Wednesday, February 10, 2016

The Overall Quality of Byzantine Manuscripts

20
Following up on my follow up on the Byzantine text, it’s instructive to see just how close the text found in the majority of our manuscripts is to the initial text in the Catholic Epistles. From the ECM supplement, here are the number of readings where the “undivided witnesses of the Byzantine text” differ from the ECM/NA28 initial text.*

Book Differences Var. Units
James 61 761
1 Peter 46 700
2 Peter 23 417
1 John 43 765
2-3 John, Jude 33 403
TOTAL 206 3,046

To put the matter positively, the Byzantine text as defined by the ECM agrees with the editors’ reconstructed text in over 93% of variant passages in the Catholic Epistles.

GA 35, a key representative
of the Byz text in the ECM
That’s quite a lot. You could even say that your NA28 is 93% Byzantine in the Catholic Epistles if you were so inclined. In any case, this high level of agreement is what led the editors to conclude that the Byzantine text is “an important witness to the early text.” Hence the re-evaluation for the ECM2 of all places where the Byzantine reading differed from the initial text of ECM1 (see ECM2, p. 34*).

*The stats are from the ECM2 supplement, pp. 10, 13, 15, 17 and Gerd Mink, “Contamination, Coherence, and Coincidence,” p. 147 n. 15.

20 comments :

  1. Peter,
    Does this really mean that the Byzantine text is a more reliable witness, or rather that the current editors of the ECM/NA are more apt to rely on internal evidence rather than external evidence?

    Tim

    ReplyDelete
  2. Tim, more reliable than what? A good number of the changes from NA27 to 28 are Byzantine supported readings.

    ReplyDelete
    Replies
    1. Peter,
      More reliable than previously viewed by the past editors of the NA. The changes themselves are not based on new manuscript evidence but a change in the philosophy of the editors. They are the ones who have given my credence to the Byzantine text.
      Their preference for the changes from NA 27 are based on the CBGM, which relies heavily on internal criteria to sort out it's supposed schematic of priority of readings.
      Tim

      Delete
    2. Tim, better to say that some of the changes are based on valuing the external evidence differently than before. So, e.g., at 1 Peter 4.16, the internal evidence combined with a greater confidence in the Byz text has led to the change away from the earliest evidence (μερει now in NA28 instead of ονοματι in NA27). In the past they probably would not have printed a Byz reading that diverged from our earliest witnesses. Now they are willing to, at least where they think the internal evidence suggests such.

      Delete
  3. Am I right to assume that this high rate of agreement between the Byzantine and NA28 is only in the Catholic Epistles, and that in the rest of the NT it's lower?

    If so, mightn't this be because there are much fewer very old papyri of the Catholic Epistles, so that the reconstructed text is bound to showcase more readings from later witnesses?

    ReplyDelete
    Replies
    1. Eric, yes, this is possible but without the data handy I can't say. The question would be how much would the difference be. Would the agreement drop to 90%? 85%? You could compare Robinson-Pierpont to your NA28 and see. They're still not that different outside the CE.

      Delete
  4. Peter,

    Just a suggestion. It may be advisable for further education of those not familiar to precisely define your term "Byzantine text". There are several streams within the Byzantine type represented in the Apostolos (Acts & Epistles) found in in the ECM and several others not represented within the ECM. The text of RP2005 or the majority K-text, is just one among several Byzantine text groupings such as Kr/f35, Kc in addition to other minority groups and smaller clusters. That is not to mention 18-20 main Byzantine groups represented in the Gospels. The "Majority" reading may not be as is often the case, reflective of all Byzantine text groups.

    Paul Anderson
    CSPMT

    ReplyDelete
    Replies
    1. Paul, the operative word in my post is "undivided." But for specifics, just see the sources cited here or my previous post.

      Delete
  5. Put differently but still positively, would it therefore be apt to say that the "undivided witnesses of the Byzantine text" (by which I assume we mean its invariant sections among the Byzantine manuscript groups?) is itself 93% conservative of what the CBGM suggests is the earliest text form in each instance in these texts?

    This isn't a rating of the Byzantine text quality in its variants, in other words; this is a claim that these stages of the manuscript traditions should not be used only as a negative criterion, because where they agree they are significantly conservative, and this familial consistency deserves credence as a witness to forms of the text that have legitimate claim. We still drop about 7% of those invariant sections as inferior to another form, to say nothing of judgments about the variant ones, as they appear in the mix with all other evidence—but they get weighed in their own right in ways they wouldn't have a generation ago.

    ReplyDelete
  6. Yes to the first paragraph, Matthew. But I'm not sure I follow the second one.

    ReplyDelete
  7. Peter: "this high level of agreement is what led the editors to conclude that the Byzantine text is “an important witness to the early text.”

    And yet the "high level of agreement" was recognized even by Westcott and Hort, although they more pessimistically (due to Codex Bezae?) evaluated the overall level of agreement to be approximately 87.5% -- this even though their conclusion regarding the Byzantine Textform was quite the opposite from current thought.

    ReplyDelete
  8. By the way, the agreement between Vaticanus and the initial text of ECM/NA28: 96%. Vaticanus is only three percentage points closer to the earliest form of the text than the Byz text. 01 = 90%. 35 = 91%. This is all in the CE of course. But it's significant. 35 is better than 01 as a whole in the CE according to the ECM!

    ReplyDelete
  9. Peter,

    The main reason GA 35 came out in the CE as closer to Vaticanus than Aleph is that the assumed upstream exemplar (non-fam. 35) is also within the CBGM model and the ECM to be a downstream textual descendent of Vaticanus. But, we obviously reject this notion. So, it depends on which MS is the assumed upstream proto-type in the models which determines the output CE numbers as may you know.

    Yes, all Greek MSS are textually closer to one another than generally known but, particular differences stand out between the "divided" Byzantines and the "divided" Byzantines as opposed to Vaticanus for example. Of course this variation brings in the texttype question of the Byzantines vs. non-Byzantines which is another subject matter.

    Paul Anderson
    CSPMT

    ReplyDelete
  10. "The main reason GA 35 came out in the CE as closer to Vaticanus than Aleph"

    I think you got mixed up, Paul. I said 35 is closer to NA28 than 01 is to NA28. I didn't compare 35 to either 01 or 03 directly (although you can).

    Also, these are direct comparisons and don't have anything to do with up- or downstream. Unless you just mean that the initial text is a reconstruction and not a material manuscript.

    ReplyDelete
  11. There seems to be a gaping hole here, in that we are not told the nature of the disagreement in the other 93% of the variant units.

    Answers to questions like this could help make the numbers meaningful:

    In what percent is a Vaticanus and Sinaiticus agreement rejected? In what percent is simply a Vaticanus agreement rejected, or Sinaiticus if there is no Vaticanus sensible variant? How many of the 93% are oddball Western (Bezae readings)that are unchanged.

    (This gets into the point that the very placement as a variants in the apparatus is based on very limited, presuppositional criteria.)

    And how many actual changes were there in favor of the Byzantine reading in these books from NA27 to NA28?

    How many of these had clear substantial significance? What happens on the 25 most significant differences (e.g. phrases missing or added)?

    Using the statistics above, without the proper background, I believe is stumbling in the dark. And could lead to very dubious claims like:

    "your NA28 is 93% Byzantine in the Catholic Epistles ...."

    Allowing that the 93% claim is at least based on a multi-source comparison, rather than the silliness that often masks as percents, as in some of the infamous 99,5% figures, or a different set of numbers that we had in the Daniel Wallace paper:

    [textualcriticism]statistical illiteracy in textual scholarship - July, 2015
    https://groups.yahoo.com/neo/groups/textualcriticism/conversations/messages/8399



    Steven Avery

    ReplyDelete
  12. Peter,
    Having re-read this post, it seems to me that you equate the Byzantine text with the text that is left after the variants that the editors chose from outside the Byzantine text. This is a faulty way to define the Byzantine text, as if the Alexandrian and Western texts only exist in variants. Any uniqueness to every textual group/family is where they vary from other families.
    What makes the Byzantine text is where it varies from the other text and the actual comparison should be made in the number of times the Byzantine was rejected compared with that same information in the Alexandrian and Wesrern. This number will reflect the actual difference between these families.
    Tim

    ReplyDelete
    Replies
    1. Tim, not exactly. I define Byz as the text of the "undivided witnesses of the Byzantine text." But for the problem of defining Byz, see the post linked in the first line.

      As for your other point, let me ask you this: when a reading is shared by Byz and one of the other text-types, which one does it belong to and why?

      Delete
    2. Peter,
      First, thanks for the clarification on how you define the Byzantine text. This definition which appears to be the second in the ECM based on the linked post is exactly what I would contend is faulty. The initial definition in the aforementioned post is closer to what I would hold with the exception that I would use ASSUMED ausgang text in lieu of established text. This assumed ausgang text based on the number and types of errors, and the early manuscripts available appears to indicate that it is most likely that this text is Alexandrian in origin. Still, while comparing text types/clusters we can only get a real idea of the percentage of difference when we compare the number and types of variants against the established/ausgang text each type/cluster presents.
      I would add, these reasons among others seem to be why Porter, Metzger, Comfort, K. Åland et al, have argued against Byzantine priority and believe that these reasons identified above do indeed give a definition of 'uniquely Byzantine' readings, those readings which tend to conflate and smooth the Text.
      Tim

      Delete
    3. Sorry, I should have said this earlier, but the title of the post here is key: it's the overall quality of Byz manuscripts. My main point is to say that what constitutes distinctive text of these manuscripts is actually quite small. Thus Byz manuscripts as a whole are reliable far more than not. As far as the NA28 editors are concerned, they are reliable 93 percent of the time. That's a lot.

      Delete