Scribal Habits in Greek New Testament Manuscripts (Part 1) - Three Questions with Andrew Wilson

An important book has been released by Andrew Wilson:  Scribal Habits in Greek New Testament Manuscripts.  I thought it would be helpful to interview the author about it.  So here he is, from Brisbane Australia:  Andrew Wilson!

1. Tell us about the background to this book. How did the project start and take shape?

Wilson:  I developed an interest in New Testament textual criticism in my late teens through studying New Testament Greek, and through a large box of books about NT textual criticism from a variety of viewpoints my dad inherited from an elderly classicist. One book that I read which was a bit beyond my level was the collected studies of E. C. Colwell (Studies in Methodology in Textual Criticism of the New Testament, Brill, 1969). He seemed to have an open mind about many questions, pointing out some problems (occasionally humorously) in the current reigning approach.

            In the early 2000s I started trying to compile a HTML critical apparatus on my own computer, and as I did so I started noticing patterns in singular readings. Maybe Colwell’s study on singular readings had subconsciously stuck in my head, but what initially interested me were the changing rates of singulars in certain manuscripts. It struck me that the rates of singular readings abruptly changed in Washingtonianus exactly where its textual character changed throughout the gospels. As I collected more singulars from other MSS, I also noticed that singulars tended to be omissions rather than additions.

            In 2005 I moved with my wife and young family to England to serve the Lord (I had been involved in full-time evangelism and Bible teaching for a few years now), and while working with  churches in Hertfordshire and Bedfordshire between 2005-2009, I started travelling to Tyndale House, Cambridge, to do one day’s study a week in New Testament textual criticism. I made friends there who were involved in research in the subject and learned a lot from them, but my main purpose was to add information from Tischendorf and von Soden’s critical apparatuses to my lists of singular readings to get a more complete picture of the patterns of singular readings.  

    In 2011 I presented a paper on scribal habits at the Society of Biblical Literature International meeting at Kings College London. The editor of Philologia Neotestamentaria, Jesús Peláez, was present at the session and after my talk he asked me if I would like to publish the paper in the journal. My paper compared all four of the traditional transcriptional canons (Prefer the Shorter, Harsher, Harder and Less Harmonized Readings) against the evidence of scribal errors from 4200 singular readings I had collected. A year or two later it was published as “Scribal Habits in Greek New Testament Manuscripts”, Filologia Neotestamentaria 24 (2011), 95-125.

            In 2016, I was emailed by the acquisitions editor for a major academic publisher asking if I would like to publish my work as a monograph. This was a surprise, and initially I did not take up the offer. But the more I looked into the state of the question, the more I felt there was room for an expanded study that also addressed objections to the evidence from singular readings. By the time I decided to pursue the study, the acquisitions editor had moved on, and not knowing the attitude of the new editor to the approach I was taking, I decided to contact Dr. Juan Hernández, the General Editor of the Society of Biblical Literature Text-Critical Series of studies and put a proposal to him. This was accepted by the editorial board and after a few more years of hard work, the study was published.

2. What are the results of the study?

Wilson:  The present study looked at over 10,000 singular and sub-singular readings in fifty-two chapters of the NT. Singulars are readings found in only one Greek manuscript, while sub-singulars are found in just two or three manuscripts. These are worth studying because they are far more likely to have been introduced by scribes at some point in the textual transmission of the New Testament rather than written by the original author. Singulars are free from presuppositions about the ‘correct’ text or its history, and are therefore thought to be the most objective guide to the types of errors scribes produced. Because singulars are far more likely to be scribal errors than original readings, and provided they are collected in sufficient numbers, they offer a sound basis for a typology of scribal mistakes.

The Shorter Reading Canon

            In the study of 5742 singular add/omit variation units, there were 2129 additions (37%) and 3613 omissions (63%). This suggests that scribes more often omit from the text than add to it. In democratic elections, a 25-percentage point gap is called a ‘landslide win’, because the losing politician is buried by the result. The result in my study showed scribes overwhelmingly tend to omit, suggesting as James Royse said, ‘the preference for the shorter reading is fundamentally mistaken’.  Every other study of scribal habits (some using different methodologies) also showed scribes tending to omit rather than add.

● Royse (1981, 2008): 130 adds to 312 omits

●  Robinson (1982): 451 adds to 678 omits

●  Silva (1985, 1992): 76 adds to 148 omits

●  Head (1990, 2004): 10 adds to 20 omits

●  Min (2005): 18 adds to 42 omits

● Hernández (2006): 57 adds to 87 omits

●  Jongkind (2007): 19 adds to 67 omits

● Wilson (2011): 1088 adds to 1712 omits

●  Paulson (2013): 76 adds to 117 omits

●  Ebojo (2014): 148 adds to 431 omits

●  Gurry (2017): 568 adds to 644 omits

●  Farnes (2019): 1 add to 4 omits

            We are yet to find a study of scribal habits in NT manuscripts that shows scribes as a whole tending to add to the text, and thus confirming the traditional canon. Prefer the Shorter Reading has been the orthodoxy for the last 250 years, based on the theory that scribes tended to add to the text. But it seems that this received wisdom has no basis in evidence or research data to back it up.

            An important finding of the study was the confirmation of Alphonse Dain who penned one of the most important statements about scribal errors ever written:  Some mistakes are more common than others. There are at least two that all copyists, without exception, frequently commit. The first is the omission of small words . . . A second constant type of copying error, long studied, is the leap from same to same. This is by far the most common error. (Les Manuscrits, 44-45)

            Dain went on to write about various types of small words that are commonly omitted, and stated, ‘In the manuscript tradition of Aelian’s text, I have noted the disappearance of approximately eight  hundred small words’ (ibid.).

            My study found the same thing: scribes tend to heavily omit single words and short stretches of text. The evidence suggested we should rephrase Dain’s Dictum slightly as follows: ‘Scribes tended to omit rather than add, their most common errors being short omissions of one or more words and cases of haplography’. Dain is right about the reality of omission: it is the dominant corruption.

            Other results flow from the fact that scribes are far more prone to omit than add. One is that conflations are far less common than ‘double simplifications’ (Hort). In my study, I estimate that for every conflation we have 10 double simplifications. For example, consider Matthew 12:15:

            πολλοι : 01, B

            οχλοι : N

            οχλοι πολλοι: all other MSS

It is clear that the reading of N is not correct. But the reading of Codex Sinaiticus and Codex Vaticanus, who omit οχλοι, is hardly much stronger. NA 28 brackets οχλοι as doubtful, but omissions are so common that the longer combined reading here is almost certainly correct. The same goes for a similar case in Matthew 8:18. Reports of ‘conflations’ are greatly exaggerated in the literature and, as a general scribal trend, the theory of conflations does not have any solid basis in evidence.

Prefer the Harsher Reading

            Dain’s Dictum has important consequences for a second canon, because scribes tend to omit single words (of virtually all parts of speech). Metzger lists Prefer the Harsher Reading as his fourth transcriptional canon in his Textual Commentary, writing under sub-heading (c) [scribes would] ‘add pronouns, conjunctions, and expletives to make a smoother text’.

However, because scribes actually do the exact opposite in the real world – omitting single words of all parts of speech – the effect of scribal activity on the text is the reverse of what Metzger states: the omissions of pronouns, conjunctions, articles, nouns, verbs, adjectives and adverbs make the text more abrupt, terse and stylistically harsh.

            In one of my chapters, I quantified the effects of scribal activity on the style of the text and found that scribes produced a terser, more abrupt, harsher style nearly three times as often as they produced a smoother style. Scribes are again doing the very opposite of what the textbooks say.

            In other words, the failure of the Shorter Reading canon also entails the failure of the Harsher Reading canon. This has significant implications for the NT text, because the current critical text not only routinely prefers shorter readings, but in many cases these are also harsher readings involving one-word variation units.

            My study is not the only one that has found that scribes tend to omit single words.  The same is true of the studies of Royse (among one-word add/omits, there were 118 additions to 227 omits), Silva (who calls them ‘function words’ and whose 1992 study had 12 additions and 33 omissions), Jongkind (whose calls them ‘verba minora’ and counts 13 additions to 22 omissions), and Min, who recommended that in cases of minor variants with equally strong external evidence, in general the longer reading is preferable. Every study with enough data to be statistically significant has shown that scribes tend to omit rather than add in cases of one-word variation units. What this means is that scribes make the text terser and harsher, instead of polishing it up.

Prefer the Disharmonized Reading

            Another corollary of scribes tending to omit rather than add is that omission often results in disharmonization. That is, in passages where the synoptic gospels have the same wording, omissions (particularly of one or two words) have the effect of making passages originally the same different.

            For an example from my study, consider Mark 2:5 (‘Jesus, seeing their faith’, from the story of the paralyzed man lowered down through the roof). Here it is in the ECM:

            6-8 a ο ιησους [Mt 9,2; Lk 5,20 v.l.] ... P88V . 13V. 1542r. CyrH. Dam

            b om. [Lk 5,20] 892

            – P88 . 13. 495

 Notice that in reading b, manuscript 892 omits ο ιησους, but as the ECM notes in square brackets after reading a, the parallel passage in Matthew 9:2 contains these words. Therefore, the omission in this manuscript in Mark is not bringing Mark into conformity with Matthew, but the very reverse: it is disharmonizing.

            I won’t bore readers with all the statistics, but in the synoptic gospels, my study found that among singular readings scribes disharmonized more than twice as often as they harmonized. Most of the harmonizations were additions, while most of the disharmonizations were omissions. This phenomenon of disharmonization was most pronounced in one word variation units and two words variation units.

            The fact that scribes more frequently disharmonize the text by their mistakes is again the opposite of the traditional canon, which states that scribes tended to bring similar passages into harmony. On the other hand, my study found that in longer variation units (10 words or longer), scribes tend to harmonize to parallels.

            Some modern text-critics argue that harmonization is the dominant textual corruption. Thus, Hugh Houghton in his recent textual commentary writes that change ‘due to the influence of another similar text or construction’ accounts for ‘the majority of variants’ (p. 15*). Dirk Jongkind and Peter Williams write the same in the introduction to The Greek New Testament, Produced at Tyndale House (see page 506). However, even Jongkind’s own study showed scribes omitting far more often than they harmonized (67 singular omissions to 30 singular harmonizations, Scribal Habits of Codex Sinaiticus, 210-13, 229-232). Omission is the dominant scribal corruption, and harmonization comes a distant second.

Prefer the Harder Reading

            In my study, over 3500 singulars of all types (adds, omits, substitutions and transpositions) were examined in twelve chapters of the NT text.  The results show scribes creating far more difficult readings than easier readings. In fact, they created over 30 times more difficult readings for every easier reading.

Summary

            The results show that all four of the transcriptional canons are connected. By frequently omitting words and phrases, scribes also tend to make the text more terse, abrupt and harsh, less harmonised, and generally more difficult.  In short, scribes tend to make mistakes and mess things up. Who’d have thought?

3. Which canon do you think will be the most doggedly adhered to despite the evidence you presented? “Prefer the harder reading”, right?

Wilson:  My opinion is that the traditional Harder Reading canon is essentially correct, but only if we pay attention to how certain scholars – notably Hort and Metzger, but even Greenlee – carefully defined and qualified it. The majority of people commenting on New Testament textual criticism at all levels today seem to have adopted a simplistic definition of the Harder Reading canon that is almost the exact opposite of its true meaning. They think that the canon says, Prefer the Harder Reading, period. This means that any blunder a scribe introduced into the text, short of nonsense, is to be preferred.

            But that is not what Hort or Metzger say. Metzger says it like this: ‘In general, the more difficult reading is to be preferred, particularly when the sense appears on the surface to be erroneous, but on more mature consideration proves itself to be correct.’ (Textual Commentary, 12*-13*). Harold Greenlee says it even more simply:  ‘The more difficult reading is likely to be the original. This statement needs to be qualified a bit, however. What we mean is that if one reading seems at first sight to be unexpected or difficult to understand, but on further study is seen to make good sense, then this reading is probably original.’ (Scribes, Scrolls, and Scriptures, 59).  Hort says, ‘with the exception of pure blunders, readings originating with scribes must always at the time have combined the appearance of improvement with the absence of its reality. . . . In itself therefore Transcriptional Probability not only stands in no antagonism to Intrinsic Probability, but is its sustaining complement.’ (Introduction, 27).

            When Hort says that Prefer the Harder reading does not stand in antagonism to intrinsic probability but is its sustaining complement, he means they are both saying the same thing in different ways. Thus, we should not prefer a reading merely because it is more difficult. Difficult readings are precisely what authors strive to avoid; most authors write to be understood. Instead, because intrinsic probability is the sustaining complement of transcriptional probability, the true meaning of the canon is this: we should prefer the reading that makes more sense, particularly when at first sight it appears to be difficult in some way.

            The worrying thing is that so many top-level biblical commentators nowadays not only argue for readings which make less sense, but they do not even try to make the case that such harder readings are intrinsically superior. The best example of this is the singular reading of Codex Bezae at Mark 1:41. Numerous critical commentaries prefer the ‘angry’ Jesus here, but the important thing to notice is that they do not make much of an attempt to show how this reading makes more sense in context, i.e. is intrinsically superior. The notable exception is Bart Ehrman who is perhaps largely responsible for the modern popularization of this reading. But at least Ehrman tries to argue that an ‘angry’ Jesus fits with some themes in Mark’s Gospel and attempts a bit of textual ventriloquism (putting reasons into Jesus’ mouth about why he got angry at a leper).

            Ehrman thus proves the point being made here: it is not enough to merely identify a reading as more difficult – you must also prove that it is also intrinsically superior. I happen to think that Ehrman’s arguments in Mark 1:41 don’t come close to satisfying the requirements of the Harder Reading canon’s qualifying clause, or that Jesus venting his anger at an unsuspecting, socially ostracized sick man is intrinsically superior (see pages 304ff in my study). But the reason Ehrman’s argument has been so successful (culturally – in convincing others to follow him), is because so many other people seem to have either forgotten the qualification clause or never truly grasped that the canon must not be read in an unqualified way. Metzger did not have a simplistic understanding of the canon, and so the ‘angry’ Jesus reading was never allowed into the NA text. The same cannot be said with certainty about many other commentators, right up the editors of the NIV who, astonishingly, adopt the ‘angry’ reading.

            One of my friends long ago put the whole matter this way: ‘prefer the dumber reading is a dumb rule’. Strange to relate, there are many Bible scholars dabbling in textual criticism who think that we should prefer the more ridiculous reading. Maybe my study will help them get a clearer picture of what scribes did and what the canon really means. 

More questions to follow in Part Two!

Dirk Jongkind versus Reality: Vaticanus' Scribe

            Dirk Jongkind, editor of the Tyndale House edition of the Greek New Testament, is featured in a series of video-lectures and discussions hosted by the Forum of Christian Leaders (FOCL), with titles such as   

           ● Introduction to the Greek New Testament, and                    

           ● Greek Textual Criticism, and

           ●  The Earliest Evidence for the Text of the New Testament, and

           ●  Studying the Manuscripts. 

            In the last-named lecture (uploaded in November of 2017), which is only 23 minutes long,  Jongkind made some claims about Codex Vaticanus which invite clarification.

            The New Testament text in Codex Vaticanus, Jongkind stated, was made by a scribe who almost never made significant mistakes:  “When you’re copying a text, you’re going to make blunders.  The scribe responsible for this one hardly made any.”

           

            This claim should be considered alongside the observations made by Greg Paulson in his 2013 thesis at the University of Edinburgh, Scribal Habits in Codex Sinaiticus, Vaticanus, Ephraemi, Bezae, and Washingtonianus in the Gospel of Matthew; on page  60 Paulson observes, “There are 97 singular readings in B in Matthew.”  [Let me break that down for newcomers:  “B” = Codex Vaticanus, and a singular reading is a reading which occurs in only a single manuscript.]    Paulson also notes that two of the readings attested exclusively by Codex Vaticanus are adopted into the Nestle-Aland compilation, in Matthew 9:3 and 26:53b.  In the Tyndale House compilation, B’s reading in 9:3 (ειπαν) is rejected in favor of ειπον, and B’s reading in 26:53b (μοι αρτι πλείω) is rejected in the Tyndale House compilation in favor of μοι αρτι πλείους.  Although I have not thoroughly consulted the Tyndale House edition to check the editors’ decisions in all 97 passages where Codex Vaticanus has a singular reading, it seems reasonable to expect that none of them were adopted.

            Most of B’s anomalous readings in Matthew are benign.  But the readings attested exclusively by Codex Vaticanus in Matthew include – just a sample here – the following:

            ● the insertion of εις την χώραν αυτων in 2:13 (repeated from 2:12)

            ● the omission of εργα at the end of 5:16

            ● the omission of μὴ δέξηται in 10:14

            ● the insertion of ουκ before αφεθήσεται (first occurrence) in 12:32

            ● the omission of και δίκαιοι in 13:17

            ● the omission of εις after αγαθου in 19:17

            ● the repetition of πληρωθη το ρηθεν δια του in 21:4

           

Sifting through three chapters from Mark and three chapters from Luke, more mistakes from Codex Vaticanus’ scribe are observed – and this does not include Vaticanus’ itacistic (vowel-switching) anomalies and unusual name-spelling:
   

In Mark chapters 6-8:
            ● 6:2:  B adds οι before πολλοι
            ● 6:17:  B* omits την γυναικα

            ● 6:20:  B omits και before συνετήρει

            ● 6:33:  B reads εγνωσαν instead of επέγνωσαν

            ● 6:38:  B transposes to εχετε αρτους

            ● 6:39:  B reads εν instead of επι

            ● 6:54:  B* omits αυτων

            ● 7:4:  B reads ραντίσωνται instead of βαπτισωνται

            ● 7:9:  B reads τηρητε instead of στήσητε

            ● 7:14:  B reads λέγει instead of ελεγεν

            ● 7:15:  B reads το κοινυν instead of ὁ δυναται κοινωσαι

            ● 7:24:  B reads ηδυνάσθη instead of ηδυνήθη

            ● 7:37:  B adds ως after πεποίηκεν

            ● 8:2:  B omits μοι

            ● 8:3:  B reads εισιν instead of ηκασιν

            ● 8:10:  B adds αυτος after εμβας

            ● 8:12:  B omits υμιν after λεγω

            ● 8:20:  B reads και λεγουσιν αυτω instead of οι δε ειπαν

            ● 8:25:  B reads εθηκεν instead of επέθηκεν

            ● 8:35:  B reads εαυτου ψυχην instead of ψυχην αυτου

            ● 8:37:  B adds ὁ before ανθρωπος

            ● 8:37:  B reads εαυτου instead of αυτου

In Luke 1-3:
            ● 1:37:  B* reads οτι ουκ αδυνατήσει twice.

            ● 2:9:  B omits φόβον μεγαν, reading, instead, σφόδρα
            ● 2:19:  B omits ταυτα after ρηματα

            ● 2:22:  B omits του after ημεραι

            ● 2:37:  B reads αφειστα instead of αφίστατο

            ● 2:47:  B omits οι ακούοντες αυτου

            ● 3:8:  B transposes to αξίους καρπους

            ● 3:17:  B reads αβέστω instead of ασβέστω

            ● 3:33:  B omits του Αμιναδαβ

            These are not exhaustive lists.  “His execution is very careful,” Jongkind said of the scribe who produced these readings.  It seems to me that while the scribe of Codex Vaticanus is certainly not the worst scribe ever (a title that must go to the scribe of Old Latin Codex Bobbiensis), his execution leaves something to be desired, and the claim that he hardly ever made blunders must be regarded as an exaggeration.

            In addition, near the end of his lecture on Studying the Manuscripts, Jongkind employed a slide which stated, “This Byzantine Text is the text printed by Erasmus and became the received text.”  That is the sort of oversimplification which Jongkind elsewhere warns his audience against making.
            There are over 1,000 significant differences between the Byzantine Text and the Textus Receptus.  Both the Byzantine Text and the Textus Receptus usually are supported by the majority of Greek manuscripts at points in Matthew-Jude where they disagree with the Nestle-Aland compilation, but the Textus Receptus has some readings which are only supported by a small number of Greek manuscripts (and in a few cases, by none).  A professor who continues to describe the Byzantine Text as if it is one and the same as the Textus Receptus, over a decade after the release of the Robinson-Pierpont Byzantine Textform, might be vulnerable to the charge of intentionally confusing and misleading his audience.



Readers are invited to check the data in this post.

The Myth of Tenacity

            On page 78 of The King James Only Controversy, author James White states:   “Once a variant reading appears in a manuscript, it doesn’t simply go away.  It gets copied and ends up in other manuscripts.”  To support this statement, White appealed to Kurt & Barbara Aland’s similar statement:  “Once a variant or a new reading enters the tradition it refuses to disappear, persisting (if only in a few manuscripts) and perpetuating itself through the centuries.  One of the most striking traits of the New Testament textual tradition is its tenacity.” – Aland & Aland, The Text of the New Testament, p. 56.

            Aland & Aland, however, only provided three examples of this “tenacity” – and two of the three are very poorly attested among later Greek manuscripts.  Their first example is the abrupt ending of the Gospel of Mark (i.e., the ending at 16:8); among extant Greek manuscripts, it is supported by exactly one later copy (304).  In addition, 304’s text of Mark is 90% Byzantine; there is a very real possibility that 304’s exemplar contained Mark 16:9-20; like 2386 and 1420 (both of which used to be cited – erroneously – as support for the abrupt ending), 304’s support for the abrupt ending of Mark may turn out to be merely a quirk. 

            What about Aland & Aland’s other two demonstrations of tenacity?  They are found in Matthew 13:57 and Mark 1:16:

            Regarding Matthew 13:57:  Aland and Aland point out that whereas some manuscripts (B D Θ 33 700) simply present Jesus saying that a prophet is without honor in [his] homeland (πατριδι), and some manuscripts (À f¹³) present Jesus saying that a prophet is without honor in his own homeland (ἰδία πατριδι), and some manuscripts (L Byz K N W) present Jesus saying that a prophet is without honor in his homeland (πατριδι αὐτοῦ), the scribe of Codex C conflated two readings, so as to write ἰδία πατριδι αὐτοῦ.  “This is typical,” Aland and Aland say – “a scribe familiar with both readings will combine them, reasoning that by preserving both texts the right text will be preserved.”

            The obvious problem with the Alands’ proposal is that Codex C’s reading in this passage is anything but “typical;” C’s reading is singular – that is, it is the only Greek manuscript that has this reading!  Far from being “tenacious,” this reading burst into existence with the scribe of C, and then went extinct.  We may have here a tenacious page of parchment, but not a tenacious reading.

            The example in Mark 1:16 is better, where it can be argued that the majority reading is a combination, or conflation, of the reading in Codex A (του Σίμωνος) and the reading in Codex D G W Θ and the Textus Receptus (αὐτου), producing the longer reading αὐτου του Σίμωνος.  Although a counter-argument can be made, let’s step back at this point and survey the evidence that has been collected to illustrate the proposal that once a reading (whether part of the original text, or the creation of a scribe) appears in a manuscript, it does not go away:

            ● The presence of the abrupt ending of Mark in minuscule 304.

            ● The reading ἰδία πατριδι αὐτοῦ in Codex C (a singular reading), and

            ● The reading αὐτου του Σίμωνος in the Byzantine Text of Mark 1:16.  

            Now consider the mass of evidence against the concept of tenacity:  the hundreds of singular readings that appears in ancient manuscripts, but of which there is no trace in later manuscripts.  How many such readings are there?  Greg Paulson wrote his 2013 thesis on singular readings in the codices Vaticanus (B), Sinaiticus (À), Bezae (D), Ephraemi Rescriptus (C), and Washingtonianus (W) in the Gospel of Matthew, and he mentioned how many singular readings – i.e., readings that do not recur in any other Greek manuscript – each one of these codices has in its text of Matthew.  Paulson’s data:
            Vaticanus:  97.

            Sinaiticus:  Scribe A:  163. 

            Bezae:  259.

            Ephraemi Rescriptus:  75

            Washingtonianus:  112.

            I emphasize that these numbers – showing that five important early manuscripts combine to produce a total of 706 singular readings – only take the text of Matthew into consideration.  If one were to extrapolate, so as to maintain a proportion of 25 singular readings per chapter of the Gospels, then we could reasonably expect that a study of all the singular readings in these five manuscripts throughout the four Gospels would total about 2,225.  Well, there we would have 1,500 non-tenacious readings – never seen before or after the one time they appear – versus the three submitted by Aland & Aland.  Suppose that two-thirds of these were nonsense-readings which scribes could reasonably be expected to regard as mistakes.  That would leave 500 non-tenacious, non-nonsense readings, just in the Gospels, just in these five manuscripts.

            And what if we consider some earlier manuscripts?   A single quotation from James Royse (from p. 246 of Scribal Tendencies in the Transmission of the Text of the New Testament, chapter 15 of The Text of the New Testament in Contemporary Research, © 1995 Wm. B. Eerdmans Publishing Co., Ehrman and Holmes, editors) may serve to tell us:  Royse provides a chart which conveys that Papyrus 45 has 222 significant singular readings; Papyrus 46 has 471 significant singular readings; Papyrus 47 has 51 significant singular readings; Papyrus 66 has 107 significant singular readings; Papyrus 72 has 98 significant singular readings; Papyrus 75 has 119 significant singular readings.  (In a footnote, Royse helpfully defines “significant singular readings” as “those singular readings that remain after exclusion of nonsense-readings and orthographic variants.”)  

            Hundreds of readings refute the Alands’ claim about tenacity.  (Over a thousand in just six papyri.)  There is no evidence that these readings were ever perpetuated after they entered the transmission-stream precisely once.  Here the case might rest, Q.E.D. 

  

            But let’s also consider a few peripheral pieces of evidence.             

            ● Origen and Luke 1:46.  Bruce Metzger pointed out (in his essay References in Origen to Variant Readings) that Origen, in his commentary on Matthew, mentioned that in some copies, the Magnificat is said to be sung by Mary, while according to other copies, Elizabeth is the singer.  Presently, no Greek manuscripts say that Elizabeth was the singer. 

            ● Tertullian and John 1:13.  All Greek manuscripts of John 1 support the use of a plural in this verse, so as to understand it as a generalization about all genuine believers.  Tertullian, however, seems to have been convinced that the original reading here was singular, and that Valentinian heretics were responsible for changing it to a plural form.  (Denis S. Kulandaisamy has written a 300-page book on this little subject.)   

            ● Epiphanius and Matthew 2:11.  The apologist Epiphanius of Salamis, in the late 300s, mentioned that the wise men opened their “wallets” (πήρας), and noted that some manuscripts – like all Greek manuscripts extant today – referred instead to their “treasures” (θησαυροὺς).  No known manuscript today has the reading πήρας.  

            ● Jerome, Isidore, Theodore of Mopsuestia, and Ephesians 5:14.  Several significant patristic writers drew attention to a textual variant in Ephesians 5:14, where instead of “And Christ will shine on you,” some manuscripts in their day read “And Christ will touch you.”  John Chrysostom seems to have alluded to this variant, too, although he did not go into detail about it.  If it is extant in any Greek manuscripts, there is no mention of them in the Nestle-Aland Novum Testamentum Graece. 

            ● Origen and Matthew 4:17.  Origen says plainly that some manuscripts do not have the word “Repent” in this verse.  All of our Greek manuscripts, however, presently support the inclusion of “Repent” in this verse; only in the Old Syriac copies and in the Old Latin Codex Bobbiensis is it missing.     

            ● Jerome and Matthew 13:35.  Jerome firmly asserts that all of the ancient copies state that these things were done to fulfill what was spoken through the prophet Asaph.  No manuscripts of Matthew today have this reading, and Jerome mentions that by his own time, the text had been altered; one alteration had yielded the erroneous reading “Isaiah the prophet.” 

            If you take in hand Amy Donaldson’s two-part dissertation Explicit References to New Testament Variant Readings Among Greek and Latin Church Fathers, you will find details about these readings, and others, which were cited by early patristic writers but which now are never or only rarely found.

            Finally, the myth of tenacity is not only refuted by empirical evidence; it is rendered superfluous by logic.  For if a reading were to enter the transmission-stream, and then fall into oblivion, and the manuscript containing it did not survive, and no one referred to it in patristic writings, how would we know?  Even if we did not have the empirical evidence that thoroughly refutes the theory of tenacity, the theory could only ever be an assumption, not a verifiable thing.

                         

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