Wednesday, September 25, 2013

Didymus the Blind and the Text of Matthew

Bart Ehrman (in his 1986 book Didymus the Blind and the Text of the Gospels) identified 163 genetically significant textual variant-units in Didymus’ text of Matthew.  In Table 1 (on page 191), Ehrman ranked the witnesses which agree with Didymus’ text according to the percentage of agreements at those points.  Codex A has the highest percentage of agreement:  80%.  Ehrman tossed out that evidence on the grounds that Codex A, due to damage, only attests to 20 genetically significant variants (in Mt. 25:6-28:20). 


However, if Codex A should be ignored on the grounds that the data from Matthew 25:6-28:20 is too small of a sample, then why are 163 short passages from Didymus a large enough sample to demonstrate affinities with manuscripts that contain 1,071 verses? Picture the genetically significant textual variant-units that Ehrman collected as bits of an extremely fragmentary manuscript of Matthew – a manuscript that can be read only at those 163 places. Obviously this represents only a small portion of the text of Matthew; even if every variant-unit collected by Ehrman was an entire verse, the total would be only 15% of the verses in the Gospel of Matthew.

This problem can be approached in terms of variant-units, as well as in terms of verses.  Ehrman completely rejected Codex A’s testimony because, he said on page 190, “It should seem obvious that since A does not preserve even one-eighth of the total number of readings under consideration (20/163), its testimony must be discounted.”  But does Ehrman’s collection of 163 variant-units preserve even one-eighth of the total number of readings under consideration in collations of manuscripts of Matthew?  I sifted through the apparatus of the 27th edition of Nestle-Aland, and noticed that textual variants are listed for at least 765 verses of the Gospel of Matthew.  I did not count the number of individual variant-units that are in the apparatus for Matthew, but I reckon that there are over 1,100 (since, although many verses have no variant-units in the apparatus, many other verses contain more than one variant-unit).  If Codex A’s testimony must be discounted because it contains only 12.2% of the variant-units under consideration when Didymus is the object of the comparison, then shouldn't Didymus’ testimony be discounted because it contains only about 14.8% (163/1,100) of the variant-units under consideration when the entire text of Matthew is the object of the comparison?

Even though Codex A is extant only for Matthew 25:6-28:20, there is no reason to imagine that Didymus’ text of Matthew was block-mixed, as if it aligned with one text-type in the last four chapters but some other text-type in chapters 1-24.  That is, there is no reason to suspect that the 80% agreement between Didymus and Codex A in 25:6-28:20 would disappear if the rest of Matthew was extant in Codex A.

If we were to treat Matthew 25:6-28:20 as a separate book used by Didymus, what text-type would Didymus’ copy of this book have?  Would it be decidedly Alexandrian, or something else?  An easy way to find out is to sift through the data, and compare how often Didymus agrees with Codex A and the Textus Receptus to how often Didymus agrees with Vaticanus and Sinaiticus (the flagship manuscripts of the Alexandrian Text).  Consider the following list:

Places in Matthew 25:6-28:20 Where Didymus Agrees with TR or A or Aleph or B:
(1) 25:6 - Didymus has ECERCESQE - agrees with TR A Aleph B
(2) 25:6 - Didymus has GEGONEN - agrees with TR A Aleph (disagrees with B)
(3) 25:15 - Didymus has IDIAN DUNAMIN - agrees with TR A Aleph B
(4) 25:16 - Didymus has EN - agrees with TR A Aleph B
(5) 25:33 - Didymus has MEN - agrees with TR A Aleph B
(6) 25:33 - Didymus has DEXIWN - agrees with Aleph A (disagrees with TR B)
(7) 25:33 - Didymus has EUWNUMWN - agrees with TR A B (disagrees with Aleph)
(8) 25:41 - Didymus has OI - agrees with TR A (disagrees with Aleph B)
(9) 25:41 - Didymus has POREUESQE - agrees with TR A B (disagrees with Aleph) 
(10) 26:15 - Didymus has PARADWSW - agrees with TR A Aleph B
(11) 26:31 - Didymus has DIASKORPISQHSETAI - agrees with TR (disagrees with A Aleph B) (Ehrman listed this as a Citation “[C]” but not as a Citation taken to be representative of Didymus’ text, and used as a base for collation.)
(12) 26:52 - Didymus has MACAIRH - agrees with A Aleph B (disagrees with TR)  (Ehrman listed this as a Citation “[C]” but not as a Citation taken to be representative of Didymus’ text, and used as a base for collation.)
(13) 26:53 - Didymus has DOKEIS OTI OU DUNAMAI - agrees with TR A Aleph B (There is a blank space in Ehrman’s book where the letter “A” should be.)
(14) 26:53 - Didymus has MOI - agrees with TR A B (disagrees with Aleph)
(15) 26:53 - Didymus has PLEIOUS - agrees with TR A (disagrees with Aleph B)
(16) 26:53 - Didymus has DWDEKA - agrees with Aleph B (disagrees with TR A)

(17) 26:53 – Didymus has LEGIWNWN ANGELWN - agrees with Aleph A (disagrees with TR B)
(18) 27:40 - Didymus has EI TOU QEOU - agrees with TR A Aleph (disagrees with B)  (Ehrman listed this as a Citation “[C]” but not as a Citation taken to be representative of Didymus’ text, and used as a base for collation.)
(19) 27:40 - Didymus has QEOU - agreeing with TR B (disagreeing with Aleph A)  (Ehrman listed this as a Citation “[C]” but not as a Citation taken to be representative of Didymus’ text, and used as a base for collation.)
(20) 28:19 - Didymus has MAQHTEUSATE - agreeing with Aleph A (disagreeing with B TR)

In these 20 units, each pair (TR+A, and Aleph+B) has the potential to score 40 agreements. Which pair scores higher: the Byzantine pair, or the Alexandrian pair?
TR: 15. A: 17. Aleph: 12. B: 12.

Combined total of TR and A = 32/40 = 80%
Combined total of Aleph and B = 24/40 = 60%


We may thus conclude that Didymus’ text of Mt. 25:6-28:20 was significantly more Byzantine than Alexandrian.

If we were to ignore the citations which Ehrman, for whatever reason, did not identify as citations taken to be representative of Didymus’ text, we would be left without #11, #12, #18, and #19, and the list would look like this:


(1) 25:6 - Didymus has ECERCESQE - agrees with TR A Aleph B
(2) 25:6 - Didymus has GEGONEN - agrees with TR A Aleph (disagrees with B)
(3) 25:15 - Didymus has IDIAN DUNAMIN - agrees with TR A Aleph B
(4) 25:16 - Didymus has EN - agrees with TR A Aleph B
(5) 25:33 - Didymus has MEN - agrees with TR A Aleph B
(6) 25:33 - Didymus has DEXIWN - agrees with Aleph A (disagrees with TR B)
(7) 25:33 - Didymus has EUWNUMWN - agrees with TR A B (disagrees with Aleph)
(8) 25:41 - Didymus has OI - agrees with TR A (disagrees with Aleph B)
(9) 25:41 - Didymus has POREUESQE - agrees with TR A B (disagrees with Aleph) 
(10) 26:15 - Didymus has PARADWSW - agrees with TR A Aleph B
(13) 26:53 - Didymus has DOKEIS OTI OU DUNAMAI - agrees with TR A Aleph B (There is a blank space in Ehrman’s book where the letter “A” should be.)
(14) 26:53 - Didymus has MOI - agrees with TR A B (disagrees with Aleph)
(15) 26:53 - Didymus has PLEIOUS - agrees with TR A (disagrees with Aleph B)
(16) 26:53 - Didymus has DWDEKA - agrees with Aleph B (disagrees with TR A)

(17) 26:53 – Didymus has LEGIWNWN ANGELWN - agrees with Aleph A (disagrees with TR B)
(20) 28:19 - Didymus has MAQHTEUSATE - agreeing with Aleph A (disagreeing with B TR)

So with 16 variant-units, each pair (TR+A, and Aleph+B) has the potential to score 32 agreements. Which pair scores higher: the Byzantine pair, or the Alexandrian pair?

TR:  12.  A:  15.  Aleph:  10.  B:  10.

Byzantine:  12+15 = 27/32 = 84%
Alexandrian:  10+10 = 20/32 = 63%

Thus once again, it appears that Didymus’ text of Mt. 25:6-28:20 was significantly more Byzantine than Alexandrian.  Ehrman’s use of the TR, however, introduced an improvable factor into the analysis.  What if the Robinson-Pierpont Byzantine Textform (2005 edition) is put in place of the TR?  In the 20-item list, RP2005 agrees with the TR every time, except in Mt. 28:19.  RP2005 does not have OUN, and thus agrees with Didymus.  With this refinement in the analysis, the comparison looks like this:

RP2005:  13.  A:  15.  Aleph:  10.  B:  10.


Byzantine:  13+15 = 28/32 = 88%
Alexandrian:  10+10 = 20/32 = 63%

Decidedly Alexandrian?? 


It looks like either Didymus used a text of Matthew that was uniquely block-mixed, so as to be primarily Byzantine in chapters 25-28, and something else in chapters 1-25, or else something is very wrong with Ehrman’s analysis.

No comments: