Polybiana Moore, John M Greek, Roman and Byzantine Studies; Fall 1971; 12, 3; ProQuest pg. 411 Polybiana John M. Moore THE FOLLOWING DISCUSSION aims to amplify and in some respects to correct, in the light of the evidence obtained from a full collation, the relationships suggested earlier between the various primary MSS of Polybius, Books I-V; MSS eliminated as derived from extant exemplars by the selective collation are disregarded.1 The second and third sections below will list and attempt to explore the information which may be derived from the format and structure of surviving MSS. The following MSS will be discussed: A Vaticanus Gr. 124, X cent. (A.D. 947?) C Monacensis Gr. 157, XV cent. D Monacensis Gr. 388, XV cent. (also Excerpta Antiqua from VI- XVIII) E Parisinus [BN] Gr. 1648, XV cent. Z Vaticanus Gr. 1005, XIV/XV cent. J Vindobonensis Phil. Gr. 59, XV cent. (contains 1.1.1-70.5; V.94.9- 111.10; rest lost) F Vaticanus Urb. Gr. 102, X/XI cent. (Excerpta Antiqua from I-XVIII) G Mediceus Laurentianus Pluto 69.9, XVI cent. (ExcerptaAntiqua from VI-XVIII) From the Constantine Excerpts: M Vaticanus Gr. 73, de Sententiis, X cent. P Turonensis 980, de Virtutibus et Vitiis, X cent. X Ambrosianus Gr. N 135 Sup., de Legationibus Gentium, A.D.1574 1 See J. M. Moore, The Manuscript Tradition of Polybius (Cambridge 1965) [hereafter MTP]; for descriptions of the MSS involved see MTP 10-20,56-60,130-33,140; the analysis which the present discussion supplements and corrects is MTP 22-35, 40-45. I am grateful to Professor D. C. C. Young for detailed discussion of a draft of this paper, and also to Professors B. M. W. Knox and F. W. Walbank and Mr N. G. Wilson for helpful comments and criticisms. 411 412 POLYBIANA I The Tradition of Books I-V The conclusions on Books I-V which were drawn from the selective collation may be schematically summed up in the following stemma; there is no attempt here to indicate dates or intermediate hypar­ chetypes, except that each 'fork' in the stemma implies the existence of a common parent for the branches in question.2 HYPARCHETYPE (Containing I-XVIII at least) ARCHETYPE SOURCE OF THE CONSTA",TINE EXCERPTS HYPARCHETYPE 'BYZANTINE TRADITION' (I-XVIII at least) ~ A F r M P X ~ C Z J D E The division of the traditIon of Books I-V into branches has been amply confirmed; from nearly 300 errors noted in A and not found in the v ACDE: KaT' aUTOv ZJ There are a considerable number of passages where ZJ have readings not found in ACDE, the majority of which appear to be errors rather than corrections; they show that in Book 1 Z and J are derived from a common parent. This conclusion was also reached by A. Diaz Tejera,6 but needs reinforcement since the key section of his proof of the relationship is based on five passages, two of which are classic omissions ex homoeoteleuto and cannot therefore be used as conjunctive errors (I. 5 .5; 20.14), and two of which are omissions of the article-one (I.46.5) shared by DE-evidence on which 1 would hesitate to base a case; the fifth is good evidence, the convincing omission of cVVaActAa~ctv at 1.34.2 (not 34.3), for which corrected readings of AD E are cited above. The remainder ofDiaz Tejera's argument definitely tends to support a common parent (e.g. the confusion in ZJ at I.5.2-3), but he also relies in part on a number of good readings; they can never be used in this sort of demonstration, and cannot on the basis of his own stemma7 prove "la conservaci6n de la tradici6n en contraste con la alteraci6n de los demas manuscritos"8-his stemma shows conclusively that they must be conjectural corrections. Diaz Tejera then lists individual peculiarities of Z and J to demon­ strate that they are gemelli, not derived one from the other. of the passages he cites for Z, f30~(J1JCLV for f301J(J~C€W (I. 10.2) is insignificant as a very frequent type of iotacism; the change of order at I.24.2 (not 3) 1'\ " C " 1'\ d . d· 1 b € LI\OV KaTct KpaTOC lor KctTct KpaTOC € LI\OV was correcte Imme late y y the first hand in Z by adding f3 and ct over the relevant words, and at 1.28.11 Z reads 1TaActL o/Laxovc read ypoct/>op.axovc; C has ypoct/>o/La~ovc, which implies a dittography of -/Laxovc (lin.) and -/LaBovc (sup.lin.), not -/LaxBovc. 1.45.7, EfJo~Bovv AZJDE: 'TTap€fJo~eovv Arc only (presumably contaminated into Ar from C; MTP 41ff). 1.58.2, Z reads rijc pl,.,.,c after correction, apparently from rqv pl,.,.,v. 1.59.12, Z 422 POLYBIANA in A.27 It has been plausibly argued not only that A tried to represent the exemplar he was copying with great care (a supposition which is amply confirmed by his practice passim) but also that the line length of A was the line length of the common ancestor of A and the alpEc8aL Z; £1TLt/>.rpEc8aL was clearly in the common parent ofCZJDE. 1.52.5, cLTapKlac is also in E. 1.54.8, P.1)SEv is the unanimous reading of ACZJDE; the variants cited by Diaz Tejera are variants of p.T}S£ three words earlier, where, for what it is worth, ACZJ all have p.r, S~, a very common accentuation of the word, particularly in AZ. Diaz Tejera 144: 1.58.8,J reads Kai paSlwc as Z.1.36.1,J has a1TCxVTwv; the "autor desconocido" of the correction mentioned is presumably Angelus Vergetius, who copied the MS; MTP 20f, 38. 1.37.4, 1TA&YLOV is in Z as well as J. Diaz Tejera 145--46: 1.50.7, 1Tapayy"o..aL ZJDE: 1TaPWY"LA£ CApe (probably A2); there is Virtually no doubt that the reading of the first hand in A was 1Tapayy"o..aL; therefore 1Tapayy£o..aL was the reading of the archetype, 1TapWYELAE is a correction, and 1TapayydAac has no MS support at all. Diaz Tejera 146-47: 1.63.3, iJ p.£'iovACZDE: ov p.£lw J The MS authority makes it clear that ],s reading cannot be anything but a conjecture, and the conjecture is not supported by the second statement of the terms of the treaty at III.27.2-9, where the period for payment is given (27.5) as ten years-the halved period correctly restored by Scaliger's emendation at 1.63.3. The two accounts of the treaty are not totally identical, but to suggest a period of 20 years in the first account and 10 in the second is to introduce an inconsistency of a totally different order from all the others and is not a tenable emendation, particularly when such a convincing alternative exists. On the details of the treaty see F. W. Walbank, A Historical Commentary on Polybius 1 (Oxford 1957) ad III. 27. 11 See J. Schweighauser, ed. Polybii historiarum quidquid superest V (Leipzig 1792) 124; F. O. Hultsch, Quaestiones Polybianae II (Progr. Dresden 1869) 3; MTP 173f and plates 2-4. 28 See the stemm a above, p.412; if AMF are affected, the lacunae must have been in the common source of all three major branches of the tradition. II Most recently by Diaz Tejera 125. JOHN M. MOORE 423 One can support this, and suggest that A was a direct transliteration of this uncial codex in the light of the following readings: III.61.4 a€!, AIPC: 8€t Aac (misreading of AILl) II1.98.9 7TO'\€LC Alpc : 7T()8. C A ac (AILI) V.14.12 aUa AIPC: a/Lct Aac (AAjM) V.67.6 KT~C€LC AIPC KV~C€LC A ac (T/Y) There are numerous errors shared by A and the 'Byzantine Tradition' which apparently sprang from incorrect readings of uncial letters, but this is to be expected in any tradition; the four cases cited, however, show the scribe of A making a mistake of transliteration, and then noticing and correcting it. If A was a direct transliteration of (0, what of vAaKTOl)JL€VOV edd.: &cpvAaKTov JL/.vov FD G VIII.5(7).2 €JLf3€A€c Suda and edd.; €JLJL€A€C FD G There are other passages where mistakes may have arisen from mis­ readings of minuscule letters, but they are more dubious. Since the mistake~ quoted were in the common parent of the Excerpta Antiqua, it appears that in Books VII and VIII this common parent was copied from a minuscule manuscript. It seems a reasonable hypothesis that the Excerpta Antiqua were selected from a single set of volumes, and the evidence of F in Books I-V shows that the excerpts from these books were taken from fP, a copy intermediate betweenW and r; the Excerpta Antiqua would then have been selected from fP and com­ panion volumes now lost. The stemma position and the evidence suggesting that fP was a minuscule MS combine to suggest that the Excerpta Antiqua were selected in the IX or X century. Excerpts were very popular at the time, witnessed most dramatically by the 53 titles of the Excerpta Historica made by order of Constantine VII Porphyrogenitus and usually referred to as the Constantine Excerpts. If the project resulting in the Excerpta Antiqua was to make a selection from eighteen or more books of Polybius, and there is evidence that some of the text was selected from a minuscule exemplar, it is likely that a complete minuscule transcription (or copy of a transcription) was used as the source, particularly at a time when so many old 81 Book XVII was probably missing by the X century; see infra. p.432. 3S In this section readings ofF are quoted from my own collation, those of the remaining manuscripts of the Excerpta Antiqua from the apparatus of Hultsch's edition except where otherwise stated; the latter readings should therefore be treated with a certain degree of caution. It may reasonably be hoped that further evidence will emerge from a full collation of these manuscripts. The agreement of FDG shows the reading of the hyparchetype of the Excerpta Antiqua; cf. stemma, MTP 171. JOHN M. MOORE 425 codices were being sought out and transcribed because of the re­ awakened interest in classical authors.33 The evidence of errors, then, which shows decisively that A and (p were gemelli, combined with the signs that A was a direct trans­ literation from W, entails the conclusion that there were two minus­ cule transliterations made from W in the IX or X century, A of books I-V, and (p with lost companion volumes containing Books I-XVIII at least. A transliteration of the whole of the surviving text of Polybius must remain for the moment strictly unproven, but is likely to have existed. It is more than probable from evidence preserved on the rare occasions where two titles overlap that the compilers of the surviving titles of the Constantine Excerpts all derived their text from the same branch of the tradition.34 Evidence in Book XIV suggests that their text was more complete than that used by the Excerpta Antiqua,35 which implies a source different from (p at least in Book XIV, and probably therefore different throughout. This conclusion is supported by errors found in ADE and F which are not shared by M or P.36 It is more probable that the compilers of the Constantine Excerpts (whose surviving titles contain material from all but five of the original forty books of Polybius) would have used minuscule trans­ literations rather than uncial MSS, expecially in view of the suggestion that the compilers 'marked up' a copy of the work to be excerpted and passed it to the scribes to have the selected sections copied in accordance with the instructions written in the margins. Accuracy would be more likely to emerge from using a minuscule 'master copy', and it seems intrinsically more likely that they would have used such a copy than a probably ancient uncial codex. I have found no convincing uncial transliteration or minuscule errors in the mistakes of the Polybian text in the Constantine Excerpts from Books I-V. The only passage worth citing is: IV.30.2 CVV€7T€KVpWCav ACZ: CVV€1T7}vpwcav all MSS of Excerpt a de Legationibus Romanorum This appears to be based on a clear example of misreading of early 33 On the whole question of transliteration from uncial to minuscule see A. Dain, Les Manuscrits 2 (Paris 1964) 126-45. 3( For such evidence as exists of common error in F and the various titles of the Con­ stantine Excerpts, see MTP 172f, and P. Pedech, Polybe XII (Paris 1961) xl f and xlvi n.lo 35 See infra p.433. 36 MTP 22. 426 POLYBIANA minuscule; all these manuscripts, however, are derived from a copy made from the lost MS I e 4 of the Escorial Library by Darmarius in the XVI century; his first copy is also lost, and we cannot be sure that the minuscule error was in I e 4.37 Scanty as is our evidence for anything which lies behind A, F and the extant manuscripts of the Constantine Excerpts, it is possible to make a few tentative observations. First, there is no indication in A which could lead one to suppose that it was the first volume of a set; there is no note such as T(fJ}LOC a', and Ephraim signs the subscription at the end with every appearance of having completed his task. This implies that Books I-V comprised the first volume of Wand that any indication that there were more to come was either missing or omitted by Ephraim. The existence of further volumes is proved by the existence of F and its position in the stemma. F itself is more informative than A. The titles of individual books usually consist of the following formula with the appropriate numeral inserted: 7ToAvf3lov ** 'TOU .J Abyov;38 7ToAvf3lov is omitted for Books III and VI, the erasure in the title ofII is approximately 7 letters long (EK has normally been restored, but is clearly too short-perhaps lc'Topu;)v?). The titles ofI and XIII are missing, the former lost with the initial few folia,39 the latter apparently missing in the exemplar from which F was copied. The title of XIV, where all excerpts from the book are copied by p2, is in the margin and abbreviated. In addition to these titles, there is a large B in the margin in the first hand at the begin­ nings of Books VI, XI, XV and XVJ.4° The 13 at VI is followed by an erasure large enough for an uncial letter, in which the traces of jj may probably be seen; that at XI has a sign after it, ,.." which may have an erasure under it; there is certainly an erasure after the B of XV and XVI, and the second has a bar over it-the normal way F indicated a numeral; compare the titles for the individual books above. It is perhaps not implausible to suggest that the iJ in the margin in 17 MTP 152-61. 38 In the erasure (**) restore JK; one can only surmise that the JK was erased at some time with the purpose of representing the text of F as complete. 3t A recent hand in F has added the title -rijc '7To},v{3lov J'7Tt'TOf.Lfjc }'6yoc iioc; there is no means of knowing now much authority (if any) this title should be allowed. 410 In discussing excerpt MSS "beginning" and "end" should be read as referring to the beginning and end of the passages selected from a book for inclusion in a particular set of excerpts. JOHN M. MOORE 427 F, perhaps short for fitfi).,{ov or filfi).,oc, indicates the start of a new volume in the exemplar from which the text was taken, particularly in view of the traces of B, which would imply the equivalent of 'Volume 2', at the beginning of the excerpts from VI. The normal Byzantine word for a volume was T€VXOC, while {3t{3Mov referred to the whole work, but it is hard to see any other interpretation for the B. These possible indications of volumes can hardly apply to F itself or to the common parent of the Excerpta Antiqua since the resultant volumes would be too short. Therefore they were presumably noted by the common parent from its source, and copied by F. The titles of Books XV and XVI pose a problem; either one 13* must be wrong-a volume consisting of only one book is almost unthinkable-or the original excerptor switched sources from one exemplar which started a volume at XV to another which started at XVI. The latter hypothesis is implausible in that there is nothing to indicate such a change of source, and it therefore seems likely that the second entry is an error. It may have arisen because the 13* was correctly placed at the beginning of XV, and the scribe copying F or an intermediate exemplar remembered the entry when he got to the beginning of the next book and wrongly added it to the regular title, thinking it had been omitted by mistake in his exemplar. There is a third remote possibility. There is evidence of damage in the manuscript from which the Excerpta Antiqua derived their text in XII, XIII and XIV; con­ ceivably, Book XV had become detached from the volume of which it originally formed a part, and so constituted a separate 'volume' at the time when the Excerpta Antiqua were selected. Whatever the truth on this detailed point, such signs as there are suggest that Polybius Books I-XVIII formed four volumes in cP in the IX or X century; the volumes contained Books I-V, VI-X, XI-XIV and XV-XVIII. The suggestion that the second pair of volumes only con­ tained four books each, while the first pair contained five each, need not undermine the hypothesis; even assuming the volumes to have been of approximately the same physical size, we do not know the extent of the original text of any of the fragmentary books, and Polybius would not have been the first writer whose books became longer as he got further into his subject. It does seem neces­ sary to assume that a volume ended with Book XVIII. As in A, so there is no indication in F or any other surviving MS of the Excerpta Antiqua 428 POLYBIANA that it is the first volume of a set; one must therefore tentatively con­ clude that the excerpts ceased at the end of XVIII. No satisfactory hypothesis has so far been advanced as to why they should have ended at this point, but consideration of content suggests the following as a possibility. By the end of XVIII the excerpts had not merely covered the history to the end of the Second Punic War, but had also dealt with the conflict with Philip V which was an almost inevitable result of relations between Rome and Philip during the Second Punic War, and had brought the narrative down to a suitable climax with the battle of Cynoscephalae and the Isthmus declaration of 196 B.C. The presence of some material dealing with other areas would be natural, granted Polybius' method of writing history. This does not, of course, rule out the idea that the compilers may either have projected a second volume to include the rest of the period covered by Polybius but never completed it, or conceivably that the selection was made but lost, though the latter suggestion is inherently much less likely. Something more may be deduced from information preserved in the surviving titles of the Constantine Excerpts. Polybius is unusual amongst the authors included there in that some excerpts are pre­ ceded by the number of the book from which they come, while occasionally there is a note marking the end of a book. This is not the normal practice of the copyist of P (to take one example), who only has books of a work numbered eight times in the excerpts from Josephus and once in Appian; there is nothing to help the identifica­ tion of the individual books from which extracts come in Thucydides, an example where the relative brevity of the excerpts might have led to more precise identification of sources than elsewhere, nor yet in Dio, where the massive amount of text excerpted might equally have made the compilers feel that some form of reference would be useful. On the other hand, the notes of book numbers are scattered throughout Polybius in P and in both titles of the de Legationibus: they are missing from M, but the present state of that codex makes it impossible to draw any sort of conclusion from this. The excerpts preserved from Polybius under other titles are so brief as to make their information valueless for the present inquiry. P marks both the end of Book X and the beginning of XI, the beginning and end of XX, and the beginnings of XXVII, XXVIII and XXXI, and the end of the Polybius excerpts (dAOC rijc tCToptac 7TOAVf3Lov p,eyaAo7ToAtTOV) at the end of XXXIX. Some or all of the extant copies of JOHN M. MOORE 429 the de Legationibus Romanorum note the following: ta' (XI) wrongly for LE' at the beginning of XV, the beginning of XXI and XXVII, the end of XXX (the note standing in fact at the end of XXIX, but no excerpts are included from XXX), and the beginning of XXXVIII. The manuscript X, which is Darmarius' copy of the lost Escorial I e 4 and the parent of all other extant copies of the de Legationibus Gentium, marks the beginning of XXVII (minutely misplaced, though clearly not a significant error), and the beginning of XXXIII and XXXVI. The only information in M is in the final subscription to the Polybius section which states that the scribe has reached the end of Book XXXIX, with the comment '~TH TOV M Aoyav. Therefore XL may well have been lost at the time the note was written, though it could merely mean £ (3/4 A; 6/7 DE; 9 Z) Kat, (om. AZDE W F, de Leg.Rom.) II.4B.3 S£E (5 AE; BD; lOZ) S,on AZDE W III. 101.2 7T£P' T6V "Avvtf3av DE; pro quo ? space in KapX'Y)Sovtovc CZ; om. A (space A W, alterna- 14) tive marg- inalia inW, ~orr III. 101.2 aKpwp£{Wv om. A (space B) A W III. 101. 10 1Tpo£J£CLv P.'Y)TE -r1jv (2/3 A) Adav (P.'Y)TE add. Al or A2 in marg.) A W III. 107. 10 7TpOXHpO (3) A (7TpOX£tpOVCL T6 SE cTpaTo7T£SoV corr. cz)a A W IV.45.2 T' £Zd(v) om. A (space 5) A Wb IV.B1.7 nvwv Kat (6 AZDE) 7TWV AZDE W V.14.12 7TpOXWPOVVTWV (1 line = ca. 20 A; 20 E) (a£Jvp.ovVT£C P.EV add. A 4E in space) AE we V.15.6 a7T'Y)AAaTT£To ( II A; 12 E) xtac ( a7T'Y)AAaTT£To Tijc , KaKovx,ac CZDA4EPC) AE WC V.15.B ovx ofov (13/14 A; 3 Z; 11 D; B E) Ep.€Tp'Y)cav AZDE W V.50.6 lp.anov (B A) A Wd a If the supplement in CZ is correct, it is noteworthy that the length is 26 letters, which could be 1 line + 3 letters in A, and A's 3-letter space is at the end of a line. One may logically suspect that the three spaces at I1I.lOL2 and 10, and also the three at V.14.12-I5.8, have linked mechanical causes, but the disposition of the lines on the pages of A does not suggest a deduction similar to that which can be made for 1.2.7-8+1.3.3. b Either U) was legible when lP was copied, but not for A, or lP corrected. C CZD adopt a supplement from r which may also have been in lP and perhaps U). d Something must be missing here, but only A actually has a space. JOHN M. MOORE TABLE I-continued Text Reference (approx. number of letter spaces indicated hy figures in parentheses) Mss Affected 435 Gap /Damage/ Illegibility Deduced in It is only to be expected that examples of actual gaps left in the text at the time of selection will be rare in excerpt MSS since the compiler would have avoided defective text as far as possible and also have made obvious corrections; nonetheless, the following spaces left in excerpt texts must be presumed to go back to the common parent of the Excerpta Antiqua and the source of M respectively: XVIII.l8.7 Kat Tau'rac (13 F; 7 D; 6 G) EXOVT€C FDGe Aaf-L/3avoVTaL (9 FD; 4 G) OUK Hyparch. XVIII.20.5 'Ep€Tptav rilc c/> (10 F; 6 D; 5 G) FDG of Exe. C 0 SE ([JlAt7T1TOC Antiqua XVIII.26.8 TatC f-LEv (17 F; 14 D; 7 G) E1T€CTL FDG XXIV.12(14).1 7TOL€tV (36 M) 0 ([JLA01TOLf-L"f)V--rOA- (XXV.9a.1) f-Lfi. (24 M) KaLpOtC Me Source of XXIV.13(15).2 T€M.WC (10 M) KaV , , Constant-€1TLTaTTWCW (8 M) Kat TOV-rWV M ine Ex- XXXVI.9.5 cP&'CKOVT€C ou Tijc (11 M) TOVC T~V cerpts (XXXVII. 1.5) , M 1TpOaLp€CLV e Readings of F D G my own from microfilm; those of M taken from Excerpta Historica ... IV, Excerpta de Sententiis, ed. U. P. Boissevain (Berlin, 1906). From these pieces of evidence a picture emerges. The text of Polybius appears to have survived into the IX or X century fairly complete apart from the probable loss of most or all of five books available in antiquity. It is more than probable that only one copy survived, since wherever we can check any two of the three main sources of our text against each other there is every indication that they had a single ultimate source, and from I.2.7-8 and I.3.3 it appears to have been the uncial codex from which A was a direct transliteration. There is ample evidence of damage or illegibility, which argues that the codex was not produced after the end of the Iconoclast controversy in the revival of Hellenism; this would hardly have left time for the volumes to become as decrepit as the evidence suggests they were. We are therefore led to infer that this one vehicle for the survival of 436 POLYBIANA Polybius was probably copied some time between the II century, the earliest plausible date at which at one may posit the use of the codex rather than the papyrus roll as the format, and the VII century, when the Byzantine ..Ad. - EKE{vwv ex hom. Text pre­ served in no Athen. 14.22 no A2C no Arc no A no A yes A yes A, de Leg. Rom. yes AF no A yes AF no ACZDE yes ACZJDE yes ACDE no ADE 21 22 22 24 27 56 39 36 38 117 31 23 23 34 a The column headed F; €lc a,cp.vptovc is repeated from below, the interval being 64 letters. C The omitted portion marked by brackets. Number Reference JOHN M. MOORE TABLE II-continued Text omitted ex hom. Text omitted by Z (added by Zl in margin) Text pre­ served in 439 Length 3 IIl.l17.6 HJ[3YJp€C OE - OUX.KOcLOVC yes ACDE 64 vi Text omittedd by J 1 1.15.9 ~ 'TaC Imo(NcEtc -l/J€VO€'ic yes ACZDE 34 V11 Text omittede by C 1 I.l9.5-20.4 Ka'Tcx)\fx[30f1-€VO, - XEtpt~€LV 'TO: yes AZJDE AZJDE 2 1.41.1 wc bTl, - K€Kpa'TT)KO'TaC no 3 II 20 5 • ll' <;-> \'p .. YJ'T'TYJU€V'T€C 0 - 7TpOC w- 4 III.55.9 5 III.103.S 6 IV.47.7 7 IV.7S.7 8 V.62.3 9 V.82.7-S fLaLOVC o,a 'TO - X€'fLWVOC lli~AWV - C'TaOLOVC 7TapaKaAOVV'T€C - llpovclav 7Tan - 7Tpoc[3aLIIELV '.. , €V O'C - 'T€'TPYJpovc €r€'TaK'TO - 7TPO€C'TT)C€ (added by Cl in margin) viii Text omitted! by DE 1 1.50.2 'TWV OE - oveac 2 I.60.4 'TWV 7T€pt - E7T{voLav 3 11.5.4 Eg€j3T)eav - 7TO'\€WC 4 H.17.10 5 II.48.4-5 6 11.49.10 7 11.56.10 8 m.11.8 9 III.22.11 E'T£ OE - aCK€'iv Oi'TLV€C - M€ya'\o7To'\{'TaLC OfLo{wC - lJ7To'8€tg€LV Aoyovc - tmOK€'f1-€VOLC f1-EV av 'TL - 7TLC'T€U€LV ) AV'TLa'TWV - T appaKLVL'TWV yes no no? yes yes no no AZDE AZDE AZDE AZDE AZDE AZDE AZDE no? ACZJ yes? ACZJ yes ACZ yes ACZ yes ACZ no? ACZ yes ACZ no ACZ yes ACZ, de Leg. Rom. 2059 53 115 49 32 55 24 28 132 46 44 54 51 101 51 43 55 39g d Note also: I.34.7, after 7TCXV7'cxX68€v, 157TO 'TWV ;:TT7T€WV - 'TOVTOVe (1.34.6) repeated in J; passage length 34, interval 227 letters. I.56.9, after 1TOA€/.dove a garbled version of I.56.6 is inserted; length of passage 47, interval 293 letters. e Note also: I.72.3, oU'Tove 1Tpawc - Xpwjdvovc twice in C; length of repeat 42. V.65.6, 1TMi8oc omitted after i1T7rEWV, 'TO 7TA~8oc added after oVTac; interval between 47. V.50.9, .plAove ('Toli {1anMwe> 8u¥ 'TOV C; the addition apparently comes from the line above; interval 39. r Note also: V.66.S, a1TOAt1TWV ¢e cpvAaKcxe repeated from above after £€A€VKdcxv; interval length 120. g.Length adjusted to take account of errors in all MSS. 440 POLYBIANA TABLE II-continued Number Reference Text omitted ex Text pre- Length hom. served in lO III. 24. 16 • Ap(5£aTac - TappaKLv{Tac no ACZ 38 11 III.30.1-2 7TPOT£POV - ij37] yes ACZ 230 12 III.33.15-16 A£pY1}'TWV - 7T~Kov-ra yes ACZ 157g 13 II1.39.6 E7T~ I-'~v - TptCXL)..{ovc yes ACZ 54 14 m.48.6 ov(5~ 3lc - 7T(f'\aL no ACZ 41 15 111.102.11 o[ 8£ - (JappMewT£pov yes ACZ 32 16 IV.23.1 Taiha 8~ - are independently derived from W, it is possible either that both here independently omitted a line of W or that the line was already omitted in W; on the generally accepted hypothesis that omissions which are not ex hom. are unlikely to occur twice, the latter is the more plausible suggestion, and entails the further tenta­ tive hypothesis that the parent of W, which may be called the arche­ type, had the same line length as Wand A. This is not a difficult hypothesis, since something close to the line length of A and the format of two columns to a page are normal for uncial codices. The omission at IV. 12. 11 by FCZDE (TABLE II.ii.I) is confirmatory evidence for the line length ofW, since it implies an omission of a line ofw by the scribe ofc]>. The omission by CZJDE (F missing) (TABLE II.iii.I) is dubious; it may be that, with an abbreviation for Kal, it could have been a line of 25 letters, and even shorter with suspensions for -wv, but one cannot be sure; the omission must have occurred at the latest in the copying of r from CP, and since the length is very close to that of W but not to that which will be suggested below for C]>, it is more likely to go back to the copying of cP from W. In the light of this, one must look more carefully at TABLE II.i.2 and 3; both these passages have been condemned as spurious. The first, however, is added in the margin by A2, the hand which apparently went through A correcting Ephraim's copy against W, as well as adding a few easy corrections of his own.53 Text which is found in C and A 2 was presumably in the margin of W, and omitted by the other MSS. The hand of the second passage is more difficult to determine since the phrase is inserted in a narrow space between the lines; it is normally ascribed to the hand of Ar (and would hence be an example of contamination from C to Ar),54 but may in fact have been added by A2. It is remarkable that both passages are the same length, and that this is the line length of A and W. From this it does not necessarily follow that they are not spurious additions to the text, perhaps added at a very early date, but it is surprising that the putative corrector twice matched the line length so accurately. 63 The readings of A 2 were very fully analysed by Th. Biittner-Wobst, ed. Polybii Historiae (BT, Leipzig 1882-89) I.vi ff and II.ix ff; cf. MTP 40f. A2 was not in the habit of adding significant phrases when emending, as opposed to correcting copying mistakes or omissions. 6& MTP 41ff. JOHN M. MOOR£. 441 Two further passages need notice at this point. TABLE II.iv.2 is an omission strongly ex hom., omitted only in F. It will be shown below that the line length of cP was almost certainly 31+, and so one cannot posit the omission of a line of cP as the cause; it is possible that cP omitted a line of cJ.) ex hom., and then added the omitted text in the lllargin, r restored the text to the right position (hence its inclusion in eZJDE), but the compiler of the Excerpta Antiqua or F did not, and also omitted it in the margin. Secondly, there is the omission of ci,ua cvccc/J'EtV JLa,uov by Fez at V.II.5, which is 19letters, and exactly one line long, though not a complete line, in A. This has normally been taken as a case of coincidental omission in two manuscripts, F and the common parent of ez, since it is hardly possible to support a com­ mon origin for FeZ independent of the source of J DE. This explana­ tion may well be right, but it is within the bounds of possibility that we have here an example similar to TABLE II.iv.2, and that only the common parent of DE restored the marginal text to its proper position. This would entail the supposition that these three words were in the margin in cP and r. Conceivably the omission at III. 107. 10 should also be considered here, but it is very doubtful evidence; see supra p.434, TABLE I with note a. The evidence supporting the thesis that A reproduces the line length of cJ.) is widespread in the tradition and cogent. Turning to a later stage of the tradition, TABLE II.iv.1 is very impor­ tant because it is not an omission ex hom., and is all the more significant since it starts in the middle of a word; it looks like the omission of a line of cP by the compiler of the Excerpta Antiqua, or of a line of the hyparchetype of the excerpts by F. Further evidence comes from TABLE lLiii.3, 4, 5 and 6 (reading 6 as 3x39); example 5 is not an omission ex hom., and was presumably therefore omitted in the common parent of eZDE, and the other omissions would have occurred there if they did not happen more than once independently. The presence of the text of omissions 4 and 6 in F shows its presence in CP. The repetition cited in note b to TABLE II.iv is also relevant here; it presumably occurred because cpaAayyoc Elc Otq..wp{ovc is the correct text at the second point; if the line length of cP were a bout 32, the two occurrences of cpaAayyoc would have been directly above each other, two lines apart, which would greatly facilitate the error. Thus we arrive at a hypothetical line length of 31-39 for CPo Confirmation of the line length may be seen in the following 444 POLYBIANA passages from the Excerpta Antiqua:55 (1) VI1.I2(11).5 TOU T67TOV­ KpaTf£Lvom. FD; length 31 letters. (2) At VI.23.11 ~ has been omitted by all manuscripts:after 7Tp6TEPOV, and wrongly added after avaxaAacOij­ vat. The interval is 35 letters, and it looks as if ~ slipped down one line. The omission in TABLE II.iii.2 is a very strong case of homoeotes (TptaKOclovc precedes the passage omitted). This may be the sole reason for the omission, which could also easily have occurred on a number of occasions independently. If the MSS had the letter T' for the numeral, however, the length of the omission becomes 46 letters, and could therefore be a case of the omission of two lines of W by q,. Letters for numerals are not found in A, which perhaps makes it less likely that they were used in W; but it is not impossible that they were, and were eliminated in A in the course of transliteration. For what the evidence is worth, a number of Polybian MSS, including P and those of the de Legationibus, use letters for numerals from time to time. The length of the omission cannot in any event be connected with the suggested line length of q,. Turning to TABLE I1.v and vi, we can rule out immediately V.l. It is an extreme case of homoeotes, and its length is only reconcilable with Wand A; the established stemma, however, rules out any direct connection between ZJ and either manuscript. The omission of 34 letters not ex hom. by the common parent of C Z (v.2) looks important, and fits remarkably well with the evidence from J, which is inde­ pendently descended from r; J has an omission of 34 (vi.I), and the repeated passage at 1.34.7 (vi, note d) is also of 34 letters; the interval between the two occurrences of this passage is too long to be useful evidence, but is not inconsistent with a line length in the thirties-it would probably imply an interval of seven lines. The second inter­ polation in J is included in the table for completeness, but is not usable eVidence since the text is garbled, and the interval too long to be useful. This leaves v.3, where the first hand in Z has copied an omitted passage in the margin; at this point Z is clearly derived from r via an intermediate exemplar which was the common parent of Z 55 The argument assumes that the format of tlJ remained standard throughout. as is likely if it were produced as a single copy of the books in question. The readings ofD given here have all been checked on microfilm; in fact. Hultsch says that the omissions are shared by all the manuscripts of the Excerpta Antiqua. though this is in a sense immaterial in the present discussion. since the stemma shows that an omission in FD must spring from an omission in either the hyparchetype of the Excerpta Antiqua or an intermediate copy made before the tradition split into its surviving branches. JOHN M. MOORE 445 and C. It may be that Z made a slip and corrected it, or that about 32 or 64 was the line length of the common parent of CZ. On the other hand, Z appears to be careful to preserve the marginalia in the exemplar he is copying, and at one point preserves three consecutive marginal glosses with different symbols to mark their proper place in the text; they are now on different pages, but the distinction of symbol argues that they were all originally on facing pages at least, if not on the same page. If Z was as meticulous as this, it is possible to suggest cautiously that this omission, which fits so well as two lines of r, was perhaps a mistake made by the common parent of CZ and rectified in the margin, and that C restored the passage to the main body of the text while Z copied it where he found it. So one may reach the preliminary conclusion that the line length of r was in the region of 32-34 letters. Leaving aside C for the moment, DE furnish by far the fullest body of evidence (TABLE ILviii). Where so many passages are omitted, the possibility of the same error being made independently more than once must always be borne in mind, but a significant number are not omissions ex hom. Those marked '?' are cases where the similarity of letter groups is only a matter of a couple of letters or so, and where suspensions might well have virtually eliminated it. The first of two startling groups of omissions in DE is the sequence 16, 17 and 18 in TABLE ILviii, where two omissions of 92 are followed by one (which is not ex hom.) of 46; this suggests a line length of 46± . In fact, if one takes a range of length of 44-55, which is wide but not unreasonable in view both of the increasing use of suspension as minuscule developed and of the fact that omissions ex hom. can easily arise if the similar letter groups fall approximately one line apart (absolute precision is not essential), half of all the examples in TABLE ILviii fall in the range. They are: 1-6 (taking 5 as two lines), 8, 13, 16-18, and the doubtful cases 11 and 12, which must surely be a multiple of lines but are both ex hom. and long enough to fit many possible patterns; in addition, 7 is only one letter away from fitting. Of these examples, two are definitely not ex hom. and two more probably not. On the other hand, almost equally striking is the sequence of 21-25, of which 23 and 24 are not ex hom.; they suggest a line length of 40± . A bracket of 38-42 includes examples 9, 10, 14, 19 (3 x 40), 21-25 (22 and 23::!:: 2x40), and 7 is only one outside this group also. Of these passages, five are not ex hom. Since the body of both 446 POLYBIANA groups falls towards the middle of the length bracket, it seems difficult to reconcile the evidence and suggest that a single line length lies behind these omissions; it seems more likely that we are dealing with two lost exemplars, one with an approximate line length of 44-55, the other of 38-42. The omissions found in DE would then have been made in the course of copying from these two exemplars. It was argued above on the basis of the evidence of CZ and J that the possible line length of r was 32-34 letters. This fits remarkably with the two passages of DE not so far discussed (TABLE II. viii.l5 and 20), both, granted, omissions ex hom., but of 32 and 33 letters respec­ tively. Since the common parent of DE eLl) apparently contained a large number of mistakes of omission, it would be surprising, though possible, to suggest that the line of descent was of three stages: r to a MS with a line length of 38-42 letters, and from there to 0 with a line length of 44-55, which would account for the common omissions of DE (i.e. those which were in their common parent Ll). It is surely more plausible to suggest that the line length of r was of the range 32-42, and thus account for the omissions in C, Z and J, and the group of shorter omissions in DE, in addition to TABLE II.viii.l5 and 20, by one hypothesis; 0 would then have made the omissions of 38-42 (better supported by examples which are not ex hom.) in copying r, and Ll those of 44-55, the line length of o. The position of Z after III.62.2 proves conclusively the existence of an intermediate stage between rand C; the omissions of C (TABLE II.vii) are of widely varying lengths and provide little useful informa­ tion. Example 1 must presumably be of one or two pages since it is an omission ex hom. of considerable size,56 but the evidence of all the other passages is inconclusive because there is such a degree of editing behind the present state of the text of C that any omission may have been deliberate; the passage at V.65.6 (TABLE Il.vii, note e) is a good case, where the change may have been a slip, but is more likely to have been a deliberate correction. 57 For what it is worth, the repetition of 56 For a parallel omission of one page see MTP 62. 51 Similarly, because of the degree of editing involved in making a set of excerpts, and the demonstrable fact that the excerptors were willing to modify what Polybius wrote, the MSS of the Constantine Excerpts have not been included in this analysis; all their omissions of more than an odd word in the first five books appear to be complete sense units, and could well have been left out deliberately; therefore, however well they fit any theory, they cannot be used as evidence. JOHN M. MOORE 447 1.72.3 suggests a line length of 42±, which is close to TABLE II. viiA (not an omission ex hom.); possibly the common parent of CZ had a line length in the region of 42-49. The omissions of D fall within the range 87-95, and the repetitions are at intervals of 94 and 98 letters (TABLE Il.ix); this suggests a com­ mon line length behind them. The evidence for B (TABLE II.x) is different. There is only one omission, and E added the omitted text in the margin. The other passages all consist of instances where E has a number of words copied by the first hand in an erasure, or has copied a section twice. The lengths of the passages in question fall all but one in the range 51-62; it is possible that E omitted a line when copying, realised his mistake, and corrected by erasing the section he had copied too soon and adding the omitted passage in the erasure, rather than following the more normal practice of adding the omitted text in the margin. The length of the text in the erasure is, then, not so much an indication of the length of the omitted passage as of the interval before the scribe realised his mistake. Since the lengths tally so well with those of the repeated passages (which are presumably one and three lines respectively), it looks as if he realised his mistake after copying the succeeding line. The point is equally valid if he had in fact copied the preceding line twice and realised his mistake. The one erasure which does not fit the suggested pattern (TABLE II.x.8) would then be accounted for by the hypothesis that he did not realise his mistake after one line but at the end of the phrase, and the actual omission (TABLE II.x.1) would be of two lines. The discrepancy in length between the omissions of D and the evidence of E means that they cannot both reflect the line length of.1 ; further, the figures are such that one cannot posit one line for E and two lines on every occasion for D-in any case, it would be implausible to suggest that everyone of D's omissions involved two lines of his exemplar'S text. There must, then, be an intermediate copy on at least one of the two branches. E's text, while clearly showing that it is part of the