Wednesday, September 19, 2018

Evaluating Ehrman’s Comparative Argument for Textual Unreliability

We have discussed at some length on this blog the value (or lack thereof) of what I called the comparative argument for the reliability of the NT text. In particular, we discussed Bart Ehrman’s three objections to it and why I found one of them significant. In this post, I want to return to that same objection, but this time I’m going to stack it against one of Ehrman’s own reasons for thinking the NT textual tradition is unreliable or at least not reliable enough.

Ehrman’s Argument

Here is Ehrman in his NT introduction:
In the earliest centuries, the vast majority of copyists of the New Testament books were not trained scribes. We know this because we can examine their copies and evaluate the quality of their handwriting, and we can assess how accurately they did their work. The striking and disappointing fact is that our earliest manuscripts of the New Testament have far more mistakes and differences in them than our later ones. The earlier we go in the history of copying these texts, the less skilled and attentive the scribes appear to have been.*
And here is his objection to the comparative argument as applied to the number of manuscripts:
Second, and more important: just because we are WORSE off for other authors than for those of the New Testament does not in itself mean that we can trust that we know what the NT authors wrote. I am a lot stronger than my five-year old granddaughter. But I still am not able to bench-press a half-ton truck. Yes, but you are MANY TIMES stronger than her! It doesn’t matter. I’m nowhere near strong enough. We have far more manuscripts of the New Testament than for any other ancient writing. But that doesn’t mean that we can therefore know what the originals said. We don’t have nearly enough of the right kinds of manuscripts.


As I noted in my previous post, this argument cuts both ways and the reason is obvious. Depending on what you’re lifting, both Ehrman and his granddaughter might be strong enough. After all, why is he trying to lift a half-ton truck rather than a 5-lb dumbbell? Ehrman doesn’t tell us.

The same issue arises when it comes to his judgment about the quality of scribal copying. Assuming for a moment that early scribes were worse than later scribes, what does that actually tell us? The answer to that depends on what exactly we’re demanding. For Ehrman, the fact that early scribes are worse than later ones seems to mean we can’t arrive at the original text. But why? Couldn’t we just as well conclude that early scribes were good or at least good enough and later scribes were better? Asked another way, about how many early variants does it take before we reach agnosticism? I have no idea.

Thankfully, we can compare the product of early and late witnesses and, when we do, we find that they both preserved an awful lot of the exact same text. (If they hadn’t, we would have trouble comparing them in the first place.)

Besides this, the places where we have the most data reveal that the majority of extant variants is actually found, not in the earliest manuscripts, but in our later manuscripts. Whether created there or not, we can’t say with complete certainty. But we can say something about the claim that our earliest manuscripts have far more mistakes and differences than our later ones. What we can say is that it’s simply not true.

Now, someone should say, “Well, that’s because we have so many later manuscripts!” That is true, but it only raises anew the problem with the original comparison. What sense does it make to compare the number of variants in earlier vs. later manuscripts when the real question is: “Have any of them preserved the original text?” As I like to ask my students, “How many good manuscripts does it take to have a good text?” Answer: one. I happen to think we have many more, but the point is that numbers aren’t the real issue.

For these reasons, I don’t see how Ehrman’s comparative argument against textual reliability gets us much further than the comparative argument for textual reliability. In both cases, we have no shared premise (as Pete Williams put it). These comparative arguments “work” only as long as we don’t  recognize this. Once we do, we see that comparison is the wrong tool for the job.

What I’ve left unaddressed is Ehrman’s other metric of scribal ability, which is the quality of handwriting. But that I will leave for my co-bloggers who can address it far better than I can.

*Bart D. Ehrhman, The New Testament: A Historical Introduction to the Early Christian Writings, sixth edition (OUP, 2016), 25.


  1. While not being a scholar myself nor trained in NT Greek I have a question. Could scribal errors actually be a positive when it comes to the reliability of the NT Text? If all fragments of the NT text had no variations would we be able to evaluate them for reliability at all? If they were all the same how would we know that we didn't have a corrupted version? If some early leader (a Christian
    Uthman ibn Affan if you will) had preserved one copy of the NT that he preferred theologically and destroyed all variants how would we know?

    As it is do the variants actually do the opposite of Ehrman's contention and actually testify of the reliability of the NT text?

    1. Dr.David Trobisch claims that all NT manuscripts we have nowadays stem from a single edition of the NT, made around 150 AD by the Church leader Polycarp. You can check his book "The First Edition of the New Testament" and the subsequent article "Who Published the Christian Bible?" I don't buy most of his arguments, but the thesis is interesting.

    2. Paul, I think the answer is "sort of." I think Ehrman would agree that most variants are simply normal transmissional issues that can be expected to occur in the hand-driven reproduction of any large work, sacred or secular. So, the situation in the NT MSS is largely, if not altogether, what we would expect to find, barring miraculous preservation of the copying of the text from all error (something God never promised and has not elected to do) or massive (and successful) conspiracy. As for a "Christian Uthman" or (to Orange Hunter's point) an edition from which all extent material derives, these scenarios, while not impossible, are unlikely, given, for instance, the geographic spread of NT witnesses (Syriac, Armenian, Sahidic, Latin, etc.) in the early centuries.

  2. It seems we can at least bound what was in the originals. The whole problem with the "we can't know anything about the originals" argument is that we act like we can. When we find a papyrus with something like a Gospel text, we can pretty quickly decide that the papyrus is Matthew, Mark, Luke, John, or other. If we don't have any idea of what an original Mark looks like, then how can we distinguish between Mark and Matthew. Why do we view Matthew as something separate rather than 'yet another' a corrupted edition of Mark if we don't have at least a bounds of what an original Mark looks like?

  3. As I read Ehrman, he does not appear to be opposed to (or "against") the comparative argument; rather, he seems to recognize that it has limitations. After all, Ehrman writes, "That is good news indeed—the more manuscripts you have, the more likely it is that you can figure out what the authors originally said." He adds, "That doesn’t mean, however, that we should give up all hope of ever discovering what the New Testament authors wrote. It simply means that there are some places, possibly a lot of places, where we will never know for sure."

    Based on this, can we say that Ehrman thinks the NT tradition is "unreliable" or "not reliable enough." And not reliable enough for what? This needs clarification.

    Peter, you use the term agnosticism here. Are you applying that to Ehrman's position? And in what sense are you using the term? Can we be 100% absolutely certain what the originals say without having the originals? Based upon the evidence, and if we are honest, then we would have to say, "No." In this sense, all of us are agnostic. Nevertheless, given the evidence, scholars reconstruct what they think the text originally said - some places with higher levels of confidence than others?

    So aren't there some common premises here after all?

    1. D., I don’t know where you’re quoting Ehrman from, so I can’t offer much of a response to that. But yes, my main point is that the level of reliability or accuracy or certainty needed is not defined and therefore not shared. Without that, who is to say how many variants are too many or how much uncertainty is too much? Too much for what? I don’t see any obvious way to reach agreement on that with Ehrman. What we can talk about is whether or not most variants are found in our earliest manuscripts. They are not. We can also talk about whether later manuscripts agree in most cases with our earliest ones. They do. Is textual criticism therefore not needed? μη γένοιτο!

    2. Peter, the Ehrman quotations that I provided are also from his NT Introduction (p. 23, 25).

      Ehrman also writes here, "I don't want to mislead you into thinking that scholars believe that we can never have any idea what Luke - or any other New Testament authors - actually wrote. For most passages, most sentences, most words, scholars are reasonable confident that we can know what was written - even if there are passages in doubt" (28-29).

      He adds, "Even though there are enormous problems in establishing what the authors of the New Testament originally wrote, in most instances textual critics are reasonably certain that they know" (31).

      Given these quotations, and the others that I provided, is it fair to say that Ehrman maintains "the NT textual tradition is unreliable or at least not reliable enough"? This seems to be mischaracterizing his position. He seems to think that it is reliable enough to reconstruct large portions - if not most - of the NT with a high level of confidence.

      You write, "Couldn’t we just as well conclude that early scribes were good or at least good enough." It sounds to me that Ehrman would agree, and this would provide a shared premise between the two of you.

      For example, let's assume that an original manuscript is copied to create manuscript X and manuscript Y, and in the process manuscript X introduces a variant that is copied 100 times. Manuscript Y, on the other hand, introduced no variants and is never copied. It doesn't follow that the variant matches the original simply because 101 manuscripts contains this variant as opposed to one manuscript lacking this variant. I assume you would agree with this? This also seems to be the point that Ehrman is making when he discusses the number of NT manuscript (p. 29) - at least in my reading of him. So pure numbers of manuscripts alone doesn't necessarily provide us with the original reading. I think that this is why Ehrman discusses five criteria on pages 29-31. Wouldn't this be another shared premise between the two of you?

      On a slightly different point, instead of speaking about the total number of copying errors, would it be more fruitful to talk about the frequency of errors?

    3. D., I would love to be wrong about Ehrman’s views and maybe his views have shifted with this latest edition. But I’m not sure. On the one hand, it’s great to see him say that for most words in Luke, we are reasonably confident. On the other hand, he says there are possibly a lot of places where we will never know for sure. Those are mixed signals. Which level of confidence is he after? But, if I’m wrong and he more confident than he used to be, that’s great.

      As for the missing shared premise, what I mean is that comparing the quality of early to late scribes isn’t a comparison that matters much. We could just as well say that late scribes are way less accurate than the printing press, but what would that tell us about our ability to reconstruct the original text? What we want to know is whether or not any of our manuscripts preserved the original text and to what degree. On any measurement, most of them preserves most of it and we know that by seeing how much text they all share in common. In the quote in my original post, Ehrman only compares the negative evidence (variants or differences) and leaves unmentioned the overwhelming positive evidence (agreement). That’s why he can frame it as a “striking and disappointing fact.” But one could just as easily see how much agreement there is as a far more striking and reassuring fact. One that still leaves plenty of work for text critics, of course.

      I hope that clarifies somewhat.

    4. Peter,

      I disagree that Ehrman only compares the negative evidence. He writes, "The good news: We have more manuscripts for the New Testament than for any other book from the ancient world. . . . This is good news indeed - the more manuscripts you have, the more likely it is that you can figure out what the authors originally said" (23). He adds, "There is, however, some further good news. Some of the manuscripts of the New Testament appear to be highly accurate copies and a few of them are very ancient" (23).

      So I read him as affirming positive evidence (agreement), but not in an uncritical form. Take my example above with manuscript X and Y. Agreement between them would raise the likelihood of reconstructing an original reading. However, would agreement that also included later copies of manuscript Y (i.e., Y1, Y2, Y3, Y4, etc.) increase this likelihood even more? Perhaps not, since the subsequent copies are dependent upon Y. But showing agreement with a third unrelated text (manuscript Z), though, could increase it. Please correct me if I am wrong. Thus, other factors must be considered, not just the total number of manuscripts.

      That seems to be the reason why Ehrman poses five criteria for establishing the text: 1. age of the manuscript, 2. distribution of the manuscripts, 3. consideration of style, 4. the more difficult reading, 5. quality of the manuscripts (p. 29-31). In my reading of him, agreement between manuscripts is beneficial, but only alongside these other methods. The problem is that too many people cite the 5000+ manuscripts as evidence without filtering out the manuscript that fail to add any possible knowledge of the originals.

      In terms of his "mixed signals," I see it as being analogous to the weather. A meteorologist can be reasonably confident about tomorrow's weather, but can never know for certain. That's the nature of inductive reasoning. Absolute confidence is unobtainable. This caution recognizes the limitations of the methods of investigation.

    5. D.
      I have to agree with Peter, Bart does send mixed signals. This is particularly true when you compare what he writes in his academic works with his popular writings.
      He has consistently maintained in his academic works that we can know with reasonable certainty what any New Testament book originally contained, yet in his popular writings, he gives the impression that we have zero chance of knowing what any New Testament writer wrote.
      Who knows what he really thinks, certainly not Bart.


    6. Timothy,

      Are you counting Ehrman's NT Introduction among his scholarly works or not? Also, would you be so kind as to provide some examples of these mixed signals?

  4. Peter Gurry,
    << The places where we have the most data reveal that the majority of extant variants is actually found, not in the earliest manuscripts, but in our later manuscripts. >>

    Are you sure about that? One might say that there are more variants in MSS made after 500 than there are in MSS made before 500, simply because there are so many more MSS extant that were made after 500.

    But it seems very probably to me that if you were to take in any any two MSS made before 500, and count the differences between them, and then take in hand any two MSS made after 500, and count the differences between (MSS with overlapping contents I mean), you would find far more differences between MSS made before 500 than between MSS made after 500.

    1. Any two? Actually, no. 33 and 35 show 86% and 88% agreement in the CBGM for the Catholic Letters and Acts, respectively. 01 and 03 show 87% and 92% in the same. But the point I take it you’re trying to make only reiterates that agreed reference points make all the difference when we’re comparing things. And therein lies the problem with the comparative arguments I’m dealing with here.

    2. Peter Gurry,
      << Any two? Actually, no. 33 and 35 show 86% and 88% agreement in the CBGM for the Catholic Letters and Acts, respectively. 01 and 03 show 87% and 92% in the same. >>

      Touche; I have to grant that *if* one only considers the parts of two early MSS where they agree the closest, *then* one can find pairs of later MSS that don't agree as closely. But if one does not cherry-pick specific portions of their texts, what is the total agreement-rates betwen B and Aleph for all parts of the NT that they share?

    3. Well, I didn’t “cherry-pick” anything. Acts and the Catholic Letters are what we have extensive data available for. But if you want to use the ECM data for parallel pericopes which is much more selective, we still get 84.2% agreement between 33 and 35 and 84.9% for 01 and 03.

    4. Peter Gurry,
      << Acts and the Catholic Letters are what we have extensive data available for.>>

      Are you saying then that there is no way to verify your claim as far as Gospels-manuscripts are concerned?

    5. Nope. I’m saying the amount of data currently available is different. That’s all.

    6. Peter, Okay; let's work with that: in a world where B and Aleph only contained the General Epistles, they would agree with each other better than some pairs of later MSS agree with each other. Are there any other pairs of MSS from before 500 about which this can be said?