Evidence of Canon Formation from Work-Combinations in NT Manuscripts?

Questions over the history of the formation of the New Testament canon continue to occupy researchers. While Dan Brown continues to promote the popular view that the Council of Nicaea or similar council determined the canon of scripture (see my post here on the history of this myth), scholarship has not submitted to such reductionism. Instead, scholars continue to study canon lists, early Christian citations, manuscript contents, and other kinds of testimonia such as early Christian statements on their own biblical theory in order to assess the historical formation of the canon. Given the state of research on these questions, a recent article in TC: A Journal of Biblical Textual Criticism by Michael Dormandy entitled, “How the Books Became the Bible The Evidence for Canon Formation from Work-Combinations in Manuscripts,” is a welcomed addition to the growing body of literature.

In his introduction, Dormandy makes clear that he is not arguing that the manuscripts he surveys are equivalent to the New Testament or the canon. Rather, he claims the combination of works in the MSS he investigates shows early Christian bibliographic practice:

The terms canonical and New Testament are problematic, because they carry more historical and theological weight than is helpful to my present purpose. This project does not discuss the rise of the word κανών to refer to a collection of authoritative works. It does not even chart the development of the concept of a bounded set of authoritative works. Rather, it concentrates specifically on the development of the bibliographic practice of combining particular works together. This raises an obvious question about the relationship between the concept and the practice: did the early Christians believe certain works were canonical because they were normally part of the same bibliographic unit, or did they regularly include them in the same bibliographic unit because they considered them canonical? This question is also outside my present scope. In this paper I merely aim to present, more comprehensively than before and all in one place, the data on the bibliographic practice and to analyze and summarize that data. In order to make this clear, I use the term collection-evident, rather than canonical, to refer to a combination that contains only works that today are considered canonical. This is because such a combination may be evidence for the bibliographic practice of combining particular works, but not direct evidence for the theological concept of canon (p. 3).

For the moment, Dormandy leaves open the question whether the bibliographic tendencies predate or postdate the earliest statements regarding the Christian New Testament or parts of it (e.g. Irenaeus) and whether the one caused or was the consequence of the other. I will return to this at the end. After a full analysis of all of the MSS up to and around 400, Dormandy categorizes them as follows:

1. “Certainly Collection-Evident”: Artifacts containing more than one work, all of which are today considered canonical. 

2. “Plausibly Collection-Evident”: Fragments, containing only one work, which, more plausibly than not, came from artifacts containing more than one work, all of which are today considered canonical.

3. “Certainly or Plausibly One Work”: Artifacts containing only one work, fragments which, more plausibly than not, came from artifacts containing one work or fragments which have no evidence suggesting that their original manuscripts contained multiple works. 

4. “Plausibly Multiwork, Indeterminably Collection-Evident”: Fragments which, more plausibly than not, came from manuscripts containing more than one work, where it is impossible to say whether or not those additional works were among those considered canonical today. This category contains mainly fragments of the shorter New Testament, that is, the letters other than Romans, 1 and 2 Corinthians, and Hebrews. It would be unlikely to make a manuscript for merely one of the short works. Although 0173 has pagination suggesting that it began with James and sufficiently little text per page that it could have contained only James, we can presume that such small, inefficient codices were not the norm. If there is evidence from pagination or column numbering or otherwise regarding the identity of the other works in the parent manuscript, the fragment can be placed into category 2 or 5, but if not, it is categorized here. 

5. “Plausibly Not Collection-Evident”: Fragments, containing only one work, but which, more plausibly than not, came from manuscripts that contained more than one work, at least one of which was not a work that is canonical today. Many manuscripts in this category are fragments of works that are too small to occupy a codex or roll alone, but which have pagination or column numbering that cannot be reconciled with any standard collection of the works now considered canonical. 

6. “Definitely Not Collection-Evident”: Artifacts containing more than one work, at least one of which is not today canonical.

And the Results? Dormandy presents them as follows:

	Second cent.	Third cent.	Fourth cent.	Fifth cent.	Sixth cent.
Certainly Collection-Evident	0.8	4.7	6.5	12.8	2.8
Plausibly Collection-Evident	0	0.5	3	4	0.5
Certainly or Plausibly One Work	6	30.8	39.5	35.9	7.3
Plausibly Multiwork, Indeterminably Collection-Evident	0	8	6.6	9.5	2
Plausibly Not Collection-Evident	0	0	2.3	1.8	0
Certainly Not Collection-Evident	0	1.7	4.3	8.4	2.6

According to Dormandy, these data offer correctives to both the open canon view and the closed canon view. Regarding the first he says:

They challenge the open canon view, because the frequent combination of (in modern terms) canonical works suggests that they were widely seen to have something in common. Of course, most of the manuscripts that survive date from a time when even open canon scholars would argue that the four-fold gospel and the Pauline corpus were stable. At the very least, however, the data presented here suggests that most book manufacturers did not doubt that the canonical works belonged together, which suggests that they did have some distinctive characteristic in common. It is far beyond the scope of this project to suggest what this might have been (p. 23). 

But equally significant to Dormandy is his final conclusion:

On the other hand, the data presented here also challenges the closed canon view. Single-work artifacts are also too numerous for us to believe that complete editions of the New Testament or four-gospel codices were the normal format for New Testament manuscripts. The data surveyed suggests that the works we consider canonical were commonly associated together, but not always in the same bibliographic unit (p. 24).

Dormandy’s work in this article is quite nuanced and careful. He’s aware that his “One Work” category could be the result of his method and that the number could be smaller. But still, most will be surprised to see how large that number is. Furthermore, for the fourth century, some will be surprised to see how close the numbers between his first and sixth categories are. Also, he walks a careful line between canon—to be defined by the modern notion or by the ancient lists—and the bibliographic practices evident from the contents of the MSS. Still, some might wonder why the modern notion of canon enters his discussion here and why he did not simply compare the results of the bibliographic practice with the biblical canon lists from early Christianity, which he notes positively earlier in the article. But I digress.

Can we know whether early Christian bibliographic practices of forming units preceded or followed their statements on the canon? Dormandy offers this comment:

The fact that this trend [bibliographies of four gospels and Paul’s letters] is so consistent suggests that the bibliographic practice is not a straightforward consequence of explicit statements of the canon: it seems unlikely that any explicit statement could have sufficiently extensive influence. This conclusion must be tentative, partly because there are few, if any, artifacts prior to the earliest statements (few, if any, collection-evident gospel manuscripts predate Irenaeus’s statement of the four-fold canon) and partly because the dating of both artifacts and canon lists is problematic (if the earliest date for the Muratorian Canon is accepted, there would be few New Testament manuscripts which precede it). Tentative as this conclusion is, however, it is still evidence that early Christian bookmakers did not have to be told by ecclesiastical superiors what was in the canon. This in turn suggests that the early Christians may have perceived particular qualities in the works that we consider canonical, even before explicit statements of the canon arose (pp. 22–23).

Dormandy’s caution is warranted due to the current dates of the MSS and the dates of the earliest statements on these collections. But if Dormandy’s tentative conclusion is true, it would show that Irenaeus’s statement on the four-fold Gospel (Haer. 3.11.8) is not novel as some scholars have interpreted it (cf., e.g., Geoffrey Mark Hahneman, The Muratorian Fragment, pp. 101–4). That is, Irenaeus’s statement of the exclusive and authoritative four-fold Gospel would be continuous with traditional Christian bibliographic practice. Indeed, we need to wrestle with such an implication of these data.

On the other hand, it seems that Dormandy’s data show that MSS do not closely conform to the canon lists and early Christian statements regarding their canonical collections. Yes, his first category evinces significant overlap, but his third and sixth categories show that Christian bibliographic practices and the early Christian NT canon lists do not frequently coincide. That is, the MSS, when analyzed according to work combinations, include more books or exclude some books that many of our early testimonia and lists have or do not have. The early trends may show that the two phenomena, bibliographic practice and canon, are becoming coextensive, but they are far from that realization in this period. Furthermore, we are still searching for the Catholic Epistles as a collection in bibliographic practice before the mid-fourth century, even though they have already entered into some of the early canon lists by this time (cf., e.g., Origen’s Hom. Jos. 7.1, Claromontanus, and Cyril of Jerusalem’s Cat. 4.). These observations suggest that the bibliographic practices and the lists of early Christians, at least outside of the Gospels and Paul (the two earliest attested collections), do not correspond closely.

In any case, Michael Dormandy has made a good and careful contribution to the discussion, and it deserves to be widely read and considered by those scholars considering the question of the history of the biblical canon.