Tuesday, October 02, 2007

IGNTP John majuscules: How reliable are the transcripts?

No I am not going to start a full-scale scaremongering on the transcripts as befell Comfort's volume on the early papyri some time ago. IGNTP John is a great project (earlier posting on this blog here), also because it has a version on the Internet. Any project that makes its database publicly accessible is of tremendous value --- but I have a few worries.

Out of curiosity I checked one folio of the electronic transcript of Sinaiticus (folio 59) with my own transcript of Tischendorf's transcript, in cases of doubt with Tischendorf's transcript itself and also with the photographs. Only one folio, and only covering John 17:22 - 19:13, and this from only one manuscript.

This is what I found (and you can check for yourself here and here):

1) column 3, line 2:
IGNTP has δεδωκε, should be δεδωκεν

2) col. 3, l. 28:
IGNTP has συνεισηλθε, should be -θεν

3) col. 3, l. 37:
IGNTP has ειπε, should be -πεν

4) col. 7, l. 45
IGNTP has ειδον, should be ιδον

5) col. 7, l. 47
IGNTP has εκραξαν, should probably be - with Tischendorf - εκραξᾱ (with superstroke). This one is admittedly difficult as it is half under a correction.

6) col. 8, l. 11
IGNTP has οφειλει, should be οφιλει

7) col. 8, l. 17
IGNTP has εισηλθεν, should be εισηλθε̄ (with superstroke).

At one point the transcript is improving on Tischendorf:
col. 6, l. 42
IGNTP has rightly αληθειας, not αληθιας

The following point is 'undecided' without a better photograph:
col. 7, l. 36
IGNTP: no ο at end of line, Tischendorf: there is an ο.

The score in terms of penalty points:
Tischendorf: 1
[I had 3 errors in my transcription of Tischendorf, and I thought each of these three more annoying than the eight by IGNTP and Tischendorf combined.]

On the positive side I am glad there is nothing major, on the worrying side I think 7 errors on a folio is a little high. However, the advantage of electronic databases is that it is possible to clean them up slowly over the years, and, after all, we have to start somewhere, don't we?

Up-date: Do read the comments. It is clear that this particular electronic transcription was unreliable and should not have been on-line. The whole electronic edition has been temporarily withdrawn so that the technical problems can be fixed (hence the link given above may not work for the moment) - at least for Sinaiticus an early unchecked transcription was apparently put on-line even though an accurate and thoroughly checked transcription was available (and was used in the production of the book). So this looks like a technical problem in putting the wrong version of the transcription(s) on-line, and is not an indictment of the reliability of the textual work undertaken by the IGNTP team. None of these problems effect the reliablity of the published book. Do read the comments for more information. (PMH)


  1. I agree with Dirk. Electronic versions are relatively easily corrected but his findings beg a question on the reliability of the printed one, though nothing major was found, but still ...

  2. Thanks Dirk,
    It doesn't look so good does it? This number of errors (especially clear errors, not worrying about readings hidden under corrections for the moment) is far too high. You could be talking about around 100 errors just in Sinaiticus. It is hard to see how this could have happened if their were two truly independent transcriptions with discrepancies carefully checked.
    Do you think there could be some contamination from a base text?

  3. OK. I've figured it out. They have contaminated the transcription by using the Textus Receptus as the collating base. This explains Dirks Nos. 1, 2, 3, 4, 6, 7 - all the obvious errors.
    This is crazy, since the problem was already pointed out re the Papyri Volume.
    It shows that the transcriptions were not made freely and independently, but only by revising the TR. If this was practiced throughout then who knows how many errors there could be in these transcriptions.


  4. Hopefully there is a good explanation of this!

  5. I've just checked the printed book (which doesn't contain a transcription of Sinaiticus) to check what happens at these points. The references to Sinaiticus in the apparatus to the text in the printed book, at each of these points, is absolutely correct.

    So this may be some kind of technical computer style of error, since the transcription of Sinaiticus that appears on the web is obviously not the transcription that was used to construct the text and apparatus in the printed book.

  6. I've updated my site at
    so that now if a verse in John with a variant reading is displayed, there is also a link to the appropriate page of the IGNT Majuscules.

  7. "So this may be some kind of technical computer style of error"

    There are a number of computer errors in the IGNTP online transcriptions, some more serious than noted here. For instance, in the Byzantine edition, in the Chrysostom transcription of John 3, verse 5 ends "eis thv basileiav tou"; similarly, verse 13 ends, "o wv en tw". In each case, the last word is omitted. The error occurs in other places as well, but only seems to show up in transcriptions, but not in the comparative apparatus. I reported the problem a month ago, and they acknowledged it, but so far it does not appear that any action has been taken.

    Casey Perkins

  8. Thus there is every reason to buy the book.

  9. Just to let you know that I have phoned David Parker to let him know about the problem.

  10. Many thanks for commenting on this. An immediate response from checking this is that the book is correct (as Peter Head has pointed out). The transcription on the web edition seems to be xml that has been compiled from an older file. We have checked the transcript files and discovered that the errors noted here were all present in an older form of the file, which was later corrected (including, by collating this against the Munster transcription -- which seems to be correct at all these points, see the transcript on http://intf.uni-muenster.de, click on prototype). But for some reason, the older file was used when making the web version. We will replace the transcript as soon as possible with the newer, correct version -- the same version which was used to make the apparatus in the print book which (happily) does not have these errors. So in fact Peter Head is perfectly correct: there was indeed a (very embarrassing) computer glitch. We are taking the site down for now, to correct this, which we will do as soon as possible.
    David Parker, ITSEE

  11. While we are at it....I have been comparing the same site's electronic Vetus Latina transcriptions in relation to the Pericope Adulterae against my own personal transcriptions made from the material at Münster, and am finding various discrepancies as well. Another computer glitch? Or are my transcriptions which I thought to be careful wrong?

  12. Given David Parker's comment, could the original post be edited with an update that calls attention to his claims, lest somebody draw a wrong conclusion before reading the comments?

  13. For a project like IGNTP that purports to be as exhaustive as possible within the limits ascribed to a given volume, what should determine when something must be rendered graphically as it is in a MS, and when something is merely an orthographic stylistic variation that doesn't demand replication.

    What I have in mind is all of the final superstrokes that are rendered instead with final nus (in the list given here, though apparently not in the printed volume). Since the superstroke stands for nu, why should that be considered a mistake? Can't I just think of it as an alternate way for a scribe to write the letter nu? And surely something like IGNTP shouldn't be expected to represent every alternate way that various letters can appear, should it? Perhaps this very issue is already covered in the project's introduction, which I admit I haven't read.

  14. Dear Maurice,
    Do let us know what discrepancies you've found on the Vetus Latina website. A brief glance at Jülicher seems to support most of our readings, but I'm happy to check the photos for any further points outstanding.
    Hugh Houghton

  15. Eric,
    I am not sure what you mean - what wrong conclusion could be drawn from the post?

  16. Eric, I must respectfully disagree. It is important to note exactly what a manuscript contains, including orthographics etc, for they play a part in analysing individual scribes and their habits.

  17. Houghton: Do let us know what discrepancies you've found on the Vetus Latina website

    Sent as feedback via the web site.

  18. Peter M. Head said...
    I am not sure what you mean - what wrong conclusion could be drawn from the post?"

    Maybe I'm reading the whole thread wrong here. But isn't David Parker saying that the errors you pointed out are strictly limited to the electronic edition of IGNTP, not the printed one (as Timo pondered on his first comment)? If that's the case, then it is definitely something worth noting in the main post--pardon me if I'm reading things wrong here.

    Timo Flink said...
    "Eric, I must respectfully disagree. It is important to note exactly what a manuscript contains, including orthographics etc, for they play a part in analysing individual scribes and their habits."

    There's nothing to disagree with really. My questions are genuine questions, not just rhetorical. I'm interested in what you seasoned text-critics think about how orthography should play into a project like IGNTP.

    So, if the nu superscript is important, what about other variant ways of presenting letters, such as the graphemes that represent various letter combinations. Should IGNTP replicate those as well? Again, this is a genuine question.

  19. Ah, I see what you mean Eric. There are a couple of issues here. Firstly, the electronic and the printed editions are not easily separated since they are two aspects of the same project.
    Secondly, David Parker himself, in a review of Comfort-Barrett in TC 4(1999) wrote as follows:
    "how many errors does it take to make a transcription worthless? The answer, scientifically, should be that even one is enough to do that. ... I would certainly be pretty worried if anyone could point to half a dozen clear errors in the IGNTP transcriptions of John papyri, or even to four."
    (http://rosetta.reltech.org/TC/vol04/ComfortBarrett-ed1999rev.html para 37) So, as he admitted, this is a pretty embarrassing issue since errors in the transcriptions necessarily damage confidence in the whole project. Of course, hopefully, it is just a computer glitch and the proper transcription files can be up-loaded quickly. When they are, no doubt we can get Dirk (and others) to do some more spot-checking. This is not point scoring, as David said in that same review: "without accuracy there is no gain".
    Thirdly, if the web version of the transcription was 'an older file' it remains problematic that the transcription seems to have been created by using the TR as a text base and adjusting it, rather than creating a new transcription (or even using a text base that is closer to Sinaiticus). This is problematic because it is not how the IGNTP says things were done.

  20. Eric, I think Peter already said it, but I'll clarify my position: no errors allowed, period. All the text-critical work hinges on accurate information on the readings. Errors on those transcriptions mean (or can mean) errors in our work. Lets say (hypothetically) that I'm preparing a study on orthographic variants. Misinformation on the transcriptions mean I write rubbish without even knowing it. So yes, even one error is a potential problem. IMO, IGNTP should aim towards perfection, now that it is put on the Net, as correcting problems is easier than on print.

  21. Eric (and others),
    I have placed an up-date to the post. Let me know what you think of that. I am certainly not trying to pick a fight with IGNTP folk.