The Coptic text of the Gospel of Thomas is paralleled in three sections by Greek fragments from Oxyrhynchus. P.Oxy. 655, in the Houghton Library at Harvard, consists of small amounts of text, which correspond to sayings 24 and 36–39. P.Oxy. 1, in the Bodleian Library in Oxford, is a fragment of a codex containing sayings 26–33 (cf. also saying 77). This latter exemplar in particular shows that—with some exceptions—there is a good deal of correspondence between the Greek and Coptic versions, and therefore that it is quite possible that substantial portions of the Coptic version of Thomas go back to a Greek original.

This is also the case with the fragment in the British Library in London (P.Oxy. 654 = Pap 1531 verso), which parallels substantial portions of the beginning of Thomas. After the incipit and sayings 1–6, however, the Greek text then runs almost completely dry. This is unfortunate, because the next saying is one of the Coptic text's most enigmatic statements: “Jesus said, ‘Blessed is the lion which the man eats, and the lion becomes man; and cursed is the man whom the lion eats, and the lion becomes man.’” (Gos. Thom. 7).¹

We have only a very small amount of Greek text to compare with the Coptic at this point; the last three lines of the fragment contain only ten or a dozen clearly legible letters.²

The top line above (l. 40) can be partially reconstructed as

, as is reflected in the Coptic

. On the next line, the ω of the editio princeps is problematic, however. To suppose with Grenfell and Hunt that this indicates an omega would—by comparison with other omegas in P.Oxy. 654—make (a) the curvature at the top of the letter rather asymmetrical, (b) the left “half” of the letter almost twice the width of the right half, and (c) the letter very wide:

It is perhaps this consideration, in conjunction with the Coptic text, which has led to Attridge's different reconstruction of the line; he proposes reading [

]. However, again, the ink at the upper-left of line 41 renders this reading difficult to the point of impossibility.

³

Having examined the manuscript using a binocular microscope, both my colleague Dr. Peter Williams and I observed that the small trace of ink at the upper-leftmost point of line 41 inclines slightly downward, as is also apparent from the image above.⁴

Attridge's proposal of a nu before the epsilon is plausible, but to suggest an omega before such a nu does not appear feasible.

Reading right to left, the tau, sigma and epsilon are clear (

). But the possibilities for the preceding three traces of ink are not numerous. We may proceed first to identify which letters are possible purely on the basis of the script; then we can narrow this down to what seems plausible in Greek; finally, we may speculate on which may be a good fit with the Coptic.

First, then, of the three traces, the first is most likely to be part of a letter distinct from the second and third, since there is no letter in our scribe's alphabet which would enable the first two traces to be taken together. In this first trace, the downward curve might speak in favor of E, K, Σ, or

. Of these, epsilon is almost certainly ruled out by the absence of evidence for the middle bar of the letter; compare the bar in the visible epsilon above. Sigma remains a possibility, but, overall, sigmas in the fragment tend to be written very small; the trace here is quite high. The second and third traces of ink lie so close together that it is difficult to imagine that they constitute two separate letters; they could form part of H, M, or N. On the other hand, alpha and iota appear typically joined up in this fragment, so AI is another option. This gives as possible the following combinations:

Second, considering the real options in Greek, many of these combinations do not inspire much confidence as word-endings before

.

⁵

The first column (those ending in -H) could point in the direction of a feminine nominative or dative singular noun or adjective, or alternatively a verb in a 2nd singular middle form or a 3rd singular aorist subjunctive. We can probably rule out the second and third columns, except for

, which might suggest an accusative singular, or

, or even—for the whole visible line—a form of

. The fourth column could similarly make for a feminine nominative plural form (noun or adjective), or some form of an aorist infinitive (especially in the case of

.

is also an option.

Third, moving on to hypothesizing as to the correspondence with the Coptic, the most likely possibility is—rather boringly—kaiv. Feminine nouns or adjectives would be somewhat out of place here: the only characters involved are lions (

= masculine) and people (

). Another verb seems unlikely immediately before

, and particularly so if one takes into account the Coptic parallel. However, the ten other instances of the word

in the fragment support such a reading here in line 41. For the sake of brevity, we will present the upper-most parts of five of the clearest examples below our fragmentary line:

If the second and third traces seem a trifle close together to form an alpha-iota here, this can be explained by the fact that the writing in line 41 appears relatively small compared to that of most of the rest of the fragment. So

is probably the best fit, since (a) it could stand without any difficulty before a form of the verb “to be,” and (b) there are three instances of

(‘and’) in the Coptic. Certainty on the matter is of course impossible, but it does seem clear that the two previous proposals are fraught with difficulty, and that the present suggestion is at least preferable to them.

Does this reconstruction help us to assess how close the Greek text might be to the Coptic? Certainly this is an enterprise which must be undertaken with caution, particularly when the Coptic has played a role in the reconstruction of the Greek. There is, however, one implication which can tentatively be drawn.

Attridge estimates the average line length in lines 32–39 to be roughly 31 or 32 letters. If this is right, then we have approximately 25 or 26 letters between

and our new

for “the lion whom the man will eat” (

). Attridge's restoration exempli gratia for this snippet of text (

) is thus now almost certainly too short (only 21 letters), and yet there seems little possibility of improving upon his suggestion without departing from the sense of the Coptic.

⁶

It appears, then, that the Coptic is very slightly briefer than the Greek at this point.

In sum, then, our fragmentary Greek text of Gos. Thom. 7 has both points of similarity to, and difference from, the Coptic. Both are clearly macarisms, and as a result, a general similarity of content is very likely—as per our comments at the beginning on the degree of correspondence between the Oxyrhynchus fragments and the Coptic. However, the different lengths of the clause discussed above (if, as suspected here, the Coptic is a touch shorter than the Greek) should also alert us to be cautious about assuming that the Coptic is simply a literal rendering of something very close to our Greek fragments. As a result, this makes attempting even a restoration exempli gratia in this case an extremely hazardous, and probably superfluous, enterprise.