Wednesday, January 16, 2008

Review of the Digital Textual Commentary of Metzger

As announced on this blog here, Bruce Metzger’s A Textual Commentary on the Greek New Testament (2nd ed. United Bible Societies/Deutsche Bibelgesellschaft, 1994) has been digitalized and released by a number of software companies. This review treats the version available from Accordance, which has been tagged so as to allow a number of valuable search capabilities (for details and orders, see here).

Now, if one has the printed version, why bother about the digital? The first reason is obvious: If you regularly use a digital UBS GNT, why not integrate it digitally with its companion volume, in order to facilitate your study and work? The digital version opens up a whole world of search possibilities that can enhance the use of this valuable resource. Thus, apart from general text-critical guidance, equivalent to the printed version, the search capacity of the digital version allows for more detailed exploration of various aspects relating to the extensive work of the committee. In addition, I think the digital version will be particularly useful when scholars want to find representative examples of various textual phenomena – examples, since the commentary contains a very limited selection of variants in the first place – to the extent that one agrees with the explanations and definitions of the committee.

In the Accordance search fields for this module one can search in the following categories: “Reference” (the passages discussed), “Titles” (words in the titles), “English Content,” “Scripture,” “Greek Content,” “Hebrew Content,” “Transliteration,” “Manuscripts,”” Syriac,” “Certainty,” “Uncial Greek.” These can be combined in advanced searches (with “AND,” “OR,” “NOT” commands). In the following, I will briefly go through some of these categories with selected examples of possible searches.

1. Reference
I start in the category “Reference,” in order to identify all the passages discussed in the commentary, as distributed book by book. Below I have counted only the main references (in a few cases a discussion of a certain passage extends to another passage):

Matthew 181
Mark 150
Luke 191
John 184
Acts 545 (!)
Romans 97
1 Corinthians 91
2 Corinthians 50
Galatians 38
Ephesians 43
Phillipians 29
Colossians 33
1 Thessalonians 19
2 Thessalonians 11
1 Timothy 25
2 Timothy 15
Titus 10
Philemon 7
Hebrews 49
James 26
1 Peter 46
2 Peter 26
1 John 39
2 John 9
3 John 3
Jude 15
Revelation 92
Total: 2028 passages

The very large number of passages in Acts (over 1/4 of the passages) is of course due to the significantly different “Western” text of Acts as represented by Codex Bezae.

2. English Content

2.1 Theological Motivation behind Textual Alteration
As is well-known, the committee frequently refers to possible theological motivation behind textual variants. The digital version facilitates a closer look at such places in the commentary. In order to identify them, I used some relevant keywords like “Christological,” “theological,” etc within the English content search field. I have only included references that are relevant in relation to scribes (i.e., not a reference to e.g., an author’s theology or the like). In a few cases the keywords overlap or occur more than once in the same passage. Apparently, in one case theological motivation is explicitly rejected, e.g., in Phil 2:7 where the variation is described as “non–doctrinal.”

a) keyword “theological” renders 12 hits concerning the following ten passages Luke 2:38; 11:4; 16:12; 24:51; 24:53; Acts 2:41: 5:32; 9:22; Rom 9:4; 1 Pet 1:22.

In Acts 2:41, for example: “The substitution in D of πιστεύσαντες for ἀποδεξάμενοι was doubtless motivated by theological concern that faith in, and not merely reception of, the word preached by Peter is prerequisite to receiving baptism.”

Of the ten passages four receive A-rating, two B-rating and four are unrated.

b) keyword “pious” [copyists/scribes/glosses, etc] renders 11 hits in Luke 23:43; John 8:8; Acts 15:29; 28:31; 1 Cor 6:11; 2 Cor 4:6, 14; Phil 3:12; 2 Thess 2:8; 1 Tim 6:5; Rev 22:21.

In John 8:8, for example: “In order to satisfy pious curiosity concerning what it was that Jesus wrote upon the ground, after γῆν several witnesses (U Π 73 331 364 700 782 1592 armmss) add the words ἕνος ἑκάστου αὐτῶν τὰς ἁμαρτίας (“the sins of every one of them”).” See my article on the subject here. Codex Π, by the way, does not contain John 8:8, so this is an error both in digital and printed version of the commentary.

Four of the eleven passages receive A-rating, one B-rating, two C-rating and four are unrated.

c) keyword “piety” renders two hits in Gal 1:3 and 2 Pet 1:2. The comment on the latter passage says: “Other readings incorporate various amplifications reflecting the piety of copyists.”

One passage received A-rating, and the other B-rating.

d) keyword “doctrine” renders 3 hits in Luke 2:33, 41 and 43, all relating to the doctrine of the virgin birth.

For example, the comment on Luke 2:33 reads: “In order to safeguard the doctrine of the virgin birth of Jesus, ὁ πατήρ was replaced by Ἰωσήφ in a variety of witnesses . . .”

One passage receives B-rating, two are unrated.

e) keyword “doctrinal” renders 9 hits in eight passages: Matt 24:36; 28:20 (footnote 6); Luke 1:46; 24:53 (in footnote 21); Acts 1:2; 9:20; Phil 2:7 (“non-doctrinal”); 1 John 5:20.

In Matt 24:36, for example, the comment reads: “The omission of the words [οὐδὲ ὁ υἱός] because of the doctrinal difficulty they present is more probable than their addition by assimilation to Mk 13.32.”

Out of the eight passages, four have A-rating, one has B-rating, three are unrated.

f) keyword “Christological” renders 2 hits in Luke 24:53 and John 3:13.

In both cases the members of the committee were split in their opinion. For example, in Luke 24:53 θεόν {A}: “During the discussions a sharp difference of opinion emerged. According to the view of a minority of the Committee, apart from other arguments there is discernible in these passages a Christological–theological motivation that accounts for their having been added . . . On the other hand, the majority of the Committee, having evaluated the weight of the evidence differently, regarded the longer readings as part of the original text.”

One passage receives A-rating, the other B-rating. This is of course strange considering the “sharp difference of opinion.”

g) keyword “Christo-centric” renders 1 hit in Luke 8.3: “The plural [αὐτοῖς] is supported by good representatives of the Alexandrian and the Western text–types; the singular (compare Mt 27.55; Mk 15.41) appears to be a Christocentric correction, due perhaps to Marcion.”

The passage receives a B-rating.

h) keyword “reverence” renders 3 hits in Luke 24:53; John 11:33; Acts 4:24.

For example, the comment on Acts 4:24 says: “The shortest form of text appears to be the oldest; the additions were doubtless made in the interest of heightening the apostles’ reverence in prayer.”

Two of the passages have B-rating, one is unrated.

i) keywords “in the interest” (followed by reference to theological motive) occur in Luke 2:41 (see above); 23:42; Acts 1:14; 4:1, 19, 24 (see above), 32; 6:10-11; 9:31; 12:2; 15:32; 1 Cor 7:5; Gal 1:3.

For example, in Gal 1:3 the comment says: “The apostle’s stereotyped formula was altered by copyists who, apparently in the interest of Christian piety, transferred the possessive pronoun so it would be more closely associated with ‘Lord Jesus Christ.’”

Two of the passages receive A-rating, four B-rating, whereas seven are unrated.

I think these preliminary search results cover this aspect of the commentary rather well, broadly speaking, at least for this brief review (the passages in question would need more analysis and detailed categorization). I am sure there are more examples and, in fact, it is very difficult in many cases to define and identify “theological motivation.” Nevertheless, the impression is that these passages, where the committee is rather explicit on the matter, are very few in number as compared to the total of over two thousand passages contained in the commentary.

In addition, we find that passages in Luke and Acts are overrepresented, and this is clearly due to the many references to the “Western” text of Codex Bezae (D05) in Luke and Acts with its “theological tendency.”

To conclude this section, I noted, to my surprise, that the keywords “intentional” rendered only two hits, and “consciously” one hit, neither having anything to do with theologically motivated alteration.

2.2 Different Opinion among Committee Members

The next brief survey in the English Content search category relates to places where the committee members were in disagreement. First I looked at those specific places where individual members expressed their minority opinion explicitly by indicating their initials (when, as explained in the preface, they felt that “the majority had seriously gone astray”).

Since the possibility to search is limited to a list of individual search words, probably due to some technical reason, this type of search (for initials) demanded special search strings like “B * M” (for Bruce M. Metzger), where * in this case stands for the period.

The search gave the following results:

Bruce Metzger (B.M.M.) twenty-six passages, two of which were A-rated; five B-rated; twelve C-rated; two D-rated; and five unrated

Allen Wikgren (A.W.) twelve passages, five of which were B-rated; five C-rated: one D-rated; and one unrated

Kurt Aland (K.A.) three passages, two of which were C-rated; and one D-rated

Carlo M. Martini (C.M.M.) one passage, which was B-rated

Matthew Black (M.B.) none

As I had suspected (but not verified) from my experience with the printed version, Bruce Metzger most often expressed his different views (more than the other members in total). As a side note, I came to understand in the Metzger memorial session at the recent SBL Meeting in San Diego that C. M. Martini frequently would act as the mediator when there was disagreement among the members, so that a decision could finally be reached in spite of heated discussions. (Aside from an irenic nature, one can suspect that Martini’s pastoral experience and good command of German and English helped in this regard.)

Secondly I looked for the keywords “majority” and “minority” wherever they refer to the majority or minority of the committee. When there is explicit reference to an alternative view of a “minority,” this generally indicates a stronger disagreement between members than the simple reference to the “majority” view/decision (which of course always supports the printed reading).

a) keyword “minority” occurs in the following twenty-two passages: Matt 6:33 (C-rating); Mark 5:21 (C); Luke 10:32 (unrated, but square brackets); 22:17-20 (B); 24:3 (B); 24:6 (B); 24:40 (B); 24:51 (B); 24:52 (B); 24:53 (A); John 3:13 (B); 5:2 (C); 8:16 (A); 10:26 (B); 13:2 (B); 13:22 (B); 1 Cor 1:2 (unrated); 2 Cor 7:8 (C); Col 1:20 (C); 2:23 (C); 1 John 5:10 (B); Rev 12:10 (unrated).

I think it is remarkable that two of these passages receive an A-rating.

b) keyword “majority” occurs roughly 400 times referring to a majority of the committee. We can infer that in the other approximately 1600 passages, the committee was unanimous in their decision.

3. Manuscripts
In this category, it is possible to search for all textual witnesses (MSS, versional and patristic evidence) cited in the commentary, which extend well beyond the GNT apparatus itself (a list of all cited evidence is available). It should be noted, however, that this function is not suitable for any general statistics of textual affiliation because of the selective nature of incuded passages and manuscript citation. Nevertheless, this search function might aid in the search for particular examples.

A search for some examples of single witnesses renders the following results:

P45 is cited 61 times in the commentary (in somewhat less passages, since it may occur more than once in a discussion – the same goes for most examples below)
P46: 269 times
P66:75 times
P72:68 times
P75: 71 times
ℵ: 1114 times
A: 778 times
B: 926 times
C: 621 times
D: 1103 times (both D05 and D06)
1739: 344 times
Ambrose: 12 times
bo mss: 37 times

The search for particular papyrus witnesses requires a spacing between the P-symbol and the number. The reason for this is because the special papyrus symbol is a different font (MSS), and is thus treated as a separate word from the number following, since Accordance does not support multiple font types for the same word.

It is possible to find combinations of MSS supporting a reading, e.g., ℵ A B C D occurs in 23 passages. It is more difficult when first hand/correctors are involved, or when one wants to find other “non-successive” manuscript combinations (e.g., all passages where B agrees with 1739). Then one can use a search string like B 1739, but it requires a lot of extra work, since the list may include passages where B and 1739 have competing readings.

4. Certainty

In this category one can search for the letter-ratings A, B, C or D. As scholars know (and often regret), there is a related but distinct system with square brackets for uncertain words in the printed text. Thus, it would have been useful if the software had allowed searches for square brackets in this category. However, it is still possible to find those places (ca. 250 passages) by searching for “square brackets” in the English Content category.

The result of the search for letter-rating is:

A: 507 passages
B: 538 passages
C: 368 passages
D: 9 passages (Matt 23:26; Mark 7:9; John 10:29; Acts 16:12; Rom 14:9; 1 Cor 7:34; 2 Peter 3:10; Jude 5; Rev 18:3)

These figures differ slightly from those indicated by Kent D. Clarke in his “Textual Certainty in the UBS’ Greek NT”, Novum Testamentum 113 (2002): 113, but, on the other hand, Clarke indicates the ratings in the edition, and not the commentary, and these figures are A: 514; B: 541; C: 367 and D: 9. If I look at a specific book and combine a search for “Certainty” with “Reference” I see that the software indicates 32 A-ratings in Matthew, whereas Clarke indicates 34.

In order to explain the discrepancy, I compared the printed commentary with the printed edition, and saw that in Matt 5:44 there are two A-letter ratings whereas there is one comment on the verse which is marked with “bis” (twice). On the other hand, I spotted an error in Matt 28:6 in the printed commentary, where an A-letter rating, present in the edition is missing (ἔκειτο). The digital version, of course, is most likely to contain all such errors and typos (eg. “Origin” for Origen in Rom 12:2) in the printed version, which is fine with me, since I think it could create more problems if the digital version attempted to correct the printed version.

Since we know that there are about 2030 passages discussed (see above), we may assume that approximately 600 passages were unrated (and these are not cited in the apparatus of GNT4). Maybe it would have been good for completeness sake to be able to search for “unrated” in the Certainty-category.

The digital version of Metzger’s Textual Commentary will not only save time for many students and scholars, but it is also likely to lead to new ways of using this standard reference work. The search capacity of the digital version does not allow any detailed studies of the textual affiliation of the witnesses since the commentary due to its nature is highly selective. Nevertheless, it does allow a more detailed exploration of various aspects relating to the work of the committee. Moreover, the digital version will be particularly useful when scholars want to find representative examples of various textual phenomena. We can be grateful to the United Bible Society and the software companies that they have made this fine resource available to us.


  1. This was interesting, not in the fact that it is digitized (we/ Logos have/has had a digital version of Metzger for years), but in the fact that this digitization is somehow different than ours. Understanding how another product works is often good fodder for ideas and improvements. In this case, I’m left wondering about certain disparities. For example, a search for “theological” in the Logos edition produces 82 occurrences in 61 places. That search also included footnotes, which the blogged version apparently does not, but a quick glance led me to think there were many more occurrences in the commentary than the 12 that the blog post search produced. “Pious” was a bit closer in hit results (ours had a few more), but even “piety” didn’t match – we had three compared to the two in the Accordance version, and no footnotes were at issue. These results made me wonder if the Accordance version was selectively tagged, or had a specific tagging philosophy. The same issue held true for searching for individual manuscripts. For example, the blog post search yield 68 occurrences of P72; our version had 136. This is likely (again) a footnote issue, as our search includes those, but perhaps that doesn't entirely explain the issue. Searching for the papyri just requires a right click in our version. Highlighting two MSS (or more) next to each other and then right clicking will only allow one to find those MSS next to each other elsewhere. For more than one MS not adjacent to another, our searching is akin to the blogged version. Generally, the two versions seem the same except for the footnotes and the right click. One advantage to Libronix, though, is that all such searches can be done simultaneously through the apparati of other texts and commentaries.

  2. Thanks Tommy,
    That was interesting.

  3. Dear msh,

    I have just browsed your comment, and may respond further, but I note that you write:

    "For example, the blog post search yield 68 occurrences of P72; our version had 136."

    Incidentally 68+68=136! This is because "P" (Papyrus) yields 68 + "72" yields 68 hits=136 hits in *68 places*, so please consider your search result again.

  4. Further response:

    MSH: "For example, a search for 'theological' in the Logos edition produces 82 occurrences in 61 places. That search also included footnotes, which the blogged version apparently does not, . . ."

    Note my remark in the introduction to that part of the investigation: "I have only included references that are relevant in relation to scribes (i.e., not a reference to e.g., an author's theology or the like)." It would be interesting if you tried again and looked carefully at all your references (82 occurences) to see how many refers to the theology of scribes. Maybe you will come up with 12. This probably explains all the discrepancies. It will be easy to check the three hits for "piety", which I also have, but I have not included the first hit, since I perceive that it does not ascribe the variation to the piety of scribes, whereas that is the case for the two other.

    Thus, you see that I have not simply done searches for the keywords, but put more effort into the matter.

    I may respond further, but that is all for now.

  5. MSH:

    On your last note: "One advantage to Libronix, though, is that all such searches can be done simultaneously through the apparati of other texts and commentaries."

    Please note that my Accordance version can also easily do that, but that was not at all the point of my review. I did not do a comparison of various competing releases of this product. I did a review from the perspective of the text-critic, in order to see how it can be used mainly in comparison with the printed version. I do not suspect there are major differences between the Logos/Libronix and the Accordance. I have a Mac and I use Accordance. And I am impressed with the many search capabilities of Accordance. But, as I have said in previous posts on similar subjects, there is a potential danger: the digital tools are amazing, but one must know 1) the various limitations of the corpus (in this case, e.g., the selectivity of the commentary); 2) what to search for; and 3) how to interpret the data.

    I really appreciate your response, and I hope I have now responded to most or all of your remarks.

  6. Ah - this is usually the issue (how counts are rung up). Thanks for that. That of course also happens in what is considered the same corpus when different teams tag that corpus. Hence I would agree that electronic searching should always be double-checked. One would suspect few differences, given that the digitized book is the same book, and typically comes from the same source (but not always).

  7. If there were only a way to get the digital version for free if you already have the printed version.

    Thanks for the excellent review. Accordance seems like a great software if you have deep pockets. I use Online Bible for Mac and Macsword, which offers for free everything the expensive basic Accordance package offers. The price of some Accordance modules are strange. $50 for tagged TR? Free with OLB/Macsword. $50 for Robinson-Pierpont? Come on, that's public domain!

  8. Great article! I just checked this blog, and I wanted to clarify a couple of things though I may be wrong. My counts differ from TW regarding John (182; TW: 184), Acts (543; TW: 545), Colossians (32; TW: 33), and Revelation (93; TW: 92) for a complete total of 2,022. Also, TW calls these "passages;" however, some of these text-critical issues appear in the same verse. While he surely knows that they are not passages per se, we should clear that up for as not to muddy the water.

    I just started my blog: where I "documented" or "declared" (seems too strong) these differences.