How (And How Not) to Define a Text-type

 In the ninth lecture in my online introduction to New Testament textual criticism, I describe text-types.  There has been a recent wave of resistance in academia to affirm the reality of text-types, on the grounds that only the Byzantine text has an archetype capable of confident reconstruction.  This resistance is due to a failure to acknowledge the proper way to define a text-type.  Instead of profiling entire collections of readings in separate genres of the New Testament (Gospels, Acts, General Epistles, Pauline Epistles, Revelation) , a constellation of 50 or less distinct readings is all that is needed to separate manuscripts into the traditionally recognized text-types (Alexandrian, Byzantine, Western, Caesarean).  

Because of pervasive mixture and each manuscript's scribe's uniqueness, once each text-type's distinct variants - the stars in the constellation, so to speak - are identified, 45 out of 50 variants, rather than 50 out of 50, sufficiently shows the type of text a manuscript contains.

This approach is applicable to the full text across a genre; it does not apply to small fragmentary manuscripts, the classification of which should be made and which should also be considered provisional.   Why provisional?  Because block-mixture is real.  It was once proposed that a small sample is sufficient to show the text-type of a manuscript:  the text-type of the extant sample was extrapolated to apply hypothetically to the non-extant portion.   The logic seemed sound:  if you open a jar and stick a spoon inside and pull it out full of grape jelly, it's reasonable to conclude that the whole jar is full of grape jelly.

But sometimes there's a jar like Smucker's Goober-Grape.  One small spoon is an insufficient basis to ascertain the jar's contents.  Some manuscript are like that.  The textual character of Codex Washingtoniensis, 032, varies widely in different segments of text.  Codex Regius, 019, shifts from being predominantly Byzantine at the beginning of Matthew to being mainly Alexandrian by the end of John.  The large manuscripts that only survive as small fragments might have been like that too.  

When we have the text of a full genre preserved in a manuscript, its text can be validly assigned to a type.  Over a century ago, Edward Ardron Hutton, assisted by F. C. Burkitt,  helpfully wrote An Atlas of Textual Criticism in which he presented (or re-presented) a valid basis for dividing groups of manuscripts' text into families.  Hutton affirmed that "The test of antiquity is decidedly against the Syrian," i.e., against the Byzantine text.  Built into his statement is the assumption that we can identify what the Byzantine text is.

Hutton observed that the same kind of close relationship seen in family 13 (see the diagram here) can exists - at a lesser degree of magnification - between larger groups of manuscripts.  He proceeded to list "Triple readings" - variation-units which are so to speak a three-horse race and the three horses are Byzantine ("S"), Alexandrian ("A"), and Western (W").

The text of a non-fragmentary manuscript can easily be assigned to a text-type, or be recognized as mixed, on the basis of Hutton's Triple Readings.  There is no need to add comets and fireflies to the constellation while the stars are blazing bright.

Matthew 17:21 - What's the Early Evidence Say?

In the Evangelical Heritage Version, Matthew 17:21 says, "But this kind does not go out except by prayer and fasting.”  The KJV, NKJV, EOB-NT, MEV, WEB, and 1995 NASB read similarly.  In the RV 1881, ASV, ESV, NIV, NLT, and NRSV, however, there is no such verse; the versification jumps from 20 to 22.  What has happened?

Bruce Manning Metzger

          Bruce Metzger did not spend many words explaining:  “Since there is no good reason why the passage, if originally present in Mathew, should have been omitted, and since copyists frequently inserted material derived from another Gospel, it appears that most manuscripts have been assimilated to the parallel in Mk. 9.29.”  (Textual Commentary on the GNT, p. 43)  His concise treatment is unsatisfactory for at least three reasons, first of which is the consideration that Matthew himself when using Mark’s Gospel (or something closely resembling it) had no discernible reason to skip over this statement of Jesus. 

          Second, the external evidence merits a closer look.  Neither the apparatus in the UBS GNT nor the Nestle-Aland NTG is sufficient.  We begin with their data, supplemented by Swanson:  verse 21 is absent in À* B Q 579 788 892* l253 it^e  ff¹ the Sinaitic Syriac, the Curetonian Syriac, Palestinian Aramaic, the Sahidic version, some Bohairic witnesses, an Ethiopic witness, and an early strata of the Old Georgian version.    Everything else favors the inclusion of τοῦτο δὲ τό γένος ούκ ἐκπορεύεται εἰ μὴ ἐν προσευχῇ καὶ νηστείᾳ (À^c reads ἐκβάλλεται instead of ἐκπορεύεται, 118 reads ἐξέρχεται, and 205 1505 l1074 read εξέρχεται) – including C D F G H K L Y O W Y Δ Σ Φ 0281 f¹ f¹³ 28 157 180 565 597 678 700 892^c 1006 1010 1071 1241 1243 1292 1342 1424 Byz Lect it^a it^aur it^b it^c it^d itf it^ff2 it^g1 it¹ itⁿ it^q Vulgate Peshitta Harklean Syriac Armenian some Georgian, and the patristic evidence is lopsided in favor of inclusion:  Origen Asterius Basil-of-Caesarea Chrysostom Hilary Ambrose Jerome Augustine.  Hort noted that daemonii is sometimes added in Old Latin witnesses.  The writer of an article at NeverThirsty stated, “The verse is not included in the newer Bibles because the older and better manuscripts of Matthew do not include it” and “Apparently in the process of copying the manuscripts, someone at a much later date copied the verse from the Gospel of Mark and added it to the Matthew account. “

 

         Now let’s zoom in on some patristic witnesses. 

          In 2010 Jonathan C. Borland presented a paper titled “THE AUTHENTICITY AND INTERPRETATION OF MATTHEW 17:21” at a gathering of the Evangelical Theological Society in Atlanta, Georgia.  He noted that 1604 2680 should be added to the list of MSS favoring non-inclusion, and that the percentage of Greek MSS favoring inclusion is 99.4%.  He also took a close look at some patristic witnesses:

          ● The author known as Pseudo-Clement, in Letters on Virginity (1:12) did not specify which Gospel he was quoting but the wording looks more like Matthew 17:21 than  Mark 9:29 when he wrote against individuals who “do not act with true faith, according to the teaching of our Lord, who hath said: ‘This kind goeth not out but by fasting and prayer,' offered unceasingly and with earnest mind.’”

          ● Clement of Alexandria, c. 200, in Extracts from the Prophets, wrote, “The Savior plainly declared to the believing apostles that prayer was stronger than faith in the case of a certain demoniac, whom they could not cleanse, when he said, ‘Such things are accomplished successfully through prayer.’”

          ● Tertullian, in de Jejun 8:2-3, without specifying whether he was citing Matthew or Mark, wrote the following:  “After that, he prescribed that fasting should be carried out without sadness.  For why should what is beneficial be sad? He taught also to fight against the more fierce demons by means of fasting. For is it surprising that the Holy Spirit is lead in through the same means by which the sinful spirit is lead out?”

          ● Origen, in his Commentary on Matthew (13:6-7) wrote, “That those, then, who suffer from what is called lunacy sometimes fall into the water is evident, and that they also fall into the fire, less frequently indeed, yet it does happen; and it is evident that this disorder is very difficult to cure, so that those who have the power to cure demoniacs sometimes fail in respect of this, and sometimes with fastings and supplications and more toils, succeed.”  And, “But let us also attend to this, ‘This kind goeth not out save by prayer and fasting,’ in order that if at any time it is necessary that we should be engaged in the healing of one suffering from such a disorder, we may not adjure, nor put questions, nor speak to the impure spirit as if it heard, but devoting ourselves to prayer and fasting, may be successful as we pray for the sufferer, and by our own fasting may thrust out the unclean spirit from him.”

          ● The Latin writer Juvencus wrote in Book 3 of Libri evangeliorum quattuor, “For by means of limitless prayers it is faith and much fasting of determined soul that drive off this kind of illness.”

          Although defenders of modern versions have claimed that “The verse is not included in the newer Bibles because the older and better manuscripts of Matthew do not include it,” antiquity in this case favors inclusion:  the oldest witness for inclusion is older than the oldest witness for non-inclusion.

          The scope of attestation also favors inclusion at least as much as it favors non-inclusion:  Western witnesses for inclusion far outnumber the Western witnesses for non-inclusion, and they are geographically widespread.

          We are left with the appeal to the “best” manuscripts as the basis for rejecting the verse.  But this is circular reasoning; the real question is “What are the best witnesses at this specific point?”, and generalizations simply do not answer that question. It is like deciding which football team wins the ballgame when the score is tied by asking which kicker has made the most field goals, instead of by actually scoring more points than the other team.  

          Third, this supposed harmonization doesn’t yield a tight harmony.  Let’s compare the text of Matthew 17:21 to Mark 9:29.  Mark wrote, τοῦτο  τό γένος ἐν οὐδενὶ δύναται ἐξελθειν εἰ μὴ ἐν προσευχῇ καὶ νηστείᾳ.  (Regarding the Alexandrian text’s non-inclusion of καὶ νηστείᾳ, see my earlier analysis.)  Metzger’s plea that Mark 9:29 was transplanted into Matthew 17 is complicated by the distinct lack of verbal similarity:

          Matthew:  τοῦτο δὲ τό γένος ούκ ἐκπορεύεται εἰ μὴ ἐν προσευχῇ καὶ νηστείᾳ.

          Mark:  τοῦτο τό γένος ἐν οὐδενὶ δύναται ἐξελθειν εἰ μὴ ἐν προσευχῇ καὶ νηστείᾳ.

          This is not a verbatim harmonization – out of 12 words (in Matthew 17:21), nine are identical – and Metzger’s comment that “copyists frequently inserted material derived from another Gospel” fails to explain why a scribe with Mark 9:29 in front of him would change 25% of its wording when inserting it into the text of the Gospel of Matthew.  It should also be noted that the kind of harmonization Metzger referred to usually involved harmonization to the text of Matthew in Mark and Luke, not the other way around (the harmonization of Matthew 9:13 and Mark 2:17 to Luke 5:32 being a notable exception).

          I propose that an early Western scribe intentionally omitted the material we know as Matthew 17:21 out of concern that readers might think that the ability of the Son of God was limited depending on whether he fasted or not.  (The same concern motivated the omission of καὶ νηστείᾳ in Mark 9:29.)  This exclusion was subsequently adopted by scribes in the Alexandrian transmission-line, which led to the reading (or non-reading) in À B Q et al.

          Matthew 17:21 should be regarded as an authentic part of the Gospel of Matthew.  The oldest evidence, the most geographically diverse evidence, and the vast majority of evidence all point in favor of its inclusion.  The NIV, ESV, etc. should be corrected accordingly.

Thanks to Jonathan Borland for sharing his insightful research.

Video Lecture: Text-types

A new lecture, 32 minutes long, about the basic concept of text-types, is online at YouTube!
This is lecture #9 in the series Introduction to NT Textual Criticism.


https://www.youtube.com/watch?v=TXq9Vn_CRkc

Lecture 9 includes, among other things,
details about Griesbach's Canons.

Hand to Hand Combat: D versus 505 in Mark 5

            Today, let’s compare the accuracy of Codex Bezae (D, 05) to medieval minuscule 505 (Harley MS 5538 at the British Library, from the 1100s).  Bart Ehrman has heralded Codex Bezae as “one of our oldest manuscripts” and D. C. Parker has given it a production date around the year 400 (although previous researched gave it a production-date somewhat later).  If a production-date around 400 or 500 is accepted, it ought to have the more accurate text, compared to a manuscript made 500 years later, right?  Nope.  The Gospels-text in Codex Bezae is one of the most inaccurate transcriptions that can be found in any Greek manuscript. 

           The main reason for this is that Codex Bezae’s text is “Western.”  The Western text is a paraphrastic text that developed in the second century.  The individuals who made it were more concerned with conveying the meaning of the text than with preserving its original form – possibly because they were transmitting not only the Greek text of the Gospels, but also the Latin text of the Gospels.   It should be understood that “Western” does not mean that this form of the text originated in the West.  The use of the term “Western” is only meant to convey that it became more popular in the West (especially in the Old Latin Gospels) than elsewhere.

            Codex Bezae is a Greek-Latin manuscript – each page with Greek text is followed by approximately the same passage in Latin on the following page – and occasionally, the Western text has been paraphrased.  (Paraphrasing could be done not only by the scribe of this particular manuscript, but by scribes much earlier in the transmission of its text.)  For example, in Mark 7:19b, the Latin text in Codex Bezae says (after the spelling is tidied up), “Sed in uentrem, et in secessum exit, purgans omnes escas.”  This is not unusual; the Vulgate text edited by Wordsworth and White says the same thing.  But on the accompanying page of Codex Bezae with Greek text, where most manuscripts (including B A L Π K M) read ἀφεδρῶνα ἐκπορεύεται, Codex D’s text reads instead ὀχετὸν ἐξέρχεται (and 7:19 begins Mark 7:19 with οὐ γὰρ εἰσέρχεται).  That is, excrement doesn’t just go into a latrine; it goes into the sewer (an ὀχετὸν is a sewer-drain).
            One might speculate that the Western paraphraser lived in a city with sewers, rather than in the countryside with latrines (or that he made the paraphrase for people who did so); the thing to see is that the Western Text has been paraphrased.

            Now let’s see who made the more accurate text:  the early Western paraphraser, whose work is echoed in Codex Bezae, or Byzantine copyists who transmitted the text found in GA 505.  (And along the way, we might ask ourselves, “Does it look like the Byzantine Text had the Western text as a source?)  The arena for today’s comparison is Mark 5:25-34 – the account of Jesus healing the woman (known traditionally as Saint Veronica) who had been afflicted with a bloody hemorrhage for 12 years.

            The usual rules for hand-to-hand combat apply:  nomina sacra contractions are not counted as variants; transpositions are merely noted; additions, deletions, and substitutions are counted letter-by-letter, and the standard of comparison is the 27^th edition of the Nestle-Aland Novum Testamentum Graece.  However, I will make an adjustment due to one especially questionable reading in the N-A compilation (near the end of the analysis).

Mark 5:25-34 in Codex Bezae

25 – D has τις after γυνη (+3)

25 – D transposes to ετη ιβ

26 – D had η instead of και at the beginning (+1, -3)

26 – D does not have παρ before εαυτης (-3)

26 – D reads ωφελιθεισα instead of ωφεληθεισα (+1, -1)

26 – D reads επι instead of εις (+2, -2)

26 – D does not have ελθουσα (-7)

27 – D reads πε instead of περι.  The missing letters have been supplied by a corrector.  (-2)

27 – D, in the course of a transposition, reads οπισθεν instead of οπιθεν (+1)

27 – D, in the course of a transposition, has και before ηψατο (+3).  A corrector has overdotted the word.  

27 – D transposes εν τω οχλω to the end of the verse

28 – D begins the verse with λεγουσα εν εαυτη instead of ελεγεν γὰρ (+14, -9)

28 – D transposes after οτι to καν του ιματίων εαυτου αψωμαι (see next two variants)

28 – D reads του instead of των (+2, -2)

28 – D reads εαυτου instead of αυτου (+1)

28 – D reads ειαται instead of ιαται (+1)

29 – D reads ευθέως instead of ευθυς (+2, -1)

30 – D transposes and adds και between επιγνους and ὁ Ιης (+3)

30 – D does not have εν εαυτω (-7)

30 – D does not have εξ αυτου (-7)

30 – D reads απ αυτου και before επιστραφεὶς (+10) 

30 – D reads ειπεν instead of ελεγεν (+4, -5)

30 – D transposes at the end of the verse to ηψατο των ιματίων μου

31 – D, in the course of a transposition, reads δε instead of και (+2, -3)

31 – D, in the course of a transposition, reads λεγουσιν instead of ελεγον (+8, -6)

31 – D reads συνθλειβοντά instead of συνθλιβοντά (+1)

32 – D reads ειδειν instead of ιδειν (+1)

33 – D reads φοβηθισα instead of φοβηθεισα (-1)

33 – D reads διο πεποιήκει λάθρα after τρέμουσα (+17)

34 – D reads Ιης after ὁ δε (+3) 

34 – D reads σθι instead of ισθι (or κα instead of και) (-1)

TOTAL:  In Codex Bezae’s text of Mark 5:25-34, there are 80 non-original letters present, and 60 original letters are missing, for a total of 140 letters’ worth of corruption.

Now let’s look at GA 505.

Mark 5:25-34 in GA 505

25 – 505 has τις after γυνη (+3)

25 – 505 transposes to ετη δώδεκα

26 – no variations

27 – no variations

28 – 505 has ελεγε instead of ελεγεν (-1)

28 – 505 does not include εαν after οτι (-3)

28 – 505 transposes after οτι to καν των ιματίων αυτου αψωμαι  

29 – 505 has ευθέως instead of ευθυς (+2, -1)

30 – 505 has ευθέως instead of ευθυς (+2, -1)

[30 – part of the page is obscured, but 505 probably does not have ὁ before Ις] (-1)

30 – 505 has ελεγε instead of ελεγεν (-1)

31 – no variations

32 – no variations

33 – 505 has επ before αυτη (+2)

34 – 505 has θύγατερ instead of θύγατηρ (+1, -1)

34 – 505 has σέσωκ- but the ending seems abnormally contracted.  I’m going to treat this as a fluke reading.  (-2)

TOTAL:  In Mark 5:25-34, GA 505 has 9 non-original letters, and is missing 11 original letters, for a total of 20 letters’ worth of corruption.

Using this passage as a sample of the accuracy of the Greek text in D, and the accuracy of the text in 505, there are 140 letters’ worth of corruption in D for every 20 letters’ worth of corruption in 505.  To put it another way:  using the Nestle-Aland compilation as our basis of comparison, Codex D’s text of Mark 5:25-34 contains seven times as much corruption as the text in GA 505.  If the Robinson-Pierpont Byzantine Textform were used as our basis of comparison, 505’s text would be essential congruent to it:  the only differences between the two texts in this passage involve movable ν, the obscured portion of the page in verse 30, and that word-ending in v. 34. 

Now let’s consider how the word “daughter” should be spelled in verse 34.  Codex D and Codex W agree with Codex B, reading θύγατηρ.  So does GA 474 (Scrivener’s “e,” Lambeth MS 1179).  However, most other manuscripts read θύγατερ, including ℵ A K L N Θ Π Δ 33 700 1424 1582.  This was the reading adopted in the past by Bengel, by Griesbach, by Scholz, and by Tischendorf (in 1872), and even after θύγατηρ was adopted in the Westcott-Hort 1881 compilation, θύγατερ was preferred by Baljon and Souter. 

            One might try to defend the reading θύγατηρ by proposing that a change to θύγατερ was introduced to bring Mark 5:34 into closer harmony with the parallel-account in Matthew 9:22.  However, when we look at Matthew 9:22, we see that two of the three uncial witnesses for θύγατηρ in Mark 5:34 (D and W) also have θύγατηρ in Matthew 9:22.  Furthermore, L N and Θ read θύγατερ in Mark 5:34 and read θύγατηρ in Matthew 9:22.  Another consideration is that it seems unlikely that a harmonizer would fine-tune spelling but not add Θάρσει from the Matthean parallel.  The lonely testimony of B for θύγατερ in Matthew and θύγατηρ in Mark, where θύγατηρ appears in the following verse, is not as strong, on a point of orthography, as the extremely broad support that θύγατερ enjoys in both passages.  The reading θύγατερ should be adopted in Mark 5:34 – as it already has been in the Tyndale House Greek New Testament.

            With this adjustment in the picture, the accuracy of the text of Mark 5:35-34 in 505 (assigned to the 1100s) is more than seven times better than the text in Codex Bezae (assigned to the 400s).

 

The Healing of Saint Veronica -
from an ancient Roman catacomb

 

            Before we leave Mark’s account of the healing of Saint Veronica, there is a variant in Mark 5:27 worth examining.  In the Byzantine Text and in the Nestle-Aland compilation this verse is exactly the same, but after the word ηψατο, in a smattering of witnesses – M f¹ 33 579 1071 and the Ethiopic version – the words του κρασπέδου (the hem of) are included here.  In Matthew 9:20, after ηψατο, we encounter this full phrase, του κρασπέδου του ἱματίου αυτου (“the hem of His garment”).  In the next verse in f¹³, κρασπέδου has usurped ἱματίου.  Turning to the parallel-passage in Luke 8:44, we find the same phrase after ηψατο:  του κρασπέδου του ἱματίου αυτου. 

            The puzzle that this data presents is this:  if Matthew and Luke both depended on Mark, how is it that they both have του κρασπέδου and Mark does not?  If we reject the reading of f¹ and its allies in Mark 5:27, then we have here a “Minor Agreement” – a passage where Matthew and Luke share a reading with each other but not with Mark. 

            Making things a bit more interesting, Codex D has του κρασπέδου in Matthew 9:20 but not in Luke 8:44 (in which case, the Minor Agreement goes poof in the Western Text.)  Furthermore, Codex D is allied in this respect with a few Old Latin copies:  Old Latin codices a (Codex Vercellensis, probably from the 370s, a.k.a. VL 3), ff² (Codex Corbeiensis secundus, a.k.a. VL 8) and r¹ (Codex  Usserianus primus, a.k.a. VL 14) also do not have these words in Luke 8:44.      

            Without exploring this puzzle further today, I note that this particular array of readings may pose some problems for researchers who want to preserve the Two-Source Hypothesis as a solution to the Synoptic Problem while maintaining the Alexandrian readings in all three Synoptic Gospels.



Readers are invited to double-check the data in this post.

Luke 11:33 - Don't Put Your Light Under a Bushel

            When someone asks, “What’s a text-critically interesting verse in the Gospels?” the typical answer is not likely to be “Luke 11:33.”  The differences between the meanings of the rival variants in this verse are not very consequential:  basically, some manuscripts have the phrase, οὐδὲ ὑπὸ τὸν μόδιον, that is, “or under a bushel-basket,” and some do not; also, near the end of the verse, in some Greek manuscripts the Greek word rendered “light” in English is φῶς, while in other manuscripts, it is φέγγος. 

            Yet it is probably safe to say that Luke 11:33 is a strong contender for the title “Verse Most Likely to Be Changed from One Critical Edition to Another.”  Here is how some recent Greek compilations have treated Luke 11:33:

            ● Nestle-Aland NTG 27:  brackets οὐδὲ ὑπὸ τὸν μόδιον and adopts φῶς.

            ● Robinson-Pierpont 2005:  includes οὐδὲ ὑπὸ τὸν μόδιον and adopts φέγγος.

            ● SBLGNT 2011:  includes οὐδὲ ὑπὸ τὸν μόδιον and adopts φέγγος.

            ● Tyndale House GNT 2017:  does not include οὐδὲ ὑπὸ τὸν μόδιον and adopts φῶς.      

           

            The vast majority of Greek manuscripts, including the two flagship representatives of the Alexandrian Text, Vaticanus (B) and Sinaiticus (ℵ), and the best Greek representative of the Western Text, Codex D, support the inclusion of οὐδὲ ὑπὸ τὸν μόδιον.  Exploring the evidence more closely, we see that ℵ A B C D K M W X Δ Θ Π Ψ and the Curetonian Syriac, the Peshitta, the Bohairic, and all Latin witnesses, are allies of the Byzantine Text; they all support the inclusion of οὐδὲ ὑπὸ τὸν μόδιον.  (The Gothic version, alas, is not extant for this part of Luke.)

 

Luke 11:33: at the end of col. 1
and the start of col. 2.

          The main witnesses for non-inclusion are Papyrus 45, Papyrus 75, L (019), Γ (036), Ξ (040), 070 (Greek-Coptic), 700*, 1241, family 1, 69, 118, and 788, and the Sinaitic Syriac, the Sahidic version, and the Armenian and Georgian versions.  While this array of witnesses may appear negligible in terms of quantity when set alongside the mountains of witnesses which favor the other reading, the age and diversity of its members are interesting:  P45 and P75 are the earliest manuscripts of this part of Luke, and the Alexandrian (P75, L, Sahidic), Western (Sinaitic Syriac) and Caesarean (f¹, Armenian) forms of the text are represented.                 

            Let’s examine the text nearby in search of similar arrays of variants supporting shorter readings.   

● (1)  In Luke 11:11, P45, P75, and B, 1241, and the Armenian version, along with the Sinaitic Syriac, support the non-inclusion of ἄρτον μὴ λίθον ἐπιδώσει αὐτῷ; ἣ καὶ ἰχθύν, which has very abundant support, with some slight variations, in A C D (which, with 124, adds αἰτήσει after ἰχθύν) F G Y K M U W X Γ Δ Θ Λ Π Ψ f¹ f¹³ 1 1582* 1424 (in ℵ, L, 28, 157, and 700, ἄρτον μὴ λίθον ἐπιδώσει αὐτῷ; ἣ ἰχθύν, lacking the καὶ) – “bread, will he give to him a stone? And if a fish.” 

            The minuscules 69 and 788 (along with 565) do not include the second part of the verse, that is, they have ἄρτον μὴ λίθον ἐπιδώσει αὐτῷ but not the rest of the verse, as if they echo an ancestor-manuscript in which the copyist’s line of sight drifted from this occurrence of ἐπιδώσει αὐτῷ to the recurrence of the same words at the end of the verse, skipping the words in between.

            Upon comparing the witnesses for the main shorter reading in 11:11 and 11:33, we see that several of them are the same:  P45, P75, 1241, Sinaitic Syriac, Sahidic, Armenian.  And a couple of witnesses for the other shorter reading in 11:33 (69 788) also support the shorter reading in 11:33.

            It should not be overlooked that 157 omits all of 11:12, and 579 omits everything before μὴ.  Also in 11:12, where the normal reading is ᾠόν (egg), P45 reads ἄρτον (bread).

● (2)  In Luke 11:14, P45 P75 ℵ B A* L 1 33 157 788 1241 1582* Sinaitic Syriac and the Armenian version are among the group of witnesses that support the non-inclusion of καὶ αὐτὸ ἧν (and it was).

● (3)  In Luke 11:44, P45 P75 ℵ B C L 33 f¹ 157 579 and the Sinaitic Syriac, Curetonian Syriac, and the Armenian, Georgian, and Sahidic versions do not include γραμματεις καὶ Φαρισαιοι, υποκριταί (Scribes and Pharisees, hypocrites).  Most manuscripts (such as A K M Γ Δ Θ Π Ψ 69 157 565 579 700 788) include the words, but D omits υποκριταί.

● (4)  At the end of Luke 11:48, P75 ℵ B D L 579 do not include αυτων τα μνημεια (“their tombs”).  Family 1 and 157 read instead τους τάφους αυτων.  (P45 is not extant for this verse.)

● (5)  In Luke 11:53:

            P75 reads Κακειθεν εξελθόντες αυτου

            ℵ B C L 33 69 579 788 1241 read Κακειθεν εξελθόντος αυτου

            P45 reads Κακειθεν εξελθόντος (apparently without αυτου); Willker mentions that the manuscript is damaged but space-considerations rule out the inclusion of αυτου.

            A K M Π W Γ Δ Ψ f¹ 565 read Λέγοντος δε ταυτα προς αυτους, which is supported by most manuscripts. 
            D Θ 157 agree with A but continue with ενωπιον παντος του λαου; this reading has considerable Old Latin and Armenian support; especially interesting is that this reading is also supported by the Curetonian Syriac and the Sinaitic Syriac. 

            Meanwhile 69 and 788 simply read Και before ηρξαντο.

● (6) and (7)  In Luke 11:54:

            P45 P75 B L f¹ 579, with Coptic support, begin the verse with ενεδρεύοντες αυτον before θηρευσαι.

            ℵ begins the verse with ενεδρεύοντες before θηρευσαι.

            Most manuscripts, including C K Π M  f¹ 157 565 700, with support from the Vulgate and the Peshitta, read ενεδρεύοντες αυτον ζητουντες at the beginning of the verse, and read ινα κατηγορήσωσιν αυτου at the end of the verse.  A W* Δ f¹³ differ only slightly at the end of the verse, reading ινα κατηγορήσουσιν αυτου. 

            D’s text is quite different:  ζητουντες αφορην τινα λαβειν αυτου ινα ευρωσιν κατηγορησαι αυτου.  This is supported by the Sinaitic Syriac and Curetonian Syriac, and is imperfectly supported by the Old Latin.

            Θ reads ενεδρεύοντες before τι θηρευσαι at the beginning of the verse; it agrees with most manuscripts at the end of the verse.

            Without attempting to offer a full analysis of all seven of these textual contests here, I offer brief explanations vindicating the longer reading in six cases:

● (1)  In Luke 11:11, the shorter reading originated when a copyist skipped a line of text but nevertheless produced a coherent sentence; the reading of P45 in verse 12 (ἄρτον instead of ᾠόν) is a vestige of the scribe’s recollection of the original longer reading.  Harmonization to Matthew 7:9 was limited to the addition of αυτου after υιος mainly in Caesarean witnesses.

● (2)  In Luke 11:14, the shorter reading in P45 P75 ℵ B 1241 et al is a slight stylistic refinement; as Metzger noted, καὶ αὐτὸ ἧν  κωφόν “appears to be a Semitism in the Lukan style.  The chance seems low that a scribe would sense a need to shift from “He was casting out a dumb demon” to “He was casting out a demon and it was dumb,” and happen to fit Lukan style.  

● (3)  In Luke 11:44, the shorter rreading – that is, the removal of the explicit identification of the scribes and Pharisees – originated when a scribe wondered why the lawyers would feel that they were being criticized by a rebuke specifically aimed at others. 

● (4)  At the end of Luke 11:48, the Byzantine reading makes explicit what is implied without an object.  Family 1 and family 13 do likewise, but their wording is different.  The shorter reading here is original. 

● (5)  In Luke 11:53, the Alexandrian Text has non-Lukan wording; κακειθεν appears only here in Luke, and εξελθόντος appears elsewhere in Luke only in 11:14 (contested by εκβληθέντος in A C L f¹³ 69) where a demon’s departure is being described.  Willker suggests that the Byzantine reading was introduced because it avoids raising the question of where was the “there;” the exact location being unmentioned in the lection that begins at 11:47.  However, the Byzantine reading raises a question of its own, that is, who is the “them” – for at 11:46, Jesus begins criticizing not the scribes and Pharisees, but the specialists in the Law.  The Alexandrian way around this problem was to rewrite the introductory phrase, which happens to correspond to the beginning of Mark 9:30.

            The Western Text’s inclusion of ενωπιον παντος του λαου echoes Luke 8:47; this reading must be extremely early (as demonstrated by support from the Sinaitic Syriac and Old Latin Codex Vercellensis), and shows that some copyists had a tendency to expand later parts of the Gospel with verbiage taken from earlier parts.     

           

● (6)  In Luke 11:54a:  the shorter reading originated when an early copyist’s line of sight drifted from the letters –οντες in ενεδρεύοντες to the same letters in ζητουντες.  In most copies descended from the exemplar that contained this mistake, it is partly corrected (via the addition of αυτον – but 28 and 1424 read instead αυτω) but not in ℵ and Θ.

● (7)  In Luke 11:54b, the shorter reading originated when an early copyist’s line of sight drifted from the αυτου after στόματος to the same word after κατηγορήσωσιν, accidentally skipping the words in between.

            The inclusion of ευρωσιν in D’s text here is interesting; D thus echoes (albeit inexactly) the end of Luke 6:7.  This again illustrates some scribe’s tendency to expand later parts of a book by introducing elements they had encountered in earlier parts.

 

Luke 11:33 in GA 1241.

           Having reviewed the above seven textual contests, and  having seen that in six out of the seven, the longer reading is reasonably defensible, we return now to Luke 11:33.  This is Luke’s record of a saying of Jesus very similar to the one in Matthew 5:15, where the μόδιον is mentioned; a closer parallel, however, is in Luke 8:16, where the reference is to hiding the lamp under a σκεύει (“vessel”), rather than a μόδιον (bushel-basket).  In 28 there is a clear attempt to conform 11:33 to Luke 8:16; minuscule 28 reads καλυπτει αυτον σκεύει η before εις κρύπτον.  In 579, the text is conformed to Matthew 5:15 toward the end, reading και λάμπει πασιν τοις εν τη οικια.  And in 118 f¹³ 69 788, at the end of the verse, the words are transposed so as to correspond to the end of Luke 8:16.   

            In short, except for minuscule 28, the harmonizations that appear in Luke 11:33 look like they have been based on Luke 8:16, not Matthew 5:15.  If a copyist were to introduce “under the bushel-basket” into a form of Luke 11:33 that did not have the phrase, the natural place to put it would be before the reference to putting the lamp in a secret place, thus corresponding to the gist of Mark 4:21 and the gist of Luke 8:16; I mean that in both Mark 4:21 and Luke 8:16, the reference to the lamp being covered precedes whatever else is said.

            If οὐδὲ ὑπὸ τὸν μόδιον is not a partial harmonization to Matthew 5:15, it is original – in which case, how does one account for its absence in P45 P75 L 788 et al, while also accounting for its presence in ℵ B A C D K Π W?  There are two factors which do this:  (1)  Scribes’ recollection of Luke 8:16, in which τίθησιν is followed immediately by αλλ’ επι λυχνίας.  (2)  A simple homoioteleuton error.  Single-letter homoioteleuton is rare but it does sometimes happen:  all that is needed for οὐδὲ ὑπὸ τὸν μόδιον to disappear is for a scribe’s line of sight to drift from the ν at the end of τιθησιν to the ν at the end of μόδιον.  Adding to the ease of such an occurrence is that the scribe would have written τίθησιν αλλ’ επι λυχνίας  (or, τίθησιν αλλ’ επι λυχνίαν) a few chapters earlier.  

            There is still the other variant-unit in Luke 11:33 to consider:  φως or φέγγος?  Φως looks like a harmonization to Luke 8:16, especially in 118 f¹³ 69 788.  Φέγγος is the rarer word, and considering the presence of φως in the parallel-passages, there would be little impetus to replace φως with φέγγος; meanwhile familiarity with Luke 8:16 would tend to elicit a harmonization from φέγγος to φως.  The reading of P45, φέγγος, should be adopted.  Here we have an ancient reading which is not Alexandrian (for P75 ℵ B 33 read φως) nor Western (for D also reads φως) nor Caesarean (for f¹ reads φως and f¹³ also reads φως, transposed).  The Byzantine Text, and the Byzantine Text alone (but with support from the back-up team of L, Γ, 124, 565 and 700), besides the usual Byzantine witnesses Α Κ Μ W Δ Λ Π etc,, displays the original reading here, defying the theory that it is merely an amalgamation of the other text-forms, and supporting the theory that the Byzantine Text contains a stratum of ancient and independent readings. 

Readers are invited to double-check the data in this post.

Holes in Hort's Case Against the Byzantine Text

            In Hort’s 1881 Introduction, he proposed that the Syrian (Byzantine) Text is derivative of the Alexandrian and Western text-types.  He based his argument partly on conflations – an argument which I addressed earlier this month – and partly on the observation that Irenaeus, Hippolytus, Clement of Alexandria, Origen, Tertullian, Cyprian, Methodius, and Eusebius of Caesarea do not support any distinctly Byzantine readings:   “Before the middle of the third century, at the very earliest,” he wrote, “we have no historical signs of the existence of readings, conflate or other, that are marked as distinctively Syrian.”  

            That statement is no longer true, thanks in part to the discovery of various papyrus copies of books of the New Testament which contain distinctly Byzantine readings.  But before we look at those readings, let’s look at a map of territory that was, at one time or another, the territory of the Roman Empire.  If we were to put a big umbrella over the headquarters of Irenaeus (Lugdunum), Hippolytus (Rome), Clement of Alexandria (Alexandria, of course), Origen (Alexandria and Caesarea), Tertullian (Carthage), Cyprian (Carthage and Rome), Methodius (Olympus, in Lycia), and Eusebius of Caesarea, and assume that all Christians under those umbrellas in the 100s-200s used a text like the text of the writer who worked in that area, does that leave any part of the Roman Empire uncovered?

Headquarters of early patristic writers,
and their vicinities.

It does.  The evidence from these writers surely has much to tell us about the text of the New Testament that circulated in the areas where those writers worked and wrote, but it would be presumptive to suppose that it can tell us much about the text elsewhere – an “elsewhere” which includes Syria, Greece, Cyprus, Crete, Dalmatia, and five-sixths of Turkey, including the destinations of all of Paul’s letters except Romans.  Even when each location is extended very far from each writer’s headquarters, the picture does not materially change.                       

A second thing to consider:  the extent to which a specific writer quotes from a specific book.  Clement of Alexandria, for example, hardly quotes the Gospel of Mark at all, except from the tenth chapter.  And according to a simple check of the Scripture-index in Vol. 6 of the Ante-Nicene Fathers, Methodius, the least famous member of this group of writers, uses Acts twice.  Who can say with confidence that two quotations, both from the 28th chapter of a 28-chapter book, can show us that Methodius’ copy of Acts did not have a Byzantine character? 

Headquarters of early patristic writers,
and their vicinities' vicinities.

Similarly, Methodius uses Mark four times.  Who can say that four quotations from Mark’s 16 chapters demonstrate that Methodius’ text of Mark did not have a Byzantine character?   Methodius does not quote from 13 of Matthew’s 28 chapters; he does not quote from 15 of Luke’s 24 chapters, and he does not quote from nine of John’s 21 chapters – and his quotations from chapters 2, 4, 9, 10, 12. and 16 are limited to a verse each.  And (still relying on the ANF index) Methodius quotes from the General Epistles a total of seven times.  Who wants to establish from seven assorted verses (most from First Peter) a picture of the character of a 22-chapter segment of the text?

This is not to say that the evidence from each of the other writers is as sporadic as it is in the case of Methodius.  But when assessing the significance of the non-use of Byzantine readings by an author, one should first establish the extent of the author’s quotations, and their precision. 

A third thing to consider when asking how much this evidence can tell us:  after an inventory has been made of a patristic writer’s clear utilizations of New Testament passages, how much of that is capable of displaying Byzantine or non-Byzantine character?  For instance, Methodius uses John 1:18 near the beginning of his composition On Free Will, and one might hope to find there some evidence of whether his manuscripts read “Son” or “God” in that verse, but he only speaks allusively about raising a hymn to the holy Father, “glorifying in the Spirit Jesus, who is in His bosom.”   If one were to pick at random a verse from the Gospels, Acts, or an Epistle, chances are less than 50% that the Byzantine form of the verse is recognizably different from the Western and Alexandrian forms.  (That is my general impression.)  Where the text-types agree, even a clear quotation does not point to a specific form of the text.

Evidence of absence?
Or an absence of evidence?

A fourth thing to consider, when asking how much this evidence can tell us, has to do with the limitations of Latin:  is a specific quotation that has been preserved in Latin capable of displaying a Byzantine reading in a form discernible from an Alexandrian or Western rival?  Irenaeus’ work is mainly preserved in Latin.  Tertullian and Cyprian wrote in Latin.  When the difference between the Byzantine and Alexandrian and/or Western rivals is deep and wide, surely the answer is “Yes,” but when things are more nuanced – involving orthography, or the presence of an article, or a matter of word-order – not so much. 

            And there is a fifth factor to consider:  the diversity of readings that have been called Alexandrian or Western.  When Irenaeus or Hippolytus or Clement of Alexandria or Origen or Tertullian or Cyprian or Methodius or Eusebius clearly utilizes an identifiable passage in the Gospels, and uses a reading that agrees with Byz and disagrees with B and D, does Hort conclude that the writer has used a distinctively Byzantine reading?  Not so fast!  For a reading to be distinctively Byzantine, it has to not only be unique from the readings in the major Greek representatives of the Alexandrian and Western text-types, but their minor representatives as well.    

Thus a lot depends on what the allies of Byzantine readings happen to be.  For example, Tertullian (in On Baptism, ch. 5) supports the inclusion of John 5:3-4.  This reading is not supported by Sinaiticus, B, or D.  But because it is supported by several Old Latin copies, it is counted as a Western reading.  In addition, Sinaiticus and Vaticanus frequently disagree with one another, but at points where one of them agrees with the Byzantine Text, that particular reading is struck from the list of distinctively Byzantine readings.  Because of the textual diversity of the Old Latin text(s), and the incessant disagreement of the major Alexandrian witnesses, the Byzantine readings are told, “Be completely unique, or else you are either a Western reading or an Alexandrian reading” and thus have their work cut out for them.  If one reading were considered the Western reading, and one reading were considered the Alexandrian reading, then Byzantine readings would often have ancient allies.

So:  while this part of Hort’s basis for positing that the Byzantine Text is relatively late might initially look like a simple matter of sifting through patristic writings and noticing that eight ante-Nicene writers never use distinctively Byzantine readings, it is not really so simple.  Hort’s approach raises four questions:

            (1)  Territory:  Do these eight writers adequately represent the whole territory of Christendom in the 100s-200s?

(2)  Abundance of Quotations:  Does each of these eight writers quote from the New Testament in sufficient abundance to facilitate meaningful comparisons of the readings in their manuscripts to the readings of different text-types?

(3)  Precision of Quotations:  Does each of these eight writers quote from the New Testament with sufficient precision to facilitate meaningful comparisons of the readings in their manuscripts to the readings of different text-types? 

(4)  Versional Limits:  In the case of Irenaeus, Tertullian, and Cyprian, is the Latin text of their writings capable of displaying variations between rival Greek readings?

(5)  Levels of Distinctiveness:  when a patristic writer appears to use a Byzantine reading, but the same reading is also found in an early witness other than Sinaiticus, Vaticanus, or Bezae, is it plausible to treat that reading as if it cannot be considered evidence for the Byzantine Text on the grounds that it is not uniquely Byzantine?

Some of these same considerations apply with similar force to the contents of early papyri.  Hort claimed that the early patristic writings show that eight early patristic writers never used distinctively Byzantine readings, but the significance of his claim shrinks and shrinks the more one considers the limits of those writers’ territory, the limits of the extent of their quotations, and the limits of the precision of their quotations.  Similarly, to the extent that the provenance of our papyri can be established, they represent one particular locale (Egypt), and most of them are very fragmentary.  Yet even with these limits, they contain some readings which agree with the Byzantine Text, and disagree with the Alexandrian and Western readings as found in Sinaiticus, Vaticanus, and Bezae.   Such readings should not exist in a world in which uniquely Byzantine readings are all the result of a Lucianic recension, as Hort proposed.         

            Hort simply could not write today what he wrote about unique Byzantine readings in 1881 – at least, not without exposing his approach as biased against the Byzantine Text.  The papyri agree with the Byzantine Text dozens of times against Alexandrian and Western rivals.  If one were to argue that these are not distinctively Byzantine readings (and thus not evidence of the early existence of the Byzantine Text) because they were found embedded in non-Byzantine texts, then one would have to face the question:  if the existence of early Byzantine readings do not demonstrate that the Byzantine Text is early (at least regarding the books in which the readings occur), then how can the early Byzantine Text be shown to exist?

            To reframe the problem:  if a Greek patristic writer in 230 quoted from Mark 5:42, 6:2, 6:45, 6:48, 6:50, 7:12, 7:30, 7:31, 7:32, 7:35, and 7:36, and in each case, he used a Byzantine reading that is not supported by Sinaiticus or Vaticanus or Bezae, Hort would have a hard time proving that this is not evidence that that writer used the Byzantine Text.  Yet Papyrus 45 has such readings. 

            If one argues that that these distinctively Byzantine readings in Papyrus 45 do not mean that the Byzantine Text of Mark is early, then one is essentially admitting that Hort’s parameters really mean nothing:  he argued that the absence of distinctively Byzantine readings are evidence that the Byzantine Text is late, but you would be saying that even the presence of distinctively Byzantine readings proves nothing about the Byzantine Text (except the obvious point that some early distinctively Byzantine readings are not the result of a Lucianic recension, which is no small point).  But whoever would go that far, and still adhere to Hort’s transmission-model (even after admitting that at least some distinctively Byzantine readings are not the result of a Lucianic recension), would implicitly submit that there is only one way to show that the Byzantine Text is early:  for an early Byzantine manuscript (made before 300) to exist.  

            Only the climate in Egypt is conducive to the preservation of papyrus, so this sets a special hurdle for the Byzantine Text to surmount:  if a manuscript with a thoroughly Byzantine form of the text were made in the 100s-200s, anywhere in Syria, Cilicia, Asia, Cappadocia, Pontus, Bithynia, Thrace, Cyprus, Crete, Achaia, Epirus, Macedonia, or Dalmatia, it could not survive unless it somehow traveled to Egypt.