Monday, August 21, 2017

Dan Wallace Responds on the ‘Embarrassment of Riches’

Last week, I wrote about the charge made against some reasoned eclectics that they are guilty of praising the large number of NT manuscripts in their apologetic but then not actually using them in their text-critical work. For the details, see here.

I had hoped the post would spark some discussion and it certainly did! It’s now at almost 100 comments. Clearly, it touched a nerve. One of the people I mentioned in my original post was Dan Wallace and I am happy that he responded on the original post. I thought his response deserved its own separate post and so I present it here, only lightly edited by Dan.

This has been an interesting discussion (which I just learned about from a friend) on the quantitative argument that I have used in public debates and lectures. I’ve read through the comments as of yesterday (and noticed, but did not read, a mass of comments posted just in the last 24 hours) and noted the objections to this argument. I think the thread can be grouped as follows:
  1. Peter Gurry calls me an apologist. 
  2. Gurry mentions that both Ehrman and Robinson have argued against the quantitative argument for various reasons.
  3. The quantitative argument in isolation is weak and misleading. It’s not 5000+ MSS in any given place, and only 424 (Greek) MSS are from the eighth century or earlier. 
  4. I am apparently speaking hypocritically when I invoke the numbers because most of these are Byzantine MSS and I presumably think the Byzantine text isn’t worth much. A good analogy would be that I consider the Byzantine witnesses to be counterfeit in thousands of places.
I’m sure I’ve overlooked some of the arguments. But these are the major ones from what I can tell. My response:

  1. I’m an apologist? Peter, I think I should be insulted! Do you really want an apologist writing the foreword to your upcoming book? I may use apologetics but I am hardly an apologist.
  2. Ehrman has argued against the phrase “embarrassment of riches” because of the lateness of the MSS that comprise most of these riches. Robinson says something similar (point 3 above; dealt with below). Ironically, the phrase is a modification of a line in a book that Bart co-authored! On p. 51 of Metzger-Ehrman, Text of the New Testament4, we read “In contrast with these figures [which deal with other Greco-Roman literature], the textual critic of the New Testament is embarrassed by the wealth of material [italics added].” This is the same thing that Metzger had written in his previous editions. So, if Ehrman claims some measure of authorship on this book (and there is no hint anywhere that the two authors disagree over any points, so far as I recall), then I am almost mimicking him on the quantitative argument. This brings us to point 3.
  3. There are three parts to this objection: quantity, content, and date. That is, presumably I use the quantitative argument in isolation from other factors. Several suggested this based on the lone slide that Gurry displayed. But that is misleading. I never mention just the quantitative argument. (Thank you, Darrell, for pointing this out to the readers!) When I lecture on this topic I typically mention the number of MSS, the date of the earliest MSS, the fact that only a few contain the whole NT, the average number of pages, and the quality of the variants. It’s a lot to cover in a 45-minute lecture!

    Robinson critiques the quantitative argument by noting that there are only 424 extant (Greek) MSS of the NT before the ninth century. (Peter, I notice that you invited him to weigh in, but you didn’t invite me. 😔) There are, of course, also versions and fathers through those first eight centuries that make some contribution to our reconstruction of the text of the NT. But the point that Robinson is making is that the vast majority of MSS—coming in the ninth and later centuries—are fuller than the mostly fragmentary MSS of the earlier centuries. Again, I return to Metzger-Ehrman: “the work of many ancient authors has been preserved only in manuscripts that date from the Middle Ages… On the contrary, the time between the composition of the books of the New Testament and the earliest extant copies is relatively brief..” Thus, Metzger-Ehrman make a twofold argument: the quantity and date of the MSS of the NT are vastly more impressive than what we have for other ancient literature.

    By the fourth century we already have the complete NT, and it is duplicated over and over again through the first eight centuries. It’s true, only Aleph is a complete NT majuscule, but Alexandrinus, Vaticanus, Cantabrigiensis, Claromontanus, Washingtonianus, 019, 022, 042, 043, 047, etc., have substantial material. Among the papyri, there are over 550 leaves (over 1100 pages) of material, though most of these leaves are indeed fragmentary. Of the majuscules and lectionaries through the first eight centuries, we have well over 5500 leaves (more than 11,000 pages). Of course, it is difficult to judge how much content these would have altogether by these mere numbers (I’ll work that out at a later date…), but 12,000 pages of text is nothing to sneeze at. Compared to other Greco-Roman literature, the NT stands head and shoulders above the rest both in quantity of MSS and date of the earliest ones. And we can argue that qualitatively it is superior to many literary remains of classical authors, as some have already pointed out. And that leads to point 4.
  4. I do not think the Byzantine text is worthless, nor do I think that it has only a confirmatory voice in textual decisions. There are times, though rare, in which I think the Byzantine witnesses virtually by themselves contain the autographic wording. Phil 1.14 is one of these: τον λογον λαλειν without θεου wedged in between is largely a Byzantine reading, though P46 agrees with it. (If I recall, the UBS editors raised this from a ‘D’ rating to a ‘B,’ probably due to Karavidopoulos’s influence.) And in Matt 24.36 I reject ουδε ο υιος as an early Alexandrian-Western interpolation (or, as Karavidopoulos put it, the Byzantine reading is an ‘eastern non-interpolation’). See my essay in Mike Holmes’s Festschrift, “The Son’s Ignorance in Matthew 24:36: An Exercise in Textual and Redaction Criticism,” pp. 182–209. Further, the verdict is still out regarding the Byzantine text since over 80% of all Greek NT MSS have almost nothing published on them except in various apparatuses; they still need published collations. The ECM on the Catholics is a great step in this direction. I fully agree with Robinson that complete collations of all existing MSS is a sine qua non for our discipline. CSNTM states that as one of our goals. Making digital images of the MSS is just the first step… Darrell Post’s thesis on 2907 is a great model on collation and the resultant implications.

    Regarding the ‘counterfeit’ argument, another way to put this is that, according to Robinson, since there is c. 94% agreement between the NA text and the Byzantine MSS, this would mean that from a reasoned eclectic position the Byzantines are only 6% counterfeit (to use the most pejorative and misleading term). To speak of ‘counterfeits’ is disingenuous because it is focusing only on the disagreements which, as Gurry noted, are but a tiny fraction of the one million variants (Peter, is this a true estimate or just a guess 😊?)
Finally, in lay presentations the major concern about the text of the NT has to do with theology. Can lay folks trust their text on the major doctrines that they thought were always part of the NT? Yes, they can. Even Ehrman would agree with this (see Misquoting Jesus, p. 252, paperback edition).


  1. Nice, substantive rebuttal!

  2. A very nice and helpful response, Dan.

  3. Dr. Wallace,
    In your reference to “94% agreement between the NA text and the Byzantine MSS” –- 94% of what? Are you referring to a percentage of the text, or to a percentage of textual contests in which Byz and NA agree against some other reading?

  4. For the record, the "94% agreement" statistic comes from me, and represents the percentage of the overall NT text held in common between the NA/UBS critical text and the Byzantine.

    1. 94% is way too high of a figure to describe agreement on any given verse among all the manuscripts that have it.

    2. Except that is not what the stated amount of agreement refers to, but only the percentage of agreement between the main text of NA27/28 and that of the general Byzantine consensus as displayed in RP2005/2010.

      If the intention is to find levels of agreement among individual MSS, the percentages would be those displayed among the various tabulations found in the several volumes of Text und Textwert.

  5. Paolo Trovato8/23/2017 8:26 am

    We can add that in almost any MS tradition, no matter how huge, what really changes (except accidentals, that is, insignificant variants) is one or more words in 5 % / 6 % / 10 % of the lines of the text. The task of the textual critic should be trying hard to propose the more rational solutions in order to reduce that 5 % or 10 % to smaller figures. And that’s why you are discussing so eagerly.

    1. Paolo Trovato8/23/2017 8:33 am

      I mean: propose the more rational solutions in order to reduce that 5 % or 10 % of UNCERTAINTY to smaller figures.

  6. Dr. Robinson,
    Thanks. I would like to have it confirmed by Dr. Wallace, though, that what he was referring to is indeed the same thing (or not) that you were referring to.

    1. Yes, I agree with Maurice that this is what he and I were talking about. The percentage of agreements of course is very difficult to assess, since it involves questions of the base text and what constitutes a variant. For example, is Mark 16.9-20 counted as one variant from the Nestle text? If you start with the Byzantine text as the base, the percentages will be different.

    2. Paolo Trovato8/25/2017 8:06 am

      Paul Maas (I quote from the translation of Flower, 1958, § 4) states: “Each witness depends either on a surviving or on a lost exemplar […]. It will be obvious that a witness is worthless (worthless, that is, “qua” witness) when it depends exclusively on a surviving exemplar or on an exemplar which can be reconstructed without its help. A witness thus shown to be worthless (cf. § 8) must be “eliminated” (“eliminatio codicum descriptorum”)”.
      In his Retrospect 1956 the very Maas adds: “The fact is that there are neither ‘good’, nor ‘bad’ witnesses, only dependent and independent ones […]. The oldest exisisting witness is always completely ‘independent’, whereas the independence of later witnesses, as against those which are earlier than themselves, must first be proved by ‘separative errors’ [i.e., significant, "indicative" errors of the oldest one absent in one or more recent ones. Note of PT]. These recentiores will for the most part, though of course not always, turn out to be dependent.

    3. I should note that my 94% figure utilizes the NA main text as published; this therefore includes all material appearing in that main text, even when bracketed or double-bracketed (the latter clearly regarded as non-authentic by the editors).

      My estimate also includes all merely orthographic differences: were these to be removed from consideration, the percentage would likely be around 96-97%.

  7. From my imperfect education, I was given to understand that the entire Byzantine text of the NT could be reconstructed from the works of Christian authors of pre 600 CE.

    Edwin St.John Allen

  8. Dan Wallace,

    Okay; now that it's clear that you meant that the Byzantine Text and the Nestle-Aland compilation agree 94% of the time, let's look at that 6% where NA and Byz disagree -- specifically in the General Epistles, the only part of NA's to be altered in the past few decades.

    Granting that 94% of the text would look the same, regardless of whether one preferred 100% Byzantine readings, or 100% Alexandrian readings, the focus naturally is upon those points where Byz and Alex disagree.

    Looking over those particular textual contests where Byz and Alex disagree, it looks to me like the NA28-editors adopted an Alexandrian reading 92% of the time. Which is to say, 92% of the time, when the editors faced a choice between an Alexandrian reading, and the reading found in the vast majority of manuscripts -- the bulk of that "embarrassment of riches" -- they adopted the Alexandrian reading, and the reading supported by 85% (or more) of that "embarrassment of riches" was rejected.

    What conclusion can be drawn from this, if not that the editors of NA28 very heavily favor the testimony of a relatively small cluster of Alexandrian MSS, and reject the Byzantine readings, when they disagree with the Alexandrian readings, at 92 out of a 100 opportunities?

    If one may make an inexact football analogy: for much of the game, both the back-up quarterback and the starting quarterback are used in the same way: they remain on the sidelines while the defense is on the field. Now suppose there was a football game in which the defense (and special teams) had to remain on the field for 94% of the game -- that is, the offense took the field for only 6% of the overall playing-time. (To keep the math simple, let's say that in Illustration World, football games last 100 minutes.) /Of those six minutes,/ one quarterback was on the field for five minutes and 30 seconds, and the other one was on the field for 30 seconds.

    In NA28 in the General Epistles, the Byzantine text -- constituting the bulk of that "embarrassment of riches" -- is in the role of the quarterback who plays 30 seconds of the game. It is the relatively few flagship MSS with an Alexandrian, or predominantly Alexandrian, text, that have been put on the field 92% of the time when the offense is on the field.

    Of course one could argue that the Alexandrian quarterback runs faster, throws more accurately, and is simply a better athlete than his Byzantine back-up. But that's not the question; the question is, does it look like the Byzantine quarterback is really considered a valuable member of the team? Would the team do anything different without him? There is that 30 seconds of playing time to consider; that is more than nothing. But the thing to see is not that the coach treated the Byzantine quarterback the same as the Alexandrian quarterback for 94% of the game. The thing to see is that during the time when the coach *could* put either the Alexandrian quarterback or the Byzantine quarterback, on the field, he opted for the Alexandrian quarterback 92% of the time.

    1. "the focus naturally is upon those points where Byz and Alex disagree." I would say that for apologetic reasons, which is what is being discussed here, it is quite the opposite.

  9. Peter Gurry,
    At those points where Byz and Alex agree, the compilers *can't* make a choice between the two.
    So when attempting to see how the compilers have treated Byz and Alex -- particularly when asking whether one has been favored, and the other rejected -- it seems perfectly logical not to look at those places where the compilers cannot choose between the two, which means focusing on the points where the compilers *can* choose between the two.

    And when we look at those points, we see that the NA28 compilers opted for an Alexandrian reading 92% of the time. Which tends to imply that the most of the "embarrassment of riches" is not being treated as if it is valuable compared to the relatively few Alexandrian witnesses; rather, in the General Epistles, the Byzantine text continues to be treated as a source of textual pollution in 92% of the text in which the compilers have the opportunity to select either a Byzantine reading or an Alexandrian one.

    1. I’m not interested in the “compilers.” They didn’t copy the NT text. Scribes did. So what options the NA editors had available to them is beside the point Dan, Maurice, and I are making about the massive overlap between Byz and NA. The point is that scribes overwhelmingly preserved the original text *on any view.* To focus on the differences between Byz and Alex in this context is to miss the massive forest for about 6/7 trees.

    2. One must also consider the other "elephant in the room": throughout their apparatuses one can find numerous variant units where the NA27/28 editors follow "M" even though other alternatives were available that would differ from the majority consensus -- most of these particularly involving their normally favored MSS. This factor should not be overlooked or minimized when considering the total amount of agreements versus disagreements.

  10. Peter Gurry,

    The whole discussion is about the compilers' treatment of Byzantine readings; to not be interested in the compilers is to overshoot the subject at hand.

    If the only answer that can be mustered against the charge that the NA28's compilers, in the General Eistles, when given the opportunity, preferred an Alexandrian reading over a Byzantine rival-reading 92% of the time, is that such an opportunity doesn't exist as far as 94% of the text is concerned, it seems to me that the charge completely stands.

    1. Paolo Trovato8/28/2017 5:16 pm

      Dear James(if I may)
      I am afraid that your simile of baseball can be somehow misleading. I take the liberty to suggest a different simile (and a different model of transmission). After the death of Christ, in different times, the Evangelists write their narratives. An unknown but authoritative faithful collects the four Gospels plus other texts and in few centuries this composite series of texts becames more and more popular and eventually a part of the canonical book of Christi fideles and different old non canonical collections lose any interest.
      At the beginning of the process of transmission we can imagine one or two or five very similar MSS copies copied, that is, edited by the compiler of the collection and/or (one of) his favorite scribe(s). Each of the first copies is copied zero/ one / two / three times, depending on the different situations. After 100 years the copies could be at least two or three hundreds. After 400 years they are thousands, but at that point it is likely that the exemplars must be approved / sponsored / spread by the Christian authorities of the time.
      Nowadays it seems that we still have some thousands of extant MSS witnesses, but they are, very likely, only 5% or 10% of the complete transmission (= the “real” or “complete tree” of Jean Fourquet, Sebastiano Timpanaro etc.) and unfortunately they don’t have a mark in the first page, which says “I am copy n. 1”, “I am copy n. 2” etc., even if it is obvious that each act of copying adds fresh innovations and copy n. 7 must be less far removed from the first (lost) copies than, say, copy 937 or copy 3498. Thus, a good simile for our situation could be that you are trying to put together the pieces of an enormous old puzzle in which the most precious pieces must be putted in the center of the puzzle, but only 5% or 10% of the pieces are extant, and the rest is missing.
      Without adding other reasons (changes of script, obsolescence of the old exemplars, control of orthodoxy…), the very fact that the diffusion of the text increases as time passes, means that, if we suppose a random loss of 90%, we can work on very few ancient copies, some more copies of average oldness and lots of recent ones. Of course, it is possible that (some of or all) *the more recent copies preserve in some passages some very important good readings which are missing or uncorrect in the old extant lines of transmission*. (Textual scholars must explain the reasons of their choices in the 5% 6% 10% of the text where a choice among readings is reasonable). Nevertheless, the criterion which says “the reading preserved by 5000 (recent) witnesses is ipso facto more correct than the one who is preserved in few (old) witnesses” is absolutely WRONG for reasons which are clearly and widely treated in any good manual of classic or romance philology since 1927 (that is, Maas 1927) to present days.

    2. Dr. Trovato,
      I am so glad that you have continued to interact in this conversation. You have certainly brought a perspective that seems to me to not have been represented previously. I am certainly much more interested in following up the references you have provided.
      Thanks again,

    3. Paolo Trovato8/28/2017 6:37 pm

      Nice of you, Tim. Thanks

  11. Dear Paolo,

    The football (not baseball) analogy fits the situation rather aptly, it seems to me.

    Theories like the one you described are nothing new; well over a century ago it was proposed that if we picture two or three of four or five transmission-lines like branches on a tree, the weight of all the fruits on that branch may be considered no greater than the weight of their branch, whether there are 10 fruits or 1,000. An easy thing to imagine as a theory; not so easy to historically verify.

    But that’s tangential to the subject at hand. You seem quite mistaken about what I am saying in this discussion. I am not saying that the reading attested by 90% of the MSS must be the original reading. I am saying that it is inconsistent to describe the Byzantine MSS as “an embarrassment of riches,” and then treat them as if they are saturated with corruptions at those points where they disagree with the Alexandrian MSS. I am not arguing here that the Alexandrian readings are wrong, nor that they are right – I am simply observing how the Byzantine MSS have been described, and them contrasting that with how they have been treated by the NA28 compilers at those points where Byz disagrees with the flagship MSS of the Alexandrian Text.

    PT: “A good simile for our situation could be that you are trying to put together the pieces of an enormous old puzzle in which the most precious pieces must be putted in the center of the puzzle, but only 5% or 10% of the pieces are extant, and the rest is missing.”

    Except I am not trying to do that in this discussion.

    PT: “The criterion which says “the reading preserved by 5000 (recent) witnesses is ipso facto more correct than the one who is preserved in few (old) witnesses” is absolutely WRONG.”

    That will be good to know if I ever enter a discussion about that criterion. But in this discussion, I am not maintaining a particular transmissional theory; I am simply observing that it is inconsistent to describe the Byzantine MSS as “an embarrassment of riches” and then, when opportunities arise to choose between the readings of the Byzantine MSS, or the readings of Alexandrian MSS, to reject the Byzantine readings in 92% of those opportunities.

    1. Paolo Trovato8/29/2017 8:08 am

      Dear James,
      thank you so much for you clarifications.
      Even if perhaps it will not concern what you are “trying to do … in this discussion”, let me add something.
      JS “ Theories like the one you described are nothing new; well over a century ago it was proposed that if we picture two or three of four or five transmission-lines like branches on a tree, the weight of all the fruits on that branch may be considered no greater than the weight of their branch, whether there are 10 fruits or 1,000. An easy thing to imagine as a theory; not so easy to historically verify.”
      Indeed I proposed as a theory what, of course, is not a rule, but has been scholarly proved in many occasion with regard to the transmission of classical literature.
      “In chapter 25 of his first “Miscellanea” [about 1480. Note of PT], Politian demonstrates that the Laurentian manuscript 49,7 of Cicero’s “Epistulae familiares”, which had one quire out of order by an error of binding, is the ancestor of a group of more recent Laurentian manuscripts in which the same disturbance in the order of the letters is found without this being explicable by a displacement of the quires. He eliminated the apographs of an old manuscript of Valerius Flaccus by analogous reasoning, as the second “Miscellanea” now prove” (Timpanaro, The Genesis of Lachmann’s Method, p. 48 transl. Most).
      Generally speaking, what is known about the transmission of Classical texts is that in many cases we find “an opposition between a single manuscript (or more rarely very few manuscripts) of the High Middle Ages or even of Late Antiquity on the one hand, and a considerable number of more recent manuscripts on the other. Cases of sort are often cited in Pasquali’s work {Storia della tradizione etc. Note of PT] and in the first volume of the “Geschichte der Textueberlieferung“ [Hunger et al. 1963-64, vol. I: Note of. G. Most]: it should suffice to recall Aeschylus, Sophocles, Isocrates and Theoprastus among Greek authors; Plautus, Terence, Seneca’s tragedies, and Statius’s “Thebaid” and “Achilleid” among Latin ones. It is perfectly natural that the more recent manuscripts should be united by a certain number of banalizations and conjectural “improvements”, or else by real corruption, from which the oldest manuscript(s) is (are) immune” (Timpanaro, The Genesis, etc., p. 171).
      By the way, I am a Romance philologist and I verified this transmissional pattern many times in mediaeval Italian, Spanish and French texts.
      The problem is that to seriously determine the filiations you must use and analyse in a first phase of work only significant errors, which normally are few and difficult to detect. The list and the meticulous study of ALL variant readings without a previous severe selection produces only noise and paralyses any progress in the study of manuscript traditions. That’s why everybody can put on the same level any transmissional theory, no matter how unlikely (e.g. the football match). Only in this way, as the Latins say, you / me / everybody can be sure to address questiones infinitae, which don’t admit clear solutions, and everybody can preserve his starting ideas