 (V.H. 585), 7.2 × 8.6 cm. For more on this manuscript, see Kruger, M. J., The Gospel of the Savior: An Analysis of P.Oxy. 840 and its Place in the Gospel Traditions of Early Christianity (Leiden: Brill, 2005)Google Scholar.

27 P.Grenf. 1.8 (V.H. 601), 6.5 × 9.5 cm.

28 P.Oxy. 1782 (V.H. 642), 5 × 5.8 cm. See also Connolly, R. H., ‘New Fragments of the Didache’, JTS 25 (1924) 151–3CrossRefGoogle Scholar.

29 V.H. 619, 5.3 × 7.8 cm. See also Prümm, K., ‘De genuino Apocalypsis Petri textu’, Bib 10 (1929) 62–80Google Scholar; James, M. R., ‘The Rainer Fragment of the Apocalypse of Peter’, JTS 32 (1931) 270–9CrossRefGoogle Scholar; and Kraus, Thomas J., ‘P.Vindob.G 39756 + Bodl. MS Gr. th. f. 4 [P]: Fragmente eines Codex der griechischen Petrus-Apokalypse’, BASP 40 (2003) 45–61Google Scholar.

30 P.Colon. inv. 4780 (V.H. 1072), 3.5 × 4.5 cm.

31 Bodl. gr. bib. d2 (V.H. 323, 1083, palimpsest), ? × 12.6 cm.

32 P.Ryl. 3.463 (V.H. 1065), 8.9 × 9.9 cm.

33 P.Oxy. 1010 (V.H. 574), 5.6 × 8.4 cm.

34 P.Oxy. 1594 (V.H. 82), 8.5 × 8.5 cm.

35 Roberts, C. H. and Skeat, T. C., The Birth of the Codex (London: Oxford University, 1987) 27Google Scholar.

36 Roberts, Manuscript, 12. Gamble, H. Y., Books and Readers in the Early Church (New Haven: Yale University, 1995)Google Scholar, takes a more moderate approach, ‘The miniature format was, if not a uniquely Christian phenomenon, one heavily favored by Christians’ (236). Roberts's original hypothesis is rejected by Kraus, Thomas J., ‘Die Welt der Miniaturbücher in der Antike und Spätantike. Prolegomena und erste methodische Annäherungen für eine Datensammlung’, SNTU 35 (2010) 79–110Google Scholar.

37 Kraus, ‘Die Welt der Miniaturbücher’, 79–110. Kraus's overall list includes 91 items, but 8 of these are rolls. The same general statistics are evident if one looks at the older list of Turner, Typology, 22, 29–30. Of all the 55 miniature codices listed by Turner (Christian and non-Christian), 49 are fourth century or later, leaving only six as potentially third century (and two of these are labeled third/fourth century).

38 For more on the distinction between miniature codices and amulets, see Kruger, ‘P.Oxy. 840: Amulet or Miniature Codex?’ 81–94; and de Bruyn, T., ‘Papyri, Parchments, Ostraca, and Tablets Written with Biblical Texts in Greek and Used as Amulets: A Preliminary List’, Christian Manuscripts: Examples of Applied Method and Approach (ed. Kraus, T. J. and Nicklas, T.; Leiden: Brill, 2010) 145–89CrossRefGoogle Scholar. Kraus, T. J., ‘Manuscripts with the Lord's Prayer—They are More than Simply Witnesses to that Text Itself’, New Testament Manuscripts: Their Texts and their World (ed. Kraus, T. J. and Nicklas, T.; Leiden: Brill, 2006) 227–66Google Scholar, argues that P.Ant. 54 is a writing exercise (and offers other alternatives besides an amulet).

39 The other is P.Oxy. 4500 (Revelation) which is listed as second–fourth century. However, a date in the second century seems much too early given that the material is parchment (this would make P.Oxy. 4500 one of our earliest NT parchment texts), and the hand is clearly on its way to biblical uncial (though it does have the smaller omicron and sigma typical of the third century).

40 If we include miniature codices of OT texts, the statistics are very much the same. Of the 30 OT miniature codices listed by Kraus, only three are dated to the third century or earlier and all are on papyrus. There is also one codex dated third/fourth century and it is on parchment. See Kraus, ‘Die Welt der Miniaturbücher’, 97–8. It is unclear how many of these OT texts are Christian, but the earliest of these, P.Ant. 7 (second century), has the nomina sacra. Thus, it may be the earliest example of a ‘Christian’ miniature codex. Not surprisingly, it is on papyrus.

41 Turner's list of miniature codices showed 45 of 55 were on parchment.

42 In terms of just NT manuscripts, no parchment MSS are found from the second century, only one from the second/third century (0189), two from the third century (0212, 0220), and two from the third/fourth century (0162, 0171). In the fourth century, the situation begins to change rapidly and we find fourteen papyrus MSS and fourteen parchment MSS. The fifth century reveals 36 parchment MSS and two papyrus MSS. See Aland and Aland, The Text of the New Testament, 76. This overall trend is confirmed by a key fourth-century reference to parchment codices by Eusebius (c. 331) in his Life of Constantine where he records the request of Constantine to have fifty copies of the scriptures made ‘on fine parchment’ (Vit. Const. 4.36). For more discussion see Lake, Kirsopp, ‘The Sinaitic and Vatican Manuscripts and the Copies Sent by Eusebius to Constantinople’, HTR 11 (1918) 32–5CrossRefGoogle Scholar.

43 Roberts, The Antinoopolis Papyri, 26–7. Roberts bases the date of P.Ant. 13 on comparisons to PSI 1164, a miniature codex of Jonah dated to the fourth/fifth century. For more on this latter text, see Youtie, H. C., ‘A Codex of Jonah: Berl. Sept. 18 + P.S.I. X, 1164’, HTR 38 (1945) 195–7CrossRefGoogle Scholar.

44 Roberts, The Antinoopolis Papyri, 24. The analysis of the scribal hand comes from new high-resolution digital photographic images made by the Sackler Library at Oxford University (special thanks to Dirk Obbink for his assistance in producing these images).

45 The term ‘biblical majuscule’ is derived from Cavallo, G., Ricerche sulla maiuscola biblica (Florence: Le Monnier, 1967)Google Scholar. A detailed discussion is found in Cavallo, G. and Maehler, H., Greek Bookhands of the Early Byzantine Period, A.D. 300–800 (London: Institute of Classical Studies, 1987) 34Google Scholar. I will consider this synonymous with the term ‘biblical uncial’ which was coined by Grenfell and Hunt; discussion in Turner, E. G., Greek Manuscripts of the Ancient World (London: Institute of Classical Studies, 1987) 21–2Google Scholar; and Roberts, C. H., Greek Literary Hands (Oxford: Clarendon, 1955) 16Google Scholar. For a detailed discussion of different terms for different hands, see Gallo, I., Greek and Latin Papyrology (London: Institute of Classical Studies, 1986) 82–9Google Scholar; Barbour, Ruth, Greek Literary Hands: A.D. 400–1600 (Oxford: Clarendon, 1981) xvi–xxxGoogle Scholar; and Cavallo, G. and Maehler, H., Hellenistic Bookhands (Berlin: W. de Gruyter, 2009) 9–17Google Scholar.

46 For more on this bookhand, see Irigoin, J., ‘L'onciale grecque de type copte’, JÖB 8 (1959) 29–51Google Scholar; and Cavallo, G., “Γράμματα Ἀλϵξανδρῖνα,” JÖB 24 (1975) 23–54Google Scholar. For extensive examples of Alexandrian majuscule (as well as photographic plates), see Porter, S. E. and Porter, W. J., New Testament Greek Papyri and Parchments, New Editions: Texts (Berlin: W. de Gruyter, 2008)Google Scholar, esp. manuscripts 1, 3, 5, 6, 12, 13, 20, and 22.

47 Cavallo, G., ‘Greek and Latin Writing in the Papyri’, The Oxford Handbook of Papyrology (ed. Bagnall, R. S.; New York: Oxford University, 2009) 149–69Google Scholar, esp. 129–31; see also Cavallo and Maehler, Greek Bookhands, 22–3. Bagnall, R. S. and Nikolaos, G., ‘An Early Fragment of the Greek Apophthegmata Patrum’, ARG 5 (2003) 262–3Google Scholar, argue that the peak of Alexandrian majuscule was actually sixth century.

48 Roberts, Greek Literary Hands, 17; Kenyon, F. G., The Palaeography of Greek Papyri (Oxford: Clarendon, 1899) 105Google Scholar; and Roberts, C. H., ‘An Early Papyrus of the First Gospel’, HTR 46 (1953) 37CrossRefGoogle Scholar.

49 Roberts refers to the latter as a ‘flat ω’ (‘An Early Papyrus’, 235).

50 Cavallo and Maehler, Greek Bookhands, 82.

51 Cavallo and Maehler, Greek Bookhands, 112–14.

52 Even later versions of Alexandrian majuscule can be found in P.Berol. 10677 and P.Heidelberg 295.

53 Kenyon, F. G., The Chester Beatty Biblical Papyri: Descriptions and Texts of Twelve Manuscripts on Papyrus of the Greek Bible: Fasciculus VII: Ezekiel, Daniel, Esther (London: Walker, 1937)Google Scholar.

54 The smaller, raised ω is also present in P.Oxy. 656 to which Roberts also compares P.Ant. 12 (esp. ll. 2, 3, and 8 of Plate 2, fragment c of the verso).

55 The υ of the Chester Beatty Daniel has hard, straight edges in the shape of a capital ‘Y’; at one point it even looks like a ‘T’ (see ισχυι in l.3). In contrast, P.Ant. 12 has a soft, flowing upsilon with a noticeable loop in the stem and curved upper arms.

56 For further discussion of this manuscript, see Fee, G. D., Papyrus Bodmer II (p66): Its Textual Relationships and Scribal Characteristics (Salt Lake City: University of Utah, 1968)Google Scholar. An earlier date (though unlikely) has been suggested by Hunger, H., ‘Zur Datierung des Papyrus Bodmer II (P66)’, Anzeiger der österreichischen Akademie der Wissenschaften 4 (1960) 12–33Google Scholar.

57 Cavallo, ‘Greek and Latin Writing in the Papyri’, 129.

58 Cavallo and Maehler, Greek Bookhands, 22–3, places this manuscript in the fifth/sixth century, whereas Aland and Aland place this manuscript in the fourth century but with a ‘?’ after the date (The Text of the New Testament, 98). Comfort, P. W. and Barrett, David P., The Text of the Earliest New Testament Greek Manuscripts (Wheaton, IL: Tyndale, 2001) 138Google Scholar, are overconfident in suggesting a third-century date for this fragment.

59 PSI 1.1 closes the lower loop more consistently, whereas P.Ant. 12 often leaves the lower loop open, making the alpha and the lambda very similar in appearance.

60 Ast, Rodney, ‘Papyri editae in memoriam Wm. Brashear I’, ZPE 139 (2002) 177–86Google Scholar, esp. 178.

61 In particular, notice l. 1 of the recto, and ll. 1 and 3 of the verso.

62 Cavallo and Maehler, Greek Bookhands, 22–3. P.Amh. 192 is likely from the same manuscript as P.Amh. 191 (Exodus). See fuller discussion of both these manuscripts in Grenfell, B.P. and Hunt, A.S., The Amherst Papyri, Part II: Classical Fragments and Documents of the Ptolemaic, Roman and Byzantine Periods (London: Oxford University, 1901) 201–2Google Scholar.

63 P.Ant. 12 also shows substantial similarities to other sixth century manuscripts: (a) P.Berol. 13994 (BKT 8.4), Exodus; (b) P.Köln 806 (Aland 0309), John; (c) P.Oxy. 1076, Tobit; P.Oxy 4845 (Aland P124), 2 Corinthians; P.Oxy 4949, Aristides; P.Vindob. K. 8023bis (Aland 0237), Matthew.

64 Exceptions include P.Oxy. 2161 (Aeschylus Dictyulci); P.Oxy. 1373 (Aristophanes Equites); P.Oxy. 1235 (Hypotheses to Menander); P.Oxy. 1182 (Demosthenes De Falsa Legatione); P.Oxy. 473 (Honorary Decree). See Turner, Greek Manuscripts, 9, for more detailed discussion.

65 Roberts, Manuscript, 14–16.

66 Roberts, Manuscript, 17. It is interesting to note that enlarged letters are also found in early Jewish texts; e.g. P.Oxy. 4443 (Esther, early second century).

67 Studies on the nomina sacra include Traube, Ludwig, Nomina Sacra : Versuch einer Geschichte der christlichen Kürzung (Munich: Beck, 1907)Google Scholar; Paap, A. H. R. E., Nomina Sacra in the Greek Papyri of the First Five Centuries (Leiden: Brill, 1959)Google Scholar; O'Callaghan, Jose, Nomina Sacra in Papyris Graecis Saeculi III Neotestamentariis (Rome: Biblical Institute, 1970)Google Scholar; Brown, S., ‘Concerning the Origin of the Nomina Sacra’, SPap 9 (1970) 7–19Google Scholar; Howard, G., ‘The Tetragram and the New Testament’, JBL 96 (1977) 63–83Google Scholar; Roberts, Manuscript, 26–48; Hurtado, Larry W., ‘The Origin of the Nomina Sacra: A Proposal’, JBL 117 (1998) 655–73Google Scholar; Tuckett, C. M., ‘“Nomina Sacra”: Yes and No?’, The Biblical Canons (ed. Auwers, J. M. and Jonge, H. J. De; Leuven: Leuven University, 2003) 431–58Google Scholar.

68 Roberts, The Antinoopolis Papyri, 24.

69 Paap, Nomina Sacra, 103.

70 It is worth observing that the other unusual instance of the nomina sacra in P.Ant. 12 is the abbreviation of ὑιός (υυ). But, this too gives us little guidance on date because it was also very rarely (and sporadically) contracted—in 123 out of the 143 manuscripts in which it occurs it is left uncontracted (Paap, Nomina Sacra, 110).

71 Roberts, The Antinoopolis Papyri, 24–5.

72 Comfort and Barrett, The Complete Text of Earliest New Testament Manuscripts, 648; Hurtado, Earliest Christian Artifacts, 39; Hill, The Johannine Corpus, 455–6. Of course, even if Roberts was mistaken about P.Ant. 12, this does not mean there were no Johannine collections in the early church. Hill argues cogently that there are other indications that these books may have been circulating together (The Johannine Corpus, 449–64).

73 Roberts, The Antinoopolis Papyri, 24 (emphasis mine).

74 I discovered this error in my own study of P.Ant. 12 and of Roberts's original work on the manuscript. As far as I know, the only other time this error has been observed has been Comfort, P. W., Encountering the Manuscripts: An Introduction to New Testament Paleography and Textual Criticism (Nashville: Broadman & Holman, 2005) 38Google Scholar.

75 As we shall see below, even if one suggested that the Gospel of John came after the epistles, there would then be too much room for Revelation, leaving space for about 10,000 extra letters. Obviously some book would have been in that space. Thus, it would be impossible for this to be a purely Johannine corpus regardless of the order of the books.

76 The number of letters per page can vary for the following reasons: (a) the center pages of a single-quire codex can hold fewer letters than the outer pages (however, as we shall discuss below, there are no reasons to think P.Ant. 12 is single-quire); (b) the scribe may unconsciously increase the number of letters as he nears the bottom of the page; (c) the scribe may put more letters on a page if he fears that he is running out of space. E.g. in P75 the scribe wrote with progressively more lines/letters in an effort to fit all his material; in P66 the scribe did the opposite and wrote with more space as he realized he had too much room remaining. In regard to the role this issue plays in the contents of P46, see Duff, Jeremy, ‘P46 and the Pastorals: A Misleading Consensus?’, NTS 44 (1998) 578–90CrossRefGoogle Scholar, and the response by Epp, Eldon Jay, ‘Issues in the Interrelation of New Testament Textual Criticism and Canon’, The Canon Debate (ed. McDonald, L. M. and Sanders, J. A.; Peabody, MA: Hendrickson, 2002) 498–502Google Scholar.

77 Turner, Typology, 79–80.

78 A helpful look at the pitfalls of hypothetical reconstructions can be found in Epp, ‘Issues in the Interrelation of New Testament Textual Criticism and Canon’, 485–515; and Kraus, T. J., ‘Reconstructing Fragmentary Manuscripts—Chances and Limitations’, Early Christian Manuscripts: Examples of Applied Method and Approach (ed. Kraus, T. J. and Nicklas, T.; Leiden: Brill, 2010) 1–38Google Scholar. For more on P72, see Wasserman, Tommy, The Epistle of Jude: Its Text and Transmission (Stockholm: Almqvist & Wiksell, 2006)Google Scholar; Haines-Eitzen, K., Guardians of Letters: Literacy, Power, and the Transmitters of Early Christian Literature (Oxford: Oxford University, 2000) 96–104Google Scholar; and Nicklas, T. and Wasserman, T., ‘Theologische Linien im Codex Bodmer Miscellani?’, New Testament Manuscripts (ed. Kraus, and Nicklas, ) 161–88Google Scholar. For discussion on what might unify the various texts in P72, see Horrell, David G., ‘The Themes of 1 Peter: Insights from the Earliest Manuscripts (the Crosby-Schoyen Codex ms 193 and the Bodmer Miscellaneous Codex Containing P72)’, NTS 55 (2009) 502–22CrossRefGoogle Scholar.

79 For a detailed examination of the origins of the Catholic collection, see Nienhuis, David R., Not by Paul Alone: The Formation of the Catholic Epistle Collection and the Christian Canon (Waco: Baylor University, 2007)Google Scholar.

80 Trobisch, D., The First Edition of the New Testament (Oxford: Oxford University, 2000)CrossRefGoogle Scholar.

81 Nienhuis, Not by Paul Alone, 29–98.

82 Hill, The Johannine Corpus, 459–60, argues convincingly that we have little reason to think that a tiny epistle like 2 John would have ever had a meaningful transmission history if it had circulated alone. Therefore, we have good reasons to think it was regularly transmitted along with the other Johannine letters (not to mention the broader Catholic corpus).

83 Parker, D. C., New Testament Manuscripts and their Texts (Cambridge: Cambridge University, 2008) 286CrossRefGoogle Scholar, observes that this order is the most common in the manuscripts.

84 These amounts are approximate and come from my own calculations using the NA27 Greek text. However, I have cross-referenced them with calculations in other works which prove to be very similar; e.g. see Trobisch, The First Edition of the New Testament, 135–6 n. 37.

85 One might argue that punctuation would increase the amount of space required, but P.Ant. 12 shows very little use of punctuation. Moreover, with the nomina sacra, these books might actually take up slightly less space. In the end, these minor issues do not affect the numbers enough to make a difference in the overall hypothesis. As for suggestions that there might be space in between individual epistles, the fact that the very first line of 2 John was begun at the end of the prior page suggests that very little space was placed between books. This would be consistent with the format of a miniature codex where space was often (though not always) conserved as much as possible.

86 As noted above, Comfort recognized that Roberts's original calculations were mistaken, but then suggested that each page of P.Ant. 12 could hold 350 letters and therefore the space prior to 2 John could contain 57,000 letters, exactly enough for the Catholic Epistles (Comfort, Encountering the Manuscripts, 38). But, it is unclear where Comfort gets the number 350 for each page when the only page we can actually calculate gives the number 405. Moreover, it is unclear how he concludes that the General Epistles will take up 57,000 letters (as noted above, my calculations suggest they take up much less).

87 Chapman, ‘The Original Contents of Codex Bezae’, 46–53; Parker, Codex Bezae, 9; Hill, The Johannine Corpus, 454–5.

88 Of course, one could hypothesize that Revelation stood at the beginning of this codex and was therefore not next to 1 John. The order would then be Revelation–James–John–Peter–Jude. This order is similar to that found in the so-called Gelasius decree linked to Pope Gelasius I (492–496). While not impossible, this order in the codex would certainly be unusual.

89 For further discussion of the position of Hebrews in the canon, see Hatch, W. H. P., ‘The Position of Hebrews in the Canon of the New Testament’, HTR 29 (1936) 133–51Google Scholar.

90 Metzger, B. M., A Textual Commentary on the Greek New Testament (Stuttgart: German Bible Society, 1994) 592Google Scholar.

91 Doctr. chr. 2.8.13.

92 Heath, D., The Text of Manuscript Gregory 048 (Vatican Greek 2061) (Upland, IN: Taylor University, 1965)Google Scholar. The place of Hebrews in 048, however, is not entirely certain; nonetheless Metzger (A Textual Commentary, 591), like Heath, also references 048 as placing Hebrews after Philemon.

93 This a reference to the order of the codex itself, not the order of the list of books found in between Philemon and Hebrews. The fact that there was a list of books separating Hebrews and the Pauline corpus may be further evidence that Hebrews was seen (at least by the compiler of this codex) as something separate from the Pauline letters.

94 Commentary on the Apostles Creed, 37.

95 Gamble, Books and Readers in the Early Church, 67, puts the upper limit at around 200 pages (100 leaves, 50 sheets).

96 Turner, Typology, 58.

97 Henrichs, A. and Koenen, L., ‘Ein griechischer Mani-Codex (P. Colon. inv. nr. 4780)’, ZPE 5 (1970) 97–216Google Scholar; Turner, Greek Manuscripts, 129.

98 This estimate of the thickness of a codex is based on the analysis found in Skeat, ‘A Codicological Analysis of the Chester Beatty Papyrus Codex of the Gospels and Acts (P45)’. Skeat concludes that P45 would have been 224 pages and 5–6 cm thick. Thus, my suggestion that a 286-page version of P.Ant. 12 would be 5–6 cm thick would be on the very conservative side.

Figure 0

Figure 1. P.Ant. 12 Recto. Figure reproduced (in print only) courtesy of Imaging Papyri Project, Oxford and Egypt Exploration Society. All rights reserved

Figure 1

Figure 2. P.Ant. 12 Verso. Figure reproduced (in print only) courtesy of Imaging Papyri Project, Oxford and Egypt Exploration Society. All rights reserved

Figure 2

Table 1. Images of P.Ant. 12 courtesy of Imaging Papyri Project, Oxford and Egypt Exploration Society. All rights reserved. Images of PSI 1.1 used by permission of Biblioteca Medicea Laurenziana, Florence

Figure 3

Table 2. Images of P.Amh. 2.192 used by permission of The Pierpont Morgan Library, New York