Wednesday, September 11, 2019

A shorter Byzantine reading in the parable of the Prodigal Son

Rembrandt’s Return of the Prodigal
I’ve written before on some shorter (not necessarily shortest) Byzantine readings and their significance for the Byzantine priority position held by my esteemed co-blogger, MAR. (See here, here, and here.) Well, I’ve just come across another such unexpected Byzantine shorter reading. This one occurs in Luke 15.21, in the parable of the two lost sons (aka the prodigal son), the younger son realizes his mistake and says:
18 ἀναστὰς πορεύσομαι πρὸς τὸν πατέρα μου καὶ ἐρῶ αὐτῷ· πάτερ, ἥμαρτον εἰς τὸν οὐρανὸν καὶ ἐνώπιόν σου, 19 οὐκέτι εἰμὶ ἄξιος κληθῆναι υἱός σου· ποίησόν με ὡς ἕνα τῶν μισθίων σου.
18 I will arise and go to my father, and I will say to him, “Father, I have sinned against heaven and before you. 19 I am no longer worthy to be called your son. Treat me as one of your hired servants.” ’
Then, when he does see his father, he gives this slightly truncated form of his speech:
21 εἶπεν δὲ ὁ υἱὸς αὐτῷ· πάτερ, ἥμαρτον εἰς τὸν οὐρανὸν καὶ ἐνώπιόν σου, οὐκέτι εἰμὶ ἄξιος κληθῆναι υἱός σου.
21 And the son said to him, ‘Father, I have sinned against heaven and before you. I am no longer worthy to be called your son.’
This is verbatim from vv. 18–19, except for the last bit which is left off. Not surprisingly, some manuscripts include it. Also not surprisingly, most editions reject the last bit, no doubt, as a harmonization to v. 19 (see Tregelles, Souter, RP, SBLGNT, NA28/UBS5, THGNT). Here is the evidence from NA28:
ποιησον με ως ενα των μισθιων σου ℵ B D 33. 700. 1241. ℓ 844 vgmss syh
omit 𝔓75 A K L N P Q W Γ Δ Θ Ψ ƒ1.13 565. 579. 892. 1424. 2542. ℓ 2211 𝔐 lat sys.c.p co
This variant is also one of many in Luke that had its grade inflated (B to A) from UBS3 to UBS4. The one exception to the strong consensus noted above is WH who have the longer reading in brackets—not so surprising given their affinity for ℵ B. What is striking is that, not only is the shorter reading  attested by Byz, but the longer reading is also attested by such important witnesses as ℵ, B, D, 33.

In addition to its good external evidence, the shorter reading has a very obvious transcriptional explanation in parablepsis. All it takes is a scribe’s eyes skipping  from σου to σου and the result is the omission of the last phrase (cf. Matt 19.9). Intrinsically, since Luke has already repeated so much of the son’s speech, we might expect him to repeat the whole of it. This kind of verbatim repetition of speech is quite common in the Bible, especially in the OT, which Luke is known for imitating stylistically.

I imagine many will reject parablepsis as less likely than harmonization (so Metzger’s Commentary). But why should such an intentional change be more likely than the equally obvious but unintentional one? Certainly, scribes harmonize to the context. But, from my experience, they accidentally omit by parablepsis even more. Any look at a large apparatus bears this out on page after page.

To make the point, we need go no further than these three verses themselves. We just need to do so in a much larger apparatus than NA. In the IGNTP Luke, I see six cases of omission in just these three verses all of which are easily explained by simple parablepsis.
  1. 71 omits αυτω in 15.18
  2. 903* omits εις τον in 15.18
  3. 1, 118, 205, 209 omit και ... σου in 15.19
  4. R* omits υιος ... κληθηναι from 15.19–21 thus omitting all of v. 20
  5. W, 713 omit ποιησον ... σου in 15.19
  6. ℓ 890 omits και ... σου2 in 15.21
The clearest parallels to our variant are 5 and 6 which are also omissions due to the repetition of σου. The omission in R* is instructive because it shows that omissions could be lengthy. 

So, my question: if these shorter readings are clearly accidental omissions, why shouldn’t we see the same in v. 21? Yes, the majority of witnesses have the shorter form, but does that in itself make it more likely? Not for me. Instead, the shorter reading has the simpler scribal explanation, one found multiple times in this same context, and is also attested by the best witnesses. Therefore, it seems to me that it should be preferred as original. 


  1. This comment has been removed by the author.

  2. Dr. Gurry writes "So, my question: if these shorter readings are clearly accidental omissions, why shouldn’t we see the same in v. 21?"

    Namely because of the flip of the evidence. In these cases we have singular or sub-singular readings. The evidence for the shorter reading in vs.21 on the other hand is nothing less than a demonstration of universal antiquity in agreement up until the printing press. As an added kicker, a potential harmonization is within the realm of probability. These combined to overthrow Aleph/B which in all honestly is the only reason we're talking about this particular variant. I would make the same proposal that Scrivener and Burgon did so long ago. Namely, to much weight has been given to these (often poorly transcribed) mss. on account of not only their exceeding great age, but also a hint (at least) of Theological bias.

    P.S. I believe you intended to write ch."15" in your list of six places instead of "5"...Respectfully

  3. Peter, I agree with your general line of reasoning to prefer an explanation by parablepsis (thus giving preference to the longer reading as more likely primary) over an explanation by intentional harmonization (thus giving preference to the shorter reading as more likely primary).

    However, I think it's also worth pointing out that this particular case is a better case for harmonization than many other variants where appeal to that is made, both because the opening half of v. 21 is so like vv. 18-19, and because v. 21 comes so closely on the heels of v. 19.

    It is the nature of eclectic TC to judge variants on a case-by-case basis, which means treating different variants differently in light of each one's own unique character. So, while a rule of thumb that parablepsis is generally more likely than any intentional change including harmonization may be good, it shouldn't be an inviolable rule, especially when one also has to consider external evidence. I would generally expect that anyone who prefers the longer reading in this case, in addition to considering internal arguments in favor of it, probably also shares WH's affinity for ℵ B. And those who don't share that affinity would probably consider the external evidence for the shorter reading overwhelming, and they would not lack a solid explanation for the variant.

  4. PG,
    This is an example that demonstrates the commitment to internal evidence and its subjective nature. Even purveyors of internal evidence cannot agree on the why. For those who hold to a documentary approach, based on external evidence, this is an easy decision, the earliest and wide ranging external evidence for the omission is overwhelming, internal evidence does not need to be considered.

    1. It is nice to fall back on external evidence when the internal conflicts so starkly like it does here. But there is never a time when internal evidence need not be considered.

    2. Certainly not for an eclectic critic😎

    3. Dr. Peter Gurry writes;

      "But there is never a time when internal evidence need not be considered."

      Agreed! Although extreme caution should be taken if one proposes to follow a reading which is not strongly attested externally. In my view at least.

    4. Quite a lot depends on what constitutes strongly attested. I would say ℵ, B, D, 33 is strong.

    5. Dr. Peter Gurry, May I ask why?

      If Aleph/B and 33 are essentially "kin" and D is the primary witness to the so called "western" text, wouldn't this imply a very limited geographical footprint?--And therefore a limited scope of influence throughout Christendom. Not to mention codex Beza is generally considered an unreliable manuscript, Aleph is very prone to singular and rarely attested errors and Vaticanus also has it's fair share of scribal blunders (above average to say the least). My personal take would be to look upon them with suspicion in such a case as this.

    6. Scope of influence doesn’t tell me anything directly about the quality of a manuscript’s text. All manuscripts have mistakes but when 01 and 03 agree that is significant. When they agree with 05 that is more significant still. 33 is like icing on the cake. Or maybe 700. 1241. ℓ 844 vgmss syh are the icing.

    7. I see. Personally I consider 01 03 05 to be manuscripts of "bad" quality due to there many obvious scribal errors. Obviously Hort would disagree. Is there more to your reasoning?

      If not, what would/could change your mind concerning these specific mss.?

      Also, do you not consider codex D a manuscript of "bad" quality?

  5. From my perspective, I have no doubt that the longer reading reflects the actual "Alexandrian archetype" (or, if preferred, a predecessor archetype of the so-called Alexandrian cluster), which archetype was subject to a certain level of editorial activity. Harmonistic expansion in the present instance would then be most likely, based on a scribal/editorial presumption that the concluding portion indeed had been accidentally omitted in their source manuscript(s) due to what seemed to them a fairly obvious haplography.

    On the other hand, I also would suggest that those critics (eclectic or otherwise) who normally favor a basically Alexandrian text really should include the longer reading as part of their preferred main text, even if placed within brackets as per W-H. Consistency in such a matter, coupled with the internal haplography argument, should easily outweigh any other less persuasive internal evidence claims.

    Of course, my own position supports the shorter reading, with no quarter given here to apparent haplography since -- as Peter noted -- such haplographic omission tends to affect at most only a limited number of manuscripts, producing errors otherwise easily corrected by later copyists. Thus, from my perspective, it would be highly unlikely for a haplographic error to multiply without correction into the vast majority of MSS.

    1. MAR: “such haplographic omission tends to affect at most only a limited number of manuscripts...”

      In fact, I noted no such thing! That is precisely the assumption I want to challenge here. Why can’t a sensible omission affect the majority of the textual tradition once it’s been created? I see no reason why it can’t, especially if it happens early enough as appears to be the case here with P75. Lectio brevior is not a canon I find persuasive enough.

    2. Perhaps I should have said "as Peter (perhaps unconsciously?) illustrated by citing occurrences from the three verses involved"?

      To repeat the examples stated:

      "six cases of omission in just these three verses all of which are easily explained by simple parablepsis" -- of which I note that none has significant support from a larger body of MSS:

      71 omits αυτω in 15.18
      903* omits εις τον in 15.18
      1, 118, 205, 209 omit και ... σου in 15.19
      R* omits υιος ... κληθηναι from 15.19–21 thus omitting all of v. 20
      W, 713 omit ποιησον ... σου in 15.19
      ℓ 890 omits και ... σου2 in 15.21

    3. Yes, that is a fair observation. But could we not come up with clear omissions that are widely attested? We have two in the first link in the first paragraph above with 1 Jn 2.23; 3.1—unless your view of transmission history keeps you from seeing those as obvious omissions!

    4. Which indeed it does, otherwise my position regarding theory and methodology would be something quite different.

  6. Interesting post... but I think the most telling example (among the six) of parablepsis is your fifth example regarding verse 19... it's the exact same phrase: ποιησον με ως ενα των μισθιων σου... and only two mss (W, 713) attest its absence in v. 19. But if your hypothesis is correct, parablepsis caused the phrase to be absent in the entire Greek manuscript tradition at v. 21 except for about 200 Greek mss. If the initial text had the phrase in both v. 19 and v. 21 and the phrase dropped out in either place due only to parablepsis, it seems odd that it ended up omitted so often in v. 21 and so rarely in v. 19. That makes me think other factors must be at play in v. 21... the phrase was simply added to harmonize to v. 19. Just my two cents.

    1. Jeff Cate, you state "v. 21 except for about 200 Greek mss." I may be wrong here, but I saw "pc" in NA27. Are there indeed 200 mss. that back the longer reading here? (INGNTP presumably being the source)

    2. That is something I thought about, Jeff. It’s just that not every variant has to have the same transmission history.

    3. I can't speak to the pc = ca. 200 mss issue, since I don't have IGNTP Lk or TuT at hand (can someone check?)

      However, it has been my experience that pc indicates a very small number of MSS in most instances, usually no more than 20-30. A number as large as 200 normally would warrant an al.

    4. I don't have the Text und Textwert Lukasevangelium volume handy to check (and I'm not sure if that's one of the Teststellen but wouldn't be surprised if it is)... but Wieland Willker's online textual commentary cites 11 specific Greek mss having the longer reading in v. 21 (01, B, D, U, X, 983, 1689(=f13), 33, 700, 1241, 2680)... and "al200" (200 being in superscript)... and correct me if I'm wrong, but I think "al200" means 200 others... and I think he bases that on data from Text und Textwert... but like I said, I don't have the TuT vol handy to double check...

    5. Perhaps Willker has a typo that should have read "al20"?

      Teunis' TNTIG citation shows 20 continuous text MSS and 6 lectionaries -- the latter which perhaps Willker did not count? This would be far more in keeping with the pc designation if so.

      NA27 13*: "pc ... a few manuscripts, other than those explicitly mentioned for a given reading, which differ from the Majority Text"

      (NA28 no longer cites pc)

    6. Before I mentioned the alleged 200 mss, I looked elsewhere in Wilker's comm to see if I was misunderstanding his abbreviation. I didn't find that he ever defined it... but I did see other examples throughout which seemed to indicate "al200" = 200 other Greek mss. Other examples mention "al350" (Lk 6:26), "al240" (Lk 9:55-56), "al129" (Lk 23:17) or "al118" (Lk 4:4)... and at Lk 2:25, he mentions "al12," and then spells out below precisely which 12 mss these are. Same with "al20" at Lk 15:16. Years ago, I corresponded with Wieland about a different variant asking where he got his data and he mentioned TuT. So if someone has access to the Lukasevangelium vol of TuT, this could clarify if there were 200 other mss with the longer reading. Hope this is helpful.

    7. This comment has been removed by the author.

    8. Hello Dr. Cate , yes it is very helpfull! My intentions weren't to nitpick, I honestly just wanted to make sure I wasn't seeing things.

    9. MAR, I think your right. I checked IGNTP today and can verify the same apparatus as Teunis has below along with; Lvt(d gat) Lvg(3 mss.) Sh Et Chyrs prodig etc..Thank you(!) to Teunis Van Topik for providing the timely reference. Unfortunately the Txt.u.Textwert {Luke vols} have been missing (for some time now) from the Library I frequent so I couldn't check.

      MAR writes "(NA28 no longer cites pc)"

      I noticed that Dr. Gurry didn't list "pc" in the initial post as well. Wasn't sure if he just trimmed the apparatus a bit or if NA28 actually omitted the citation (I opted for ubs5 instead).

      I suppose someone might email Wie and ask him? I would volunteer but I've never corresponded with him, although I'm sure he'd like to know if it is indeed a typo.

      On another note, anyone know the reason for the different acronym; TNTIG as opposed to IGNTP?

    10. A mystery to be sure (and my copy of TuT is 900 miles away at this point).

    11. Finally able to check TuT, and indeed Willker was correct: TuT reports 207 MSS reading the longer addition, and one MS having the longer edition but omitting με (in contrast, 1444 MSS have the shorter reading, and 5 MSS omit more than that phrase due to homoioteleuton).

      The mystery to me is why NA27 claimed the 207 MSS to be only pc rather than what properly should have been al.

    12. MAR, my only guess would be the date. NA27 is ©1994. Isn't Txt.u.Textwert Lukan. ©1999?

    13. That may be, but data compilation for TuT had commenced long before the time of its publication and would have been available to the NA27 editors. Also, NA27 remained in print after 1994 in various corrected editions until NA28 appeared in 2012, as I recall.

    14. Well, considering it's half way (15:21) through Luke. Maybe they hadn't collated that passage yet and (apparently) never happened to touched on it in their later corrected editions. As you stated above, "A mystery to be sure".

  7. The TNTIG, 1987, has: S B D U X 33 213 348 349 443 577 700 983 1194 1195 1215 1216 1241 1630 2643 L80 L253 L890 L950 L1663 L1761.

    Griesbach-Schulz, 1827: B D 73 89 234 235 al. 8 Mt v [=GA V]

    Mills-Kuster, 1710: Cant. [= D] Laud. 2 [=51] Seld. 2 [54] Vien [76]

  8. I like the idea of cataloging shorter readings that NA28 and other critical text editions adopt that could be easily explained by parablepsis.

    A blog post here dedicated to that, with comments continually being added and the list periodically updated, would be one way to do that. It wouldn't be the most efficient. But it would have the advantage of being at an existing website that already has an active audience made up of people who would helpfully contribute to the catalog.