Skip to content

An Introduction to the Coherence Based Genealogical Method in NT Textual Criticism

April 5, 2018

For several years, scholars and students have puzzled over the Coherence-Based Genealogical Method (CBGM) developed in the Muenster Institute for New Testament Textual Research.  There is now a clear and helpful guide:  Tommy Wasserman and Peter J. Gury, A New Approach to Textual Criticism:  An Introduction to the Coherence-Based Genealogical Method (SBL Press, 2017), the publisher’s online catalog page here.  As I read the book in pre-publication stage, and am one of those who endorsed it, I’m a bit tardy in giving this notice of it.

The authors patiently take readers step by step, explaining how the CBGM relates to previous approaches, and then introducing and illustrating the specific procedures involved in the application of the CBGM.  I think that the authors have made the CBGM as clear as possible for those new to it, but readers will still have to follow their discussion carefully and slowly (best to have a large pot of coffee to hand!).  The CBGM both builds on more familiar text-critical steps, and also introduces some new ones, along with some new concepts and some technical terms that will be new to most.

Those of us who engage in study of the textual transmission of the NT (or any other early text) will now have no excuse for treating the CBGM as terra incognita.

From → Uncategorized

  1. WordPress Reader permalink

    After reading this post, I googled the CBGM and found this article, which I thought helped explain it somewhat too:

    Click to access JETS_59-4_675-89_Gurry.pdf

    I’m a complete outsider to this field, but what strikes me is how similar theoretical and computational ideas must underlie the various models of change over time in texts, in DNA, in pottery, in linguistics, etc.

    • Thanks for this reference. Gurry is, of course, one of the authors of the recent introductory book on the CBGM.

  2. Donald Jacobs permalink

    Thanks, now we know where to go for a coherent explanation of this new method.

  3. Robert permalink

    I believe this is the method I learned from Joel Delobel in Leuven nearly 30 years ago. Didn’t seem all that difficult to me. In fact is was a rather refreshing departure from the numerous characterizations of text types I had been exposed to previously. But maybe I’m missing something?

  4. Dr. H.,
    This book is definitely the most understandable explanation of the CBGM to date. Gurry and Wasserman raise a few questions about the methodology as well. Gurry raises a few more in some follow up articles.
    I am glad to see not only the clarity of the presentation, but the critical engagement that seemed to be lacking in the main, excepting Jongkind, Carlson and Porter/Pitts.


  5. From comments I’ve seen elsewhere the issue of the ‘starting point’ when using the CBGM seems to be hard to grasp, in particular comments about the CBGM favoring one particular text type. As the CBGM is ultimately an iterative method it is perhaps hard for anyone not well-versed in statistical techniques (and I don’t count myself as well-versed) to understand how the CBGM can produce useful results when it apparently ignores various kinds of classification (ms ages, text types, etc.) that have been the backbone of TC in the past.

    • David, You really should read the new book by Wasserman or some other good explanation of the CBGM. You don’t seem to have grasped it correctly.

      • I have the book, and on P136 I find a definition stating that the ‘initial text’ “represents the hypothetical witness from which all the extant witnesses derive.” Ok, but the ‘initial text’ produced by the CBGM is influenced by the choices regarding the data used as input to the CBGM, and so the CBGM output can be biased towards the texts used as input, e.g. if only Byz texts are used as input then the CBGM will produce a text that may also be Byz, but not necessarily the same as any of the input texts.

        The point that I was clumsily trying to make is that as the CBGM input is text, and the output is also text, it is possible to use the output as input in another ‘run’ of the CBGM (and so on), with each output being an ‘initial text’ that may be closer to the ‘original text’ than before. Of course, using the output as an input in another run may simply be impracticable, and if there is very little change between input and output it is unlikely to be worth it, but in theory any ‘bias’ in the CBGM could be eliminated by feeding the output back as an input.

        Now, given that much of the CBGM is not (cannot be?) automated, what I am suggesting is very unlikely, and instead anyone using the CBGM has to be aware of ‘garbage in, garbage out.’ If I’ve still completely messed up what I am trying to say then I apologize. 😦

      • David: As the developers of the CBGM point out, the method doesn’t remove the decision-making of textual critics, but rests upon it. At every point where there is variation, it is first necessary to judge which variant reading is more likely to be “original”, i.e., to account best for the other variants. Then, the cumulative effect of these “local” decisions is put into the computer and the programme generates a “textual flow” display showing how the text of the writing developed across time and transmission. So, yes, a different set of decisions about “local” variation-units would generate a different picture.

  6. David Harwood permalink

    I’ve read some articles about CBGM, and know there are two new books about it. For one thing, the estimated dating of texts is not considered, as I understand it. Am I mistaken?

    Also it’s somewhat hard to believe that a relation of ancestry can determined by considering the numbers of apparent local derivations. I would assume that a copyist or editor is employs as many antecedents as he may have.

    What am I misunderstanding?

    Thank you,


    • David: The crucial thing is that the CBGM doesn’t trace relationships of manuscripts, but relationship/flow of text. Manuscripts are listed that embody forms or stages of development in a given text. So, e.g., an early stage/form of a given text may be well represented in a manuscript of a comparatively late date.

Comments are closed.

%d bloggers like this: