Followers

Friday, November 30, 2018

Hand to Hand Combat: B, Aleph, and 1295 in Luke 2:1-12


            How much more reliable are ancient manuscripts than medieval manuscripts?  Today, as the Christmas season approaches, we will look into that question by comparing Luke 2:1-12 – a passage about the birth of Christ – in the forms in which it appears in Codex Vaticanus (early 300s), Codex Sinaiticus (mid-300s), and minuscule 1295 (800s).   
            Minuscule 1295 is a Gospels-manuscript housed at the National Library of France, accessible online as Supplement grec 1257.  Let’s compare its text of Luke 2:1-12 to the same passage in the Nestle-Aland compilation (27th ed.).  (The usual ground-rules for hand-to-hand combat are in play:  contractions of sacred named don’t count as variants; transpositions are noticed but not counted (unless a change to the text other than transposition occurs); kai-abbreviations and similar features are not counted as variants; bracketed word in the NA text are considered to be part of the text.  Calculations are made for all variations, and for non-trivial variations.)       


Minuscule 1295 Compared to NA27

1 – no variations
2 – 1295 reads η after αυτη (+1, -0)
3 – 1295 reads ιδιαν instead of εαυτου (+5, -6)
4 – 1295 reads Ναζαρετ instead of Ναζαρεθ (+1, -1) 
4 – 1295 reads πολην instead of πολιν (+1, -1)
5 – 1295 reads μεμνηστευμένη instead of εμνηστευμένη (+1, -0)
5 – 1295 reads γυναικι after αυτω (+7, -0)
6 – 1295 reads επλισθησαν instead of επλησθησαν (+1, -1)
7 – 1295 reads ανεκληνεν instead of ανεκλινεν (+1, -1) 
7 – 1295 reads τη before φάτνη (+2, -0)
8 – no variations
9 – 1295 reads ιδου before αγγελος (+4, -0)
10 – no variations
11 – 1295 reads εστι instead of εστιν (+0, -1)
12 – 1295 does not have και before κειμενον (+0, -3)      

This yields the following raw totals:  24 non-original letters are present and 14 original letters are absent, yielding a total of 38 letters’ worth of deviation from NA. 

When trivial variations are removed from the equation, six variants remain:
            2 – 1295 reads η after αυτη (+1, -0)
            3 – 1295 reads ιδιαν instead of εαυτου (+5, -6)
            5 – 1295 reads γυναικι after αυτω (+7, -0)
            7 – 1295 reads τη before φάτνη (+2, -0)
            9 – 1295 reads ιδου before αγγελος (+4, -0)
12 – 1295 does not have και before κειμενον (+0, -3)          

Thus, if trivial variations are set aside, 1295’s text of Luke 2:1-12 contains 19 non-original letters, and is missing 9 original letters, for a total of 28 letters’ worth of corruption.

Now let’s see how Codex Vaticanus did:

Vaticanus Compared to NA27

1 – B reads εξελθε instead of εξελθεν (+0, -1)
2 – B reads Κυρεινου instead of Κυρηνιου (+2, -2)
3 – no variations
4 – B reads Γαλειλαιας (+1, -0)
[4 – B reads Δαυειδ, twice, but this is not reckoned in the calculations because the word is normally contracted.]
5 – B reads εγγυω instead of εγκυω (+1, -1)
6 – no variations
7 – B reads ετεκε instead of ετεκεν (+0, -1)
7 – B reads ανεκλεινεν instead of ανεκλινεν (+1, -0)
8 – no variations
9 – B reads σφοδρα instead of φοβον μέγαν (+6, -10)
10 – no variations
[11 – B reads Δαυειδ, but this is not reckoned in the calculations because the word is normally contracted.]
12 – B does not have το before σημειον (+0, -2)

This yields the following raw totals:  11 non-original letters are present, and 17 original letters are absent, yielding a total of 28 letters’ worth of deviation from NA.  

When itacisms and similar minor orthographic variants are removed from consideration, only two variants remain:
            9 – B reads σφοδρα instead of φοβον μέγαν (+6, -10)
            12 – B does not have το before σημειον (+0, -2).

Thus, when trivial variations are set aside, Vaticanus’ text of Luke 2:1-12 has 6 non-original letters, and is missing 12 original letters, for a total of 18 letters’ worth of corruption. 

And now for the examination of the text of Codex Sinaiticus in Luke 2:1-12:

Sinaiticus Compared to NA27

1 – À reads εκιναις instead of εκειναις (+0, -1)
1 - À reads Αγουστου instead of Αυγουστου (+0, -1)
1 - À reads απογραφεσθε instead of απογραφεσθαι (+1, -2)
2 – À reads αυτην instead of αυτη (+1, -0)
2 - À reads απογραφην instead of απογραφη (+1, -0)
2 - À transposes so as to read εγενετο πρωτη
3 – À does not have παντες (+0, -5+)
3 - À transposes so as to read εκαστος απογραφεσθε (+1, -2)
3 - À reads εαυτων instead of εαυτου (+2, -2)
4 – À reads την before πολιν (+3, -0) 
5 – À reads απογραφεσθαι instead of απογραψασθαι (+2, -2)
6 – À reads τεκιν instead of τεκειν (+0, -1)
7 – À reads επι instead of εν (+2, -1)
8 – À reads ποιμαινες instead of ποιμενες (+2, -1)
[9 – À reads Θυ instead of Κυ, but the correction may have been made while the codex was still in production.]
9 - À reads επελαμψεν instead of περιελαμψεν (+2, -4)
10 – À reads αυτοις instead of αυτους (+1, -1)
10 – À reads φοβισθε instead of φοβεισθε (+0, -1)
11 – À reads εστιν instead of εσται (+2, -2)
11 - À reads πολι instead of πολει (+0, -1)
12 – À reads ημιν instead of υμιν (+1, -1)
12 – À reads σημιον instead of σημειον (+0, -1)
12 – À reads ευρησεται instead of ευρησετε (+2, -1)
12 – À reads εσσπαργανωμενον instead of εσπαργανωμενον (+1, -0)
12 – À does not include και κειμενον (+0, -11)
12 – À reads επι instead of εν (+2, -1)

This yields the following raw totals:  25 non-original letters are present, and 42 original letters are absent, yielding a total of 67 letters’ worth of deviation from NA.

When itacisms, transpositions, and minor orthographic variants are removed from consideration, eleven variant-readings remain:
            ● 2 – À reads αυτην instead of αυτη (+1, -0)
            ● 3 – À does not have παντες (+0, -5)
            ● 3 - À reads εαυτων instead of εαυτου (+2, -2)
            ● 4 – À reads την before πολιν (+3, -0) 
            ● 7 – À reads επι instead of εν (+2, -1)
            ● 9 - À reads επελαμψεν instead of περιελαμψεν (+2, -4)
            ● 10 – À reads αυτοις instead of αυτους (+1, -1)
            ● 11 – À reads εστιν instead of εσται (+2, -2)
            ● 12 – À reads ημιν instead of υμιν (+1, -1)
            ● 12 – À does not include και κειμενον (+0, -11)
            ● 12 – À reads επι instead of εν (+2, -1)

Thus, when trivial variations are eliminated, Sinaiticus’ text of Luke 2:1-12 has 16 non-original letters, and is missing 28 original letters, for a total of 44 letters’ worth of corruption. 

And now, let’s go to the podium!

Vaticanus took the gold in this contest – which is not surprising, considering how highly it was esteemed when the Westcott-Hort compilation of 1881 – the grandmother of the modern Nestle-Aland compilation – was assembled.  Vaticanus’ text has 28 letters’ worth of deviations from NA, and the only significant variants – in v. 9 and v. 12 – constitute only 18 letters’ worth of corruption.

1295 goes home with the silver:  its text of Luke 2:1-12 has 38 letters’ worth of deviations from NA, and its six non-trivial variants constitute 28 letters’ worth of corruption (four added words, one word-substitution, and one omitted word). 

Sinaiticus takes the bronze:  À has 67 letters’ worth of deviation from NA in the text of Luke 2:1-12; its eleven significant variants constitute 44 letters’ worth of corruption.  This does not say much for the reliability of the copyists who worked in Sinaiticus’ transmission-stream:  compared to the copyists in 1295’s transmission-stream, the copyists in Sinaiticus’ transmission-stream managed to produce a text of Luke 2:1-12 that had almost twice as much corruption, in less than half the time.  We may safely conclude – if the Nestle-Aland compilation is considered a very close approximation of the original text – that the age of a manuscript is no guarantee of accuracy, at least as far as this passage is concerned.

●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●   

Post-script:  A Brief Textual Commentary on Luke 2:1-12
Luke 2:1-12 - the text of the
Complutensian Polyglot (1514)
with a few alterations.

1295’s text of Luke 2:1-12 differs from the passage in the Byzantine Textform only in regard to itacisms and the spelling of the word “Nazareth.”  We have used the Nestle-Aland compilation in the preceding comparison, for the sake of convenience – but its accuracy is not granted automatically.  Let’s briefly investigate the six significant differences in Luke 2:1-12 between 1295 and Vaticanus – which happen to also be the six significant differences between the Byzantine Textform and the Nestle-Aland compilation in this passage. 

● Luke 2:2 –η should be read after αυτη.  The reading in the Alexandrian Text is a simple case of haplography.

● Luke 2:3 – Between ιδιαν πολιν and εαυτου πολιν, the former has a parallel, though distant, in Matthew 9:1; no evident impetus exists to change from ιδιαν to εαυτου.         

● Luke 2:5 – The inclusion of γυναικι has a clarifying effect.  The Peshitta does not support the inclusion of this word.    

● Luke 2:7 – The word τη before φάτνη was removed because some early scribes considered it question-raising, inasmuch as Luke’s narrative has not mentioned a stable or animals; nor is a reason given to expect just one manger to be in the place where Mary gave birth.  Observe how the KJV and NKJV do not translate the word, although the Textus Receptus includes it.  

● Luke 2:9 – The recurrence of the word ιδου (here, and in 2:10) seemed too repetitive to an early copyist.  There is no impetus to add the word, especially so close to its appearance in 2:10.  Inclusion is supported not only by A D K Δ but also by the Old Latin, Vulgate, and Peshitta. 

● Luke 2:12 – Byz, A, K, et al do not have και before κειμενον.  The word is a natural expansion in the Alexandrian text.   

(A text identical to the Byzantine Textform, except for the readings recommended in verses 2, 3, and 5, would be closer to the Nestle-Aland compilation than the text in any of the three manuscripts considered today.)


Readers are invited to check the data in this post, and to explore the embedded links to additional resources.





Tuesday, November 27, 2018

Hand to Hand Combat: B and Aleph vs. 6 and 2401

How reliable are the manuscripts Vaticanus and Sinaiticus, compared to medieval manuscripts? Ask that question to promoters of translations based on the Nestle-Aland compilation – versions such as the NIV, NLT, ESV, and CSB – and the answer you receive will very probably be something like, “Vaticanus and Sinaiticus were made in the 300s.  The older a manuscript is, the closer it is to the original document; thus, the text in these two ancient manuscripts is more accurate than what one finds in medieval manuscripts.”
That seems reasonable, right?  Yes indeed.  On the other hand, it seemed reasonable for centuries to think that the sun revolves around the earth.  All the textbooks said so.  There is just one way to tell whether what seems reasonable is factual:  scientific testing.
When it comes to testing the accuracy of Vaticanus and Sinaiticus, we have a problem:  what shall one use as the standard of comparison?  If the Nestle-Aland Novum Testamentum Graece is used, there is a problem:  the Nestle-Aland compilation is the same as the UBS compilation; the UBS Greek New Testament’s Introduction acknowledges that its editors began “on the basis of Westcott and Hort’s edition of the Greek New Testament,” and Westcott and Hort (back in 1881) acknowledged that they esteemed Vaticanus and Sinaiticus so highly that they (the editors) were willing to reject their agreements only tentatively, even when their readings opposed all other Greek manuscripts.  So there is a bit of a circularity problem when an echo is used as the standard by which to measure the quality of the voice from which it came.      
Another option might be to use the Robinson-Pierpont Byzantine Textform as the standard of comparison – but then the objection would arise that such a standard would give an unfair advantage to the medieval manuscripts.  So, acknowledging the echo-problem, let’s put Vaticanus and Sinaiticus and two medieval manuscripts in a boxing-ring – today we leave the usual arena in the Gospels, and turn to Colossians 3:1-11 – and see which manuscript’s text is more accurate.                 
The two medieval manuscripts going up against the two “oldest and most reliable” heavyweights are minuscule 6 and minuscule 2401.  Let’s take a brief look at the medieval challengers.
Minuscule 6 has a distinguished history:  it was one of the collection of manuscripts cited by Stephanus in his printed Greek New Testament in 1550/1551.   Its production-date is not certain; in the 1800s, Scrivener consider it to be from “xi or later” but the production-date given in the Nestle-Aland Introduction is “XIII” – the 1200s.  Stephanus cited it as witness #5 (“ε′”).  The Nestle-Aland compilers gave it special treatment, listing it as a “Frequently Cited Witness” in Acts and the Pauline Epistles.  It is one of the few manuscripts that does not include the words “in Ephesus in Ephesians 1:1.
          The description of minuscule 6 in Scrivener’s Plain Introduction (1894 edition) is brief:  “In text it much resembles Codd. 4, 5, and 75.  12mo, 5½ ´ 4½, ff. 235,” – supplemented by book-prologues, chapter-lists, chapter-numbers in side-margins, chapter-headings, Eusebian section-numbers in the margins in the Gospels, and a liturgical calendar of lections with St. Chrysostom’s liturgy; the writing-material is parchment.  Scrivener continues:  “This exquisite manuscript is written in characters so small, that some pages require a glass to read them.”  Robert Waltz provides some additional information about MS 6 at the Encyclopedia of New Testament Textual Criticism, including the observation by Wisse that 6’s Gospels-text is affiliated with a subgroup of family-Π.  I would say, too, that some parts of its text have a special closeness to the text of 1739.
          Minuscule 2401 is part of the Goodspeed Manuscript Collection at the University of Chicago, where it is nicknamed “The Theophanes Praxapostolos.”  It was produced in the 1100s.  In addition to the books of Acts, the Pauline Epistles (including Hebrews), and almost all of the General Epistles (pages from Second Peter are missing), it has book-summaries (part of the Euthalian Apparatus) and stichoi-counts.    

          Before we investigate the text of Colossians 3:1-11 in minuscules 6 and 2401, let’s consider how Vaticanus and Sinaiticus each compare to the Nestle-Aland compilation.  (As in past comparisons, the contractions of sacred names are not counted as variants, and words that are bracketed in NA are counted as part of the text.  Calculations will be made of the raw total amount of variation, and of non-trivial variation.)           

Vaticanus Compared to NA27

3:1 – no variation
3:2 – no variation
3:3 – no variation
3:4 – B reads μων instead of υμων (+0, -1)  (A close examination of the online digital image of this page of the manuscript shows that the copyist wrote ΖΩΗΜΩΝ.)  
3:5 – no variation
3:6 – no variation
3:7 – B does not include επι τους υιους της απειθειας (-24) 
3:8 – B reads νυνει instead of νυνι (+1)
3:9 – no variation
3:10 – no variation
3:11 – no variation

Number of non-original letters:  1
Missing original letters:  25
Total number of letters lost or added:  26

It looks like the Nestle-Aland compilation is practically a transcript of Codex Vaticanus, until we reach verse 7.  That’s a significant non-inclusion.  (The NIV, by the way, presently does not include επι τους υιους της απειθειας in its base-text; the reading is only mentioned in a footnote.)  The variant in 3:8 is an orthographic triviality, so we can conclude that Vaticanus contains 25 letters’ worth of non-trivial variation from the Nestle-Aland compilation in Colossians 3:1-11. 

Let’s see if Codex Sinaiticus’ text is better. Sinaiticus has some corrections in this passage and it is not easy to tell with complete confidence whether or not a correction was made before the manuscript left the scriptorium, or at some later time.  I will simply follow the main uncial text, and mention the corrections. 

Sinaiticus Compared to NA27

3:1 – À has εν instead of τω (+2, -2)  (Each letter in εν has been marked over with “/” and τ and ω have been written above the line.)
3:1 – À does not have εστιν (+0, -5)  (The word has been added above the line by a later corrector.)
3:2 – no variations
3:3 – no variations
3:4 – À reads υμις instead of υμεις (+0, -1)
[3:5 – À’s scribe initially did not write υμων after μελη; the word is added above the line.  The addition is not counted as part of the text of À.]
3:5 – À reads πορνιαν instead of πορνειαν (+0, -1)
3:5 – À reads πλεονεξειαν instead of πλεονεξειαν (+1, -0)
3:6 – À reads απιθιας instead of απειθειας (+0, -2)
3:7 – À reads υμις instead of υμεις (+0, -1)
3:8 – À does not read και υμεις (+0, -8) (The words have been added in the side-margin.) 
3:9 – no variations
3:10 – À reads επενδυσαμενοι instead of ενδυσαμενοι (+2, -0)  
3:11 – À does not have τα after αλλα (+0, -2) (The word has been added by a corrector above the line.)

Number of non-original letters:  5
Missing original letters:  22
Total number of letters lost or added:  27

When we remove trivial orthographic variants from the picture, and if we give the corrections in verse 8 the benefit of the doubt by assuming that it was made before the codex left the scriptorium, then the list of disagreements between À and NA boils down to just five –

            3:1 – À has εν instead of τω (+2, -2) 
            3:1 – À does not have εστιν (+0, -5) 
            3:8 – À does not read και υμεις (+0, -8)
            3:10 – À reads επενδυσαμενοι instead of ενδυσαμενοι (+2, -0)  
            3:11 – À does not have τα after αλλα (+0, -2)

This yields 21 letters’ worth of non-trivial variation.  Together, Vaticanus and Sinaiticus contain 46 letters’ worth of non-trivial variation in Colossians 3:1-11.  Now let’s look at this passage in minuscules 6 and 2401. 

MS 6 Compared to NA27

3:1 –  no variation
3:2 –  6 does not have της [At least I did not see της in the microfilm-images from the National Library of France.  Digital images might clarify this point, and I invite others to investigate.]  (+0, -3)
3:2 – 6 does not have τω before Θω (+0, -2)
3:3 – no variation
3:4 – 6 reads ημων instead of υμων (+1, -1)   
3:4 – 6 reads υμων after μελη (+4)  
3:5 –  6 reads ειδωλαλατρεια instead of ειδωλαλατρια (+1, -0)
3:6 – no variation
3:7 –  6 reads αυτοις instead of τουτοις (+1, -2)
3:9 – 6 appears to read πεκδυσαμενοι instead of απεκδυσαμενοι (-1)  [I suspect that the letter is present in the MS but not visible in the microfilm.]
3:10 – no variation
3:11 – 6 reads πασι instead of πασιν (-1)

Number of non-original letters:  7
Missing original letters:  10
Total number of letters lost or added:  17 

When we remove trivial orthographic variants from the picture, then the list of disagreements between 6 and NA27 boils down to the following:

3:2 –  6 probably does not have της (+0, -3)
3:2 – 6 does not have τω before Θω (+0, -2)
3:4 – 6 reads ημων instead of υμων (+1, -1)  
3:4 – 6 reads υμων after μελη (+4) 
3:7 –  6 reads αυτοις instead of τουτοις (+1, -2)
3:9 – 6 appears to read πεκδυσαμενοι instead of απεκδυσαμενοι (-1) 

which yields 6 non-original letters and 9 missing original letters, for a total of 15 letters’ worth of non-trivial variation (or less, depending on whether or not της  is in verse 2 and depending on whether the α in απεκδυσαμενοι is there or not).
      
Now let’s turn to our final combatant. 

2401 Compared to NA27

3:1 – no variations
3:2 – no variations
3:3 – no variations
3:4 – 2401 reads ημων instead of υμων (+1, -1)  
3:4 – 2401 does not include συν αυτω (-7)  (This non-inclusion is supported by Codex A.)
3:5 – 2401 reads υμων after μελη (+4)  (This reading is supported by Codex A.)
3:6 – no variations
3:7 – 2401 reads αυτοις instead of τουτοις (+1, -2)
3:8 – no variations
3:9 – no variations 
3:10 – no variations
3:11 – 2401 does not include βαρβαρος (+0, -8)

There are no trivial readings in Col. 3:1-11 in 2401, so the raw data and the final totals are the same:  its text has 6 non-original letters, and is missing 18 original letters, for a total of 24 letters’ worth of non-trivial variation.

Final score:
Letters’ worth of non-trivial variation in Vaticanus:  25
Letters’ worth of non-trivial variation in Sinaiticus:  21
Letters’ worth of non-trivial variation in 6:  15
Letters’ worth of non-trivial variation in 2401:  24

Conclusion

This little two-on-two contest does not verify the popular axiom “The older the manuscript, the better the text.”   Vaticanus is slightly older than Sinaiticus, and both Vaticanus and Sinaiticus are at least 600 years older than minuscule 6.  Yet, using NA27 as our proxy for the original text, the young minuscules 6 and 2401 introduce, combined, only 39 letters’ worth of non-trivial deviations from the original text, while ancient Vaticanus and Sinaiticus introduce 46 letters’ worth of non-trivial deviations from the original text.

As an additional exercise, suppose we possessed a manuscript that read exactly like the Robinson-Pierpont 2005 Byzantine Textform in Colossians 3:1-11.  Here is how it would compare to the NA27 compilation: 

Byzantine Textform (RP2005) Compared to NA27

3:1 – no variation
3:2 – no variation
3:3 – no variation
3:4 – Byz reads ημων instead of υμων (+1, -1)  
3:5 – Byz reads υμων after μελη (+4) 
3:6 – no variation
3:7 – Byz reads αυτοις instead of τουτοις (+1, -2)
3:8 – no variation
3:9 – no variation 
3:10 – no variation
3:11 – no variation

            Thus, in Colossians 3:1-11, the Byzantine Text, with six non-original letters present and three original letters absent, is closer to the original text than any of the manuscripts in today’s contest  if the Nestle-Aland compilation, which relies heavily on the “the most reliable manuscripts, is used as the standard of comparison.   

Friday, November 23, 2018

Matthew 6:13 - How Does the Lord's Prayer End? (Part 2)


          Let’s continue looking into the textual contest at the end of Matthew 6:13:  the basic question is, are the words “For yours is the kingdom, and the power, and the glory forever, Amen” – ὅτι σοῦ ἐστιν ἡ βασιλεία καὶ ἡ δύναμις καὶ ἡ δόξα εἰς τοὺς αἰῶνας ἀμήν – part of the original text, or not? 
Matthew 6:13b in Codex L.

          As we saw in Part 1, over 98% of the Greek manuscripts that have this verse include these words (including Codex W) – but it is not included in the important manuscripts Vaticanus, Sinaiticus, Bezae, the fifth-century uncial fragment 0171, and the fragmentary palimpsest Codex Z (Dublinensis, 035).  It is not supported by the core representatives of family-1, but it is included in family-13.  The Gothic version and the Armenian version (and others) have it, but most representatives of the Old Latin version, as well as the two most ancient copies of the Middle Egyptian version, do not have it.  The early patristic writers Tertullian, Origen, Cyprian (in Treatise 4), Ambrose, and Cyril of Alexandria (in Catechetical Lecture XXIII) comment on the Lord’s Prayer but do not mention this reading, and several others (Hilary of Poitiers, Caesarius, Gregory of Nyssa) do not mention it when they use this verse – but it was quoted by John Chrysostom (c. 400), and it was quoted in Apostolic Constitutions (380), and it appears to have been utilized by the author of the Didache – an exceptionally early composition (early 100s). 
          The unknown author of the Latin composition Opus Imperfectum in Matthaeum (early 400s) also quoted Matthew 16:13b, in Homily 14, in a way which shows that he read it in his text of Matthew.  This reference is interesting, not only because it gives additional Latin support to the reading, but because it augments the theological range of the support for the reading; while Chrysostom was thoroughly orthodox, the author of Opus Imperfectum was not. 

          With all that in mind, we now turn to some evidence which is not featured in the usual textual apparatuses:  amulets.  The Lord’s Prayer was ubiquitously used in church-services, but it was also applied to a different purpose in some parts of the Roman Empire:  when Christians made small amulets containing snippets of Scripture, one of the most-used passages in such items – after the beginnings of each of the Gospels – was the Lord’s Prayer.  
          Let’s take a look at some of these amulets and the text(s) they contain.
          Papyrus Ct.YBR 4600, housed at Yale University, was made sometime in the 500s-700s.  It consists of a single sheet of papyrus, blank on the reverse side.  It contains Matthew 6:9b-13, but is torn down the middle; as a result about half of the text is not extant.  In verse 13, it appears that the word “Lord” was added after “Lead us not into temptation.”  After “But deliver us from evil,” the doxology does not appear, but rather than come to a close, the text continues with one more line, which says, “To our Lord.”  It seems debatable whether “To our Lord” should be regarded as a closing-title, or as a phrase introducing some non-extant continuation.

          Papyrus Oxyrhynchus LX 4010 is assigned to the 300s.  It is a liturgical prayer, and instead of beginning with “Our Father” it has an introductory portion (which cannot be confident deciphered due to extensive damage, but which seems to address God as the All-powerful Master and God of all comfort).  In v. 12, this witness reads ωσπερ instead of ως.  It does not have the doxology after “but deliver us from evil.”  After “our debtors” and “into temptation,” the remainders of two lines have been obliterated.  It appears that after writing, “but deliver us from evil,” the writer repeated the words “deliver us.”  It is possible that more text followed on another page.
          Á3 (Talisman 3), also known as BGU 3.954, was excavated at Herakleopolis Magna, Egypt, and is assigned to the 500s.  It contains – or contained – an apotropaic prayer (i.e., a prayer for protection and health), and it begins by addressing God as the All-powerful Master. (It is similar in this respect to P. Oxy. LX 4010.)  An individual named Silvanus prays for protection from demons, and from illness, and then introduces “the Gospel-prayer” – Matthew 6:9b-13. The word “Lord” was added after “Lead us not into temptation.”  After that, the document is damaged but enough has survived to show that it contained a doxology that included the words “the glory forever.”  This is followed by snippets from John 1:1 and Matthew 1:1 and a final petition for health.     
          (Unfortunately, as Brice Jones explains in his book New Testament Texts on Greek Amulets from Late Antiquity, this witness was part of the cargo that perished in 1899 when the ship transporting it from Egypt to Europe caught fire in the Hamburg harbor.  Fortunately the document had been meticulously described by researchers Ulrich Wilken and Charles Wessely.)
         
          Á6, also known as Papyrus Iandani I.16 is an Egyptian document from the 400s or 500s.  It includes the text of the Lord’s Prayer, and includes a doxology – “for yours is the glory forever and ever.”  It has other material besides the Lord’s Prayer:  it also features the opening lines of Matthew 1:1, part of Matthew 8:1 (or Luke 9:37), part of Luke 11:1-2, snippets from Psalm 90, and more – all rather garbled, but as Brice Jones has noted, Ernest Schäfer helpfully diagnosed the cause of the mix-up and rearranged the text that the novice copyist was attempting to write.   

          Á13, also known as Papyrus Duke inv. 778, made in the 500s, is a double-sided amulet; Psalm 91 (in Greek) is on one side and the Lord’s Prayer is on the other side.  A detailed description of this document can be found in an article in the 2004 Bulletin of the American Society of Papyrologists by Csaba La’da and Amphilochios Papathomas.  (For a picture and transcript, see this link.)  The words “Your will be done” and “as we forgive our debtors” are absent. The word “Lord” appears after “Lead us not into temptation” (a feature also seen in Á3).  After “Deliver us from evil,” there is an expanded doxology:  “Through the only-begotten” – at this point there is a hole in the papyrus which might have contained the contracted word “Son” – “for yours is the glory and the power and the all-holy Spirit, now, always, and forever and ever.  Amen.”  (“Amen” is written as ϘΘ, two Greek letters which have a numerical value of 99, the same as the word ἀμήν (α = 1, μ = 40, η = 8, ν = 50).  In the papyrus, the Θ is only minimally extant.)  This concluding phrase is similar to the way in which Gregory of Nyssa concluded his catechetical lecture on the Lord’s Prayer, and it also resembles a liturgical formula, which we will consider soon.
          Another witness that should not be ignored is the Gnostic composition called the Prayer of the Apostle Paul.  This text is written in Coptic on the flyleaf of Nag Hammadi Codex 1 (the Jung Codex), which is about as old as Codex Sinaiticus.  Like the texts on apotropaic amulets, this brief composition is a request for protection and health.  It contains some phrases strongly reminiscent of Saying 17 in the “Gospel of Thomas” (which, in turn, resembles First Corinthians 2:9, which is mainly an adaptation of Isaiah 64:4) and Philippians 2:9.  The Gnostic (probably Valentinian) author concludes the prayer with, “For yours is the power and the glory and the praise and the greatness, forever and ever, Amen.” 
          The Prayer of the Apostle Paul includes the request, “give healing for my body when I ask you through the Evangelist,” and although this has been interpreted as a reference to Paul, I suggest that “the Evangelist” means the same thing as “the Gospel-prayer” in Á3.

          Isidore of Pelusium (late 300s-450) also quoted the doxology, twice, in the course of commenting on the Lord’s Prayer, in his Fourth Book of Epistles, #14, To Eutonius the Deacon (cf. Migne PG Vol. 78, col. 1073 and 1076).  This is especially interesting since he resided in Alexandria before taking up monastic responsibilities at Pelusium; one would expect him to have used instead a text that agreed with the flagship manuscripts of the Alexandrian Text.
          Other evidence joins Isidore in confirming that the non-inclusion of Matthew 6:13b was not the only reading known in Egyptian transmission-lines.  Codex L and minuscules 33 and 1241 are among the MSS that support inclusion.  Minuscule 892, widely considered the most Alexandrian of all Gospels-minuscules, also includes the doxology.  
          When one considers that Sahidic and Fayummic versions support “For yours is the power and glory forever, Amen,” a case can be made that the non-inclusion of the doxology is essentially a Western reading than has Alexandrian support:  The Alexandrian witnesses À B and 0171 and Cyril have the shorter reading, as does the Middle Egyptian version; however, that seems to be the extent of support for non-inclusion among  Alexandrian witnesses.  Meanwhile, among the Western witnesses for non-inclusion are D, at least seven Old Latin copies, Tertullian, Cyprian, Ambrose, the Vulgate, and Peter Chrysologus.   
          (It may be noted that practically next door in 6:15, À is allied with mainly Western witnesses, favoring another non-inclusion (the non-inclusion of τὰ παραπτώματα αὐτῶν.) 
         
          So:  which reading accounts for the creation of its rival?  Though this question expresses a standard text-critical canon – the paramount text-critical canon – it is in some cases an oversimplification, because some readings created by copyists (especially shorter readings) may arise due to factors that are not suggested by rival readings.  
          In the case at hand, it should be clear to everyone that the use of doxologies was very widespread, not only in the writing of personal prayers, but also in prayers offered in church-services.  The amulets provide examples of the former; the Didache provides examples of the latter.  It should also be clear that the Lord’s Prayer was subject to adaptation – sometimes via expansion, and sometimes via abridgement.  (The prayer in Matthew 6:9-13 was vulnerable to abridgement via harmonization to the prayer in Luke 11:2-4, just as the prayer in Luke was vulnerable to expansion to the prayer in Matthew 6.)  Additional examples of prayers with doxologies being offered in church-services are found in the liturgies that have been handed down from antiquity.  Let’s consider two examples.
          ● In what is known as the Liturgy of Saint James, after the priest recites embellished citations of the first two clauses of Matthew 6:13, he is to say out loud, “For yours is the kingdom, and the power, and the glory forever,” and then the congregation is to reply with “Amen.” 
          ● In what is known as the Anaphora of Saint Basil, the priest says, “For Yours is the dominion, the kingdom, the power, and the glory of the Father and the Son and the Holy Spirit, now and forever to the age of ages.”  And further along in the service, the recitation of the Lord’s Prayer occurs just before the breaking of the bread in the communion-service:  the congregation is to recite the Lord’s Prayer up to the end of the phrase, “Deliver us from evil,” and then the priest is to  recite the doxology, with an embellishment (or “embolism”), saying, “For Yours is the kingdom and the power and the glory, of the Father and the Son and the Holy Spirit, now and forever and to the ages of ages.”
         
(Additional liturgical examples of prayer-conclusions which ascribe glory to God in one way or another – such as “In the name of Your only-begotten Jesus Christ, through whom to You is the glory and the strength in the Holy Spirit to all the ages of the ages, Amen” and “Through Your only-begotten to Thee is the glory and the strength in Holy Spirit, now and to all the ages of the ages, Amen” – can be found in the Sacramentary of Sarapion of Thmuis, which is generally considered to have been produced in the mid-300s.)
         
Matthew 6:13b in 892.
In the late 1800s, Dean Burgon, conscious of this liturgical custom, proposed that it is so ancient that it accounts for the rival reading in Matthew 6:13.  Several pages of Causes of Corruption are devoted to the exploration of this variant-unit.  Burgon pointed out that if the doxology was added to the text of Matthew after previously existing as a liturgical formula, then instead of seeing, “For Yours is the kingdom and the power and the glory forever, Amen,” we would see, “For Yours is the kingdom and the power and the glory, of the Father and the Son and the Holy Spirit, now and forever and to the ages of ages” – which, although it is what we see in the Anaphora of Basil and the Liturgy of Saint James,  is not what we see, except in two manuscripts (157 and 1253).  The idea that Matthew 6:13b was based on a liturgical formula, Burgon argued, is just the opposite of what ought to be concluded:  instead, a variety of liturgical formulas are based on Matthew 6:13b. 
          Burgon proceeded to propose that the loss of Matthew 6:13b was related to the custom of having the congregation recite the Lord’s Prayer up to the end of “deliver us from evil,” and then having the priest alone recite the doxology.  An early copy with a mark beside it – intended to mean that the doxology was not to be read aloud by the congregation – could easily be misconstrued by a professional copyist to mean that the doxology was not to be written by the scribe.  And thus the whole phrase failed to be perpetuated in a transmission-line that branched out into the base-texts of the Old Latin version(s), and the Alexandrian text-form represented by À B Mae, and the text used by Origen. 
 
Matthew 6:12-13 in Codex K.
        
Such an occurrence would have to happen extremely early – sometime in the 100s.  But this is completely feasible.  As evidence that liturgical formulas were in use in liturgies in the 100s, we may turn to Irenaeus’ work Against Heresies, Book One, chapter 3:  remarking on false teachers’ heresies about Aeons, he mentions that “We ourselves, when at the Eucharist, pronounce the words “to aeons of aeons”” – that is, the words “forever and ever.”  We see the same kind of expression in
Á3, Á6, Á13, and the Gnostic “Prayer of the Apostle Paul.”
          The words “and ever” in the liturgy are a natural expansion – but an expansion of what?  When we see, in the Didache, a description of the Lord’s Prayer being used at a communion-service in the 100s, and when we see, from Irenaeus, a reference to the phrase “forever and ever” being used at the communion-service in the 100s, it seems reasonable to conclude that Irenaeus is referring to an expansion of the same Matthean doxology described in the Didache.
          Thus the table was set, so to speak, in the 100s for a scribal mechanism that could cause the accidental loss of Matthew 6:13b in exemplars that were in the ancestry of both the Old Latin version, and a significant segment of the Alexandrian text-stream, and a form of Matthew used by Origen.  This accounts for the absence of the doxology in the quotations by Latin writers reading Old Latin copies (such as Tertullian, Cyprian, Ambrose, and Augustine), and for the absence of the doxology in copies used by Origen. 

CONCLUSION

          Since the days of Erasmus, it has been alleged that the doxology in Matthew 6:13 is an accretion that slipped into the text from the liturgy.  In the 1700s, Bengel expressed this idea in his Gnomon (Vol. 1, pages 192-195), pointing out that the medieval writer Euthymius, in the course of criticizing the Bogomils for not using the doxology, claimed that it “the choral conclusion added by those who were the divine illuminators and guides of the church.”  (Without closer study, I cannote tell if Euthymius really criticized the Bogomils for avoiding a reading which he admitted was an accretion, or if Euthymius meant nothing more than that Matthew 6:13b was used as the framework for part of the liturgy by those to whom the liturgies are attributed.)  Bengel’s research on the passage can be consulted in the text-critical appendix of his 1734 Η Καινη Διαθηκη and in his Apparatus Criticus.
          More recently, Bruce Metzger claimed that the evidence “suggests that an ascription, usually in a threefold form, was composed (perhaps on the basis of 1 Chr 29.11-13) in order to adapt the Prayer for liturgical use in the early church.” (Compare Hort’s Notes on Select Readings on this variant-unit.)
          This is a convenient explanation for defenders of the Alexandrian text, but the same thought, in the minds of early scribes, would be an effective impetus for the removal of the doxology from the text of Matthew.  Picture a Christian in the second century attending church-services in which the congregation recites the Lord’s Prayer up to the end of “Deliver us from evil,” and the priest (or elder) proceeds to recite the doxology.  In the mind of some participants, the words recited by the congregation – and only those words – were perceived as the prayer; the doxology being regarded as a liturgical flourish.  If such a participant were to proceed to become a scribe, it would be very easy for him to conclude, when encountering the doxology in Matthew 6:13 in an exemplar, that the scribe of his exemplar had mistakenly included a liturgical flourish in the text – and all the more easily considering that there is no such doxology in Luke 11. 
          Of course no one on earth can demonstrate that this happened in the second century; yet the implication of this theory – that Mt. 6:13b is original – is supported not only by a vast proportion of manuscripts (over 98%), but by the significantly earliest witness (the Didache), and by widespread witnesses (Chrysostom, Isidore of Pelusium, Codex W, L, Δ, 892, 1192, 2812, the Peshitta, the Gothic version, the Armenian version, the Opus Imperfectum, etc.).  The non-inclusion of “For Yours is the kingdom, and the power, and the glory forever, Amen” should be considered an early Western reading that was adopted in Egypt but which failed to gain wide acceptance in other Greek transmission-lines (not unlike the non-inclusion of τὰ παραπτώματα αὐτῶν in Matthew 6:15).



Readers are encouraged to explore the embedded links for additional resources.