Textual Stability of NT Writings
One of the most interesting studies I’ve read recently is by K. Martin Heide (Munich University), entitled “Assessing the Stability of the Transmitted Texts of the New Testament and the Shepherd of Hermas,” which appears in a forthcoming book, The Reliability of the New Testament: Bart Ehrman and Daniel Wallace in Dialogue, ed. Robert B. Stewart (Minneapolis: Fortress Press, 2011), pp. 111-145.
Essentially, Heide does a quantitative analysis of the extent of textual variation across early centuries in NT manuscripts, and then the same with manuscripts of the extra-canonical text, The Shepherd of Hermas. (Because the very earliest manuscripts of all these texts are fragmentary, Heide focuses on passages extant in the earliest witnesses, using them as test-passages.)
His calculations for the NT manuscripts included show an average textual stability of 92.6% (of the words in the test passages), the spread of results for individual manuscripts ranging from 87.1% to 99.7%.
To achieve some perspective, Heide then conducts the same sort of analysis of extant manuscripts of Shepherd of Hermas (the single most frequently copied extra-canonical text, and more frequently copied than most canonical texts in the first several centuries). The average stability is ca. 86%. Heide notes that this doesn’t “even reach the worst value of the New Testament text, as reprsented by P45 [the Chester Beatty Gospels codex].”
Heide’s judgement is that “the reproduction of the New Testament writings was subject to greater scrutiny” (136), and in the case of Shepherd and also other extra-canonical texts (e.g., Epistle of Barnabas, Protevangelium of James), “Theological interests and piety destabilised the text of these manuscripts far beyond the stability of the text of the canonical gospels” (138).
Comments are closed.
Larry Hurtado's Blog 
Good to know. Thanks!
What was the total sample size of these early fragments of each text? As with any study based on statistical samples there must be some statistical margin of error. I’d be interested in any peer-review results concerning the methodology, too. When history goes all science-y it inherits science’s tools as well as its balances and balances.
My other thought is that these numbers reflect surviving manuscripts, not manuscripts produced. What kind of assumptions might we make with regard to orthodox destruction of errant canonical manuscripts versus the fate of copies of the Shepherd of Hermes?
All in all, pretty interesting.
Well, the size of the text-portions (extant in earliest papyri) varies in the study from ca. 110 to over 1000 words. When the study is published it will be perfectly possible for other scholars to engage its method and results.
It would be unwarranted to make any assumption about some sort of systematic destruction of extra-canonical texts such as Shepherd of Hermas, which was highly regarded and widely copies & used, even/especially among emergent “orthodox” circles.
Actually, I was thinking along the lines of systematic destruction of the “canonical” texts but your point about the early popularity of the Shepherd is well taken.
I know of no evidence of “systematic” destruction by Christians of texts regarded as scripture. There were attempts by some Roman authorities to do so, but only in the mid to late 3rd century CE.
We have only a small portion of the manuscripts that were likely produced in the first three centuries. But it appears that what we have is random evidence, and so, actually, carrying strong probative force, precisely because they haven’t been selected for preservation.
Did the study isolate individual NT books to see if any of them were independently less stable in transmission than Shepherd of Hermas?
Yes, in that most of the earliest papyri manuscripts are remnants of a copy of one gospel. And Heide provides a manuscript-by-manuscript analysis of the level of stability.