Over at The Textual Mechanic blog, Timothy Mitchell has posted a review of God’s Library. I’m happy to say that the review is mostly positive, although Mitchell does mention some “glaring problems,” “contradictions,” and “circular argumentation” that he detects in the book. I’d like to take this opportunity to try to clarify a couple points he raises.
The main source of the “contradictions” seems to be that on Mitchell’s reading, the book “argues that the differences in codex construction, the style of stitching, the use of single or multi-quires, ‘do not, however, have any bearing on the dates of our surviving copies’ (p. 36).”
But that’s not exactly what the book says. The section from which the quoted words on page 36 are drawn is titled “The Relationship Between Single-Quire Codices and Multi-Quire Codices.” And the specific point under discussion is this particular relationship. The sentence following the words Mitchell quotes makes this explicit: “These observations do not, however, have any bearing on the dates of our surviving codices. That is to say, it is not at all the case that single-quire codices are invariably older than multi-quire codices” (emphasis added). So, the quoted words are not a statement about codex construction in general, but rather a more limited intervention into a series of claims by earlier scholars about the alleged antiquity of single-quire codices as a group, as an endnote makes clear (p. 291, note 31): “[Hugo] Ibscher claimed that no multi-quire codex predated the fourth century, a position that few papyrologists would accept today (Ibscher, “Der Kodex,” 11–12). Nevertheless, I think Ibscher was correct that the multi-quire codex was a conceptual outgrowth of the single-quire codex, and indeed the technology of the multi-quire codex presupposes the prior existence of single-quire codices. Turner’s apparent doubts on this point (The Typology of the Early Codex, 99) do not seem warranted.”
So, the point is simply that, although the single-quire codex must, as a technological development, precede the multi-quire codex, this fact does not really help us assign dates to any particular surviving codex. A similar issue arises again when Mitchell writes:
“If codicology has no ‘bearing on the dates’ then the similarity in shape and binding between the Bodmer papyri should not be marshalled as evidence of a later date for P. Bodmer II or any other undated manuscripts (see also p. 202-206, et al.).”
As I noted above, the quotation about “no bearing on the dates” refers specifically to the question of single- vs. multi-quire construction as a standalone criterion. The book nowhere makes such a broad claim about codicology. That said, I am in fact suspicious of most efforts at codicological dating for the same reason that I am suspicious of much of the practice of palaeographic dating, namely the lack of securely dated comparanda brought to these discussions. Eric Turner himself noted this problem in The Typology of the Early Codex (p. 3): “An objection will at once come to mind. . . Papyrus manuscripts (codices included) are arranged chronologically on the basis of the dates assigned to their handwriting. If a system of codex-formats erected on the basis of handwriting dates is used to correct these dates, the argument proceeds in a circle and can have no validity. The objection is a just one.”
Thus, like palaeographic similarity, codicological similarity can establish possible dates for manuscripts when securely dated comparanda are brought into the discussion. A codex with a relatively secure date (either a terminus post quem or a terminus ante quem) can establish that it is possible that other codices with similar formats or handwriting were also produced around the same time. But, because graphic similarity and similarity of format do not necessarily equate to chronological similarity, this kind of comparison can’t rule out other possible dates for the undated artifact.
So, what God’s Library does argue in regard to some of the Bodmer Papyri, including P.Bodmer II (LDAB 2777), is that a combination of factors (palaeography, codex format, binding technique, suspected archaeological provenance, etc.) can raise the possibility of a later dating than has traditionally been proposed for some of the codices. The wording of the book is quite explicit and careful on this point. Thus, on p. 199 of God’s Library, P.Bodmer II (often dated “circa 200 CE” or the like) is considered alongside other small square-format codices among the Bodmer Papyri with similar handwriting, especially the Crosby-Schøyen codex (LDAB 107771) and P.Bodmer XX+IX (LDAB 220465): “In terms of both its handwriting and its format, P.Bodmer II is completely at home aside these two manuscripts, at least one of which (P.Bodmer XX, the Apology of Phileas) was copied no earlier than 305 CE. It seems quite possible that the antiquity of P.Bodmer II has been exaggerated” (emphasis added).
Thus, while palaeography and codicology cannot demand a later date for P.Bodmer II, the two in combination do show convincingly that a later date cannot be excluded. So, the book uses palaeographic comparisons to extend the range of possible dates for several codices.
Mitchell concludes his review with a question: “Readers will likely come away from God’s Library scratching their heads and asking themselves, ‘Can palaeography be used to date undated manuscripts, or not?’”
God’s Library is indeed highly critical of the way that palaeography has been used to assign dates to early Christian manuscripts, but the conclusions of both the discussion of palaeography and of the book as a whole offer pretty direct answers to this question: “Although paleography, when practiced in a disciplined manner involving close comparison with securely dated examples of handwriting, can establish a range of possible dates for an undated literary manuscript, it can never be conclusive” (p. 72). . . “Paleography can provide a general guide to the date of a manuscript of the Roman era, but it simply cannot deliver the precision and certainty that some of its practitioners claim” (p. 270).