STOP Early Journal Content on JSTOR, Free to Anyone in the World This article is one of nearly 500,000 scholarly works digitized and made freely available to everyone in the world by JSTOR. Known as the Early Journal Content, this set of works include research articles, news, letters, and other writings published in more than 200 of the oldest leading academic journals. The works date from the mid-seventeenth to the early twentieth centuries. We encourage people to read and share the Early Journal Content openly and to tell others that this resource exists. People may post this content online or redistribute in any way for non-commercial purposes. Read more about Early Journal Content at http://about.jstor.org/participate-jstor/individuals/early- journal-content . JSTOR is a digital library of academic journals, books, and primary source objects. JSTOR helps people discover, use, and build upon a wide range of content through a powerful research and teaching platform, and preserves this content for future generations. JSTOR is part of ITHAKA, a not-for-profit organization that also includes Ithaka S+R and Portico. For more information about JSTOR, please contact support@jstor.org. ON THE INTERPOLATION OF CERTAIN HOMERIC FORMULAS By George Melville Bolling In a series of articles 1 I endeavored to show that a high degree of correlation exists between three classes of facts: (1) the significant omissions in the Vulgate papyri; (2) fluctuations in the testimony- given by our manuscripts; (3) positive or negative indications that the lines were not in the edition of Aristarchus — a correlation to be explained only on the assumption that our manuscripts all descend from an edition published in Alexandria about 150 B.C., and agreeing line for line with the edition of Aristarchus. There remained a small residuum of cases which did not show the expected correlation. The existence of such a residuum would have been nothing surprising, and might well have been allowed to stand without explanation. There occurred to me, however, an explanation which seemed both simple and convincing. I stated it, therefore, in a form too brief, as the sequel has shown. Cauer in an otherwise favorable discussion of the first of my articles has rejected, Grundfr. 3 , pages 46-48, this explanation entirely. The importance of the questions involved in the theory as a whole, not a wish to insist upon the correctness of a position I have taken, is the cause of my returning to this matter. In the Vulgate text of the Iliad as established by Wolf, speeches are introduced by certain formulas, among which we may distinguish a "brief type" and a "full type." Most of the examples may be placed under two headings. I. Here in the brief formulas there stands immediately before the speech a statement in which is employed not a simple verb of saying but a verb or phrase which either merely implies speaking, or indicates in addition the tone, the contents, or the purpose of the speech. Typical examples are: A 137 "> 4> 98 afliiXlKTOV 8' OTT a.KOV(rav E 471 ivB av ^apinjSwv /xdXa vciKtcrtv 'Eicropa 8lov 1 Cf. Amer. Journ. Phil., XLII (1921), p. 253 n. [Classical Philology XVII, July, 1922] 213 214 George Melville Bolling Other examples are: T 38 A 241 Z 325 $ 480 (ytiKtat .... eireecrcu, B 224 (vtucte pWic), K 158 (wrpvve re vtiKtak r avr-qv), 9 398 A 185 Q 143 (corpw/e) , M 342 (irpotei KripvKa), K 476 (Aiop.r)Sti 8il%tv), I 224 8t[8iKT 'A\i,\rja.), ^ 734 (aviaraTO /cat KarepVKt), B 55 K 302 (riprvvtro (5ov\r)v), X 430 Q 747 761 (e£ypXt yooio), T 364 $ 272 (a^co^ef), £2 703 (/ca>/cu<7ef re ... . 7e7cofe re), E 358 (iroWa \iacropkvq .... fiTtev), A 130 (7owaf ecrtfyi') , Z 45 (Xa/3coy eXXtcrcreTO yowuv), O 660 (Xto-crero .... yovvovptvos), T 304 ^ 42 (ripvuro), E 463 (/ceXeue). The contrasting full formulas add to such expressions, a verse con- taining an explicit verbum dicendi. Typical examples are: A 336 co 368 toi)s 8e iSoiv vtiKtaatv aval; avbpwv 'Ayapkpvwv /cat atas 4>a>vqoas eVea WTtpoevra irpoo-qv8a'

aro pWov II 553 avrap 'Axatovs topcre MerotTtdSeco ITaTpo/cX??os \axs T ' e7retra ards pecw epne'i Xet/3e 8( olvov ovpavov eiaavi8wv, /cat c^tov^cras 'tiros T]v8a as contrasted with II 231 (vxtT, eirura ara.% peaw ep/cei' Xet|3e 8e olvov ovpavov elaaviScov Ala 8' ov Xd0e Ttptracipavvov. The list may be closed with a passage which, like these last, makes a good transition to the next section: T 255 tvxtro' tol 8' apa 7rdcT€s iir' avTcxfriv ei'aro tnyfj 'Apyeloi /card polpav, 6lkovovt(s fladiKrjos tv^apevos 8' apa uttcv 186iv ets ovpavov tvpvv. 1 Or KtpToiikav ejros r/iia. v.l. of rivks in Sch. T. For similar examples after other verbs, cf. Y 181 B 1S4 * 442 a 193 353. Interpolation of Certain Homeric Formulas 215 II. Too familiar to need exemplification (A 15 f. is the first example) is the fact that a certain amount of material may stand between the formulaic verse and the speech it introduces. If such material is extended, the poet may place before the speech another formulaic line, thus creating another sort of full formula as in: r 386 yprft 8k piv tiKvla iraXaiy tvu irpoaitnrev elpoKopqi, fj ol AaneSaipovi vauTawctj ijcrKeiv etpia /cctXd, paXicrra 8k p.iv (piXkectte. rfj p.iv ktica.p.kvr) irpotre^cocee Si' ' A.^poSiTT)' Classifying the examples according to the first verb I may note after wpoa-ecprj B 795, ayopevuv O 285 2 253, ixpaiveiv p.r\ri.v H 326 I 95, y.vdov 8.KOVOV 496 eXwv eW'urcreTO yovvcov $ 73, odvpero X 81, $nu&i> X 37, eKUKve T 286, ave N 480 $ 330, urpwev N 46 94 P 219 326 585, upae P 74 T 82, KeXeve O 552, opo/cXa Q 252. In S 323 the simile placed thus seems unique, and the Catalogue of the Myrmidons II 168-97 is introduced according to this pattern but on a large scale. 1 When the second formulaic verse is not em- ployed, we have examples of a brief type such as: N 215 tov Se ■Kpoa'&frr) Kpeiwv hvocrixOiov elaapevos (pdoyyijv 'AvSpalpovos vu Ooavri os iracy TlXtvpaivi Kai aiwcivij KaXvS&vi AtrcoXoicrtJ' avaaat, Qtbs 8' &s tUto 8i)p.y Of these it is right to cite only the more noteworthy, E 786 I 253 (cf. Sch. A) M 268 P 555 * 18 ft 724. III. There remain a few examples which do not come under either of the previous headings, 4> 213 434-35 478-80, but which seem unnecessarily full and will be discussed later. From these examples I conclude that both types of formulas are common in the poems, and that in very many of the passages the types might be interchanged without causing any difference that we can appreciate. It is therefore our duty to accept in each case whatever the external evidence indicates as the reading of the arch- type of the manuscripts and of Aristarchus. I have no wish to eliminate the full type entirely. That is the view against which Cauer's argument is chiefly directed and I agree heartily with his 1 1 do not include O 145 because the speech 146-48 is not the one implied in the KaKkuaaTo of line 143. 216 George Melville Bolling conclusion. "Dass es im 2. Jhdt. vor Chr. irgendwo einen Text gegeben habe, der hierin das einheitliche Bild der Reinheit von allem Entbehrlichen geboten hatte ist ganz unwahrscheinlich." But I also believe — and it is here we differ — that there are seven of these verses which on external evidence should be regarded as interpola- tions. Against that position the fact that other verses of this sort are attested in papyri has no force: those that are well attested are genuine, those that are not are interpolated. The MSS. evidence regarding other lines is, when properly interpreted, in favor of my conclusion: it shows that a number of brief formulas have been turned into full formulas by the interpolation of single lines, thus raising a strong presumption that the same thing may have happened in other cases. The clearest examples are I 224 a M 162 a N 218 a X 330 a none of which have been taken into our texts. The last is found in but one manuscript (Ludwich's P) of the fourteenth century; M 162 a (a very thoughtless addition) and I 224 a occur in a few MSS. not earlier than the thirteenth century; and while N 218 a is better attested, the evidence is overstated in Allen's commentary, as a comparison with Ludwich shows. Fortunately, in the last case papyrus evidence is abundant, the line not being found in P. Berol. 46, P. Brit. Mus. 732, P. Morgan, nor in the Syrian palimpsest. M 162 a is also not in P. Morgan, while there is no papyrus available for the other two passages. Next come K 191 P 585 $ 434 480 which are rejected by Bekker 2 (except 434), La Roche, Leaf, Ludwich, Ameis-Hentze- Cauer. They are all omitted 1 by a considerable number of MSS. including A; all except $ 434 by G and T; all except P 585 (for which its testimony is lacking) by S of the tenth or eleventh century. The Syrian palimpsest omits the only line 480 for which we have its testimony, and this line is known to have been absent from the edition of Aristarchus. There is no papyrus evidence for these lines, but the correlation elsewhere observed gives us every reason to concur in the judgment of these previous editors. Here I should place P 219 the first of these disputed lines. It is omitted by S T and six other of Ludwich's MSS., a mark has been placed against it in A 1 I disregard additions by second or third hands; they can easily be found by any who consider them important in this question. Interpolation of Certain Homeric Formulas 217 for reasons that can only be conjectured, and it seems to have been unknown to Eustathius. This in itself is strong evidence, but nevertheless these five editors have all allowed the line to stand. Its omission by P. Berol. 230 must, I think, now turn the scale. The remaining six examples 1 are the ones that do not exhibit the correlation expected, and in this respect $ 73 resembles them- Until the collation of V 32 by Allen it was supported by all MSS. but was known to have been absent from the edition of Aristarchus; there is no papyrus evidence. The line was rejected by Bekker 2 , La Roche in his school edition, 2 Ludwich, and Ameis-Hentze-Cauer; Leaf retained the verse, but was properly troubled by the conflict between the testimony of Aristarchus and of the manuscripts. The other passages are: r 319 : om. P. Berol 263; hab. P. Brit. Mus. 126, 136 et codd. T 389 : om. P. Tebt. 427, P. Oxy. 542; hab. P. Brit. Mus. 126 et codd. A 369 : om. P. Oxy. 753 et A 1 ; hab. P. Brit. Mus. 136, codd cett. N 46 : om. P. Brit. Mus. 732 et F' 1 ; hab. P. Paris, P. Morgan, codd. cett. N 480 : om. P. Brit. Mus. 732, iv iroWois oi ^iperai? Sch. T; hab. P. Morgan, et codd. P 326 : om. P. Berol. 9783; hab. codd. My explanation is that these are interpolations, but interpolations which have been unusually successful — especially if we reckon (and I think we must) with the probability that the isolated variants in the MSS. are merely accidental. 4 This explanation is based on the fact that in some cases, cf. A 196-97, 6 E 420 2 441 X 363 the fluctuation in the MSS. has almost disappeared; in other words some interpolations have been successful enough to spread to most of the manuscripts. I do not see our right to deny the possibility that a few interpolations should be slightly more successful than these. They are of course hard to 1 1 do not consider * 213 because the fluctuation in the MSS. is probably due to haplography; nor I 95 P 74 which are omitted by single manuscripts. There is as yet no papyrus evidence for these lines. 2 La Roche 2 (1878), p. 162, argues from the variants along lines that recall Didymus, ap. Sch. A at T 327 TtKy.i\(iiov Se ttjs SiaaKfmjs to ical krkpoit ep€ 73 by Didymus; note in contrast that for the less successful <£ 480 we can show merely that it was unknown to Aristonicus. The probability of an explanation increases in proportion to the improbability of other hypotheses that can account for the same facts. Of such hypotheses I can suggest but three, and all seem to me untenable. The first is that the omissions are purely accidental. That this may be true of one or two examples cannot be denied, but as an explanation of the group as a whole it attributes too much to chance. The probability cannot be estimated mathe- matically because the chances of omission are not equal for each line in the poem; but I may suggest a rough parallel. Draw 15,693 squares on a sheet of paper, and blacken say 40 of them distributed at random; and then allow various individuals to touch the sheet blindfold. In 100 trials (approximately the number of lines acci- dentally omitted in the papyri) what chance is there that a blackened square will be touched seven times ?' This explanation might in the second place be combined with a suggestion that these lines are 1 About one in fifty million, I am told. Interpolation of Certain Homeric Formulas 219 peculiarly likely to be omitted; but I cannot find any grounds to warrant the assumption of such a peculiarity. Finally it might be assumed that someone had eliminated these lines intentionally. A critic who did so, would not have confined himself either to these particular lines, or to this particular type of supposed interpolation. As we come upon his handiwork here, traces of it should appear elsewhere. Until they are found the cause assumed cannot be regarded as a vera causa. Our Vulgate is the outcome of Alexandrian criticism. As Cauer says, these critics (so far as we can tell) did not in this matter follow consistently any fixed principles. In other words, they were guided by the evidence of their manuscripts, as they understood it — by which restriction I express a suspicion that they were apt to attach too much weight to mere quantity of manuscripts, to follow the majority of the MSS. with some allowance for a difference between codices meliores and codices deteriores. A study of the variants with which they had to deal will show that they had problems of the same sort — cases in which both a full formula and a brief formula had more or less warrant in the same passage. In such cases I believe we must recognize a strong presumption in favor of the brief formula. Aristarchus, as we shall see, decided in favor of that side usually, but not always; the moderns have always followed Aristarchus. That is partly due to a lack of clearness as to what the printed text is supposed to represent; partly it is because Aristarchus is supported by our manuscripts. The latter fact is irrelevant: the manuscripts have shown what lines were in Aristarchus, and can show nothing more. A distinction between the text of 150 and the text of 550 B.C. will avoid the former difficulty — and my use of "interpolation" shall change accordingly from now on. I may note first that my denial of T 389 to Aristarchus, brings his text into agreement with P. Hibeh 20, dated ca. 280-240 B.C., which omits this line. We have no other evidence, and must deny that the line stood in the Attische Ur-exemplar. In B 55 Aristarchus has a brief formula tovs 6 ye avyKaXecras irviavfji> riprvvero ^ovkriv. Someone — the scholia say Zenodotus — had a longer text that brought a verbum dicendi immediately before the speech: aiirap ewel p' ijyepdtv 6nr)yepkes t' eytvovro ToiffL 8' aviaTO' p.ei8r)ar in t 6v6p.a£tv> The example is a good parallel to 434 f. Zenodotus read — athetized is a misinterpretation — in the description of the Trojan bivouac : 489 Tpoioiv avr' ayoprjv Troirjaaro 4>ai8t.p.os "E/crcop voi vtuiv ayayihv 7roTap.4S tin Sii'ijecrt kv Ka.6a.pCji, o6i 5ri venvwv dualvtTO x&P *- 492 e£ lirirojv 5' aTrofiavTts kirl x&bva pvdov clkovov 497 "KtKXvTt p.tv TpcDes ktX." The brief formula is abundantly defended — against Duntzer, 1 Zenod., page 164 — by the examples cited above, especially A 137 «>$ 98 a^eiXiKTw 5' oir' aKovaav. The text could certainly not be ques- tioned, did not Aristarchus read four additional lines: G 493 tov p "Ektoop ayoptve 5ii#iXos' h> 5' apa x« l P' M Z 318 «7X°s ex' ivSeKCL-mrixv Trapoidt 8e Xapirtro dovpos =Z 319 <"XA"? X^X/ceii?, Ttpi Si xpuceos Set w6pKr)s' =Z 320 T& 6 y' ipti.aa.p.a>o% eirea Tpwecci ptTr/vda <*>B 109 i shall mention furthermore only 2 the fact that Aristarchus' atheteses of B 791-95, and <1> 471 change full formulas to brief ones, and there is the possibility that they rest upon MSS. evidence. If so, Aris- tarchus was treating $ 471 precisely as Ludwich treats N 218 a — only he did not have at his control such perfect mechanical devices for the expression of his ideas. The editor who aims at reconstruct- ing the text of 550 b.c. or earlier, should treat 493 ff . in like fashion. For Zenodotus' text must be taken to represent his MSS. 8 and is 1 According to Amois-Hentze, Anhang, p. 103, Duntzer expressed the opposite opinion in Arintarch, p. 95. 2 1 suspect also * 287 which makes Poseidon alone the speaker in contradiction to the closing formula with its ws e'nrbvTe. The only support to this view in the speech is line 290 which was athetized by Aristarchus, while lines 290-92 were athetized by Seleucus and not found in the Cretan edition. I hope to return to this passage in a larger context. 3 Aristonieus' statement to the contrary is known to be a guess, and I think his guesses of this sort can be shown to be worthless; they are certainly malicious. Interpolation of Certain Homeric Formulas 221 consequently the earliest form of the tradition known to us. The motive for the intrusion of the extra lines is clear, and often paral- leled. So also is the practice of using borrowed material. The interpolation is more elaborate than the others but then it is made at a much earlier date; and, if Aristarchus is right, B 791-95 furnishes a good parallel for this. On the other hand had the lines stood in the text from the beginning no one would have been tempted to remove them. That there is a certain cleverness in the interpolation there is no occasion to deny; though Cauer, in the revision of Ameis- Hentze, has found a flaw to indicate. Interpolations, I have assumed, of this sort were made shortly after Aristarchus. We know that they were made still later, and to find them also before his time must, I think, strengthen my assump- tion. I wish to examine rapidly the evidence from the Odyssey also. The examples arranged as before are : I of the brief type, /? 26J y 54 € 444 f 118 t 493 527 ft 183 339 v 127 £ 413 p 239 a 111 v 97 148 X 26 w 530; of the full type /? 324 5 25 ( ?) 769 k 265 430 482 X 56 396 ( ?) v 37 o 45 a 400 v 198 <6 248 361 x 410. 1 II. The more important of the brief type 401 y 42 476 k 225 | 461 ; of the full type 160 f 24 V 158 236 346 k 37 v 167 tt 180 399 a 326 v 375 co 425 453. The latest stratum of interpolations is represented by k 225 a , o 44 a (in a formula that is already of the full type) and a lll a . The last line has been taken into the text by Ludwich and by the Oxford edition but not by Blass, page 179, nor by the Ameis-Hentze-Cauer edition which agrees with Bekker 2 and La Roche. No papyrus covers any of these passages. 2 Next come k 265 430 482 all rejected by Bekker 2 , La Roche, Ludwich, Ameis-Hentze-Cauer, and Blass, Interpolationen, pages 116, 118, 119, because of the fluctuation of the MSS. There is again no papyrus evidence. Finally Aristarchus athetized o 45, cf. Blass, page 159. There are differences between this and the conditions observed in the Iliad — but our material, manuscripts, papyri, and scholia, is also different. The Ohio State University 1 Also SetfiUrKero 6)VT}o>i>7i 227 '. 2 The fifteenth volume of the Oxyrhynchus Papyri has since come to hand, and a Ilia i s not found in P. Oxy. 1820.