Selective Reading and Unsubstantiated Criticism

A Response to David Pastorelli: La mise en oeuvre de la cohérence prégénéalogique dans le cadre de la Coherence-Based Genealogical Method: évaluation critique. BABELAO 10-11 (2022) 169-188

David Pastorelli claims to have detected a major flaw in the CBGM: a bias in favor of the Byzantine witnesses due to a “dysfunction of pre-genealogical coherence” (p. 187). He considers pre-genealogical coherence to be “paramount in the implementation of the method”,[1] while genealogical and stemmatic coherence are only mentioned in passing. The claim of the philologists using the CBGM, the editors of the ECM in the first place, who assure that philological assessment of variants is at least as important as the coherence-related calculations, is disregarded altogether. Having thus reduced the object of his criticism to a handy format Pastorelli tries to show pre-genealogical coherence to be a “fallacious employment of percentages and averages”.[2] Instead of using the CBGM, he recommends a return to the text-type theories of the 20th century which he considers to represent facts that need no further discussion.

The CBGM is a computer-aided philological method that in a fair scientific debate must not be reduced to one of its elements. Had Pastorelli read Mink’s introductions to the CBGM more carefully, he could not put forth a sentence like this:

Pre-genealogical coherence is the type of coherence most important for establishing the initial text.[3]

Moreover, he would not confuse the terms “substemma” with “local stemma” or textual flow diagrams with the global stemma.[4] Nevertheless, let us see whether Pastorelli’s criticism of pre-genealogical coherence as such is valid.

For the main part of his critical review Pastorelli refers to the chapter on pre-genealogical coherence in an introduction to the CBGM by Wasserman and Gurry.[5] In this chapter Wasserman and Gurry demonstrate how pre-genealogical coherence could be applied to two variant passages:

Mark 1:1 om. υἱοῦ τοῦ θεοῦ

Matthew 16:27

Using data available at <> Wasserman and Gurry show that the pre-genealogical coherence of the attestations of Mk 1:1b and Mt 16:27b and c is weak. This means that for several or all of the respective witnesses their close relatives do not share the same variant. This is correctly interpreted as a symptom of multiple emergence of these variants.

Pastorelli criticizes this method in three points.

1) The calculation of pre-genealogical coherence

Pastorelli constructs an example of two witnesses x and y differing from each other in two passages. The agreement rate of these two would be zero for the variant passages comprised by the sample. Then a third witness z would be added. This witness would differ from x and y at 98 passages. On this basis the agreement rate for x and y would rise to 98%. Concerned by this result Pastorelli asks, “What should the proportion of witnesses and variant passages be?”[6] – The answer is simple. As Mink has repeatedly emphasized, all relevant evidence has to be taken into account, and that is, in the case of the Greek New Testament, the total of variants yielded by a full collation of all Greek witnesses included in the critical apparatus of a writing.[7] The CBGM does not claim to produce useful results for a selection of two or three manuscripts. Pre-genealogical coherence is not about extrapolating on the basis of samples.

2) The delimitation of the variant passages

Pastorelli criticizes the delimitation of variant passages in the ECM apparatus as arbitrary without substantiating this proposition with a single example. Instead, he refers to an unsubstantiated statement of Bengt Alexanderson: “This is all arbitrary, a ‘place of variation’, a reading, a variant, a passage can be anything.”[8]

The terms “reading” and “variant” are well defined for the ECM:

“A reading is the generic term for the wording of a textual unit in which a manuscript is distinguished from one or more or from all other manuscripts. A variant refers to one of at least two readings of the same textual unit which is grammatically correct and logically possible in its context. Errors are readings which do not fulfil these criteria. [...] Alternative and orthographically possible forms of the same variants are classed as orthographica.”[9]

The delimitation of variant passages and, correspondingly, the segmentation of a critical apparatus is a complex editorial task. Mink says,

“Places of variation are places in the text where variants appear. At least two different variants occur in a place of variation [...]. A place of variation may comprise more than one word, but it can also be the space between two words. Ideally, it covers a logical unit of variation. This means that mutually interdependent changes to a text should belong to one unit of variation (e.g. if a subject and correspondingly the predicate are put in the singular). A unit of variation can also be postulated when a group of words presumably belonged together in a copyist’s view (e.g. if a word group consisting of article/particle/noun shows changes in different combinations for the article/noun and for the particle). Sometimes, very pragmatic considerations might be adduced to determine a unit of variation, so as to enable the comparison of all texts at a certain place. Places of variation may also overlap. In one place of variation the question may be e.g. whether a rather large group of words has been omitted or not; yet another instance of variation may result from variants within that group of words whenever it was not left out.”[10]

Any editor who ever constructed a critical apparatus will agree that the delimitation of variant passages has a subjective element. Still, it is possible to derive quantitative data as a basis for pre-genealogical coherence from a comparison of the included witnesses at all variant passages since the underlying database contains a statement for each witness at each variant passage as either containing one of its variants or being deficient.

3) Witnesses may contain mixed texts

Pastorelli points out that the textual character of 579, 037 and 032 changes due to block mixture. It is true that this was not taken into account by Wasserman and Gurry. This does not call into question, however, their overall result, namely the lack of pre-genealogical coherence of Mk 1:1b and Mt 16:27b, because the attestations of these variants do not comprise these witnesses.

One argument that Pastorelli puts forth against taking account of pre-genealogical coherence is the use of threshold values in Wasserman’s and Gurry’s treatment of Mk 1:1 and Mt 16:27. Once again Pastorelli tries to turn a tool made for a completely different purpose against the CBGM. He states,

The most important impact of pre-genealogical coherence in statistical terms is the arbitrary decision to fix a threshold value below which the witnesses are ignored in the comparison. For a given manuscript, this threshold value is its percentage of agreement with the majority text. For example, this percentage of agreement for 09 is 96.3% which does not allow to take more than three witnesses into account: 07 (97.0%), 1341 (96.8%), and 031 (96.4%).[11]

As if it was the only means available for the study of pre-genealogical coherence, Pastorelli here refers to a clustering tool developed in the context of Parallel Pericopes, a special ECM volume regarding selected pericopes in the Synoptic Gospels:[12] <>. The data basis and the purpose of the two clustering tools offered on this site are clearly explained on the start page:

The two clustering tools may be used to compare groupings found through “Text und Textwert” with ones based on the full collations for “Parallel Pericopes”.[13]

A third tool called “Find Relatives” is introduced on the same page as related to pre-genealogical coherence:

“Find Relatives” applies the “Parallel Pericopes” groupings to attestations of variant passages. It is designed to show an important aspect of the Coherence-Based Genealogical Method (CBGM): pre-genealogical coherence in attestations.

All three tools are explained by comprehensive guides accessible from their individual interfaces.

Pastorelli does not care about such complexities when he launches his attack against the threshold value in the lists of relatives produced by the Parallel Pericopes clustering tool. Neither does he care about tools developed after 2011 in the context of the ECM. In 2013, the second edition of the Catholic Letters appeared along with a suite of CBGM tools related to these writings here. In 2017, Acts appeared along with an online counterpart here. The same applies to Mark, which appeared in 2021. For Acts and Mark, CBGM tools are available here. Had Pastorelli cared to look at these tools he could have seen that none of the lists of relatives provided for each included witness is ever cut off due to a threshold value.

In 2016, Wasserman and Gurry used the Parallel Pericopes clustering tool for their demonstration of pre-genealogical coherence in a brief introduction to the CBGM. Pastorelli’s criticism against the CBGM is based on this introduction and the clustering tool published in 2011. He ignores the CBGM tools and related literature published since. It is probably due to such selective reading that Pastorelli thinks a Byzantine variant to be automatically preferred by ECM editors, just because its attestation is pre-genealogically coherent:

Well, as far as the Byzantine text-type features the strongest homogeneity, a characteristically Byzantine reading definitely shows the strongest coherence and automatically obtains the preference (reading a).[14]

However, about ten percent of the majority readings listed in the ECM apparatus have coherent attestations but still are deemed secondary, because internal evidence argues against them. The ECM commentary on Mk 1:1/12-16 is a good example for a discussion of all evidence relevant for the decision in favor of a majority reading.[15] Coherence provides only one argument among many others.

The reasons for a preference of 36 majority readings in Acts are spelled out in general terms in the introduction to ECM Acts:

Since the Textus Receptus was overcome by the scholarly textual criticism of the 19th century, there is tenacious negative bias against the Byzantine majority text. Wherever well-known, older textual witnesses like Vaticanus and Sinaiticus, and even more so in combination with a papyrus, stand against the majority of minuscules, the decision against the majority text was often made easily, without seriously considering the quality of the variants in question. Therefore, the editors of the present edition have taken two facts as paramount.

First, it is often overlooked that in the vast majority of variant passages only a few witnesses differ from all the others. As a rule, the popular witnesses from the 4th/5th centuries and, if extant, from even earlier papyri, agree with the majority of all witnesses. This implies that at all these passages the old age of the majority text is not in doubt.

Second, it is necessary to distinguish consistently between a manuscript and the text transmitted in it. “Recentiores non deteriores” is a principle widely accepted in editing philology, but in New Testament scholarship it was applied only to a few younger manuscripts featuring similar textual peculiarities as Vaticanus and Sinaiticus. For the reason given above, it is undoubtedly true that the textual tradition as a whole goes back to a very early period and that the coherent transmission of the majority of all textual witnesses provides a strong argument for, not against, the variant in question.

If the bias against the text of the majority of all witnesses has been overcome, then the variants transmitted by the majority will appear in a different light, even if some early witnesses read differently. It can then be considered with due impartiality whether or not a majority reading does in fact follow the tendency towards the fuller, easier, more smooth variant. There can be no doubt that this tendency exists, but it applies to the transmission on the whole, not only with scribes of younger manuscripts. It is true that variants of this kind accumulated in the majority text, but in more than a few cases the more difficult variant is in the majority text. Moreover, the editorial team of the ECM sees a strong external criterion in favor of the majority reading where a variant with A-related attestation is confirmed by the majority, because this points to a continuous transmission since the early period.

As a consequence, the text of ECM Acts agrees with the majority variant in 36 out of 52 cases where textual decisions were made against NA28. There are only two cases where a decision was made against the majority variant in NA28.[16]

Moreover, Pastorelli purports that by preferring variants because of pre-genealogical coherence users of the CBGM had re-introduced the number of witnesses as a criterion for the assessment of an attestation.[17] This contention is just as unsubstantiated as is the purported automatic preference of pre-genealogically coherent attestations.

Another false proposition refers to the Byzantine text. Without looking for a confirmation by ECM editors Pastorelli cites Wasserman and Gurry saying that we still see the Byzantine text form as a text-type.[18] It cannot be denied, of course, that the late Byzantine text has reached a relatively stable form, but this is not a valid reason for a partial return to the text-type theory. If we abolish the concept of text-types, it follows that we can no longer use the term “Byzantine text” as defined by Metzger. It has become obsolete to consider the Byzantine text form as “based on the recension prepared near the close of the third century by Lucian of Antioch”,[19] but many still see it as the last text-type standing. Having emerged from the recension hypothesis, however, the term “text-type” should be given up completely.[20]

If we use the term “Byzantine text” it serves as a short form of “late Byzantine majority text”. If we do so, we are aware that it means the last phase of a process whose beginnings are marked by manuscripts like Vaticanus and Sinaiticus. The transmission of the Greek New Testament forms a continuum whose overall structure is still calling for an adequate description. The goal is to understand how the late Byzantine majority text developed and which factors were at work in this process. New research is due starting from the genealogical relationships between variants and an exploration of the relationships between their witnesses.



Novum Testamentum Graecum. Editio Critica Maior

Ed. by the Institute for New Testament Textual Research.

Vol. I The Synoptic Gospels, Part 2 The Gospel of Mark, ed. by Holger Strutwolf, Georg Gäbel, Annette Hüffmeier, Marie-Luise Lakmann, Gregory S. Paulson, and Klaus Wachtel. Stuttgart: German Bible Society 2021.

Part 2.1: Text, Part 2.2: Supplementary Material, Part 2.3: Studies.

Vol. III Acts of the Apostles, ed. by Holger Strutwolf, Georg Gäbel, Annette Hüffmeier, Gerd Mink, and Klaus Wachtel. Stuttgart: German Bible Society 2017.

Part 1.1: Text Chapter 1-14, Part 1.2: Text Chapter 15-28, Part 2: Supplementary Material, Part 3: Studies.

Vol. IV Catholic Letters, ed. by Barbara Aland, Kurt Aland†, Gerd Mink, Holger Strutwolf, and Klaus Wachtel. Stuttgart: German Bible Society, 2nd rev. edition 2013.

Part 1: Text, Part 2: Supplementary Material.

Parallel Pericopes. Special volume regarding the synoptic gospels, ed. by Holger Strutwolf and Klaus Wachtel, Stuttgart: German Bible Society 2011.

Metzger, Bruce M.: The Text of the New Testament. New York/Oxford: Oxford University Press 31992.

Metzger, Bruce M., and Bart Ehrman: The Text of the New Testament. New York/Oxford: Oxford University Press 42005.

Mink, Gerd: Contamination, Coherence, and Coincidence in Textual Transmission, in: The Textual History of the Greek New Testament. Changing Views in Contemporary Research, hg. v. Klaus Wachtel und Michael W. Holmes, (SBL Text-Critical Studies 8) Atlanta 2011, p.141-216.

Mink, Gerd: Problems of a highly contaminated tradition: the New Testament. Stemmata of variants as a source of a genealogy for witnesses, in: Studies in Stemmatology II, ed. by P. van Reenen, A. den Hollander and M. van Mulken, Amsterdam [u.a.] 2004, [13]-85., Corrigenda here.

Text und Textwert der griechischen Handschriften des Neuen Testaments, ed. K. Aland et al. Vol. IV Die Synoptischen Evangelien, 1 Das Markusevangelium; 2 Das Matthäusevangelium; 3 Das Lukasevangelium (Berlin/New York: De Gruyter, 1998-1999). Vol. V Das Johannesevangelium, 1 Teststellenkollation der Kapitel 1-10 (Berlin/New York: De Gruyter, 2005).

Wasserman, Tommy and Peter Gurry: A New Approach to Textual Criticism: An Introduction to the Coherence-Based Genealogical Method. Atlanta: Society of Biblical Literature 2017.



[1] Pastorelli 174: “capitale dans la mise en oeuvre de la méthode”.

[2] Pastorelli 180: “utilisation fallacieuse des pourcentages et des moyennes”.

[3] Pastorelli 170: “La cohérence pré-généalogique est le type de cohérence le plus important pour l’établissement du texte initial.”

[4] Cp. Pastorelli 172.

[5] Wasserman/Gurry 2017, p. 37-58.

[6] Pastorelli 178: “Quelle proportion entre témoins et lieux variants faut-il avoir?”

[7] Cp. Mink 2011, p. 145-146.

[8] Pastorelli 179.

[9] ECM IV.1, p. 27*; ECM III.1, p. 24*; ECM I.2.1, p. 16*-17*.

[10] Mink 2004, 27-28.

[11] Pastorelli 181: “L’impact le plus important de la cohérence pré-généalogique en terme statistique est la décision arbitraire de fixer un seuil en dessous duquel les témoins sont ignorés dans la comparaison. Pour un manuscrit donné, ce seuil est son pourcentage d’accords avec le texte majoritaire. Par exemple, ce pourcentage d’accords pour 09 est 96,3%, ce qui ne permet de prendre en compte plus que trois témoins : 07 (97,0%), 1341 (96,8%) et 031 (96,4%).”

[12] ECM: Parallel Pericopes, ed. Holger Strutwolf and Klaus Wachtel. Stuttgart: Deutsche Bibelgesellschaft 2011.

[13] The tool called “T&T Mss. Clusters” is based on Text und Textwert der griechischen Handschriften des Neuen Testaments, ed. K. Aland et al. Vol. IV Die Synoptischen Evangelien, 1 Das Markusevangelium; 2 Das Matthäusevangelium; 3 Das Lukasevangelium (Berlin/New York: De Gruyter, 1998-1999). Vol. V Das Johannesevangelium, 1 Teststellenkollation der Kapitel 1-10 (Berlin/New York: De Gruyter, 2005).

[14] Pastorelli 182: “Or, dans la mesure où le type de texte byzantin présente l’homogénéité la plus forte, une leçon byzantine caractéristique détient, à coup sûr, la cohérence la plus forte et elle reçoit automatiquement la préférence (leçon a).”

[15] ECM I.2.3, p. 9-10 or here.

[16] K. Wachtel: Notes on the text of the Acts of the Apostles, in ECM III.1,1 p. 30*-31*. By the way, this text also comprises a report about the application of CBGM procedures, including pre-genealogical coherence, to Acts.

[17] Pastorelli 182-183.

[18] Pastorelli 182.

[19] Metzger, Text of the NT, 31992, 212. – The 4th edition of Metzger’s “Text of the New Testament”, co-authored by Bart Ehrman, says about the term “Byzantine text” that “its final form represents a slowly developing tradition, not one that sprang up immediately at one time and place” (42005, S. 279).

[20] For a brief discussion of the text-types as defined by Metzger for Mark see Wachtel, Notes on the Text of Mark, in ECM I.2,3 p. 1-7.

Add Comment
This is good stuff. Will it be published. (Please publish it in a print journal.)
Posted on 17/10/23 08:36.

Recent Bloggers Recent Bloggers

Darius Mueller
Posts: 2
Stars: 5
Date: 17/06/24
marie-luise lakmann
Posts: 3
Stars: 12
Date: 29/05/24
Katie Leggett
Posts: 1
Stars: 3
Date: 04/05/24
Greg Paulson
Posts: 24
Stars: 68
Date: 19/01/24
Klaus Wachtel
Posts: 2
Stars: 3
Date: 13/03/23
Theodora Panella
Posts: 2
Stars: 6
Date: 09/03/20
Jan Graefe
Posts: 2
Stars: 0
Date: 18/09/19
Siegfried G. Richter
Posts: 1
Stars: 3
Date: 01/07/19
Nikolai Kiel
Posts: 1
Stars: 4
Date: 06/06/19