In an article I published earlier this year, I suggested that surviving examples of single quire codices provide evidence for thinking that the Beatty-Michigan codex of the Pauline epistles (P46) may have contained more leaves than is usually supposed. The first editor of the codex, Frederic Kenyon, attempted to calculate the original size of the quire by using the surviving page numbers. He concluded that the book originally consisted of a quire of 52 bifolia (= 104 folia = 208 pages), which would not have had sufficient space to contain the set of 14 letters typically found in later manuscripts of Paul’s letters. In the article, I pointed out that some of our better preserved single-quire codices are not completely symmetrical in regard to the numbers of inscribed pages in the two halves of the codex. This led me to the conclusion that “page numbering of single-quire codices is not necessarily an exact guide to the total number of bifolia that originally made up the quire.”
One more piece of evidence for that conclusion is a phenomenon that I neglected to mention in the article: Single-quire codices can have more than one sequence of page numbers, and the changes in sequences can be unpredictable. An example of this phenomenon is the Crosby-Schøyen codex (a.k.a. Mississippi Coptic Codex I a.k.a. the Savery Codex, a.k.a. Schøyen MS 193 a.k.a. LDAB 107771). This square-format single-quire papyrus codex was reasonably well preserved when it appeared on the antiquities market in the 1950s. It was missing its outer folia, but the blank inner margins of these leaves were left intact at the spine. These can be seen reasonably well in the image below.
There was also a leather spine strip protecting the outside of the spine, so we can say with confidence that the codex, when complete, consisted of a stack of 35 sheets. Thirty-three of these sheets were true bifolia, and two of them were stubbed singletons, meaning that the codex consisted of 68 folia (=136 pages). These pages carry at least five separate texts: Melito’s sermon on the passover, 2 Macc. 5:27-7:41, 1 Peter, Jonah, and an unidentified text. We also find multiple sets of page numbers in the codex. An example of the different sequences can be seen in this second photo taken before the codex was disassembled. 2 Macc. 5:27-7:4 ends on a page numbered ⲝⲋ (= 66) and 1 Peter begins on the facing page numbered with a new sequence beginning with ⲁ (= 1).
The first surviving page number in the codex is ⲓⲑ + ⲕ (that is, 19 + 20), but we can tell from the fragments of the damaged outer folia of the codex that this is not folio 10 of the quire, as we might expect had the pagination began with “1” on the recto of the first folio. Instead, pages 19 + 20 fall on folio 13, meaning that 3 folia (= 6 pages) preceded the beginning of this sequence of page numbering. We can summarize the schemes of page numbering in a table:
|Folio in Quire||Title of Text||Pagination in Codex|
|1 recto – 3 verso||Content unknown||[page numbering not preserved]|
|4 recto – 26 recto||Melito, Peri pascha||pages numbered [ⲁ]-ⲙⲉ (-45)|
|26 verso – 37 verso||2 Macc. 5:27-7:41||pages numbered ⲙⲋ-ⲝⲋ (46-66)|
|38 recto – 54 recto||1 Peter||pages numbered ⲁ-ⲗⲅ (1-33)|
|54 verso – 62 verso||Jonah||pages numbered ⲁ-[ⲓⲍ] (1-)|
|63 recto – 68 verso||Unidentified text(s)||[page numbering not preserved]|
So, we see three distinct series of page numbers. The first extends across two tractates, but the following two tractates each begin with a new set of page numbers starting at 1. And there are even more oddities. The numbering of the pages in 2 Macc. 5:27-7:41 contains an error in that two successive pairs of pages are numbered ⲛ-ⲛⲁ (50-51). And Jonah actually begins on the last numbered page of 1 Peter (folio 54 recto = page ⲗⲅ = 33), but then the verso of that page (the second page containing the text of Jonah) begins another new pagination sequence with the number ⲁ (= 1)!
The use of multiple sequences of page numbers in this single-quire codex provides another example of why we need to exercise some caution when we use a single sequence of page numbers to try to reconstruct the original number of folia for a damaged single-quire codex like P46.