Thursday, October 05, 2017

γεινομαι not γινομαι in Luke

Luke 6.36 in P75
One of the claims we’re making in the THGNT is that at the time of the New Testament there was a distinction (or at least a partially preserved distinction) between short and long [i], with the latter sometimes represented by ει. In due time we’ll publish more data backing up this claim. Here I’ll just start with the spelling of the word γίνομαι which in Luke we believe should be spelled γεινομαι, or blending later accent with earlier letters (we explain how accents and letters are separate ‘layers’ in our edition) γείνομαι. Here are some data on early spelling.
  • Luke 6:36 γειν P75(vid) 01 02 03 05 but γιγν 032 
  • Luke 9:7 γειν P75(vid) 01 03 05 032 but γιν 04 
  • Luke 11:26 γειν P75 02 03 05 032 but γιν 01 
  • Luke 12:40 γειν P75 02 03 05 032 but γιν 01 
  • Luke 12:54 γειν P45 P75 01 02 03 05 032 
  • Luke 12:55 γειν P45 P75 01 02 03 05 032 
  • Luke 13:17 γειν P45 P75 02 05 032 (03’s reading γεν- better explained from γειν- than γιν) 
  • Luke 15:10 γειν P75 01 02 03 032 
  • Luke 19:19 γειν 01 02 03 05 
  • Luke 20:33 γειν 02 03 032 
  • Luke 21:7 γειν 01 02 03 032 
  • Luke 21:28 γειν 01 02 03 04 05 032 
  • Luke 21:31 γειν 01 02 03 032 but γιν 04 
  • Luke 21:36 γειν 01 02 03 05 but γιν 04 032 
  • Luke 22:26 γειν 01 03 05 (γεν P75 02 032) 
  • Luke 22:42 γειν P75 01 02 03 but γιν 032 (γεν 05) 
  • Luke 23:8 γειν P75 01 02 03 05 032
  • The earliest witnesses P45 and P75 always support γειν
  • γιν is only supported by 01 04 032, of which we know that 01 has an overwhelming preference for iota in many instances where other mss have epsilon-iota. 04 is fifth century and sometimes supports epsilon iota and 032 may not be as early as the rest and still favours γειν more often than γιν.
  • Against the 3 relatively weak witnesses for γιν we have 7 for γειν.
  • In only 6 of the 17 occurrences in Luke is there earlyish support for γιν. In the rest there is none.
γεινομαι was the normal spelling in Luke. It’s not a misspelling, but a prestigious koine spelling used by careful scribes to bring out the long vowel which arose when the second gamma of the Classical form γιγνομαι was dropped. You can call it a ‘historic spelling’ if you like and claim it has nothing to do with pronunciation, but that just makes the scribes smarter that they were able to preserve into the fourth and fifth centuries spellings representing pronunciations which were no longer current.

And finally
There is one incredibly overused word in this context, which is the word itacism. We can only claim that such has occurred when we understand the standard and conventions which scribes were seeking to attain and are able to demonstrate that they missed it. Itacism certainly occurs often enough in some mss, e.g. 01, but many instances when this is claimed are nothing of the sort.


  1. I'm curious: does the same situation hold in Acts? I'd expect an authorial orthographic preference to carry over from one work to the next, although it's possible that if Luke had been using different amanuenses, there could be a difference.

    1. Different source material might also explain differences like this.

  2. I can paste a reply I've already used on FB:

    No. There are a number of spelling differences between Luke and Acts of a consistent nature. In part this reflects the lack of such early witnesses for Acts (no P75). We've been conservative relative to the traditional spellings and only allowed the ει spellings when they meet a certain bar. This means that in adopting these spellings no one can accuse us of being overly radical. The ει spelling was too thin in Acts 12:5; 19:26; 23:10; 24:2 for us to be comfortable adopting it throughout the book, so we didn't. Having said that Acts 21:14 with 01 02 03 05 supporting γειν against 04 for γιν, shows we could have taken our spelling reform further.

    And from our Introduction (placed at the end of the edition, so you can get quickly into reading the Greek text), pp. 510-11:

    "Some lexemes with ει for etymological ī were only accepted in Luke: δανείζω, ἐπικρείνω, κλείβανος, λείαν, λειμός, Σειδών, σεινιάζω; for Luke we also allowed the comparatives ἐντειμότερος and μεικρότερος. In a number of lexemes our spelling in Luke has been different from that of Acts, a result which may well reflect more about the manuscripts that have survived than about the compositional form of these texts."

  3. Thanks!

    Another question on a similar matter: how does the THGNT handle the imperfect of μελλω in John? The single- and double-augment forms (e.g., εμελλεν and ημελλεν) both have early support in most of the gospel, but Bruce Morrill has shown in his full collation of Jn 18 that all of the earliest witnesses read ημελλεν in Jn 18:32.

  4. Definitely eta when all early witnesses agree.

  5. Peter,

    Thank you for the enlightening post on γιμομαι, I mean γεινομαι. Joey above already asked about Acts and your answer is fine.

    Now, "What of the high literary Koine Greek found in Hebrews?" Also, Does the rest of the NT show the same affinities or is that a 'mixed bag?"

  6. Hebrews is an easy one to check, with merely four instances where γειν- could be seen (7:12, 18; 9:22; 11:6).

    P46 has all instances with γειν- as opposed to γιν- (7:18 beginning of word isn't extant, but likely had γειν-);
    03 (Vat) has γειν- in 7:12 and 18, which were then corrected to γιν- by corr2, with 9:22 and 11:6 as γιν-;
    01 (Sin) has γιν- in 7:12, 18, and 9:22, then γειν- in 11:6;
    finally 02 (Alexandrinus) has γιν- in 7:12, then γειν- in 7:18, 9:22, and 11:6.

    7:12: γιν- x2 (01/02); γειν- x2 (P46/03)
    7:18: γιν- x1 (01); γειν- x3* (*P46vid/02/03)
    9:22: γιν- x2 (01/03); γειν- x2 (P46/02)
    11:6: γιν- x1 (03); γειν- x3 (P46/01/02)

    So even here we see a somewhat preference for γειν- as opposed to γιν-, with 01 having the most instances of γιν- instead of γειν- (explained by the observation that 01 suffers from itacism quite frequently).

    Will the THGNT therefore also be looking at having more instances of υμειν? Or would this be explained as being itacisms in the manuscripts that have it as opposed to υμιν?

    1. S. Walch,

      Thank you for the reply and answer to my question. Interesting the changes from Classical to Koine Greek.

  7. This comment has been removed by the author.

  8. S. Walch,
    No we didn’t print γειν in Hebrews.

    Note the final sentence from p. 510 of our introduction (highlight added):

    "On the one hand, γεινεται in Luke 12:54 and 12:55 has the support of all the earlier manuscripts (P45 P75 א A B D W) and γειν* forms dominate the earliest manuscripts throughout the whole of Luke’s Gospel. It was therefore an easy decision for the editors to spell the lexeme this way. On the other hand, in other cases the editors found the choice much more difficult. In some instances we sought a compromise. For example, we have permitted εἱλάσκεσθαι and εἵλεως in the text, but for the cognate nouns have retained ἱλαστήριον or ἱλασμός, since the nouns have less attestation for the ει-spelling than the verb and adjective. In other cases we did not feel we could accept such spellings. For instance, we noted the strength of support for ει for ī in πίνω, σιγάω, and forms of *τιμ* ‘honour’, but were not sufficiently impressed to accept them generally into the text. If we have had a bias in this matter, it has been towards conventional spellings."

    We did not do an exhaustive sampling of υμειν and ημειν, but based on the samples we did do, they never came (or near) our bar for evidence. Bezae likes those spellings and you’ll find them across a range of witnesses, but we would need a consistent preponderance within witnesses across a corpus to print it. I found they just weren’t coming close.

    In looking at Hebrews, we need to discount Vaticanus after Hebrews 9:14 (replacement leaves).

    The case for γειν in 11:6 is good. In the other cases γιν has 01 04 06, and in 7:12 also has 02. This is where scribal habit can be interesting because 06(original hand) has lots of cases of long [i] represented by ει (e.g. Rom 3:4), so its testimony against ει bears more weight. Another reason to reject ει in 7:12 and 7:18 is that we gave very little weight to Vaticanus.

    Vaticanus is utterly brilliant in consistently distinguishing long and short vowels.

    Have a look at this section from 1 Corinthians in B. Every single short and long case of [krin] is distinguished. It manages to avoid hypercorrection across the NT in all 303 occurrences of short vowel in this verb and related nouns. However, that’s also the problem with Vaticanus’s testimony. It’s too clever, and often will introduce the long-short vowel distinction against the entire rest of the manuscript testimony.

    5:12 τι γαρ μοι τους εξω κρεινειν· ουχι τους εσω υμεις κρεινετε. 13 τους δε εξω ο θς̅ κρινει εξαρατε τον πονηρον εξ υμων αυτων· 6:1 τολμα τις υμων πραγμα εχων προς ετερον κρεινεσθαι επι των αδικων και ουχι επι των αγιων· 2 η ουκ οιδατε οτι οι αγιοι τον κοσμον κρινουσιν· και ει εν υμιν κρεινεται ο κοσμος· αναξιοι εστε κριτηριων ελαχιστων· 3 ουκ οιδατε οτι αγγελους κρινουμεν. μητι γε βιωτικα· 4 βιωτικα μεν ουν κριτηρια εαν εχητε τους εξουθενημενους εν τη εκκλησια τουτους καθιζετε· 5 προς εντροπην υμιν λαλω· ουτως ουκ ενι εν υμιν ουδεις σοφος ος δυνησεται διακρειναι ανα μεσον του αδελφου αυτου· 6 αλλα αδελφος μετα αδελφου κρεινεται και τουτο επι απιστων· 7 ηδη μεν ουν ολως ηττημα υμιν εστιν οτι κριματα εχετε μεθ εαυτων·

    I hope this explains why we’ve been ‘conservative’ and just given iota in these cases in Hebrews (though certainly have lots of respect for the alternative).

    I think the fact that we have many novel (i.e. old!) spellings in our edition, but have not just adopted such spellings wholesale should give you some sense of how we’ve really just tried to be conservative towards the documents after a full consideration of scribal habit.

    1. Thanks for your reply, Peter. Certainly does explain the THGNT position.

      Will there be a published study on the usage of ει/ι going through it all in more detail?

      Looking through P46, in the 24 extant instances of the forms of γινομαι, the MS consistently has γειν- and not γιν- as the core spelling. We then also have the curious case of P15, which has γινεσθε in 1 Cor 7:26 corrected to γεινεσθε (also the main reading of the MS in 7:36).

      Would these be cases of where, like the scribe of Vaticanus, the scribes in these two other MS we're being 'clever' in their spellings of words, or were just in fact doing their job of copying their exemplar?

      A possible reflection of the accepted spelling of Paul, perchance? I expect a study would answer such a question :)

    2. Sorry, that should be 1 Cor 7:23, not 7:26.

  9. Peter,

    This is enlightening, to say the least. It would appear that there may be a problem not only terms of copying, but in terms of pronunciation of the diphthongs and those of the vowels. This would account for the differences in spelling, but maybe not all spelling.

  10. I don't think ει is a diphthong at the time of our early NT witnesses, but it may be used to represent etymological long [i] as well as things which were spelled ει in the Classical period. It is also used often to represent Semitic long [i] in many proper names in the NT. Typically people have inferred from the fact that there is some vowel length confusion in documents from the time of the NT, that vowel length distinctions had been lost, whereas I think it better explains the data to say that vowel length confusion is evidence that some people, in some places, on some occasion didn't distinguish vowel length. This does not mean that there were no people anywhere who ever distinguished vowel length. No one can read Codex Vaticanus with an eye to the question of vowel length and not conclude that it almost universally distinguishes long and short [i].

  11. Peter Williams,

    << Luke 22:26 γειν 01 03 05 (γεν P75 02 032) >>


    << P75 always support γειν >>


  12. James, I obviously meant that P75 always supports γειν as opposed to γιν.

    S. Walch, Though I'd considered doing a full published study of all the ει/ι data, I don't think I'll have time, so will put some out in articles (I've got one on ει for Semitic [i] coming out in a Festschrift) and blog posts. Also, I don't see the need for me to do this since there are plenty of interested people out there. Our edition stakes out a provisional position on various of these decisions, but I hope that others will add their own studies so that we can refine conclusions.

  13. Peter and Dirk,
    What steps were taken during the editing to ensure that you're not simply adopting localized orthography used where a majority of the MSS of the first few centuries survived; i.e., how does one avoid having a built-in tendency to adopt whatever reading was found in Egypt, if one favors readings from the oldest MSS, which are oldest because the low-humidity climate of Egypt let them survive the longest? Were there any checks using later MSS?


  14. James,
    (1) The ει for long [i] spellings are found in papyri from Egypt (P45, P66, P75), but also in manuscripts which, though they might come from Egypt, one has no particular reason for thinking they do (01 03 04). The spellings are also found in Greek-Latin bilingual manuscripts which almost certainly do not come from Egypt (05 06, epistles only for the latter of course).
    (2) According to F.T. Gignac (A Grammar of the Greek Papyri of the Roman and Byzantine Periods, Vol. 1, Phonology, Milan: Istituto Editoriale Cisalpino, 1981), p. 191, Egyptian papyri do not generally distinguish short and long [i] and ει and ι interchange rather freely in Egyptian papyri. It is in fact in Asia Minor that, according to him, the long [i] was preserved for longer.
    (3) The pattern of contrast between short and long [i] in Vaticanus is so deliberate and so different from other extant manuscripts, that it is not credible that it forms part of a widespread pan-Egyptian spelling system.
    (4) Later (e.g. Byzantine) mss regularly use ει for Semitic long [i], but not for etymological Greek long [i]. This can be readily explained on the basis of the view that they normalized Greek spelling to the way they were trained to write the language, but generally left alone the many Semitic names spelled with ει, since they wouldn't have known any reason to change them.
    (5) Koine inscriptions show ει = long [i] was not localized to Egypt.
    (6) We only think that γινομαι etc. is normal because that's what was found in the mss that came west after the fall of Constantinople. At the time when western cultures started learning Classical and NT Greek they didn't have early manuscripts. We know that in Classical times it was spelled γιγνομαι. The second gamma didn't just disappear. At first it left its mark on pronunciation. The spelling γινομαι which predominates in NT mss after the 5th century is logically the third step in a pronunciation change: gignomai --> gi:nomai (colon marking long vowel which initially compensated for the lost consonant) --> ginomai (with short vowel).

    We, of course, want to weigh old mss, and not simply adopt a mechanistic preference for the old. That's where looking at the patterns of scribal habit (particularly Dirk's studies) and of spelling (in which I've taken a great interest) shows that any one witness can be highly misleading. We've sought to take in all this what seems to us to be the most methodologically and theologically conservative approach. Hence the focus on giving weight to early mss, but recognizing human fallibility and hence demanding multiple attestation and a wider pattern of spelling.

  15. From Hort, writing to A. A. Vansittart in 1865: “We are obliged to have an appendix of alternative orthographical readings. With this salve to our consciences, we can print boldy in the text. So far as I know, there is only one thing on which we are agreed, yet are not prepared to dare, in the way of new spellings; that is γείνομαι and its compounds and perhaps γεινώσκω. In the commentary we shall doubtless be less prudent; in this text meant for vulgar use Westcott demurs, and I cannot say he is certainly wrong.”

    The THGNT has boldly gone where WH would not!

  16. Yes, but we also had the advantage of the papyri. One could make the case that at the time of WH the documentary evidence was in general a toss up. Although there were some passages which clearly favoured γειν there wasn't quite enough evidence to justify the inconsistency which THGNT has embraced.

    For instance in Luke we have:

    21:28 γειν 01 02 03 04 05 032
    21:31 γειν 01 02 03 032 but γιν 04
    21:36 γειν 01 02 03 05 but γιν 04 032

    Of this, 03 on its own is almost worthless, since its scribal habit is to introduce ει for etymological long vowels. Likewise 05 and 032 aren't great on this aspect of spelling. 04 is inconsistent. So you're just left with 01 and 02.

    In THGNT we have generally avoided inconsistency when there are relatively small swings in numbers of early witnesses. I think that WH were right to do so in their day, but to flag up the question in their Notes on Orthography.

    I think that what really shifts the balance in Luke is the arrival of P45 and P75 and the cumulative pattern of absence of early γιν.

    Then there are other papyri in other passages which come to the aid of supporting the longer spelling. If I were to make a prediction it would be that the discovery of further papyri will only make the support for ει for long [i] improve.