Scribal Habits in Greek New Testament Manuscripts (Part Two) - Ten More Questions for Andrew Wilson

Andrew Wilson

 Here again is Andrew Wilson, engaging some of my questions about the implications of his important book Scribal Habits in Greek New Testament Manuscripts.

●Publishing a book challenging four of the most basic guidelines of a field of research was ambitious – almost like redrawing the compass on a map. Which canon do you think will be the most quickly discarded now that Scribal Habits in New Testament Manuscripts has been published?

            The evidence from my study, Royse’s study, Dain’s work, and every single other study of scribal habits since Royse for the last 45 years has shown that scribes generally tend to omit, and the result inevitably is a shorter, terser and harsher, less harmonized text. The interesting thing, to me, was that most of the people producing these studies that challenged the canons (Royse, Hernandez, Jongkind, myself) did not initially set out to do so; it was instead an innocent, unintended and accidental discovery.

            By contrast, NT textual criticism as traditionally and currently formulated holds to the exact opposite beliefs, in theory and, equally importantly, in practice, as is seen in our current critical text, and in most modern English Bibles based upon it. So, how will textual critics respond to this study?

            I once saw a sign in a train in England that quoted the famous economist John Maynard Keynes: ‘When the facts change, I change my mind’. What do you do, sir?’ However, as John Kenneth Galbraith (another economist) wrote: ‘faced with the choice between changing one’s mind and proving that there is no need to do so, almost everyone gets busy with the proof’.

            When I published the 2026 study, with evidence from over 10,000 singular and sub-singular readings, my thinking was that it might change some people’s minds. After all, some people only  change their mind when the evidence is overwhelming.  But maybe that’s not the way the human mind works. Most people get uncomfortable and defensive when they are shown that they hold to a set of false beliefs. Very few people are prepared to say, ‘I was wrong’. I suspect most textual critics will just ignore my study, while some will perhaps try to explain its data away.

            However, as Aldous Huxley stated, ‘Facts do not cease because they are ignored’ and as US President John Adams said, ‘Facts are stubborn things, and whatever may be our wishes, our inclinations, or the dictates of our passion, they cannot alter the state of facts and evidence’.

            There have been two sorts of responses to my study. Some people have been positive and encouraging, while others ignore it (or, for all I know) are already sharpening their pencils and preparing their rebuttals. The problem, however, is that they not only have to disprove my work, but also the results of a dozen other methodologically varied studies, and Dain himself.

            New Testament textual criticism is, as I argued once in an online article, a mixture of science, art and religion, and is like some other political or social or religious debates, such as climate change. Some people (usually on the extremes) do not change their minds based on evidence – they either don’t believe in global warming at all, or they are climate catastrophists. No amount of evidence changes these people’s minds, because their view is not based on evidence, but ideology. The same is true of modern gender debates; people who believe that men can become pregnant are not moved by any amount of biological facts that are set before them.

But in between the extremes, there are a silent majority of people who are open to carefully evaluating evidence and prepared to follow where it leads. The same is true of NT textual criticism: there are extremists at either end – KJV-onlyists, TR advocates and some Byzantine-prioritists at one end of the spectrum and unrepentant Hortians at the other. No amount of evidence makes any difference for these people.

            The equal longest chapter of my study deals with many of the attempts over the last 40 years to argue that the methodology or evidence or conclusions from Royse’s study (and those studies like his) are somehow flawed. It is true that any new and revolutionary idea like Royse’s deserves to be carefully cross-examined. So these counter-studies have used a variety of methodologies to further test scribal habits.

            Some have argued that there was some problem with the singular reading method, others have argued that the data does not back up Royse’s claim that scribes tend to omit, some argue Royse did not quite understand the canon, and some argue that the matter is unclear and that scholars should adopt an agnostic attitude to the question. But none of these responses to Royse’s work have uncovered any grave defect in his method or problems with his data. In fact, the extra data from these further studies has only added extra evidence to the mountain of data showing that scribes tend to omit. In attempting to critique Royse, they have actually strengthened his case.

            The fact that the canons are mostly false is downright inconvenient. It means that the current critical text, which is heavily based on the preference for the shorter, harsher, disharmonized reading, is wrong in many, many places. Who wants to admit that, especially after telling the world that our NT text is as close to the original text as it is possible for scholarship to attain?

            So, I don’t actually have a very optimistic attitude about textual critics changing their minds on the canons. On the other hand, I think the openness to evidence seen in the ECM and CBGM programs is a sign pointing in the right directions. So we will see. But the greatest scientists (like Einstein) and economists (like Keynes) had the humility to admit it when they got things wrong, and strangely, the world today sees this as added proof of their true greatness.

● The involvement of a proofreader in the production of a manuscript inevitably improves the text before a manuscript leaves the scriptorium. As a whole, was the Byzantine transmission stream better at retaining text than the Alexandrian text-stream? And should it therefore be generally favored as a base-text by translators as a result?

To consider the effect of scribal activity on the text, look at the additions and omissions in the NA 28 apparatus in the first 16 verses of Matthew 6:

● 6:4 omit αυτος (himself): ‎א B L Θ pc NA || D W Maj

● 6:4 omit εν τω φανερω (in secret): א B D pc NA || L W Θ Maj

● 6:5 add στησαι (to stand): D || א B L W Θ Maj NA

● 6:5 omit αν: א B L NA || W Θ Maj

● 6:6 omit εν τω φανερω (in secret): א B D pc NA || L W Θ Maj

● 6:8 add ο θεος (God): א B || D L W Θ Maj NA

● 6:10 omit ως (as) D || א B L W Θ Maj NA

● 6:10 omit της (the): א B W pc NA || D L Θ Maj

● 6:13 omit the doxology (kingdom, power, glory): א B D pc NA || L W Θ Maj

● 6:14 omit γαρ (for): D L || א B W Θ Maj NA

● 6:14 add εν τω (in the): Θ || add 3W: L pc || א B W Maj NA

● 6:15 omit τα παραπτωματα αυτων (their trespasses): א D pc NA || B L W Θ Maj

● 6:15 add υμιν (to you): D pc || א B L W Θ Maj NA

● 6:16 omit οτι (that): א B D pc NA || L W Θ Maj

            If you count, you will see that the NA text has the shorter reading 12 times out of 14 variation units, while the Byzantine text has the longer reading 10 times. Further, you will notice that 6 of the NA shorter readings are one word long. The NA text is heavily inclined to the shorter reading found in a few MSS. This snapshot will be seen wherever you look in the NT: the Alexandrian MSS (and the NA text) are much shorter.

            It is pretty obvious that Alexandrian priority and Prefer the Shorter Reading are very nearly inseparable. They stand or fall together. In practical terms, it is very hard to believe that scribes tend to omit and to hold to a preference for Alexandrian manuscripts. You can believe in one or the other, but not both.

            To see why, consider the ratios of additions to omissions from studies of scribal habits:

● Vaticanus has 12 additions to 41 omissions in my study (a 78% omission rate,

about the highest ratio of omissions to additions of any MS),

● Sinaiticus has 65 adds to 101 omits (a 61% omission rate).

● P75 had 11 adds to 34 omits (75% omit rate) in Royse’s study,

● P46 had 52 adds to 161 omits (75% omit rate) in Royse’s study.

            Someone may respond that singular readings just show the idiosyncratic errors of these MSS, not the quality of its underlying text. But when we look at the sub-singular readings of P46 and B, or P75 and B, we see the exact same phenomenon of omission. Thus, among the sub-singular readings of B and P75 in Luke and John, there are 4 additions and 11 omissions (plus two omissions of prepositional prefixes or suffixes). Among sub-singulars of B and P46 there are 3 additions and 15 omissions (plus one omission of a prepositional prefix). And when we look at sub-singular readings of B and Aleph we see the same thing.

            So, not only are our Alexandrian uncials characterised by many omissions from the text, but so are their ancestors (the papyri), and when they join together in groups of documents, they do the same thing, and when we look at larger combinations of such ‘Alexandrian’ manuscripts, they all do the same thing. This tendency to omit is ‘out of character’. Omission is their signature error, which they have inherited from their genetic ancestry, and they continue the tradition of their fathers.

            So, if we follow the “earliest and best” MSS like Vaticanus and Sinaiticus (and quietly ignore studies of scribal habits), we are not only disbelieving Dain’s Dictum (the most common error of scribes is omission), we are also forgetting Hort’s Dictum (“knowledge of documents should precede final judgements on readings”) by failing to acknowledge that the early Alexandrian MSS have a characteristic error:  omission. And we are also ignoring another important principle of Hort’s: evidence of Groups of Documents (See the huge section of his Introduction dealing with this class of evidence, pp 187-271). This is what happens when we measure Hort’s preferred MSS by Hort’s own rules and principles: they are characterised by shorter readings, the most common scribal error.

            Whenever we see shorter readings, we should be suspicious, but when we see them found in Alexandrian MSS alone, or with very few others, we can be almost certain it is an error.

            The question becomes, How many principles, dictums (some in ALL CAPS), rules of textual criticism and statistically significant studies do we wish to ignore or break to defend our pet MSS? Can we give any reason, other than repetition and tradition, for why these Alexandrian MSS are the so-called best manuscripts? I’m struggling to think of any, other than the fact that they are slighter older MSS.

            Does this mean we should prefer Byzantine readings? I can only speak for myself, and here I must reveal my underlying ideology, but my starting point in NT textual criticism is not some theory of manuscript transmission, or some history of the text (there is so much we do not know about the history of the text that this is ridiculous).  My starting point is the 98% of the New Testament text that is virtually certain. The 98% shows me that Jesus is the Son of God who proved it by rising from the dead, and who states that the Scriptures are the Word of God and they are true (Mark 7:13, John 10:35, 17:17). This means that I am not going to follow any readings that are clearly and obviously wrong . If I may adapt Daniel Mace’s ‘crisp dictum’ (Metzger), no manuscript is as old as common sense (nor is any group of manuscripts, however numerous, as old as common sense).

            When I look at some Byzantine readings, like the placement of the Roman doxology (Rom. 15:25-27) after Romans 14:23, or the Byzantine reading in James 2:18 (‘show me your faith from your works’ instead of ‘show me your faith APART from your works’), I can only shake my head, snigger, and wonder how any thinking Christian could adopt such errors and blunders that even the TR refused to follow. And if the Byzantine text is wrong in some obvious places, it can be wrong in other places.

            So if the Alexandrian MSS are better described as being of exceptional brevity rather than ‘exceptional purity’ (Hort). Does this mean that we automatically switch and treat the Byzantine MSS as exceptionally pure? I do not think so.

            Some Byzantine-prioritists will ask: so how did the Byzantine text become the great majority? Obviously, the answer is NOT by some edition of the text like what happened with the Latin Vulgate (an edition or ‘recension’). More likely, in the same way as the Masoretic text (with all its obvious errors) became the numerically dominant majority text of the OT: by adoption, that is, by scribes choosing a certain manuscripts (with their good and bad readings) as exemplars, so that later copies of these multiplied and, under high-quality copying, stayed pretty much the same.

            Therefore, while I think the evidence shows that the Byzantine text is generally preferable in add/omit variants, every case needs to be decided on its merits. In short, I do not follow Hort, but neither do I think that NT textual criticism should swing from one extreme to the other and follow Burgon – I am more inclined to Scrivener’s way of thinking.

            Our current critical text, NA 28 , consistently prefers the shorter reading, in line with the traditional canon. Thus, according to my count, in Matthew NA 28 prefers a shorter reading 418 times (58%) and a longer reading 279 times (39%) and a middle reading 20 times (3%). If we were to reverse these ratios to reflect scribal reality, so that longer readings were about 60% and shorter readings 40%, we would have to make about 140 changes to the text in Matthew alone. If we were to extrapolate that out to the entire NT, then we are looking at a ballpark figure of about 1000 changes to the text, most of which would move in the direction of the Byzantine text.

If you will forgive me abandoning the language of academic niceties, our current critical text has suffered death by a thousand cuts. That means, too, that our modern English Bibles translated from this text (ESV, NIV, NET, NASB, etc.) have left out hundreds of words, phrases and whole verses that should be reinstated.

            Admittedly, most of the correction that the current critical text needs is fine-tuning:  the addition of single words like pronouns, articles, conjunctions and verbs or nouns whose meaning was merely implied. These changes will not make much difference to the meaning of the New Testament text, but they affect its style.

● Applying what you’ve learned: should John 6:11 include τοις μαθηταις οι δε μαθηται?

            I argue for this reading in my study (pages 312-14), for four reasons.  First, there is an obvious possibility of homoioteleuton here. The leap from τοις to τοις produced the shorter reading. In my study, in line with Dain’s Dictum, homoioteleuton presents itself as a major cause of scribal omissions.

            Second, intrinsically, the shorter reading makes little sense. It is saying Jesus himself distributed the bread to a crowd of around 10,000 people, and after that did the same with the fishes. This would have taken about six hours, unless he used a miracle of teleportation (no commentator suggests this). What is the point of having disciples if you cannot put them to work to do stuff like this? Further, the Lord deliberately draws the disciples into the miracle, testing their faith in verses 5-9, getting them to make the people sit down in verse 10 and telling them to gather up the fragments afterwards in verse 12. He is inviting their participation in the miracle in order to strengthen their faith. Why then would he not get them involved in the central act of food distribution? The shorter reading does not make good sense.

            Third, some commentators argue that the longer reading is a harmonization to Matt. 14:19, but if you look at the parallel passage, Matthew uses different wording. This is what I call a Pseudo-Harmonization. If the rationale for scribal harmonization is that scribes had memorized the wording of the New Testament, and therefor supplied words from other parallels (as is argued in the textbooks by Metzger and Ehrman, Text of the New Testament, 127 and the Alands, Text of the New Testament, 290), why did the scribes all not only misquote Matthew and mix up his wording, but all hit upon the same non-parallel wording? That is an amazing coincidence. Based on the evidence that omission is more common than harmonization, we should again prefer the longer reading.

            Four, modern critics are no doubt swayed by the external evidence for the omission:  three early papyri (28, 66, and 75) as well as Sinaiticus, Vaticanus, Alexandrinus and various other witnesses. This is a powerful line-up of evidence. However, this example illustrates a point made in other cases by James Royse: this omission would ‘surely be condemned as Unproblematic scribal errors’ if it was found in manuscripts other than Sinaiticus and Vaticanus (Royse, “Treatment of Scribal Leaps”, 540). What are the chances that the shorter reading here is the result of an early Alexandrian error that found its way into these later witnesses, and coincidentally occurred in some other scattered manuscripts? I don’t think that possibility can be ignored or easily dismissed.

● Should Mark 16:18 include και εν ταις χερσιν?

            My opinion, in line with patristic sources (Papias and Irenaeus), but also based on internal evidence, is that Mark’s Gospel shows clear signs of double-authorship, i.e. that Mark wrote down Peter’s recollections of Christ. If so, the longer ending of Mark has no real problems in terms of style and diction (if Mark finished off what Peter, the primary witness, could not, for whatever reason). The manuscript evidence for its inclusion is very strong (only two manuscripts out of 1600 do not have it) and the variety of different endings are best explained by pseudo-scholarly attempts to bridge the discontinuity at 16:8-9.

            As to whether the words in 16:18 are original, there does not appear to be any parablepsis that might explain the shorter reading as an omission, and the external evidence for the shorter reading is far stronger and broader (early Byzantine and Western MSS vs a few later Alexandrians). Therefore, I think the shorter reading is probably original. But I’m open to good arguments.

● Should Matthew 10:8 include νεκρους εγειρετε?

            I think the text should probably include the words. First, because homoioteleuton would easily explain the loss of the words (look at all the textual confusion here because of similar endings), second because scribal omissions are far more common than additions, third, because although the reading seems hard (did the disciples really have the power to raise the dead?), yet the Lord was in the habit of telling them to have great faith, and Peter and Paul raised the dead in the book of Acts, and fourth because the reading has very broad attestation – among the Latin, Syriac and Coptic versions and even some early Byzantine MSS (W, P).

● Should future textbooks tell readers that, ceteris paribus, the longer reading should be preferred?”

            If they are being honest, definitely. But it is highly unlikely to happen anytime soon.

● It occurred to me that some scholars are sure to object that you’re resorting to anecdotal evidence to challenge their common sense canons. But you used over 10,000 examples. Would you say the onus of proof has shifted to those who insist that the shorter reading, and the harsher reading, and the less harmonious reading, and the more difficult reading, should be adopted?

            The traditional canons are not based on common sense at all, once you start thinking about them. The easiest mistake to make is to omit, and other simple blunders. All the other canons flow from this common-sense reality. The fact that scribes are performing a long, laborious task that requires great concentration means that, inevitably they are going to make mistakes that introduce errors and difficulties.

● Your book tends to reinforce Clark’s idea that the text gradually got shorter, not longer – like a traveler’s wardrobe losing a sock or other garment on each trip. So, should William A. Strange’s qualified endorsement of the Western Text of Acts be taken more seriously? Why or why not?

            I’m not familiar with Strange or his view, but I think Hort’s line about ‘knowledge of documents’ applies equally to the Western text, or even more than it does to Alexandrian manuscripts. Bezae has a well-known character, and if we may adapt Clark’s analogy, it has the character of a somewhat weird, wandering minstrel who embellishes and changes his ballads at a whim.

            What is not so well known about Bezae is that it adds and omits in all sorts of ways.  Everybody knows about its longer additions in Acts, but it is also guilty of enormous amounts of omissions. My figures were 161 additions, and 152 omissions. It not only has large paraphrases and additions, but it also has huge numbers of other more ordinary errors, like small omissions. It is the most eccentric text of the NT and therefore one that is to be treated with extreme caution.

● With four canons essentially reversed or rendered null, what implications are there for variant-units where the Textus Receptus and Alexandrian Text agree on a longer reading against the Byzantine Text’s shorter reading (like in, say, Acts 4:12)?

Available on Amazon

            I find these cases of Alexandrian-TR alignments quite interesting, and once tried to compile a list of them (I don’t know where that went). As to the particular verse, it is a pretty obvious case of Homoioteleuton, which means that the longer reading is highly likely correct. Most Byzantine priority proponents would point out, I’m sure, that some Byzantine streams (K^c and K^r) contain the words, so even they would probably agree.

● What are you going to do for a sequel? Or will Gillian burn your computer first?

            Gillian does not mind computers; it is books she has a problem with, and does not like me bringing into the house. As for sequels, I am working on one or two projects, but we will have to wait and see.

Thank you for a very engaging conversation and may your future studies bear good fruit.

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!

The Red Gospels of Ganjassar - A Member of Family 1

Ganjasar Monastery

The Red Gospels of Ganjassar, an Armenian Gospels manuscript from the late 1100s that has resided in Chicago in the Goodspeed Manuscript Collection since 1941, used to be at the Ganjasar Monastery (and possibly at Ani before that).  Its name is due not to the color of the ink in which it is written but to the red frames of its miniatures.  (The term “miniature” in this context itself does not have anything to do with the size of the illustrations but rather with the red lead (minium) used as pigment to frame them.)

Considering its unusual name, and recognizing that today is International Ginger Day, I am pleased to present the remarkably gifted Phoebe McArthur, a.k.a. PhoebeIsGinger, the influencer, Commander of the Ginger Defense League, to accompany a couple of its more interesting features.  Here she is posing with fol. 213r, which features text from Luke 19. 

Thanks to Alin Suciu for help  identifying the text.

The text of the Red Gospels is primarily Byzantine but there is heavy Caesarean mixture – the most obvious Caesarean feature being the placement of the story of the adulteress (usually found after John 7:52 in Greek manuscripts) at the end of John, as in family-1.  It has been classified as a member of Group Z (as in Zohrabian).  Although the text is normally formatted in two columns per page, the second page containing the pericope adultera is written in a single column filling the page.

The miniature of the Annunciation is similar to the same miniature found in Matenadaran 2374 – in which the pericope adultera appears in a highly unusual form.

The Canon-tables are very nice, though bound in the wrong order.  Several pages of the beginning of the Gospel of Matthew are missing; the extant text begins in 8:34. 

Alice Taylor has suggested (in Treasures in Heaven:   ArmenianIlluminated Manuscripts) that the illustration of Christ’s crucifixion favors the idea that the artist Abas was connected to churches in Georgia or Greece, inasmuch as the chalice is shown catching both blood and water.

Page-views of the entire codex can be accessed at the Goodspeed Manuscript Collection website.

PhoebeIsGinger with the pages of the Red Gospels containing the story about
Jesus and the adulteress following the end of the Gospel of John.

 

Revelation 3:14-16 - Quirks in Sinaiticus

          Although Codex Sinaiticus has been hailed as “The World’s Oldest Bible” and was one of the two Alexandrian manuscripts very heavily favored by textual critics of the 1900s, fans of both the traditional text and modern compilations agree that its text of Revelation is not very good.  Despite the stern warning in chapter 22 against changing the words of the book, the scribe of Sinaiticus made numerous changes, including two interesting ones in Jesus’ address to the Laodicean church in chapter three.

          Let’s compare Revelation 3:14-16 in À with the same passage in the Robinson-Pierpont Byzantine Textform and the Nestle-Aland NTG compilation.

Byz:  Καὶ τῷ ἀγγέλῳ τῆς ἐν Λαοδικείᾳ ἐκκλησίας γράψον, Τάδε λέγει ὁ Ἀμήν, ὁ μάρτυς ὁ πιστὸς καὶ ἀληθινός, ἡ ἀρχὴ τῆς κτίσεως τοῦ θεοῦ· Οἶδά σου τὰ ἔργα ὅτι οὔτε ψυχρὸς εἶ οὔτε ζεστός· ὄφελον ψυχρὸς ἦς ἢ ζεστός.  Οὕτως ὅτι χλιαρὸς εἶ, καὶ οὔ ζεστὸς οὔτε ψυχρός, μέλλω σε ἐμέσαι ἐκ τοῦ στόματός μου.

NA: Καὶ τῷ ἀγγέλῳ τῆς ἐν Λαοδικείᾳ ἐκκλησίας γράψον· Τάδε λέγει ὁ ἀμήν, ὁ μάρτυς ὁ πιστὸς καὶ ἀληθινός, ἡ ἀρχὴ τῆς κτίσεως τοῦ θεοῦ· οἶδά σου τὰ ἔργα ὅτι οὔτε ψυχρὸς εἶ οὔτε ζεστός. ὄφελον ψυχρὸς ἦς ἢ ζεστός. οὕτως ὅτι χλιαρὸς εἶ καὶ οὔτε ζεστὸς οὔτε ψυχρός, μέλλω σε ἐμέσαι ἐκ τοῦ στόματός μου.

À:   Καὶ τῷ ἀγγέλῳ τῆς ἐν Λαοδικίᾳ ἐκκλησίας γράψον· Τάδε λέγει ὁ ἀμήν καὶ ὁ μάρτυς ὁ πιστὸς καὶ ἀληθινός ἡ ἀρχὴ τῆς ἐκκλησίας τοῦ θῦ· οἶδά σου τὰ ἔργα ὅτι οὔτε ψυχρὸς οὔτε ζεστός. ὄφελος ψυχρὸς ἦς ἢ ζεστός. οὕτως ὅτι χλιερὸς εἶ καὶ οὔτε ζεστὸς οὔτε ψυχρός, εἶ παυσε  τοῦ στόματός σου.

          Setting aside mere orthographic differences, the most interesting variants here in À are the reading (n v.14) ἐκκλησίας τοῦ θεοῦ (instead of  κτίσεως τοῦ θεοῦ) and (in v. 16) εἶ παυσει  τοῦ στόματός σου (instead of μέλλω σε ἐμέσαι ἐκ τοῦ στόματός μου).

          How did such readings – unique to Sinaiticus – originate?  Let’s take them one at a time.  It’s conceivable that ἐκκλησίας was in the scribe’s mind because the word occurred a few lines previously in the text in verse 13, at the beginning of a line (as it does here in v. 14), and he repeated the word carelessly.  But another theory, proposed by Juan Hernández Jr. on p. 111 of Codex Sinaiticus:  An Early Christian Commentary on the Apocalypse?, is that the scribe who invented this variant had a theological motive.  Hernández wrote:

    Juan Hernandez Jr. (Bethel University)

           “The first Christological reading to surface is in Revelation 3:14.  The title of the risen Christ, ‘the beginning of the creation of God,” κτίσεως τοῦ θεοῦ, is altered to the ‘beginning of the church of God, ἐκκλησίας τοῦ θεοῦ.  The change eliminates the possibility of placing Jesus within the created order and is conspicuous against the backdrop of the fourth century, defined as it was by its pitched theological battles over the precise nature of the Son. . . . The eradication of such language in Codex Sinaiticus appears to indicate that the wording was a problem,  The title of Revelation 3:14 was thus harmonized to the title of Colossians 1:18, where Jesus is the head of the church (κεφαλὴ τῆς ἐκκλησίας).”

            A dogmatic motive might be argued in the opposite direction:   if the expression “ἀρχὴ τῆς κτίσεως τοῦ θεοῦ” were construed to mean that Jesus was distinct from creation, an Arian scribe wishing to perceive Jesus as merely the first created entity, appointed by the Father to rule the rest of creation, could introduce church of God in place of creation of God in order to convey simply that he was the founder of the church.   

            In the same essay (p. 112), Hernández shares some thoughts on the singular variant in 3:16, calling it “one of the boldest textual changes to the portrait of Jesus in the Apocalypse.”  He writes, “Instead of vomiting the Laodiceans out of his mouth, μέλλω σε ἐμέσαι ἐκ τοῦ στόματός μου, the text of Sinaiticus reads that Jesus was ‘to stop their mouth(s)’ παῦσει τοῦ στόματός σου. . . . “Passages that refer to Jesus eating, sleeping, being angry, or struggling at Gethsemane posed a serious theological problem and were the subject of rigorous debate.  Presumably a passage where Jesus threatens to vomit would have been just as problematic, supplying a possible rationale for the change here.”  This is conceivable but merely conceivable – an alternative is that the scribe wished to arbitrarily change Jesus’ vivid language into his own wish to quiet mediocre believers.  In either case he was very reckless. 

The Laodiceans - and subsequent believers with attitudes like theirs - thought they were more than adequate.  But they were wrong.  It is not our own desire to be happy that produces worthwhile living.  It is not simply in our own financial success that we find lasting satisfaction.  It is not in surveying the  ephemeral substances of this world that we gain insight into the meaning of our lives.  It is in Jesus Christ, who rules over creation, that we find genuine wealth, removal of our sins, and true vision.    

 

Rising Star: Nelson Hsieh Interview, Part 2

Nelson Hsieh

Returning to answer more questions in his interview on assorted text-critical issues, here is Nelson Hsieh, Assistant Professor of New Testament at New Orleans Baptist Theological Seminary:

Q:  Nelson, when did the archetype of family 35 first appear, more or less?

 

Hsieh: Wilbur Pickering’s argued that Family 35 (von Soden’s Kr group) is “ancient” (going back to the first few centuries after the New Testament), “independent” of von Soden’s Kx group (the dominant Byzantine tradition), and “non-recensional,” but I remain unconvinced in light of the (fairly) objective evidence: among the 231 manuscripts identified as Kr, these manuscripts:

1. exhibit over 98% textual agreement

2. are consistent in having an elaborate lectionary apparatus (including an explainer in 28 manuscripts entitled, Δήλωσις πῶς δεῖ εὑρίσκειν τὰ καθημέραν εὐαγγέλια: “explanation of how one must find the daily gospel readings”)

3. omit the Eusebian canon tables (except in five manuscripts) 

4. place obelii next to the Pericope of the Adulteress 

5. generally show late-Byzantine calligraphic hands 

6. have few itacistic spelling errors

7. most are written on parchment during a time when paper was in common use [See  Georgi Parpulov, “Kr in the Gospels,” in The New Testament in Antiquity and Byzantium, ed. David Parker and Holger Strutwolf, ANTF 52 (de Gruyter, 2019), 203–13.]

8. among manuscripts that can be securely dated, no Kr manuscripts exist from prior to the fourteenth century (prior to 1304 to be exact). [Parpulov also redates minuscule 35 to fourteenth century (as opposed to the eleventh century); see Parpulov, “Kr in the Gospels,” 212-13.]

Georgi Parpulov favors a late-thirteenth / early-fourteenth century origin for Kr, in contrast to von Soden, who put the origin of Kr in the eleventh century, although he expressed this with uncertainty. [See Hermann von Soden, Die Schriften des Neuen Testaments, I. Teil: Untersuchungen, II: Abteilung: die Textformen, A. die Evangelien, 2nd ed. (Vandenhoeck & Ruprecht, 1911), 763.] The uniformity of these palaeographic, paratextual, and textual features suggest a recension (i.e., a deliberate revision).

However Pickering’s protest is fair: “If anyone wishes to claim that [Family 35] is [recensional], I request that they state who did it, when and where, and that they furnish evidence in support of the claim.” [Wilbur Pickering, God Has Preserved His Text! The Divine Preservation of the New Testament, 6th ed., 2024, https://www.family35nt.org/en/God-Has-Preserved-His-Text-v6.pdf. Page 338. Emphasis added.] But Pickering is asking for a level of evidence that doesn’t exist for any textual grouping of New Testament manuscripts. The argument that Family 35 is recensional is based on palaeography, paratext, and extremely high textual agreement. 

External evidence is slim, but there are some hints towards possible origins of Family 35:

(1) the provenance of Kr  manuscripts tends to be in and around Constantinople and Greece (especially Mount Athos). Thus, Parpulov rejects von Soden’s origin of Kr exclusively in Constantinople and suggests that Kr  was circulated through a network of late Byzantine scribes and their patrons in and around Constantinople and Greece. 

(2) Kr manuscripts have a good number of named scribes, who added various poems and metrical epigrams that show up in a few other Kr  manuscripts, but not the whole Kr  tradition, which suggests private and personal contact with some other scribes. This, then, also goes against von Soden’s suggestion that ecclesiastical authorities produced and disseminated Kr . [ Von Soden, Die Schriften des Neuen Testaments, Vol. I.2, 805.] All of these named scribes contributed to the growth and distribution of Kr manuscripts, although none of them alone would have had the authority to unilaterally impose such a revision upon others. 

(3) Parpulov suggests that Kr probably was not based on a single exemplar/archetype; if such were the case, we would expect at least one scribal colophon (among 231 Kr manuscripts) to make such a claim. We would expect something similar to the so-called “Jerusalem colophon” found in some middle-Byzantine manuscripts; the colophon claims that its manuscript was “copied and corrected from ancient copies from Jerusalem, which were preserved on the holy mountain.” [See Tommy Wasserman, “The Greek New Testament Manuscripts in Sweden with an Excursus on the Jerusalem Colophon,” Svensk Exegetisk Årsbok 75 (2010): 77–107.]

(4) Probably the best guess (made by both Parpulov and von Soden) is that there was some sort of lectionary influence upon Kr manuscripts since the extensive lectionary material is the most distinctive paratextual feature of Kr manuscripts (even having an explainer of the lectionary system entitled, Δήλωσις πῶς δεῖ εὑρίσκειν τὰ καθημέραν εὐαγγέλια: “explanation on how one must find the daily gospel readings”). However, since we do not have a critical edition of the Byzantine Gospel lectionaries, it is not yet possible to understand the nature and extent of the influence of the lectionaries upon Kr manuscripts.

Thus, while we cannot satisfy Pickering with a specific person, place, and date for the creation of Kr  – yet there are enough hints to say the following: Kr arose among a private network of several named scribes operating in and around Constantinople and Greece during the late-thirteenth and early-fourteenth century. Put negatively: Kr probably does not derive from a single/authoritative archetype, thus probably no ecclesiastical authority created or disseminated the Kr revision. Unless further external evidence is found (such as a new scribal colophon), our best guess is that the Byzantine Gospel lectionary influenced the text of Kr manuscripts.

Q: What are your thoughts about Snapp’s canon “Prefer the less specific reading”? 

 

Hsieh: According to Snapp, one of his principles of equitable eclecticism is: “A variant which involves a rare, obscure, or ambiguous term or expression is more likely to be original than a rival variant which involves an ordinary or specific term or expression..” [James Snapp, New Testament Textual Analysis (Independently Published, 2023), 403-4.] Elsewhere in the same work, Snapp discusses “thoughtful/intentional changes,” which might be motivated by “a desire to augment/clarify the meaning of the original text … [therefore], a non-specific reference might be exchanged for a more specific reference.” [Snapp, New Testament Textual Analysis, 15-16.] Elsewhere, Snapp says that there was “a tendency to augment the specificity of Old Testament references. … copyists did not try to make a text more difficult – but sometimes they did, when attempting to make it clearer or more specific.” [Snapp, New Testament Textual Analysis, 323-24.]

Scattered throughout his work, Snapp provides examples of this tendency:

• Matthew 1:22-23 describes Jesus’s birth as happening, “so that the word from the Lord through the prophet might be fulfilled,” then quotes from Isaiah 7:14. NA28’s text has no mention of Isaiah; Codex Bezae, several versions, and Irenaeus (in part) add “Isaiah.” According to Snapp, “this makes the text more specific.” [According to NA28, the following added “Isaiah”: 05, it (= all or a majority of Old Latin witnesses), Syriac (Sinaitic, Harklean, and Curetonian with minor variation), one Sahidic manuscript, and Irenaeus in Latin (in part). Irenaeus quotes from Matthew 1:22 three times in the Latin edition of his Against Heresies (3.16.2; 3.21.4; 4.23.1); Isaiah is added only in the second instance (Adv. haer. 3.21.4). This is not absolutely clear evidence that Irenaeus’s text of Matthew had “Isaiah”; perhaps Irenaeus himself accidentally added it. See Adelin Rousseau and Louis Doutreleau, eds., Irénée de Lyon: Contre les Hérésies, Livre III, SC 211 (Les Éditions du Cerf, 1974), 292, 410; Adelin Rousseau, ed., Irénée de Lyon: Contre les Hérésies, Livre IV, SC 100 (Les Éditions du Cerf, 1965), 692.]

• Matthew 2:5-6 quotes from Micah 5:2 and in NA28, the quotation is introduced with: “as it is written through the prophet.” Codex Sinaiticus has a marginal notation incorrectly identifying this quotation as from Isaiah. The Old Latin Codex Vercellensis (VL 3), dated to the second half of the fourth century, also incorrectly identifies the quotation from Micah 5:2 by adding “Isaiah” into the main text. 

• Matthew 2:15 quotes from Hosea 11:1 and in NA28, the quotation is introduced with: “so that the word from the Lord through the prophet might be fulfilled.” In Codex Sinaticus, there is a marginal notation which reads: εν αριθμοις (“in Numbers”), referring to the Old Testament book of Numbers. Matthew’s quotation matches the Hebrew Masoretic text of Hosea 11:1 and differs from the Septuagint (Matt 2:15 reads ἐκάλεσα τὸν υἱόν μου, while LXX Göttingen reads μετεκάλεσα τὰ τέκνα αὐτοῦ). The scribe of Codex Sinaiticus is not entirely wrong; NA28 notes that Matt 2:15 could also be an allusion to Numbers 23:22 (“God brings them out of Egypt”) or Numbers 24:8 (“God brings him out of Egypt”). [There are many other marginal notes in Codex Sinaiticus which mark OT quotations: Matt 1:23 quoting Isaiah 7:14 has the marginal note ησαιου (“from Isaiah”); Matt 2:6 quoting Micah 5:1 incorrectly has ησαιου (“from Isaiah”); Matt 2:15 quoting Hos 11:1 has εν αριθμοις (“in Numbers”); Rom 3:4 quoting Ps 50 has εν ψαλμω Ν (“in Psalm 50”); Rom 3:10-18 quoting Psalms 13 and 52 has ψαλμω IΓ κ(αι) NB (“in Psalm 13 and 52); Rom 4:3 quoting Gen 15:6 has γενεσεως (“from Genesis”); Rom 4:7-8 quoting Ps 31 has ψαλμω ΛΑ (“in Psalm 31”); Rom 4:17 quoting Gen 17:5 has γενεσεως (“from Genesis”); Rom 4:18 quoting Gen 15:5 γενεσεως (“from Genesis”); Acts 2:34-35 quoting Ps 109 has ψαλμω ΡΘ (“in Psalm 109”); Acts 3:22 quoting Deut 18 has δευτ (“(from) Deuteronomy”); Acts 3:25 quoting Gen 22 incorrectly has δευτερ (“(from) Deuteronomy”); Acts 4:25-26 quoting Ps 2 has ψαλμ B (“(in) Psalm 2”); Acts 7:49-50 quoting Isaiah 66 has ησαιας (“Isaiah”); Acts 13:33 quoting Ps 2 has ψαλμω B (“in Psalm 2”); Acts 13:35 quoting Ps 15 has ψαλμω (“in a psalm”); Acts 13:41 quoting Hab 1:5 incorrectly has ιωηλ (“Joel”); Acts 15:16-18 quoting Amos 9 has αμως (“(from) Amos”); 1 Pet 2:22 quoting Isa 53 has ησαιας (“Isaiah”); and 1 Pet 3:10 quoting Ps 33 has ψαλ ΛΓ (“(in) Psalm 33”). The scribe errs four times in Matt 2:5; 2:15; Acts 3:25; 13:41 (or three times if we allow Matt 2:15 to be an allusion to Numbers). For further discussion of these, see Ulrich Schmid, “Diplés und Quellenangaben im Codex Sinaiticus,” in Von der Septuaginta zum Neuen Testament: Textgeschichtliche Erörterungen, ed. Martin Karrer, Siegfried Kreuzer, and Marcus Sigismund, ANTF 43 (de Gruyter, 2010), 85-87.]

• Matthew 13:35 quotes from Psalm 78; in NA28, the quotation is introduced with: “so that the word through the prophet might be fulfilled.” However, the original scribe of Sinaiticus, 038, families 1 and 13, minuscule 33, and manuscripts known to Jerome add “Isaiah” (see discussion of this variant in question #2 above). According to Snapp, once again, scribes “make the text more specific, identifying the prophet incorrectly as Isaiah.” [Snapp, New Testament Textual Analysis, 192. Bold text original.]

• Snapp also says: “Isaiah’s name was probably added in Matthew 13:35 in some early copies by Western copyists who liked their quotations to be specific.” [Snapp, New Testament Textual Analysis, 324.]

• Matthew 21:4-5 quotes from Zechariah 9:9; in NA28, the quotation is introduced with: “so that the word through the prophet might be fulfilled.” Bruce Metzger identifies a marginal note in 021, minuscule 42, several versional witnesses (Old Latin a, c, h; some Boharic manuscripts), and Hilary as adding “Zechariah.” Metzger also identifies versional witnesses (four Vulgate manuscripts, some Boharic manuscripts, and the Ethiopic) as incorrectly adding “Isaiah.” [Bruce M. Metzger, A Textual Commentary on the Greek New Testament, 2nd ed. (Deutsche Bibelgesellschaft, 1994), 44.]

• Mark 1:2-3 quotes from Malachi and Isaiah; the Byzantine text begins Mark 1:2 with: “as it is written in the prophets,” whereas NA28 begins Mark 1:2 with: “as it is written in Isaiah the prophet.” The difference is: “the prophets” vs. “Isaiah the prophet.” According to Snapp, “the reading ‘in Isaiah the prophet’ was created by an early copyist to make the statement more specific, naming the more prominent prophet who was being quoted.” [Snapp, New Testament Textual Analysis, 191.]

From these examples, Snapp extrapolates a general tendency: “in the Gospels, when a prophet is quoted but not specifically named, it is not rare to see that his name is provided.” [Snapp, New Testament Textual Analysis, 191.] While all this information is interesting and revealing of scribal habits, I cannot agree with Snapp’s rather broad observation and alleged scribal habit. In four out of the six examples, the addition of a specific name is a singular reading in the Greek tradition, only has versional support, or is only found in the margin and not the main text:

• The main text of Codex Bezae, several versions, and Irenaeus (in part) in Matt 1:22-23. 

• The margin of Codex Sinaiticus and the main text of Codex Vercellensis in Matt 2:5-6. 

• The margin of Codex Sinaiticus in Matt 2:15. 

• The margin of 021, the text of 42, several versional witnesses, and Hilary in Matt 21:4-5. 

Only in Matthew 13:35 and Mark 1:2 does the more specific reading have substantial support. In other words, four of the six examples are eccentric outliers. Thus, from two examples and four eccentric examples, I do not think we can extrapolate a broad scribal tendency to take unspecified quotations and make them more specific by providing the name of a prophet (sometimes even an incorrect name). If anything, this tendency to supply specific names is seen more in the versions than in Greek manuscripts.

But these observations are drawing attention to a phenomena that does deserve more attention. There are probably more than just these six examples, although additional examples are likely to have little Greek manuscript support (or only versional support) since they are not already known to us at this point.

Q: In Matthew 13:35, 01 includes the name “Isaiah” even though Psalm 78 is the material being cited.  Some influential textual critics of past generations have supported this reading as original, but others consider it so difficult as to be nonsensical.  If “Isaiah” falls off the nonsense cliff here, so to speak, why not reject “in Isaiah the prophet” in Mark 1:2?

Hsieh: What happened with Codex Sinaiticus at Matthew 13:35 is interesting and revealing of scribal habits. But first we need to look at the full data: yes, the original scribe of 01 reads “Isaiah,” despite the quotation that follows coming from Psalm 78. But the reading is also corrected by deleting “Isaiah”: NA28 and UBS6 have the deletion as coming from Sinaiticus’s first corrector, while codexsinaiticus.org has “S1”: a correction during the original production of the manuscript, either while copying (in scribendo), or after copying. A look at the manuscript reveals the use of cancellation dots and small diagonal strike-through lines:

Codex Sinaiticus at the deletion in Matthew 13:35 (fol. 207r)

This style of deletion matches the earliest corrector(s) in having only one dot per letter. [Dirk Jongkind, Scribal Habits of Codex Sinaiticus (Gorgias, 2007), 252; Peter Malik, “The Earliest Corrections in Codex Sinaiticus: A Test Case from the Gospel of Mark,” The Bulletin of the American Society of Papyrologists 50 (2013): 229.]Thus, in deciding on the original text of Matthew 13:35, Sinaiticus is a witness to both the addition of “Isaiah,” as well as its omission – both during the original production of the manuscript. But the deletion of “Isaiah” is the final reading provided by Sinaiticus during its original production, so the mistake was recognized and corrected. 

According to NA28, the reading adding “Isaiah” is also attested by families 1 and 13, as well as the “queen of the cursives” (minuscule 33); S. C. E. Legg’s edition also has 038, minuscules 174, 230, 543, 713, 788, 826, 828, 983, 1573, 1689, one Vulgate manuscript, and some Ethiopic codices in support of the “Isaiah” reading. There are some good witnesses in support of the “Isaiah” reading (families 1 and 13 and minuscule 33), but not enough to overturn the overwhelming evidence for the omission of “Isaiah” (01 first corrector, 03, 04, 05, 032, 579, 700, 892, 1241, and the Byzantine tradition). [Among minuscule manuscripts, some of the “best” in the Gospels include 33, 69, 579, 700, 892, and 1241. See Gregory R. Lanier, “Dating Myths, Part Two: How Later Manuscripts Can Be Better Manuscripts,” in Myths and Mistakes in New Testament Textual Criticism, ed. Elijah Hixson and Peter J. Gurry (IVP Academic, 2019), 130. Minuscule 69 is not extant for Matthew 13:35; minuscule 33 supports the “Isaiah” reading, while minuscules 579, 700, 892, and 1241 support the omission of “Isaiah.”]

In addition, we can explain on transcriptional grounds how the reading adding “Isaiah” arose in Matthew 13:35. The text being copied is ὅπως πληρωθῇ τὸ ῥηθὲν διὰ τοῦ προφήτου (“so that the word of the prophet might be fulfilled”). The mention of “the prophet” alongside “fulfill” could easily trigger “Isaiah” in the mind of a scribe, especially since similar phrasing is used elsewhere in Matthew:

• 1:22 fulfilling Isaiah 7:14 (τὸ ῥηθὲν ὑπὸ τοῦ κυρίου διὰ τοῦ προφήτου = “the word from the Lord through the prophet”). 05 and a few others add “Isaiah.”

• 2:15 fulfilling Hosea 11:1 (τὸ ῥηθὲν ὑπὸ τοῦ κυρίου διὰ τοῦ προφήτου = “the word from the Lord through the prophet”)

• 2:17 fulfilling Jeremiah 31:15 (τὸ ῥηθὲν διὰ Ἱερεμίου τοῦ προφήτου = “the word through the prophet Jeremiah”)

• 2:23 fulfilling something in “the prophets” (τὸ ῥηθὲν διὰ τῶν προφητῶν = “the word through the prophets”)

• 4:14 fulfilling Isaiah 9:1-2 (τὸ ῥηθὲν διὰ Ἡσαΐου τοῦ προφήτου = “the word through Isaiah the prophet”)

• 8:17 fulfilling Isaiah 53:4 (τὸ ῥηθὲν διὰ Ἡσαΐου τοῦ προφήτου = “the word through Isaiah the prophet”)

• 12:17 fulfilling Isaiah 42:1-3 (τὸ ῥηθὲν διὰ Ἡσαΐου τοῦ προφήτου = “the word through Isaiah the prophet”)

• 21:4-5 fulfilling Zechariah 9:9 (τὸ ῥηθὲν διὰ τοῦ προφήτου = “the word through the prophet”)

• 27:9 fulfilling Jeremiah – although the quotation is actually from Zechariah 11:13 (τὸ ῥηθὲν διὰ Ἱερεμίου τοῦ προφήτου = = “the word through Jeremiah the prophet”)

• Different but still relevant is Matthew 2:5-6, where a quotation from Micah 2:5 is introduced with γέγραπται διὰ τοῦ προφήτου (“as it is written through the prophet”).

Alongside these examples is our text under consideration, Matthew 13:35, which reads πληρωθῇ τὸ ῥηθὲν διὰ τοῦ προφήτου (“so that the word through the prophet might be fulfilled”). It is completely understandable how a scribe may have accidentally and unconsciously added “Isaiah,” given that the three preceding occurrences of πληρωθῇ τὸ ῥηθὲν διὰ τοῦ προφήτου have the exact same wording as Matt 13:35 and Isaiah was explicitly identified (Matt 4:14; 8:17; 12:17). These three preceding passages all explicitly mention Isaiah and are without textual variation (according to Swanson’s volume on Matthew). 

Matthew 1:22-23 could also be mentioned since Isaiah 7:14 is fulfilled, although the text of Matthew does not explicitly name “Isaiah” except in 05 and a few other witnesses. Perhaps it is a testament to the quality of Sinaiticus’s original scribe(s) that he alone corrected his mistaken addition of “Isaiah” in Matthew 13:35 (whereas families 1 and 13, and minuscule 33 have no corrections).

We can add the testimony of Eusebius and Jerome. [For Eusebius’s and Jerome’s text-critical comments on Matthew 13:35, see Amy M. Donaldson, “Explicit References to New Testament Variant Readings Among Greek and Latin Church Fathers” (PhD diss, University of Notre Dame, 2009), vol. 2, 368-72. Available here: https://curate.nd.edu/articles/thesis/Explicit_References_to_New_Testament_Variant_Readings_among_Greek_and_Latin_Church_Fathers/24732099] Eusebius, in his commentary on Psalm 77 says that “in the accurate copies” (τοῖς ἀκριβέσιν ἀντιγράφοις), Isaiah is omitted. Jerome claims to have found the reading “Asaph” (the author in the subscription of Psalm 78) “in all the ancient copies, but people in their ignorance changed it.” Jerome conjectures that scribes didn’t know who Asaph was, so they changed the reading to the more familiar name Isaiah. According to NA28, the Asaph reading is only attested in manuscripts known to Jerome and is thus found in no extant manuscripts; the only evidence we have is that Jerome claimed to have found the Asaph reading “in all the ancient copies.” But the Asaph reading clearly seems like an improvement to the text. However, we see Jerome attempting to deal with the problem based on transcriptional probability (a scribe being unfamiliar with Asaph changed the text to Isaiah) rather than tossing out the “Isaiah” reading as nonsensical.

Now regarding “Isaiah” in Mark 1:2, the situation in Matthew 13:35 is not comparable. They are different in the three ways:

(1) In Mark 1:2, the “mistaken” reading in NA28 has γέγραπται ἐν τῷ Ἡσαΐᾳ τῷ προφήτῃ (“as it is written in Isaiah the prophet”), which feels like an “error” because Mark first quotes from Malachi 3:1, then from Isaiah 40:3. Thus the NA28 reading with “Isaiah” in Mark 1:2 is partially true to the context. The Byzantine reading γέγραπται ἐν τοῖς προφήταις (“as it is written in the prophets”) removes the “error” of NA28’s reading. In contrast, Matthew 13:35 only quotes from Psalms and has no quotation from Isaiah. Thus the “Isaiah” reading in Matthew 13:35 is not even partially true but completely wrong. In summary: while the readings with “Isaiah” in Mark 1:2 and Matthew 13:35 are both “harder” readings because they appear to be a mistake, the context of Mark 1:2 at least has a quotation from Isaiah, whereas Matthew 13:35 does not.

(2) The two readings in Mark 1:2 are different syntactically, so that an accidental change is unlikely (plural: ἐν τοῖς προφήταις, “in the prophets” vs. singular: ἐν τῷ Ἡσαΐᾳ τῷ προφήτῃ, “in Isaiah the prophet”). Thus, Mark 1:2’s variants almost certainly must have been an intentional change (in either direction). In contrast, the two readings in Matthew 13:35 both contain the singular “prophet” and the difference is the addition or omission of one word; the longer reading adding “Isaiah” could have been an unintentional addition, influenced by and harmonized towards earlier passages, which contain the same exact phrasing (πληρωθῇ τὸ ῥηθὲν διὰ τοῦ προφήτου) and add “Isaiah.”

(3) The better witnesses attesting to the “Isaiah” reading in Mark 1:2 and Matt 13:35 are different: In Matthew 13:35, the “more specific reading” (including “Isaiah”) does not have the support of 01 (since 01 first corrector removes “Isaiah”), 03, and several “good” minuscules (700, 892, 1241). Whereas these five manuscripts switch sides to the more specific reading including “Isaiah” in Mark 1:2. The “better” manuscripts which stay consistent in supporting the more specific reading in both Matt 13:35 and Mark 1:2 are family 1 and minuscule 33. 

Perhaps this reveals a “cult of best manuscripts” in favoring 01 and 03, but other evidence (mentioned above) factors into the decision: on transcriptional probability, the accidental addition of “Isaiah” in Matthew 13:35 is understandable given how often Isaiah occurs in Matthew’s fulfillment formulas. In contrast, accidental change in Mark 1:2 is unlikely, given that a change from the plural (ἐν τοῖς προφήταις) to the singular (ἐν τῷ Ἡσαΐᾳ τῷ προφήτῃ) seems more deliberate (in either direction). And the literary contexts also differ substantially: Isaiah is quoted in the near context of Mark 1:3, so that an initial “apparent” difficulty with the “Isaiah” reading in Mark 1:2 is not really problematic on further reflection. In contrast, Isaiah is not quoted at all in the near context of Matthew 13:35 (only in the far context of Matthew 13:14-15 is Isaiah mentioned); “Isaiah” is a truly difficult reading in Matthew 13:35, but for reasons of external evidence (lacking the support of 01, 03, 700, 892, 1241) and likely accidental error in adding “Isaiah,” the more difficult “Isaiah” reading should be rejected in Matthew 13:35.

Q: The relatively new Evangelical Heritage Version (2017) has an interesting New Testament base-text, incorporating significantly more Byzantine readings into the text that the ESV, NIV, NLT, etc.  Is this going to start a trend?

 

Hsieh: Bible translations (like Greek critical editions) are highly monopolistic, so that a few translations and editions dominate the market. While it is interesting that a new translation adopts more Byzantine readings, I find it hard to imagine more mainstream translations doing the same, unless they follow the new trends in the Tyndale House Greek New Testament (THGNT) and Editio Critica Maior (ECM), both of which adopt more Byzantine readings. [See my forthcoming chapter, “Two Ships Sailing Toward Byzantium,” in the 2026 conference proceedings of the Center for the Study of New Testament Manuscripts (CSNTM) Text & Manuscript conference (forthcoming with Hendrickson Publishers).] And even if they consult these new editions, the translation committees may not agree with all the changes.

But even if revised versions of the ESV, NIV, etc. follow the new critical texts, changes happen slowly as committees meet annually (if even that often), so it could be many, many years before these new Byzantine readings show up in translations. But the new trend will probably eventually make its way into mainstream translations. It will take time and the ECM will not be complete until the mid-2030s, with Nestle-Aland 30 probably published in the late-2030s. [This estimate is based on Martin Karrer’s editing of ECM Hebrews, which has a ten-year grant from the German government, running from 2025 to 2035.]

Q: Will the Nestle-Aland compilation ever adopt the reading of 01 and 03 at Matthew 27:49 and thus put the last nail in the coffin of the doctrine of Biblical inerrancy?

Hsieh: This famous “Western non-interpolation” tortured Westcott & Hort into printing the longer reading in their main text in double brackets rather than relegating it into the margin, despite the longer reading being a clear harmonization to John 19:34. The longer reading is found in the reasoned eclectic’s “favorite” manuscripts: Codices Sinaiticus and Vaticanus (also 04, 019, 036, 1010, some Vulgate manuscripts, Christian Palestinian Aramaic in part, one Ethiopic manuscript, Middle Egyptian Coptic, and 030 with some modifications to match John 19:34 more closely). Sinaiticus and Vaticanus supporting the harmonization should not be alarming: all manuscripts have errors and even Sinaiticus and Vaticanus err with harmonizations. Willem Wisselink has shown that Codex Vaticanus harmonized in 31 percent of the 1,489 variations he investigated; this was the lowest number among manuscripts he surveyed (P45, P75, 01*, 01 corrector, 02, 05, 032, 038, 045, 33), but it was definitely not zero harmonizations! [

 Willem Wisselink, Assimilation as a Criterion for the Establishment of the Text (Uitgeversmaatschappij J. H. Kok, 1989), 78.]

An intentional change in Matthew 27:49 could go either way: (1) a scribe could have removed the longer reading because of the problem of chronology and a contradiction between Matthew and John. [Brooke Foss Westcott and John Anthony Hort, “Appendix I: Notes on Selected Readings,” in The New Testament in the Original Greek, Volume 1: Introduction and Appendix (Macmillan, 1881), 22; Philip Wesley Comfort, A Commentary on the Manuscripts and Text of the New Testament (Kregel Academic, 2015), 176.] John has the spear piercing Jesus’s side after his death, whereas the longer reading in Matthew 27:49 would have the spear piercing Jesus’s side before his death. Or, (2) Bart Ehrman suggests the longer reading was added as an anti-docetic “orthodox corruption,” proving that Jesus had real flesh and blood. [Bart D. Ehrman, The Orthodox Corruption of Scripture: The Effect of Early Christological Controversies on the Text of the New Testament, 2nd ed. (Oxford University Press, 2011), 228-29.]

Another suggestion from Ulrich Schmid is that the longer reading was an early reader’s note in the margin of manuscript, which eventually made its way into the main text of a manuscript. [Ulrich Schmid, “Conceptualizing ‘Scribal’ Performances: Reader’s Notes,” in The Textual History of the Greek New Testament, ed. Klaus Wachtel and Michael Holmes (Society of Biblical Literature, 2011), 60-61.] Schmid’s argument is based partly on how the version of the spear incident in Matthew 27:49 is not an exact replica of John 19:34 – the beginning is substantially different (ἄλλος δὲ λαβὼν λόγκην in the long reading of Matthew 27:49 vs. ἄλλ᾽ εἷς τῶν στρατιωτῶν λόγχῃ in John) and in the long reading of Matthew 27:49, the word ἔνυξεν is placed earlier, ἐυθύς is omitted, and the word order for blood and water is reversed (030 “corrects” the last two of these differences).

Matthew 27:49 would not be the only place in Matthew where material from John’s Gospel has intruded into Matthew:

• At Matt 27:35, a good number of manuscripts (037, 038, families 1 and 13, 1243, 1424, Harklean Syriac) have: ινα πληρωθη το ρηθεν δια του προφητου διεμερισαντο τα ιματια μου εαυτοις και επι τον ιματισμον μου εβαλον κληρον (“so that the word through the prophet might be fulfilled: ‘They divided my garments among themselves and cast lots for my clothing’”). This comes from John 19:24.

• At Matt 28:18, a few manuscripts (038 and Syriac Peshitta; 1604 with minor variation) have: καθως απεστηλεν (απεσταλκεν 1604) με ο πατηρ· καγω αποστελω (αποστελλω 1604) υμας (“just as the father sent me, so also I send you”). This comes from John 20:21.

Matthew 27:49 would be the only instance where 01, 03, and 04 have a Johannine intrusion, but none of these harmonizations should be accepted as original. Even early manuscripts have harmonizations (as Wisselink demonstrated).

Therefore, I find it unlikely that Nestle-Aland will adopt the longer reading in Matthew 27:49, especially given the small number of witnesses in favor of the longer reading and the aversion to harmonizations by the Nestle-Aland editors. We await ECM Matthew and its results, but the Coherence Based Genealogical Method (CBGM) will probably find the shorter reading to have better coherence given its wide attestation.

Biblical inerrancy rests on a stronger foundation than one textual variant. Among the large amount of literature on inerrancy, I find Paul Feinberg’s chapter to be carefully argued. [Paul D. Feinberg, “The Meaning of Inerrancy,” in Inerrancy, ed. Norman L. Geisler (Zondervan, 1980), 265–304, 468–72.]

] And there is not even a strong text-critical argument to accept the longer reading of Matthew 27:49, which then introduces a potential contradiction. The support of Codex Sinaiticus and Codex Vaticanus for the longer harmonizing reading is out of the ordinary, but only shows that all manuscripts are flawed. 

**********

Thanks, Nelson, for sharing your thoughts with us!