The Invisible Variant in Luke 2:15

           Although promoters of an assortment of Bible versions frequently insist that footnotes give their readers plentiful information about textual variants, many variants, especially those which scarcely affect the meaning of the text, are not mentioned in any footnotes in any major English versions.  An example of this appears in Luke 2:15:  in the middle of Luke’s Christmas narrative, after a multitude of heavenly hosts finishes praising God (regarding the famous variant in Luke 2:14, see this post), and as the shepherds decide to investigate the town of Bethlehem, immediately following οἱ ἄγγελοι, the Byzantine text has the words καὶ οἱ ἄνθρωποι (“and the men”) before οἱ ποιμένες (“the shepherds”).

          This is an invisible variant:  the KJV’s base-text (the Textus Receptus) includes καὶ οἱ ἄνθρωποι, and the ESV’s and NIV’s primary base-text (the Nestle-Aland compilation) does not; the WEB’s base-text is Byzantine; the NRSV’s base-text is (almost always) Alexandrian.  Yet this phrase – “the shepherds said to one another” – is identical in English in all four of these English versions (and in the EOB-NT, CSB, EHV, and NET).  One would never realize from the renderings in almost all English versions that one base-text has three more words.  Wayne A. Mitchell’s New Heart English Bible, which points out the variant in a footnote, is an exception, as are some editions of the KJV with a similar marginal footnote.

                This variant has become invisible in more ways than one.  It was in Tregelles’ 1860 Greek New Testament (albeit within single-brackets), but although the Tyndale House GNT (2017 edition), is, its editors say (in the Preface, p. vii) “based on a thorough revision of the great nineteenth-century edition of Samuel Prideaux Tregelles”), not only is καὶ οἱ ἄνθρωποι not included in the text, but there is no apparatus-entry to inform readers of the existence of this variant.  Likewise, although this variant was initially included in the apparatus of the UBS GNT, in the fourth and fifth editions it has mysteriously vanished without a trace. 

          Which reading is original?  Metzger acknowledged, in his Textual Commentary on the Greek New Testament (1971, corrected edition 1975), that “The fact that the longer reading is characteristically Lukan in style argues strongly in its favor.”  Nevertheless the longer reading was rejected by a majority of the UBS editorial committee which “preferred to make a decision on the basis of preponderance of external evidence.” 

          This statement from Metzger is not easy to defend when one looks at what the preponderance of external evidence is.  À B L W Q (which moves οἱ ἄγγελοι to precede εἰς τὸν οὐρανὸν) X f1 565 (which also omits εις τον αγγελοι earlier in the verse) 700 and 1071 appear to be almost the only manuscripts that support the non-inclusion of καὶ οἱ ἄνθρωποι.  The Vulgate, the Sahidic version, and the Peshitta support the shorter reading, but versional evidence is inherently tenuous in this particular case because more than one translator could independently decide that the sense of the passage could be sufficiently rendered without translating the appositive phrase here, as can be seen from various English versions.   The Harklean Syrian (made in the early 600s) appears to support ἄνθρωποι καὶ οἱ ποιμένες.  Καὶ οἱ ἄνθρωποι is supported by Codex A, Codex D, Γ, K (017, Cyprius), M, Δ, P (024), S (028), U, Y, and Ψ.   Codex M ends a page exactly after οἱ ἄγγελοι, and καὶ οἱ ἄνοι begins the first line on the next page.  Minuscules 33 157 892 1010 and 1424 are among the many minuscules that support the longer reading, as does the Gothic version.  Codex D’s word-order is apparently unique in this verse; D begins verse 15 with Και εγενετο │ως απηλθον οι αγγελοι απ αυτων │εις τον ουρανον και οι ανθρωποι.  Small gaps appear in D before Και εγενετο and before και οι ανθρωποι.  Codices C, N, and T are lacunose here. 

Codex Delta supports
the longer reading.

        At the outset of the third chapter of Causes of the Corruption of the Traditional Text of the Holy Gospels, John Burgon briefly commented on this variant, pointing out that the scribe of À omitted the οἱ before ἄγγελοι, “whereby nonsense is made of the passage (viz. οἱ ἄγγελοι ποιμένες).”  Burgon considered the shorter reading to have originated due to homoeoteleuton elicited by the six clustered-together recurrences of οἱ in this verse.  More recently, Michael Holmes seems to have agreed, inasmuch as καὶ οἱ ἄνθρωποι is included in the text of the SBL-GNT.  I consider Burgon’s observations completely valid, and Metzger’s appeal to the “preponderance of external evidence” is basically code for “the pro-Alexandrian bias of the editors.”  Καὶ οἱ ἄνθρωποι should be included in future compilations.

 

Matthew 12:47 and Homoioteleuton

Matthew 12:45b-50 in the
ESV Reader's Gospels.
Where's verse 47?

           In the English Standard Version, in the passage about the visit of Jesus’ mother and brothers in Matthew 12:46-50, there is a strange feature:  there is no verse 47; it is in the footnotes rather than in the text.  The ESV’s footnote says, “Some manuscripts insert verse 47:  Someone told him, “Your mother and your brothers are standing outside, asking to speak to you” – except in the ESV Reader’s Gospels, which has neither verse-numbers nor footnotes; its text goes directly from the end of verse 46 to the beginning of verse 47.
            In the Christian Standard Bible, meanwhile, Matthew 12:47 is included in the text, and a footnote says, “Other mss omit this verse.”  Before examining the text-critical reasons for the disagreement among these English versions, let’s first consider how inconsistently the CSB describes the manuscripts in its footnotes and headings.  For where its text retains Matthew 12:47, the CSB describes the manuscripts that disagree as merely “Other,” but following Mark 16:8, its editors have added a line (as if to tell the readers where to aim their scissors), and interrupted the text with a heading, “[Some of the earliest mss conclude with 16:8]” – but both notes refer to essentially the same small cluster of manuscripts.

            What manuscripts omit Matthew 12:47?  Primarily Codex Vaticanus, Codex Sinaiticus, Codex Regius (L, 019), Codex Tischendorfianus IV (036, from the 900s), and minuscule 579, allied with the Sahidic version, the Sinaitic Syriac and the Old Latin Codex Bobbiensis.  A few other copies lack Matthew 12:47, but if those four had contained it, their testimony would be considered trivia.

            And of the over 1,600 Greek manuscripts of Mark, two uncials omit Mark 16:9-20:  Codex Vaticanus and Codex Sinaiticus (and both show signs of their copyists’ awareness of the absent verses, as I have shown already, here and here).  Codex L echoes a later stage of a text used in Egypt; it has the “Shorter Ending” after Mark 16:8, and then verses 9-20), and minuscule 579 also has the “Shorter Ending” between Mark 16:8 and 9 (though, unlike Codex L, without short notes to introduce each ending).  And in the versions, the text of the Sinaitic Syriac stops at 16:8, and the copyist of Old Latin Codex Bobbiensis, after mauling verse 8, included the Shorter Ending, poorly transcribed. 

            The same cluster of witnesses is in view in both cases – but the unsuspecting reader of the Christian Standard Version meets them masked as “Other mss” where their testimony is rejected, and as “Some of the earliest mss” where the passage is bracketed. 

            Now let’s turn back to Matthew 12:47, and see why it is missing not only in the ESV but also in that cluster of manuscripts in which the best Greek uncials and worst Syriac and Latin copies appear to keep close company.  What we have here is a simple case of homoioteleuton (also spelled homoeoteleuton).  Verse 46 ends with the Greek words ζητοῦντες αὐτῷ λαλῆσαι (“seeking to him to speak”), and verse 47 ends with the words ζητοῦντες σοι λαλῆσαι (“seeking to you to speak”).  An early copyist – working at a point in the transmission-stream early enough to be echoed by the Alexandrian branch represented by ﬡ, B, and L, on the one hand, and by the Western branch represented by the Old Syriac and Codex Bobbiensis, on the other – accidentally skipped verse 47 when his line of sight drifted down from the λαλῆσαι at the end of verse 46 to the λαλῆσαι at the end of verse 47.

            It is not hard to see how this happened.  Meanwhile, consider what the ESV’s editors must believe about the transmission of this passage.  If the ESV’s non-inclusion is correct, then the copyists of virtually all other Greek manuscripts – C D W Z Δ Θ 28 33 157 892 (which adds προς αὐτον; see Willker’s comments for details), 1424, the family-1 group, the family-13 group, and well over 1,500 minuscules, and hundreds of lectionaries – and over 10 Old Latin copies (including Codex Vercellensis, from the 370’s), the Vulgate, the Peshitta, the Armenian, Georgian, and Ethiopic versions, and copies cited by patristic writers including Jerome and Augustine (and Augustine cited the verse in two different forms, echoing two different Old Latin transmission-lines), were all using the wrong exemplars at this point.

Another plain case of
homoioteleuton in Codex B.

            I hope my readers will understand by this last paragraph that I am not arguing for the majority reading merely because it is the majority.  Today we know of one early Middle Egyptian manuscript – Mae-2, that is, Codex Schøyen 2650, from the early 300s – which agrees with ﬡ and B in non-inclusion of Matthew 12:47.  But if we had 50 copies in the same transmission-line, the argument would not vary, just as the discovery of 50 more copies of the Vulgate would not make much of an impact.  It is not a matter of number, but of the relative plausibility of the competing models of the text’s transmission-history. 
            The evidence demands that the scribal error that caused the loss of this verse happened very early – early enough to echo in a limited part of the Alexandrian transmission-line, and in a limited part of the Old Latin and Syriac transmission-lines.  But early parablepsis is parablepsis nonetheless.  Matthew 12:47 should be included in the text, and if there is a footnote, let it tell the reader why the verse was lost in the transmission-line of ﬡ and B, instead of just giving enough information to perpetuate confusion.

_______________

The English Standard Version is © 2001 by Crossway, a publishing ministry of Good News Publishers.  Used by permission.  All rights reserved.

The Christian Standard Bible® is Copyright © 2017 by Holman Bible Publishers.  Used by permission.  Christian Standard Bible® and CSB® are federally registered trademarks of Holman Bible Publishers. 

The Holy Bible, English Standard Version® is Copyright © 2001 by Crossway, a publishing ministry of Good News Publishers.  ESV Text edition:  2011.  Used by permission.  All rights reserved.  

Mark 2:16 and Homoioteleuton

            At the end of Mark 2:16, there is a textual contest:  do the scribes and Pharisees ask why Jesus eats with tax collectors and sinners, or do they ask why Jesus eats and drinks with tax collectors and sinners?

            The Greek manuscript support for the non-inclusion of “and drinks” is rather thin:  B, D, and W (that is, Codex Vaticanus, Codex Bezae, and Codex Washingtonianus).  They are joined by six Old Latin copies.  That, at least, is all the support listed in the textual apparatus of the fourth edition of the UBS Greek New Testament, once one removes the forgery 2427 (“Archaic Mark”) from the picture.  Metzger laid the spin on pretty thick in his Textual Commentary, claiming that the shorter reading was thus “strongly supported.”  This is quite an exaggeration, inasmuch as there is only one non-Western witness in the bunch.

            Meanwhile, the external support for the inclusion of “and drinks” is massive:  it includes not only the hundreds of manuscripts stacked up behind the apparatus’ reference to the Byzantine Text, but also Papyrus 88 (from the 300’s), Codex A, family 1, 33, 157, 892, three Old Latin copies, the Peshitta, the Gothic version, and one Sahidic manuscript. The Armenian and Georgian versions also include a reference to drinking.  This array of witnesses is widespread, both geographically and in terms of textual groups.      

Codex W has Western
affinities in this part of Mark.

            But what about the text-critical axiom lectio brevior potior – the shorter reading is to be favored?  That is simply not a well-grounded premise, and the damage that it has done needs to be undone.  As Juan Hernández Jr. has acknowledged, “The pioneering studies of Colwell and Royse on the papyri demonstrated that the general tendency during the earliest period of textual transmission was to omit.”  The Alexandrian text should thus be compared to a ship which reaches its port lighter than when it embarked because the weight of the barnacles that attached themselves to the hull during the voyage was less than the weight of things that the crewmen dropped overboard. 

            It may seem reasonable to suppose that copyists operated on the principle of “When in doubt, don’t leave it out,” but that idea is not observationally grounded.  Furthermore, even when Griesbach advocated a preference for the shorter reading over 200 years ago, he included a qualification which is relevant to the contest in Mark 2:16:  we should prefer the fuller reading before the shorter (unless the latter is seen in many notable witnesses) if homoioteleuton [that is, the occurrence of words or phrases with similar endings, in close proximity to one another] might have provided an opportunity for an omission.

            Here is the Byzantine text at the end of Mark 2:16 and the beginning of 2:17:  . . . Τί ὅτι μετὰ τῶν τελωνῶν καὶ ἁμαρτωλῶν ἐσθίει καὶ πίνει; Καὶ ἀκούσας ὁ Ἰησους λέγει αὐτοῖς . . . . 

            Notice the homoioteleuton:  the letters ει καὶ occur near the end of verse 16 and again at the beginning of verse 17.  No theory of a semi-harmonization to the parallel-passage in Luke is needed to account for the difference between the Byzantine reading and the reading in Codex Vaticanus; the simple explanation is that a copyist in the transmission-line of B’s text accidentally skipped from the first occurrence of -ει καὶ to the second occurrence of -ει καὶ.  Thus, in this case, the shorter reading is explained by the longer reading. 

            Harmonization did occur in this verse in some manuscripts.  In Codex Sinaiticus, the end of Mark 2:16 is conformed to Matthew 9:11, so as to rephrase the question not as “Why does He eat with tax collectors and sinners” but instead as “Why does your Teacher eat with tax collectors and sinners.”  Thus we observe not just one, but two mechanisms by which the reference to drinking could be lost:  accidental parablepsis, and deliberate harmonization to the parallel in Matthew.

            In over 70 manuscripts (including Codices C, L, and Δ) the reference to “your Teacher” has been inserted from Matthew – but the reference to drinking has not been removed; nor was the Greek text adjusted to match the words in Luke 5:30.  Either the copyists who created this reading were simultaneously harmonizing and unharmonizing – making the text resemble Matthew 9:11, and then making it different – or else the reference to drinking was there to begin with in their exemplars.  Their Alexandrian exemplars.

            What about the parallel in Luke 5:30?  If Byzantine copyists had wished to conform the words at the end of Mark 2:16 to the parallel in Luke 5:30, the natural way to do so would be to write ἐσθίετε καὶ πίνετε – for those are the words in Luke 5:30.  This was indeed done in a relatively small number of manuscripts (including, notably, Codex Σ and minuscules 565, 700, and 1241), but it is not the Byzantine reading.        

            In addition, if harmonization is to be suspected, then one may observe that it was not necessary for copyists to interrupt their copying and consult the parallel-passages in Matthew and Luke in order to have a basis on which to make a textual adjustment.  A basis for conformation-via-shortening is built into Mark 2:16, inasmuch as the beginning of the verse mentions Jesus eating, but not drinking. 

            In conclusion:  the reading with καὶ πίνει explains its rivals.  What we have here is a simple case of the loss of two words due to homoioteleuton.  This loss probably occurred in the Western Text, and then influenced the text of a single Alexandrian manuscript (Codex Vaticanus). 

  is
(a)  the reading that explains its rival in an uncomplicated way,
(b) attested in manuscripts representing different text-types,
(c) attested in multiple versions, and
(d) attested in a papyrus (P88, from the 300’s). 

           So:  the reading καὶ πίνει

            With this internal and external evidence pointing in favor of the inclusion of καὶ πίνει in Mark 2:16, it is tempting to suspect the survival of the shorter reading in Mark 2:16 in modern compilations (including the new Tyndale House edition of the Greek New Testament) is due to two factors:  an overestimation of Codex Vaticanus’ resistance to Western readings, and an unscientific reluctance to acknowledge Byzantine readings as original.    

            Finally, a couple of small details noted by Wieland Willker may add something to the case; he mentioned that in Codex K, in Luke 5:30, the words καὶ πίνετε are absent.  This illustrates the same mechanism which caused the loss of the words καὶ πίνει in Mark 2:16; the only difference is that the homoioteleuton in Luke 5:30-31 involves different letters (-ετε καὶ).  He also mentioned that seven medieval Byzantine manuscripts lack καὶ πίνει in Mark 2:16.  Here one of the canons of equitable eclecticism may be effectively applied:   If a variant occurs sporadically in witnesses greatly separated by age and textual character, this may indicate that the variant was liable to be spontaneously created by copyists, rather than that it was transmitted by distant transmission-streams.

____________

A Textual Commentary on the Greek New Testament 

© 1971 by the United Bible Societies, Stuttgart, Germany.

Mark 11:26 - How It Disappeared

            Homoioteleuton (also spelled as homeoteleuton) – a term which means “same ending” – was something that the copyists of New Testament manuscripts had to vigilantly guard against.  It was not unusual for the lines of the text in their master-copies to end with similar letters, and if a copyist didn’t pay attention, he might accidentally lose his place.  Usually if this happened, it would turn the sentence he was writing into nonsense, and a proof-reader would catch the mistake.  (When text is skipped in this way, the cause of the mistake is called parablepsis.)  But sometimes, when a parableptic error did not result in a nonsense-reading, words or phrases and even entire sentences could disappear without notice. 

            The copyist of Codex Vaticanus made a parableptic error in John 17:15; where the copyist should have written the Greek equivalent of, “I do not pray that You should take them out of the world, but that You should keep them from the evil one,” he skipped seven words, as his line of sight drifted from one occurrence of the word αυτους (“them”) to the recurrence of it later in the sentence, and thus wrote, “I do not pray that You should take them from the evil one.”  Fortunately someone caught the mistake later on.

            In Mark 11:26, something similar happened which caused the entire verse to disappear in the Alexandrian Text, and the same mistake occurs in some other manuscripts as well.  The Greek text of Mark 11:25 ends with τα παραπτώματα ὑμων (“your trespasses”) and 11:26 ends with exactly the same words.  A copyist whose line of sight drifted from the end of verse 25 to the end of verse 26 could skip the entire verse without realizing it, and, since the sentence would still make sense, the mistake could slip by a proofreader who was not very familiar with the text of Mark.               

Codex Bezae (D) has the passage,
with the last word of verse 25 missing.

            Relatively few manuscripts fail to include Mark 11:26, but two of the ones that lack the verse are Vaticanus and Sinaiticus, which were the primary sources of the base-text of the 1881 Revised Version, and which are still the main basis of the Nestle-Aland/UBS compilation.  That is why, in the NIV, NLT, and ESV, this verse is not in the text.  (The HCSB and NASB currently retain the verse in the text, within brackets.  The KJV, NKJV, MEV, and WEB have it in the text without brackets.)  
          However, Codices Alexandrinus (A) and Bezae (D), which are only slightly later, both include Mark 11:26 – and because Codex D has the passage with some minor variations and unusual spelling (and for other reasons), it is clear that these two manuscripts are a considerable distance apart from each other in terms of their textual relationship, while À and B are historically very close to one another.  (So much so that in the margins of the book of Acts, they share a unique form of chapter-numbers that almost certainly was transcribed from the same source.) 

Mark 11:24b-27a in the Peshitta
(Syriac).

          Other witnesses that attest to a fourth-century text also weigh in with support for the inclusion of Mark 11:26:  Codex Argenteus, made in the 500’s, is the main witness to the Gothic version which was translated by Wulfilas in the mid-300’s, at about the same time Codex Sinaiticus was being made.  Codex Argenteus includes the verse.  The Peshitta (Syriac) version, probably made in the late 300’s, includes the verse.  The Latin Vulgate, made in 383 by Jerome, who stated that he consulted ancient Greek manuscripts to ensure that the text was well-grounded, includes the verse.  Almost all Old Latin manuscripts, the descendants of Latin manuscripts made before the Vulgate dominated the Latin copying-tradition, include the verse as well.
          In addition, although Vaticanus and Sinaiticus are old, they are not as old as the manuscripts that were used by Cyprian.  Cyprian served as bishop of Carthage, in North Africa, from about 249 until he was martyred in 258 during the persecution that happened there during the reign of Valerian.  Thus his manuscripts of Mark were at least 75 years older than Vaticanus, and at least 100 years older than Sinaiticus (using the generally accepted production-dates for each, c. 325 for Vaticanus and c. 350 for Sinaiticus).  In Cyprian’s Treatise XII, Book 3, in chapter 22, Cyprian briefly undertakes the task of demonstrating from Scripture that when Christians are wronged, we must pardon and forgive.  He gets right to the point:

          

Cyprian of Carthage.

“In the Gospel, in the daily prayer:  'Forgive us our debts, even as we forgive our debtors.’  Also according to Mark:  ‘And when you stand for prayer, forgive, if you have ought against any one; that also your Father who is in heaven may forgive you your sins. But if you do not forgive, neither will your Father which is in heaven forgive you your sins.’  Also in the same place:  ‘In what measure you mete, in that shall it be measured to you again.’”

          When Cyprian refers to “the daily prayer,” he is of course referring to Matthew 6:12.  But if he had intended to quote Matthew 6:15 (which follows the gist of Mark 11:26), there would have been no need for him to then say, “Also according to Mark.”  Although Cyprian immediately jumps back to the Sermon on the Mount (i.e., to Matthew 7:2), there seems to be no reason for any conclusion other than that Cyprian was quoting from Mark 11:25-26 in this part of his composition.  

            Bruce Metzger, representing the editors of the UBS compilation in which Mark 11:26 is absent, claimed that “the words were inserted by copyists in imitation of Mt. 6.15.”  However, the wording in Matthew 6:15 is discernibly different in the UBS text:  Εαν δε μη αφητε τοις ανθρώπους, ουδε ὁ πατηρ υμων αφησει τα παραπτωματα υμων.  Compare this to Mark 11:26:  Ει δε υμεις ουκ αφίετε, ουδε ὁ πατηρ υμων ὁ εν τοις ουρανοις αφήσει τα παραπτωματα υμων.  It would take a special kind of copyist to introduce a harmonization that contained non-harmonious readings. Metzger’s rationale for the Alexandrian reading did not persuade J. K. Elliott, a British specialist in New Testament textual criticism, who maintained that in this particular case, the Textus Receptus (along with over 95% of the Greek manuscripts of Mark) has the original reading.  The absence of the verse in a smattering of later Byzantine manuscripts demonstrates its vulnerability to the kind of homoioteleuton-induced loss which could be overlooked by early copyists as easily as it could be overlooked by later ones. 

           In addition, the Byzantine Text of Matthew 6:15 is not what we see in Mark 11:26 either; it reads Εαν δε μη αφητε τοις ανθρώπους τα παραπτωματα υμων, ουδε ὁ πατηρ υμων αφήσει τα παραπτωματα υμων.  Seven out of Mark 11:26’s seventeen words are not in Matthew 6:15, and seven out of Matthew 6:15’s seventeen words are not in Mark 11:26.  This is more naturally explained as one of numerous original parallel passages than as a copyist’s attempt to introduce a harmonization, considering that it does not add much to the verbal harmony of the passage.

Mark 11:25-27a in Codex Alexandrinus.
Notice the two lines (underlined in yellow) which begin identically.

          Two small observations may be worth mentioning.  First:  Codex A’s format shows exactly how verse 26 could be accidentally skipped; τα παραπτωματα υμων appears at the beginning of two nearby lines.  Second:  in Codex D, the word υμων is missing at the end of verse 25.  This may be a deliberate omission on the part of a copyist who, having seen that the repetition of the exact phrase τα παραπτωματα υμων made the text vulnerable to loss, decided that the loss of one word was worth a reduction of the risk that a future copyist using Codex D as an exemplar would skip an entire verse.

            So, although Mark 11:26 is not found in our two oldest Greek manuscripts of the surrounding verses, earlier evidence from Cyprian confirms the presence of the verse in North Africa in the mid-200’s.  Further evidence from Codex A, Codex D, the Peshitta, the Gothic version, the Vulgate, and the Old Latin shows that Mark 11:36 was in widespread use in the late 300’s.  The presence of the verse in these branches implies its presence in the trunk, so to speak.       

            One must posit a deliberate corruption by a non-harmonious harmonizer if one is to regard Mark 11:26 as non-original.  The alternative is that an early copyist made a careless mistake.  In such a contest of competing hypotheses, Heinlein’s Axiom should be in play:  do not insist on villainy to account for what can be explained by incompetence.

Mark 9:29 - Conclusion: "and fasting" is Original

From this review of the evidence, it seems clear that the text of Mark 9:29 with και νηστεια was read throughout the early church, even in Egypt, as far as can be confirmed by the external evidence.  It also seems clear that a theological impetus can readily be provided for the excision of these two words:  orthodox copyists with a high Christology would be taken aback by the implication, real or imagined, that Jesus needed to fast in order to successfully exorcise a particular kind of demon.  It would not be hard to foresee that an unbelieving critic of the faith, coming into possession of a copy of Mark with και νηστεια in 9:29, would raise the question, “Why did Jesus, the Son of God, if he is superior to angels, need to fast in order to get some fallen angels to do as he told them?”  A bold copyist with an apologetic agenda could convince himself that the words must be a corruption, and on that basis decline to perpetuate them.

A slightly more complex apologetic rationale for scribal excision also may have existed:  in parts of the early church represented by the tradition expressed in Apostolic Constitutions, regular fasts were to be observed on Wednesdays and Fridays, and fasting was to be avoided on Mondays, Thursdays, and Sundays.  The fasts on Wednesdays and Fridays were said to commemorate Christ’s betrayal and His sufferings.  But this may have provoked a question:  what regular fasts did Jesus observe, before His betrayal and sufferings?  Without the words και νηστεια in Mark 9:29, there is no evidence in Mark that Jesus regularly fasted during His ministry.  The question would thus be rendered superfluous.

A copyist driven by a similar apologetic motive could remove the words in order to lower the risk that an unbelieving critic might pose a question such as the following:  Jesus affirmed in Matthew 11:19 that he came eating and drinking.  How, therefore, did he cast out a demon which can only be exorcised with prayer and fasting?

A much simpler explanation is also at hand:  και νηστεια could be lost via a parableptic error elicited by homoeoteleuton.  That is, an early copyist who was not familiar with the text accidentally skipped from the και in Mark 9:29 before νηστεια to the και at the beginning of Mark 9:30, thus carelessly losing the two words in between.  The possibility of this kind of mistake might not naturally occur to readers of printed texts in which Mark 9:30 begins with κακειθεν, but when reading Codex W (from Egypt), in which Mark 9:30 begins with the non-contracted και εκειθεν, the possibility must be acknowledged. 

The very same kind of careless mistake has caused the loss of και ανεσθη at the end of Mark 9:27 in Codex W, in Old Latin k (which, besides being the only Latin witness for the non-inclusion of και νηστεια in 9:29, is probably the most unreliable extant manuscript of the Gospel of Mark in any language), and in the Peshitta; according to the NA-27 apparatus this is also the likely reading of P45, which suggests that at this point Codex W and P45 echo an ancestor.

In a contest between the early church’s Christology, and the early church’s customs regarding fasting, the former had a much heavier impact on scribal habits.  Copyists were far more likely to remove a short phrase which (they reasoned) risked giving the impression that Jesus was unable to exorcise certain demons without fasting previously, than they were to insert a short phrase which would risk giving readers exactly that impression. 

If και νηστεια was not accidentally lost (or, if it was, and subsequent copyists in Egypt faced exemplars with rival readings in Mark 9:29), the perceived scandal at the thought that the King of angels needed to fast in order to exorcise a certain kind of demon, was enough to convince an early copyist in Egypt that the responsible thing to do was to protect readers from misinterpreting the text by removing the problematic words (or, if the words had already been accidentally lost, and a copyist faced rival readings in his exemplars, this line of reasoning would be a major basis for the adoption of the shorter reading). 

I would also draw attention to the reading of Codex B in the subsequent verse, 9:30.  Instead of παρεπορευοντο, B* reads επορευοντο.  (So does D.)  B*’s reading was adopted by Hort and Tregelles, but their judgment has been rejected by subsequent textual critics.  Figuring that παρεπορευοντο is indeed the original text in 9:30, the reading in Codex B may be considered evidence that the text of B in this passage has undergone editing, which may have included theologically motivated editing.