1 The reason for this is probably that this once highly 6 On this, cf. Jean-Paul Audet, "A Hebrew-Aramaic
Introduction 3. 1 Clement: fol. 51b-70a 4. 2 Clement: fol. 70a-76a 5. A list of the opopara tup (3l/3ALccp nap efSpaiocq
("names of the books of the Hebrews"), with the Hebrew and Aramaic titles in Greek transcription, followed by the Greek title of the respective book: fol. 76a6 6. Didache: fol. 76a-80b 3. The Direct Tradition a.The Bryennios Manuscript (H) Sometimes in antiquity the Didache (as we have seen in the preceding sections) was altogether forgotten (as in the West), and sometimes (as in the East) only a rather indirect knowledge of it was retained; its confusion with the Apostolic Constitutions shows that the writing itself was no longer familiar.1 Therefore it was a first-class sen sation when, at the end of the nineteenth century, an (almost) complete text of the Didache surfaced in a Byzantine manuscript — a text that attracted the interest of researchers at its first publication and introduced a new era in the history of the Didache's influence. The discoverer of the manuscript was the Byzantine metro politan Philotheos Bryennios.
7. The letter of Maria of Cassoboloi to Ignatius of Antioch: fol. 81a-82a 8. Twelve letters of Ignatius (recensio longior): fol.
82a-120a; the text of the letters of Ignatius is fol lowed by the colophon (see above) Finally, there is:
9. a discussion of the genealogy of Jesus: fol.
120a-120b.
The text of the Didache is thus found in folios 76a80b. Bryennios published the Didache text in 1883, ten years after the discovery of the manuscript.7 The whole manuscript was transferred in 1887 to the Greek patriar chate of Jerusalem, where it remains today (under the signature Kd>8. na:rp. 54).8
Bryennios,2 born in Constantinople in 1833, studied theology in East and West, and then became a teacher in the theological schools of the Byzantine Church. In 1875 he became metropolitan of Serrae in Macedonia, and in 1877 transferred to Nicomedia. In 1873 he found, in the library of the Monastery of the Holy Sepulchre
("Jerusalem Monastery") in Constantinople, a parch ment manuscript of 120 leaves, bound in leather. It turned out to be a manuscript collection dated by the scribe, who calls himself Aewp vordpLoq Kai dXe'irrjq
("Leon the notary and sinner"),3 in colophon folio 120a.
He completed the manuscript on Tuesday, 11 June 1056 (of our reckoning). The manuscript is written in a care ful and handsome minuscule script.4 Only one hand is discernible, and obvious mistakes in orthography are relatively few.
The Didache text begins at folio 67a, line 20, with the superscription: bidaxp tcoe 5u>5ei<a dnoaroXup ("teach ing of the twelve apostles"). A "second" superscription begins at line 21 (preceded by an initial A): Aibaxp
Kvpiov 8cd tup 5w5ei<a dnooToXup rolq e&peocp ("the Lord's teaching through the twelve apostles to the nations"). The text then continues on the same line with odoi 8vo ("two ways").
The end of the Didache text is not contained in H. The situation9 is this. The scribe filled each page with 23 lines of text, but folio 80b breaks off in the middle of This manuscript contains:
1. Ps.-Chrysostom Synopsis Veteris et Novi Testamenti (cf. PG 56.313-86): fol. la-38b5 2. Epistle of Barnabas: fol. 39a-51b 6 On this, cf. Jean-Paul Audet, "A Hebrew-Aramaic
List of Books of the Old Testament in Greek Transcription,"/TS, n.s. 1 (1950) 135-54; Charles
C. Torrey, "Ein griechisch transkribiertes und interpretiertes hebraisch-aramaisches Verzeichnis der
Bucher des Alten Testaments aus dem 1.
Jahrhundert n. Chr. (aus dem Englischen iibersetzt 1 The reason for this is probably that this once highly respected writing was not included within the canon. On this, cf. the remarks of Rordorf and Tuilier, Doctrine, 11-12.
2 See the autobiographical note in Schaff, Church Manual, 289-96.
3 The name "Leon," however, is uncertain. Cf. the photograph of the colophon in Schaff, Church
Manual, 6.
4 Details in Bryennios, Ai5a\f ce'-ccf 5 For this synopsis, cf. Bardenhewer, Geschichte, 3.352;
Altaner and Stuiber, Patrologie, 328; Quasten, Patrology, 3.472. und mit Anmerkungen versehen von Otto EiBfeldt)," ThLZ 77 (1952) 249-54.
7 Editio princeps: Bryennios, AidaxV8 There is a photocopy in the appendix to Harris,
Teaching.
9 On this whole subject, see Audet, Didache, 73-74.
text of the Didache (except for the final lines) that we have thus far. That alone gives it special value. But what about the quality of the text in H?15 Here judgments diverge widely. Erik Peterson considers the Bryennios text to be a late recension of the Didache, and thinks that the high regard in which it is traditionally held is a fun damental fault in scholarship.16 If Peterson were right, we could neither make an edition of the Didache nor write a serious commentary on it. To give expression, by way of example, to the counterposition, we note the completely different judgment of Rordorf and Tuilier.17
They consider the Didache text of H to be one that rests on a source from the fourth or fifth century.18 That is possible, but it still would say nothing decisive about the quality of the H text. A real decision about that would be possible only if we possessed a greater wealth of alter native manuscript material for the Didache than we have in our hands at this time. The few lines in P. Oxy. 1782, the versions, and the indirect tradition give us a picture of the rather severe instability of the textual tradition of the Didache (understandable for a writing whose trans mission was not protected by canonical validity), but they do not permit us to make a definitive judgment about the value of the form of text found in H. line 16, with the words eiravoi Tan* vetpeAtiv tov ovpavov
("coming upon the clouds of heaven"), after which the scribe sets a period; the remaining half line and the fol lowing seven lines are left blank. (The next document, the letter of Maria of Cassoboloi, begins on the next page.)10 There is no notation after the words quoted to indicate that the copyist considered the text complete at tov ovpavov.11 One must conclude that the copyist by no means thought that the text of the Didache ended with the words tov ovpavov, but apparently the source only went as far as these words. Therefore, seeing correctly that the conclusion of the Didache was no longer avail able, the scribe left the remaining lines on the page empty — hoping, perhaps, that the end of the writing could at some point be added on the basis of a different source. That the document does not end with the words endvo) mv vecpeAdiv tov ovpavov is obvious from the structure of the Didache apocalypse (16.3-8), as well as from the paraphrasing conclusions to this section in the Apostolic Constitutions and the Georgian version.12
Moreover, in my opinion one should not draw any con clusions about the format and type of writing in the source on the basis of the seven and a half empty lines left by the copyist,13 as Audet has done.14 The scribe sim ply left the rest of the page empty, based on the correct supposition that the conclusion of the document he was copying was missing, but without knowing for certain how much text had been lost. The entire procedure also shows that the copyist was a thoughtful worker who did not simply make a mechanical copy of the source.
If we examine that text without permitting ourselves to be overcome by pessimism or optimism, the results are ambivalent. On the one hand, it appears that H has been distorted by some later glosses and interpolations
(although these are only brief expressions or passages: cf. at 1.4a; 2.5b; 10.5; 13.4), but that its text can also be corrected in some places with a greater or lesser degree
The Bryennios manuscript offers us the only complete 10 This long blank space is contrary to the scribe's practice elsewhere: cf. Audet, Didache, 73-74. editions of the Didache, and our Didache exegesis, prevent us from recognizing the total insecurity of the manuscript tradition. It is still customary to explain the text as if we had, in the manuscript dis covered by Bryennios, something like the authentic version of the Didache. In reality, it appears to me that the Bryennios text represents a late recension of the Didache" (181). "We can only describe the aporiae, but cannot communicate the original text" (182).
17 Rordorf and Tuilier, Doctrine, 102-10.
18 Rordorf and Tuilier, Doctrine, 104-7. Audet {Didache, 32) had made a similar estimation before hand.
11 For the usage in this regard, see Audet, Didache, 74.
12 On this, see below.
13 Audet concludes (Didache, 75) to a separate Didache codex that was available to the copyist.
14 Audet, Didache, 75-78. He believes it was a minia ture codex with uncial script.
15 The text of 1 and 2 Clement, the letters of Ignatius, and Barnabas appear to be the result of revision. Of course, that in itself says nothing about the text of the Didache-, see Audet, Didache, 25-26.
16 Peterson, "Probleme," passim. Peterson thought that the H text went back to a late recension (perhaps stemming from ascetic circles [150, 181, and pas sim], and more precisely from Novatianist circles
[162-63]). What we find in H is not the text of the Didache but a subsequent revision of the text. "Our
19 For this whole question, see Niederwimmer, "Textprobleme," 114-30. Wengst, in his edition, offers a text that deviates rather sharply from H.
20 See below, at 9.3-4; and Niederwimmer, "Textprobleme," 125.
Introduction lines each.24 Folio 1 contains the Greek text of Did. 1.3c, of assurance (cf. at 1.3d; 2.7b; 3.4c; 5.1 end; 9.3a and 4a;
10.3 end; 10.4a; 12.1a). On the other hand, the absence of the interpolation after 10.7 (myron [oil] prayer) shows that H goes back to a text that, at least in this passage, is better than the textual tradition of the Coptic or the source for the Apostolic Constitutions.19 All this suggests that the H tradition should neither be damned nor sim ply passed on as the authentic text of the Didache. The Bryennios manuscript seems to me to present a Didache text that must remain subject at every point to inquiries of a text-critical nature, but one should not feel com pelled to the desperate conclusion that what we have before us is a late and tendentiously falsified recension of the original writing. There are demonstrable changes in individual passages, and they may be suspected else where; but that the H text is the result of a thorough, tendentious recension is not demonstrable. A final judg ment would be possible only if the textual basis of the Didache could be expanded. Everything we can say about ovx'c KCtt Ta edvr\ tovto ttolovou> . . . ("do not even the nations do that . . .") to 1.4a, airooxov twv aapKe[i]KWV eitifivixeiuv ("avoid the fleshly and bodily passions").
Folio 2 contains Did. 2.7b, eXey^eig irepl <Xv be irpooev- %ti] . . . ("some . . . for whom you shall pray . . ."), to 3.2a, r\ opyr) 7rpoq tov ipovov ("anger [leads to] murder").
These two leaves must come from the same book, which originally contained the whole text of the Didache (and perhaps more besides). The leaves are covered with irregular uncials written in brownish ink.25 "That the writer was a person of no great culture is clear also from his spelling and division of words (e.g., eiui)vp.6L(j)v, u/x|etq)."26 In folio 2 recto, line 20, after the word oov, the copyist has completed the line with wedge-shaped symbols written alternately in red ink and in the usual bl own (nine of these are still visible, but the symbols must have extended to the end of the line). In the follow ing line (21) there are three repeatedly interrupted hori zontal lines (the first and third in red ink); in the middle, and extending beyond the lines, one can discern the symbol 4 written in the ordinary brown ink.27 It is natural to assume that the copyist wished to designate the beginning of a new section in line 21 (probably on the basis of the source) by means of the horizontal lines there,28 while the symbols in line 20 serve to show that no text is missing at this point.29 the original home of the archetype of H can be reduced to the inference that, at Did. 9.3-4, it inserted the expression uXotopcx (fragment), which was characteristic of the liturgical language of Egypt; from this in turn one can infer that the archetype of H originated in Egypt.20
/ b. Papyrus Oxyrhynchus 1782 (P) In 1922 B. P. Grenfell and A. S. Hunt edited21 two frag mentary parchment leaves of a miniature codex22 from the end of the fourth century.23 These leaves (hereafter
"P") are now to be found in the Ashmolean Museum in Oxford. The pages, which are worn and badly discolored in places, especially at the edges, are 5.5 cm. by 4.5 cm.
(fol. 1) and 5.7 cm. by 4.8 cm. (fol. 2), and are written on both sides. Folio 1 recto has seven short lines, while folios 1 verso and 2 recto and verso have eight short
21 Papyri, no. 1782, pp. 12-15.
22 For miniature codices of this type, see Grenfell and Hunt, Papyri, 12; Audet, Didache, 26 n. 2; Leiv
Amundsen, "Christian Papyri from the Oslo Collection," SO 24 (1945) 126-28; Eric G. Turner,
The Typology of the Early Codex (Haney Foundation Series 18; Philadelphia: University of Pennsylvania
Press, 1977) 22, 29-30.
23 "It may perhaps date from the fourth century rather than the fifth" (Grenfell and Hunt, Papyri, 12-13).
24 See immediately below for the signum dispositions in those on the other pages.
26 Grenfell and Hunt, Papyri, 13.
27 It is also possible to discern a brownish horizontal line and (above and below it) two red horizontal lines between lines 19 and 20. One gets the impres sion that these horizontal lines were inserted by the copyist at a later time.
28 Audet differs (Didache, 55). He does not presume a signum dispositions here: "It is much more natural to suppose a projected addition, omitted after the fact for one reason or another."
29 The new section would begin just where we also place the beginning of a new chapter. I find it quesfol. 2 recto.
25 The letters in fol. 1 recto are somewhat larger than
be -Kpoaev^et ovg be ayairpoetg virep Tpp pvxp 20 oov > > >*?• > > (in' 7rpooev^p' ovg be ayairpoetg We need not mince words regarding the significance of these two leaves for the manuscript tradition of the Didache. They are some 650 years older than the Bryennios manuscript and represent the oldest manu script witness to date for the Didache text. One should also keep in mind that these two leaves come from approximately the time when the compiler of the Apostolic Constitutions was copying the Didache into book 7. vnrep Tpp pvxw oov
(3.1) TCKVOV pOV
T6KP0P pOV
ipevye airo (fol. 2v.) PairoJI iraPTog 25 Trpayparog tpevye curb iraPTog tvopoiov avrov The text presented by P is important, in the first instance, as evidence for the so-called interpolatio Chris tiana ("Christian interpolation") or sectio evangelica
("gospel section"). In folio 1, P gives us, of all things, a bit of text from the very passage that is sometimes regarded (wrongly, I believe) as a later gloss.30
Otherwise, the deviations from the H text (in the brief passage contained in P) are numerous enough to strengthen the thesis stated above concerning the insta bility of the text of the Didache. In table 2, I illustrate this situation by setting the P text alongside the H text
(whereby the order of lines is retained only in the pre sentation of the P text).31
TTOPTjpOV Kai iropppov KCti air6 itavroc; opoiov (3.2) pp yLvov opyLAogpp 7eipov opyeu Aog eiretbp obp yet 17 opyp irpog top ipopop 66x776i yap r\ opyp irpog top ipovov. . .
I will discuss only the most important textual variants.32 At some points, P appears to have the better text: in line 2, P reads tovto (as does Apostolic Con stitutions) against to avro in H. The reading in P and Apostolic Constitutions is preferable; H apparently assimi lates to Matt 5:46 or Luke 6:33. In lines 4-5, P reads yuAelre (again with the Apostolic Constitutions), which is preferable to H's dycardTe. One may ask whether H again reveals an influence from the Synoptic Gospels
(Matt 5:44/Luke 6:27). In lines 13-15, P has an older variant of the interpolation that has entered the text of the Didache at this point.33 H shows what is already a later form (apart from aTreyov, which is preferable). In lines 16-17, P reads irepi wv <5e (as do the Canons of the H P (1.3c) ovx'l kolI ra e§vrj to avro Table 2 (fol. lr.) ovxc eai Ta e i)pp TOVTO (3d) iroiovoLv; vnelq be dyanare TTOLOVtJiP Vp ecg be <pcAeLT Tovq ixtoovvrac; v/jidg kocl ovx
e Tovg ptoov Tag vpag eat ovx e£ere ex exftpov (fol. lv.) Apostles and the Epitome), probably rightly (against H's
7repl 5e cor').34 The P text also reveals a series of sec ondary elements, however: lines 8-12 are certainly a later addition to the interpolation in Did. 1.4. I will have more to say below about the successive expansions the Didache text underwent at this point. In lines 24-26, P
1 Ppop aeov e tl pe 5et not ovPTa puoat oov to irpa 7r[p]co top 7raPTw
has (as a clarification) airo rravToq repaypctToq rrovrjpov instead of H's d7r6 rravToq irovripov (which is also secured by Canons of the Apostles; Apostolic Constitutions reads died iravroq Kaxov, as does the Epitome). The (1.4) airexov 7W aitooxov tu>v oapKLKUP eai oupaTiKUP oapKe[i]Kuv e eindupLuiP'
Tndvpeiwp (2.7b) .. . eAey^ecg-irepi be (fol. 2r.) eAey^eLg irepi w tionable whether one can read the symbol in the 33 See at Did. 1.4 below. middle of line 21 as a numeral (as Hans Lietzmann 34 Barnabas, Doctrina apostolorum, and Apostolic supposes: "Notizen," ZNW 21 [1922] 238). Constitutions do not offer a witness here.
30 On this, see below.
31 For P I cite the text of the editio princeps, but with an altered numbering of the lines at the end.
32 See also Audet, Didache, 28; Niederwimmer, "Textprobleme," 122-23.
Introduction trend toward clarification appears also in Doctrina apostolorum: ab homine malo. Both these clarifications are textually secondary, but that in P is at least correct in its tendency. In lines 26-27, P omits drirb navToq. In lines
29-30, P's eireLdr) 6617761 (against H, Canons, Epitome odrjyel 7dp) is probably a later assimilation to Did. 3.4,
P apparently rests on an archetype different from that of H. Neither of the two forms of the text is always supe rior to the other. It is interesting that P (perhaps on the basis of its source) appears to indicate the first signs of a division of the Didache into chapters, as is suggested by the signum dispositionis in line 21.
5, 6.