The Longer Reading in Matthew 25:13

           A textual variant in Matthew 25:13 may shed some light on a mechanism that elicited some expansions in the Byzantine Text.  In the EOB-NT, Matthew 25:13 reads, “Watch, therefore, for you do not know the day or the hour that the Son of Man is coming.”  The words “that the Son of Man is coming” are framed by “<” and “>.”  The WEB, based on the Majority Text, says similarly, “Watch therefore, for you don’t know the day nor the hour in which the Son of Man is coming.”  The KJV reads similarly, and the Textus Receptus agrees with the Byzantine Textform at this point.  In the EHV, Matthew 25:13 only says, “Therefore, keep watch, because you do not know the day or the hour.”  There is no footnote in the EHV to indicate the existence of the longer Byzantine reading. 

          The ESV, CSB, NIV, and NASB all end the verse at the word “hour.”  The NLT, apparently abandoning its base-text, continues with “of my return.”

          What’s going on here?  Did Matthew write the words ἐν ἧ ὁ υἱὸς τοῦ ἀνθρώπου ερχεται or not?

          Short answer:  Not.

          The Byzantine/Majority Text supports the inclusion of “in which the Son of Man is coming,” but the Peshitta does not.  Codices A, D, L, W, Δ, and Σ end the verse with ὥραν (hour).  So do some minuscules, including 33, the first hand of 157, 892, and the first hand of 1424.   The Alexandrian codices À and B weigh in for the shorter reading, and so do P35 and Codex D, and patristic witnesses such as Chrysostom, Athanasius, and Augustine.  The Vulgate and the Old Latin also solidly support the shorter reading here.

          To perceive what has happened here, it is helpful to know that Matthew 25:1-13 was the lection assigned to the 17th Saturday after Pentecost in the Byzantine lectionary.  (It is also a lection in Lectionary 846, to be read in honor of female virgins and martyrs.)  When this segment is read separately from the rest of the chapter, the final sentence was expanded to tell listeners what day and hour were referred to (perhaps using Mt. 24:42 and 24:44 as a model). 

          This expansion can be seen happening in Byzantine manuscripts.  In Codex Y (034), the verse ends ὥραν in the text, but someone – apparently the same person who supplemented the manuscript for lection-reading – added in the margin, “εν η ὁ υς του ανου ερχεται.”  There’s the longer variant.

          Bruce Metzger’s dismissal of the longer reading is correct, but his explanation for its existence (as a “pedantic addition”) seems to show little appreciation for the influence of the lectionary on the Byzantine Text.  When Metzger wrote his Textual Commentary, he was all-in on Hort’s now-defunct theory of the Lucianic Recension.   A more mature Metzger would probably adjust his wording, acknowledging the longer reading as having been made under the influence of lectionary-usage.

          When was the longer reading introduced?  Probably sometime after Codex A (400s), and before 017 (Cyprius) (800s) and the marginalia in 034 (800s, if the marginalia is of the same date as the main text).  [Update:  Andy Vogan has observed that 07, assigned to the 700s, also has the longer reading.]  Someone influenced by a lectionary, wishing to benignly introduce an expansion at the end of Mt. 25:13 to wrap up a lection, created the longer reading, and it was so edifying that so many scribes adopted it that it eventually became the majority reading.  The removal of such intrusions into the text can be achieved relatively easily by filtering the majority text against the Alexandrian Text, the Western Text, and the text of family Π.

 

Papyrus 45 vs. Codex K in Luke 12

P45 - the fragment in Austria

It’s time for some text-critical hand-to-hand combat!  Which manuscript has the more accurate text of Luke 12:23-33:  Papyrus 45 or Codex Cyprius? 

Before we find out, let’s consider each manuscript’s background:  Papyrus 45 was produced in the 200s.  Papyrus 45 has undergone extensive damage, but enough has survived to show that when it was in pristine condition, it contained all four Gospels and Acts (if not more).  For much of its text, P45 is the earliest extant manuscript, particularly for the Gospel of Mark and the parts of the book of Acts that it contains.

After the initial publication of P45’s text of Mark, it was classified as “Caesarean” or “Pre-Caesarean” but these categories were shown to be inaccurate in a study undertaken by Larry Hurtado, who contended that P45’s text of Mark does not have strong affinities with any known text-type and is instead related most closely to Codex W’s text of Mark.  Almost the whole manuscript resides at the Chester Beatty Library in Dublin, Ireland, and can be viewed at the website of CSNTM, except for a fragment in Austria (see picture).       

            Codex Cyprius (K, 017) resides in Paris at the Bibliothèque Nationale de France.  It contains all four Gospels.  It can be viewed (and downloaded) at the Gallica website (search there for “Grec. 63”). Its production-date has been assigned to the 800s.      

           

            In the following comparison, the same ground rules will be applied to the text of both competing manuscripts:  first, deviations from the text of Nestle-Aland (27th edition) are noted, and then filtered.  Transpositions are noted but not counted as variants that cause an adulteration of the text.  Variants involving different formats of nomina sacra are not counted as variants.  Itacisms, cases of movable-nu, and benign consonant-exchanges are counted in the initial total, but the totals are then recalculated without them.  (The three uses of iota-adscript in P45 – τηι, δόξηι, and αγρωι in verses 27-28 – are not counted either.)  

            Let’s look at Papyrus 45 first:

23 – P45 omits γαρ before ψυχη (-3)

24 – P45 adds τα πετεινα του ουρανου και before τους κόρακας (+22)

24 – P45 reads αποθήκαι instead of αποθήκη (+2, -1)

24 – P45 reads αυτα instead of αυτους (+1, -3)

25 – P45 transposes:  προσθειναι επι την ηλικίαν αυτου

25 – P45 does not read ενα after πηχυν (agrees with NA27)

26 – no variants

27 – P45 does not read δε after λέγω (-2)

27 – P45 does not read πάση (-4)

28 – P45 transposes:  σήμερον τον χόρτον οντα

28 – P45 reads αμφιέζει (agrees with NA27)

28 – P45 reads ουν after πόσω (+3)

29 – no variants

30 – P45 reads επιζητει (agreeing with Byz against B and À) instead of επιζητουσιν (+2, -5)

31 – P45 reads του Θυ (agreeing with Byz against B and À) instead of αυτου (+5, -5)

32 – P45 reads ηυδόκησεν instead of ευδόκησεν (+1, -1)

33 – P45 reads ανέκλιπτον instead of ανέκλειπτον (-1)

33 – P45 reads ενγίζει instead of εγγίζει (+1, -1)

Sub-totals:  P45 contains 37 non-original letters, and is missing 26 original letters. 

Total amount of letters’ worth of alterations:  63.

Totals (without orthographic variants in the equation):  P45 contains 33 non-original letters, and is missing 22 original letters.  Total amount of letters’ worth of alterations: 55.

Luke 12:24b-25 in Codex K
(with verse-numbers added)

            Now let’s compare the same passage in Codex K:

23 – K omits γαρ before ψυχη (-3)

23 – K reads εστι instead of εστιν (-1)

24 – K reads τρεφη instead of τρεφει (+1, -2)

24 – [In K, the ι in πετεινων appears to have been added as a correction.]  (-1)

25 – K transposes:  προσθειναι επι την ηλικίαν αυτου

25 – K reads ενα at the end of the verse (+3)

26 – K reads ουτε instead of ουδε  (+1, -1)

27 – no variants

28 – K transposes and adds:  σήμερον εν τω αγρω οντα (+2)

28 – K reads αμφιεννυσι instead of αμφιεζει (+5, -3)

29 – K reads η instead of και (agreeing with P75) (+1, -3)

30 – K reads επιζητει instead of επιζητουσιν (+2, -5)

31 – K reads του Θυ instead of αυτου (+5, -5)

31 – K reads παντα after ταυτα (+5)

32 – no variants

33 – no variants

Sub-totals:  Codex K has 25 non-original letters, and is missing 24 original letters.  Total amount of letters’ worth of alteration:  49.

Without itacisms and benign consonant-exchanges in the equation, K’s text contains 23 non-original letters, and is missing 20 original letters.  Total amount of letters’ worth of alteration: 43.  (The total drops to 42 if we don’t include that correction in verse 24.)

            Codex K wins:  the score is 63 to 49 with all variants considered, and the score is 55 to 43 with accommodation given to orthographic variation.  (Each point is one letter’s worth of deviation from the original text; the lower score wins.)

            This tells us some things about the relative rates of variation in the transmission-lines that produced P45 and Codex Cyprius.  If we put the production-date of the Gospel of Luke in A.D. 62, and the production-date of P45 in 225, then the copyists in P45’s transmission-line introduced .3 letters’ worth of deviation every year, not counting merely orthographic changes.  In comparison, positing the production-date of Codex Cyprius at 850, the copyists in K’s transmission-line introduced .055 letters’ worth of deviation every year, not counting merely orthographic changes.  In other words, the copyists in K’s transmission-line introduced 5.5 times fewer deviations each year, on average, than the copyists in P45’s transmission-line.

            Interesting though it may be to thus see that a manuscript from the 800s has a better text of Luke 12:23-33 than a manuscript from the 200s (using the Nestle-Aland text as the basis of comparison – if the Robinson-Pierpont Byzantine Textform were used for the comparison instead, Codex K’s text would be nearly identical to it), some other data may be even more intriguing: 

● P45, K, and Byz (i.e., the Byzantine Text) do not include γὰρ in v. 23.

● P45, K, and Byz all transpose to read προσθειναι ἐπὶ τὴν ἡλικίαν αὐτου in v. 25.

● P45, K, and Byz all read ἐπιζητει instead of ἐπιζητουσιν in v. 30. 

● P45, K, and Byz all read του Θεου (or του Θυ) instead of αυτου in v. 31.

Thus, in these eleven verses, the Byzantine Text is allied with an early papyrus in four places against the Alexandrian reading.  In three of these four cases, the Byzantine reading is not in agreement with the Western reading in Codex D either.  This data implies that Hort’s approach, which was based on the theory that non-Alexandrian, non-Western readings did not exist before the late 200s or early 300s, was fundamentally flawed.  As of the publication of the text of Papyrus 45, it was no longer tenable.  Yet virtually the same compilation that Hort made in 1881 (with only about 660 differences) continues to be the standard New Testament base-text for modern versions such as the NIV, ESV, CSB, and NRSV – largely because the implications of the readings of the papyri discovered in the 1900s have not been clearly thought through.      

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

Sinaiticus Versus Cyprius: Which is More Accurate in Matthew 5:1-20?

          The axiom, “Prefer the reading with the oldest manuscript-support” is one of the standard guidelines of New Testament textual criticism.  It is natural to assume that an old manuscript’s text is more reliable than a young manuscript’s text, because the passage of time allowed more copies to be made, and every time a copy was made, copyists had another opportunity to make mistakes.

          However, guidelines are not rules.  The accuracy of a manuscript’s text depends on how accurately copyists reproduced the contents of the exemplar, and the exemplar’s exemplar, and so forth, all the way back to the autograph (that is, the original document).

          To test the validity of the practice of giving a manuscript special value (or “weight”) just because it is older, let’s have a contest between Codex Sinaiticus (À [the Hebrew letter Aleph], 01) – hailed as “The World’s Oldest Bible,” made in the mid-300’s – and Codex Cyprius (K, 017), a medieval Gospels-manuscript.  (William H. P. Hatch proposed in 1937 that “It is altogether probable that Codex Cyprius was copied about 1000 A.D.,” but other researchers have assigned it to the 800’s.) 

          In the 1800’s, Codex K was regarded as an important manuscript, but this estimate of its value changed after the publication of Westcott & Hort’s compilation in 1881.  In 1901, F. G. Kenyon stated, referring to Codex K, “It is one of the nine extant complete uncial copies of the Gospels, but as it is as late as the ninth century, and contains the normal α-text [that is, the Byzantine Text], it is not of remarkable value.”  The people who made the fourth edition of the United Bible Societies’ compilation of the Greek New Testament seem to have agreed with Kenyon, inasmuch as they did not even mention Codex Cyprius in the apparatus.  In Bruce Metzger’s handbook The Text of the New Testament, the description of Codex Sinaiticus fills over four pages, while the description of Codex Cyprius consists of one sentence, occupying three lines.

          If we were to evaluate the accuracy of À and K using the Byzantine Text as the standard of what constitutes an accurate text, there can be no doubt that the comparison would demonstrate that the text in Codex K is far superior.  But what happens when the standard is, instead, the Nestle-Aland compilation?  When we strictly compare the text of Matthew 5:1-20 in Codex À to the text of NA27, we get the following results:

Codex À omits the letter ε eight times (in verses 3, 5, 8, 10, 13a, 13b, 15, and 19).

Codex À omits the letter ς twice (in verse 13).

Codex À omits the letter ι once (in verse 20).

Codex À omits 14 words (in verse 19, consisting of 62 letters).

Codex À has the letter ε six times where the Nestle-Aland text has the letters αι (in verses 12, 13, 14, 15, 19, and 20).

Codex À has the letters ου where the Nestle-Aland text has the letter ω one time (in verse 11).

Part of the Beatitudes
from Matthew 5 in Codex Cyprius.

          Thus, if we treat NA27 as if it represents the original text, we conclude that Sinaiticus deviates from the original text via the loss of 11 individual letters, and via the loss of the fourteen-word phrase ος δ’ αν ποιήση και διδάξη, ουτος μέγας κληθήσεται εν τη βασιλεία των ουρανων (that is, in English, “But whoever shall do and teach them, that one shall be called great in the kingdom of heaven”), and via seven vowel-interchanges (itacisms) that yield a net loss of 13 original letters.  In these 20 verses, it thus appears that somewhere along the way from the original text of the Gospel of Matthew to the pages of Codex Sinaiticus, the original text lost 14 words and, in addition, 24 letters.

          In Codex K, the following deviations from NA27 are observed in Matthew 5:1-20:

Codex K omits the letter ν six times (in verses 4, 11a, 11b, 15, and 16).

Codex K has ο instead of α (in verse 1), ο instead of ω (in verse 3), and ε instead of αι (in verse 14), for a total loss of four letters via itacistic interchange.

Codex K omits the word αυτοι (in verse 7), the word οι (in verse 9), and the word τους (after προφητας in verse 12).  (Αυτοι is in the side-margin, and this is probably a first-hand correction; nevertheless I did not include it as part of the text in my calculations.)

Codex K adds the word ρημα (in verse 11), and the word και (after εγω in verse 13).

Codex K reads βληθηναι instead of βληθεν (in verse 13).

          Thus, if we treat NA27 as if it represents the original text, we conclude that Codex K deviates from the original text via the loss of six letters, and via the loss of three words, and via three itacistic changes that result in the loss of four original letters, and via the addition of two words (ρημα and και, for a total of seven letters), and via one substitution (βληθηναι instead of βληθεν in verse 13).

           

          Side by side:

          À lost 14 words (consisting of 62 letters), and also lost 24 letters, with 1 one-letter accretion.

          Κ lost 3 words (consisting of a total of 11 letters), and also lost 10 letters, with accretions totaling 10 letters.   

          Thus if we assign a penalty to each manuscript for each word or letter that deviates from the original text, whether due to subtraction, addition, or substitution, À receives 39 penalty-points, and K receives 23 penalty-points.

          If we simplify the comparison, and give each manuscript a penalty-point for each letter that deviates from the original text, whether it subtracts an original letter, or adds a non-original letter, Codex Cyprius receives 31 penalty-points, and Codex Sinaiticus receives 87 penalty-points.    

          There seems to be no way to avoid the conclusion that in Matthew 5:1-20, the text of Codex Cyprius – a manuscript with an essentially Byzantine Text of the Gospels – has descended from the autograph with much less corruption than the text of Codex Sinaiticus, even though Codex Cyprius had a much longer transmission-history.  This amply demonstrates that the idea of assigning special value to a particular manuscript because of its age is not well-grounded.

          (Readers are invited to double-check this data.  Contracted nomina sacra were not treated as errors.  Codex Sinaiticus has its own website, and fully indexed page-views of Codex Cyprius are available at the website of the Center for the Study of New Testament Manuscripts.  Codex K can also be viewed and downloaded at Gallica.)