Hebrews 1:1-6, Papyrus 46, and the Byzantine Text

Heb. 1:1-7a in Papyrus 46.

            Which contains the more accurate text of Hebrews 1:1-6:  the Byzantine Text (which some pro-Alexandrian scholars say emerged as late as the 800’s), or Papyrus 46, the earliest substantial Greek manuscript of the book of Hebrews? 

            Using the text of the 27th edition of the Nestle-Aland Novum Testamentum Graece as the basis of comparison, here are the disagreements in the Byzantine Text:

2 – Byz has a transposition (τοὺς αιωνας ἐποίησεν instead of ἐποίησεν τοὺς αιωνας).

3 – Byz has δι’ εαυτου before καθαρισμὸν but NTG does not.

3 – Byz has a transposition and the word ημων after ἁμαρτιων in the middle of the verse. 

            That’s it:  the addition of 12 letters, and two transpositions.  For the rest of Hebrews 1:1-6, the Byzantine Text and Novum Testamentum Graece are identical. 

            Here are the disagreements between Papyrus 46 and Novum Testamentum Graece:  

1 – P46 has ημων added above the text-line after πατράσιν.  (+4, but since this is a correction, and may have been added after the initial production of the manuscript, it will not be counted in the total.)

2 – P46 has ημειν instead of ημιν. (+1)

2 – P46 does not have και. (-3)

3 – P46 does not have αὐτου after δυνάμεως.  (-5)

3 – P46 has δι’ αυτου before καθαρισμὸν.  (+7)

4 – P46 has τοσούτων instead of τοσούτω.  (+1)

4 – P46 has κριττων instead of κρειττων.  (-1)

4 – P6 does not have των before ἀγγέλων.  (-3)

            When numerical values are assigned to the variants – nothing for benign transpositions, +1 for the presence of a non-original letter, and -1 for the absence of an original letter – this data yields the following results:  in Hebrews 1:1-6, the Byzantine Text deviates from the original text by two transpositions and 12 letters’ worth of corruption (all additions).  Papyrus 46’ text, meanwhile, deviates from the original text by 21 letters’ worth of corruption (9 non-original letters added; 12 original letters omitted).

            What does this tell us? 

            First, it demonstrates that the Nestle-Aland compilers did not adopt the reading with the oldest manuscript-support several times in this passage, particularly in verse 3, where P46 has δι’ αυτου and the Byzantine Text virtually concurs by reading δι’ εαυτου.  The KJV, MEV, and NKJV read “by Himself” in this verse – following the sense given by the oldest manuscript and by the majority of manuscripts – and the CSB, NIV, NASB, and ESV do not.  Readers of the ESV and NASB are not given a footnote at Hebrews 1:3 to inform them that their English translation disagrees with the oldest and most widely attested reading there – succinctly refuting the idea that their footnotes always point out where manuscript-differences affect translation. 

            Thus they have no reason to look into the variant and see that the Alexandrian reading – the lack of δι’ εαυτου (or δι’ αυτου) – is easily explained as a parableptic error, caused when an early copyist’s line of sight wandered from the end of αυτου or εαυτου to the identical letters in the next phrase.  Notably, Michael Holmes, the compiler of the SBL-GNT, did look into this variant-unit, and included δι’ αυτου in the text.  He has not been accused of holding an idiosyncratic view because of this – so far.

            Second, it tells us that if the Byzantine text of Hebrews 1:1-6 did not achieve a stable form until the 800’s, then the scribes who perpetuated it before then exercised a remarkably high level of precision and discipline:  in 736 years (assigning the production of the book of Hebrews to the year 64), collectively they introduced 12 letters’ worth of corruption – or, if δι’ αυτου is the original reading in verse 3, only five letters’ worth of corruption – plus two benign transpositions.  Meanwhile in the Alexandrian transmission-stream, it took 161 years (putting the production of P45 at 225) before a professional copyist produced a manuscript in which Hebrews 1:1-6 contained 21 (or 14, if δι’ αυτου is accepted as original) letters’ worth of corruption. 

            If, like the editors of the New Living Translation, one rejects δι’ αυτου, and, like the NLT’s Coordinating Editor for the New Testament (Philip Wesley Comfort), one accepts a very early production-date for Papyrus 46 – during the reign of Hadrian (117-138) – then it follows that the scribes in the transmission-stream of P46 required 74 years to introduce 21 letters’ worth of corruption into the text of Hebrews 1:1-6.  (Let’s express this as an annual corruption ratio:  74:21, or .284 letters’ worth of corruption per year.) 

            If one were to assume that the Byzantine Text did not emerge until 800, this would imply that the scribes in its transmission-stream up to that point (with a corruption-ratio in Hebrews 1:1-6 of 736:12, or .0163 letters’ worth of corruption per year) were more than seventeen times as accurate, year for year, as the scribes in the transmission-stream of Papyrus 46. 

            On the other hand, if one theorizes that the Byzantine Text of Hebrews 1:1-6 achieved a relatively stable form no later than A.D. 400, then – still using the NTG as the basis for the comparison – the Byzantine copyists would have a theoretical corruption-ratio of 336:12, or .0357, in Hebrews 1:1-6.  This would mean that the Byzantine copyists were eight times as careful as the ones in Papyrus 46’s transmission-stream, but at least this is not as implausible as the first theoretical scenario.

And now, a statistical excursion. 

If one were to posit

            (a) the NTG’s text of Hebrews 1:1-6 as the original text, and

            (b) the existence of the Byzantine Text of Hebrews 1:1-6 in a manuscript made in 225, and

            (c) the existence of Papyrus 46 in the same year, then one could estimate that Byzantine copyists produced .0533 letters’ worth of corruption annually in Hebrews 1:1-6, and that the copyists in Papyrus 46’ transmission-line meanwhile produced approximately .093 letters’ worth of corruption annually in Hebrews 1:1-6.  If one were to picture copyists with the annual corruption-rate displayed in the transmission-stream of Papyrus 46, but producing instead the Byzantine text of Hebrews 1:1-6, then at a rate of .093 letters’ worth of corruption in this passage annually, beginning in A.D. 64 and working for 129 years, they would produce the Byzantine Text of Hebrews 1:1-6 by the year 194. 

            If the first premise in this hypothetical scenario were slightly adjusted, so that δι’ αυτου is accepted as part of the original text in verse 3, then if the copyists who produced the Byzantine Text of Hebrews 1:1-6 introduced corruption at the same rate as the scribes in the transmission-stream of Papyrus 46, then the Byzantine scribes would require 53 years to produce the Byzantine Text of Hebrews 1:1-6, and their work would be identical to the Byzantine Text of Hebrews 1:1-6 by the year 117.  Toss on another 20 years – a decade for each transposition – and the Byzantine Text of Hebrews 1:1-6 emerges, not by 800, but before 150. 

            If one were to ignore P46’s three itacistic corruptions, and assign its production to 225, then its corruption-rate would be 18 letters over 161 years; that is, .112 letters’ worth of corruption annually – or, if δι’ αυτου is accepted as part of the original text in verse 3, then the rate is 11 letters over 161 years, that is, .068 letters’ worth of corruption annually.  Byzantine scribes working at the same rate would produce the Byzantine text of Hebrews 1:1-6 (with 12 letters’ worth of corruption) by the year 240 (working over 176 years) – or, if δι’ αυτου is accepted as part of the original text in verse 3, then if they had the same corruption-rate as the scribes in the transmission-line of P46, they would create five letters’ worth of corruption in 74 years, and the Byzantine Text of Hebrews 1:1-6 would thus emerge around the year 140, or, with two decades for the two transpositions, around 160.  I build nothing on this statistical comparison, but I find it interesting.

Nestle-Aland 28: Much Ado About Even Less

          In the previous post, I concluded that although it has been claimed that the new edition of the Nestle-Aland compilation of the Greek text of the New Testament reflects the editors’ acknowledgement of “much greater value of the Byzantine manuscripts” than before, such a thing is practically indiscernible in the compilation in James, First Peter, and Second Peter – the first three books of the General Epistles, which is the only part of the New Testament in which NA28 introduces text-critically derived alterations to the text.  Previous, NA27 and the Byzantine Text disagreed in those three books 187 times; now NA28 and the Byzantine Text disagree 181 times.  That’s not much of a shift.

          But what about the Epistles of John, and the Epistle of Jude?  Does the 28th edition of the Nestle-Aland compilation have more agreements with the Byzantine Text in those four books?  Let’s find out, investigating all of the textual changes from NA26 (1979) to NA28 (2013) in First John, Second John, Third John, and Jude.  Red text signifies that NA28 rejects a Byzantine reading that was in NA26 and NA27.  A dot accompanies each adoption of a Byzantine reading.

FIRST JOHN

1:7:  δε is not included after εαν near the beginning of the verse, agreeing with 1739 but disagreeing with the Byzantine Text.

3:7:  Παιδια instead of Τεκνία at the beginning of the verse, agreeing with A and 1739 but disagreeing with the Byzantine Text.

● 5:10:  αυτω instead of εαυτω, agreeing with B, A, and the Byzantine Text.  Here NA28 reverts to the reading that was in NA25.

● 5:18:  εαυτον instead of αυτον, agreeing with À, 1739, and the Byzantine Text.

SECOND JOHN

● v. 5:  γράφων σοι καινην instead of καινην γράφων σοι, agreeing with B and the Byzantine Text, but disagreeing with a wide array of witnesses including À A 1505 and 1739.  Here NA28 reverts to the reading that was in NA25. 

● v. 12:  ᾐ πεπληρωμένη after υμων, instead of πεπληρωμένη ᾐ, agreeing with the Byzantine Text.

THIRD JOHN

● v. 4:  τη is not included before ἀληθεία, agreeing with À, 1739, and the Byzantine Text.

JUDE

v. 5:  ἅπαξ has been moved; ἅπαξ πάντα appears between υμας and οτι (NA27 read  πάντα οτι after υμας.  The new reading disagrees with the Byzantine Text, in which υμας is followed by ἅπαξ τουτο οτι.

v. 5:  Ιησους instead of [ο] κύριος, disagreeing with the Byzantine Text, but agreeing with the Vulgate.  Although it is sometimes claimed that textual variants have no impact on Christian doctrines, James White recently acknowledged, “That’s interesting:  Jesus delivered the people from Israel.  That’s got some pretty important theological ramifications to it.  There’s no question about that.”  (Of course he meant to refer to Egypt, not Israel.)  

v. 18:  οτι is not included before επ’ εσχάτου, agreeing with À and B but disagreeing with Papyrus 72, 1739, and the Byzantine Text.  The Byzantine Text here reads οτι εν εσχάτου.   

● v. 18:  του is not included after εσχάτου, agreeing with the Byzantine Text.   

           Added up, that comes to five new disagreements, and six new agreements.    

           In the 27^th edition of the Nestle-Aland compilation, there were 92 disagreements with the Byzantine Text in these four books.  But now, in the 28^th edition, the editors’ newfound appreciation for the Byzantine Text has caused that number to plummet to 91.  Combined with the results from James, First Peter, and Second Peter, this means that the number of times the Nestle-Aland compilation disagrees with the Byzantine Text in the General Epistles has dropped from 279 to 272. 

Nestle-Aland in Mark 11: Alexandrian or Eclectic?

          In the preceding three posts, I observed that even though the Nestle-Aland/UBS compilation of the Greek New Testament is often described as an eclectic edition, based on hundreds and hundreds of manuscripts, in three sample-chapters it is almost entirely Alexandrian, and it hardly contains any distinctive Byzantine readings which represent the vast majority of extant Greek manuscripts.  To be precise, in Galatians 1, NA is .3% distinctly Byzantine; in Luke 15, NA is 1% distinctly Byzantine, and in NA’s 2,812-letter compilation of John 20, there is only one letter (and a bracketed letter, at that) which is in the Byzantine Text and not in Vaticanus (B) or Sinaiticus (À).

          Let’s make another investigation; this time, we will look at the eleventh chapter of the Gospel of Mark, which describes Jesus’ triumphal entry into Jerusalem, the cleansing of the temple, and some other incidents that occurred during the final week of Jesus’ ministry.  In Reuben Swanson’s comparison of the readings of some major manuscripts of the Gospel of Mark, chapter 11 is divided into 59 text-lines.  In 45 of those text-lines, NA agrees precisely with B.  (One of those lines is repeated, however, due to a printing error, so the real total is 58 text-lines, of which B agrees precisely in 44.)   Out of the remaining 14 text-lines, three agree precisely with À (after nomina sacra contractions are taken into consideration).  This means that if we are to find distinctly Byzantine readings in Mark 11 in the Nestle-Aland compilation, they will be somewhere in the remaining 11 text-lines.  Here is what we find there:

● At the beginning of verse 1, NA adopts À’s spelling of Βηθφαγη (disagreeing with B’s Βηδφαγη and with the Byzantine Text’s Βηθσφαγη – though the Byzantine Text is divided at this point; RP2005 has Βηθφαγη in the margin) and then rejects À’s inclusion of εις (agreeing with B and with the Byzantine Text).  Thus, while the line as a whole agrees with relatively few manuscripts (such as W Δ f1), each component agrees with either B or À. 

● In the second half of verse 1, NA disagrees with B’s reading το (disagreeing with των which is read by À and the Byzantine Text), and then agrees with B’s reading of the next word, Ελαιων, disagreeing with À’s Ελεων.  The line as a whole thus agrees with the Byzantine Text; however, neither component agrees with the Byzantine Text distinctly; each component agrees with either B or À. 

● At the end of verse 3, NA adopts B’s spelling ευθυς (instead of the Byzantine reading ευθεως) but then adopts the word-order found in À (disagreeing with B).  NA also adopts the word παλιν near the end of verse 3; the word is not in the Byzantine Text.  Next, NA adopts και απηλθον (read by B and À) instead of απηλθον δε, which is read by the Byzantine Text.  NA also does not include τον before πωλον, although τον is read by À.  Thus, this text-line does not entirely agree with B, or with À, or the Byzantine Text.  Swanson lists only one manuscript – Codex L – that has the combination of readings selected in NA.  Yet, taken individually, each component agrees with either B or À.

● In a text-line which ends with the first three words of verse 6, NA rejected the spelling found in B, À, and in the Byzantine Text (ειπον), adopting instead ειπαν, which is read by a small but respectable cluster of manuscripts (including A, L, Δ, and Π).  This component stands as a non-Byzantine reading which disagrees with À and B.

● Midway through verse 7, NA reads επιβάλλουσιν (agreeing with B and À against the Byzantine Text’s reading επέβαλον) but after the word ιμάτια, NA reads αυτων (agreeing with the Byzantine Text and disagreeing with B’s reading εαυτων and À’s reading αυτω).  The text-line as a whole thus agrees with relatively few manuscripts (including Codices C and L) and its last featured reading (αυτων after ιμάτια) is a Byzantine reading not supported by B or À.

● In the next text-line (in which verse 8 begins), NA adopts εκάθισεν (agreeing with B and the Byzantine Text but disagreeing with À’s reading εκάθισαν), and then adopts και πολλοι (agreeing with B and À but disagreeing with the Byzantine Text’s πολλοι δε), and then, after ιμάτια, adopts αυτων (agreeing with À and the Byzantine Text but disagreeing with B’s reading εαυτων).  The line as a whole thus agrees with relatively few manuscripts (including Codices C, Δ, and 579).  However, each component agrees with either B or À.    

● In the second half of verse 11, NA disagrees with À’s reading οψε (reading οψιας instead, agreeing with B and with the Byzantine Text), and also disagrees with B’s non-inclusion of της ωρας (thus agreeing with À and with the Byzantine Text).  Thus, as a whole, this text-line agrees with the Byzantine Text.  Individually, however, each component agrees with either B or À.

● At the beginning of verse 21, NA rejects the spelling of Ραββει, adopting instead Ραββι and thus agreeing with the Byzantine Text.

● At the beginning of verse 25, NA adopts στήκετε as the third word in the verse, thus disagreeing with B and with the Byzantine Text (which read στήκητε)  and with À (which reads στητε).    

● At the beginning of verse 30, NA includes το after βαπτισμα (agreeing with B and À but disagreeing with the Byzantine Text) but does not adopt B’s spelling of John’s name (Ιωάνου), agreeing instead with À and the Byzantine Text, which read Ιωάννου.  Then NA disagrees with À’s inclusion of the word ποθεν.  As a whole, this text-line agrees with relatively few manuscripts (including Codices A, D, and L).  Individually, each component agrees with either À or B. 

● At the end of verse 33, NA adopts οχλον instead of λαόν, thus agreeing with B and À against the Byzantine Text.  But then NA rejects À’s reading παντες, adopting instead απαντες which agrees with B and with the Byzantine Text.  But then, NA rejects B’s spelling of John’s name (Ιωάνην), reading  Ιωάννην instead.  And next, NA includes the words οντως οτι, agreeing at this point with B but disagreeing with À (which has only οτι) and with the Byzantine text (which transposes these two words).   The text-line as a whole thus agrees with none of Swanson’s witnesses except for a corrector of À.  Taken individually, each component agrees with either B or À.

          Thus, when the variant-units are examined individually, the distinctly Byzantine readings in Mark 11 in the Nestle-Aland compilation consist of the following:

(1)  In verse 7, NA reads αυτων (agreeing with the Byzantine Text and disagreeing with B’s reading εαυτων and À’s reading αυτω), and  

(2)  In verse 21, NA rejects the spelling Ραββει, adopting instead Ραββι and thus agreeing with the Byzantine Text.

          Thus, in Mark 11, the impact of the Byzantine Text is felt by the absence of one letter (ε) in verse 7, and by the absence of one letter in verse 21.  Reckoning that the text of Mark 11 in NA consists of 563 words, and that the Byzantine Text’s contribution to the compilation is discernible in two words, this implies that .4% of the Nestle-Aland compilation of Mark 11 is distinctly Byzantine.  Or, calculating that the text of Mark 11 in NA consists of 2,752 letters, one could say that the Byzantine Text’s existence is manifest in less than .02% of the Nestle-Aland compilation.  The rest originates with other witnesses, primarily Vaticanus, Sinaiticus, and Codex L.

Postscript:  In the course of this series of posts, a point has been raised in the comments:  because the Byzantine Text and the Alexandrian Text agree so frequently, the Byzantine Text’s existence cannot be manifested in 100% of the compilation, but can only be expressed at those points where the Alexandrian and Byzantine texts disagree.  That is undoubtedly true, but how can one answer the question being asked – To what extent is the Nestle-Aland compilation an eclectic text rather than an Alexandrian text? – if not by identifying non-Alexandrian readings (especially Byzantine readings) and seeing how much of the compilation they constitute?  Nor does it affect the answer to the question:  whether one looks at the whole compilation, or only at the parts where the Alexandrian and Byzantine Texts disagree, it is obvious that the Nestle-Aland compilation contains hardly any readings that are found in the Byzantine Text and not in the Alexandrian Text.     

Nestle-Aland in Luke 15: Alexandrian or Eclectic?

          Is the Nestle-Aland compilation basically a slightly tweaked presentation of the Alexandrian Text, relying very heavily on Codex Vaticanus and Codex Sinaiticus?  Or is it an eclectic text based on thousands of manuscripts?  In the previous post, I investigated Galatians chapter 1 and found that the text in NA27 can be almost completely derived from readings in Vaticanus and/or Sinaiticus.  The Byzantine Text’s unique contribution to Galatians 1 amounts to .3% of the text.  What about the text in the Gospels?  Let’s investigate, using Luke 15 as a sample-passage – a chapter known for its parables:  the lost sheep, the lost coin, and the prodigal son.

          Sifting through the horizontal-line comparison prepared by Reuben Swanson for the text of the Gospel of Luke, we find that out of 51 lines of text, Swanson reports that NA and B agree in 39 of them.  Out of the remaining 12 lines of text in which NA and B do not agree, the NA compilation agrees with Sinaiticus in six of them.  This leaves six text-lines in Luke 15 in which NA does not consist of the contents of either B or À.  Here are those six deviations:   

● In a line in which verse 10 begins, NA reads γίνεται χαρα, adopting the word-order found in B and À but not adopting their spelling. (They both read γεινεται χαρα.)  Apparently, only one manuscript used by Swanson (minuscule 579) has this reading; the Byzantine Text reads χαρα γίνεται instead. 

● At the beginning of verse 14, NA reads λιμος ισχυρα, adopting the second word in agreement with B and À but rejecting their spelling of the first word (λειμος).  The occurrence of λιμος and ισχυρα side-by-side here appears to be attested in only a few manuscripts, one of which is Codex L.  The Byzantine Text reads λιμος ισχυρος instead.    

● In the middle of verse 15, NA rejects the spelling of B and À (where both read πολειτων), adopting instead the spelling used in Codex L and in the Byzantine Text (πολιτων).

● Near the end of verse 17, NA rejects the reading found in Papyrus 75, B and À (λειμω ωδε) and the shorter reading found in the Byzantine Text (λιμω), and the transposition supported by other manuscripts (including Codices D, N, and Θ), adopting instead the reading found in Codex L (λιμω ωδε).  (The transposed reading, whether ωδε λειμω or ωδε λιμω, explains its rivals:  when ωδε was accidentally skipped after the preceding word δε, it was lost in the Byzantine Text, and in the Alexandrian Text, after the loss was detected, the word ωδε was moved, as a practical preventative measure, to the other side of λειμω or λιμω.)  

● At the end of verse 21, NA rejects the reading found in B and À (and in Codex D and a minority of minuscules including 700), ποίησόν με ως ενα των μισθίων σου (that is, “Make me as one of your hired servants,” the same phrase found in verse 19), adopting the shorter reading supported by almost all other Greek manuscripts, including Papyrus 75, Codex L, and the Byzantine Text.  (More Greek manuscripts support the variant in B and À here than support the non-inclusion of Mark 16:9-20; yet this variant does not even receive a footnote in translations such as the ESV, HCSB, NIV, and NASB.)

● At the beginning of verse 25, NA adopts the word-order in B (ουτος ο υιος μου), and then adopts the next variant from À (ανέζησεν, instead of B’s εζησεν) – a combination found in only a few manuscripts, including Papyrus 75 and Codex L.  NA also rejects the Byzantine Text’s inclusion of the word και.

          Thus, out of the six text-lines in Luke 16 that do not agree with either Vaticanus or Sinaiticus, Novum Testamentum Graece adopts a reading found in the Byzantine Text in two of them.  The effect of the Byzantine Text upon the text in Luke 16 thus amounts to the removal of one letter in the À-B text in verse 15, and the removal of seven words at the end of verse 21. 

          One could argue that the real impetus for these deviations was Codex L – a strongly Alexandrian witness in Luke – rather than the Byzantine Text, inasmuch as Codex L’s readings were adopted in both of these places and in verses 14 and 25.  Nevertheless, presuming the maximum discernible impact of the Byzantine Text, out of Luke 16’s 2,703 letters in the Nestle-Aland compilation, had the Byzantine Text not been consulted, the text in verse 15 would be one letter longer, and the text in verse 21 would be 27 letters (that is, seven words) longer, yielding a total of 2,731 letters.  Thus, the impact of the Byzantine Text upon the text of Luke 15 in the Nestle-Aland compilation may be said to be discernible in 1% of the text.