An inquiry into the transmission of the plays of Euripides

Citation preview

AN INQUIRY

INTO

THE TRANSMISSION THE PLAYS OF EURIPIDES

BY

G.ZUNTZ,

F.B.A.

CAMBRIDGE AT THE UNIVERSITY 1965

PRESS

OF

THE

SYNDICS

PUBLISHED BY OF THE CAMBRIDGE UNIVERSITY

PRESS

Bentley House, 200 Euston Road, London, N.W. r American Branch: 32 East 57th Street, New York, N.Y. 10022 West African Office: P.O. Box 33, Ibadan, Nigeria

UNIVERSITY

PRESS

1965

The publication of this book wa&made possible by grants from the Marc Fitch Fund and the Jowett Copyright Trustees.

Printed in Great Britain at the University Printing House, Cambridge (Brooke Crutchley, University Printer) LIBRARY

OF CONGRESS

CARD

NUMBER:

BOOK

IS GRATEFULLY

DEDICATED TO

© CAMBRIDGE

THIS

CATALOGUE 65-162II

A. TURYN

CONTENTS Preface

page xiii

Select Bihliography

XVll

xix

Introduction

I The Relationship between L and P

I

1

The State of the Qp.estion

I

2

Triclinius' Metrical Marginalia reproduced in P

6

3 P, in the Alphabetic Plays, a copy of L

II The work of Triclinius in L and its Effects upon P 1 The Problem 2 Metrica (a) Triclinian changes of the division of cola recurring in P (b) The metrical work of Triclinius in L and its antecedents (c) On the ancestry of the Annotated Plays in P 3 Triclinius' Dealings with the Text in Land their Effects upon P (a) Preliminary Helena Some passages in the Cyclops A note on inks Supplices , (f) Selected passages Heraclidae Herakles (b) (c) (d) (e)

~

13 16 16 19 19 27

35 38 38 39

51 57 62 83 87 ~

lphigenia Taurica lphigenia Aulidensis Electra (g) The problem of the Bacchae Vll

92 96 102

I IO

CONTENTS

CONTENTS

III

The Character and Origin of MSS. L and P 1 2

IV

page 126

Characterization of MS. L Characterization of MS. P

4 Hypothesis and Text of Rhesus

144

5 The Text of the Triad in L and P (a) 'Byzantine readings', ancient (h) 'Byzantine readings', recent (c) Specific features of the Moschopulean text (d) Specific features of the Thoman text (e) The relation of the LP text to Moschopulos and Thomas (f) The origin of the triad in L and P 6 The Origins of MSS. L and P (a) How manuscript P was copied (h) The ancestry of manuscript L

151 151 157 160 162 164 170 174 174 180

Results and Analogies l Triclinius the Critic

193 193

Triclinian Analogies Aeschylus

201 204

3 A Specimen: Helena, 625-97

V The Helena Papyrus (TI) l

The Text in Antiquity (a) The edition by Aristophanes of Byzantium and its effects (h) An early 'Selection'?

2

The Text in the Middle Ages (a) One archetype--or two? (h) The marginal commentary (c) One miracle--or two?

1 35

140

3 From Thessalonica to Florence Addendum

Indices I

II

Index Locorum Euripideorum Index of Manuscripts

III

Index of Papyri

IV

General Index

2II

217 218

Transcript

2 Reconstruction of the Papyrus

220

3 Comment on the Reconstruction of TI

221

The Alternation of Speakers 5 From Archetype to Original 6 The Original Text of Helena 630 ff.

230

4

viii

page 249

1

126

3 On the Hypotheseis in L and P

2

VI Main Stages of the Transmission

236 245

IX

2 49

LIST OF PLATES The plates are hound in at the end of the hook

I Cod. Pal. Gr. 287, fol. 120v: Iphigenia Taurica 102-59. II III

Conv. Soppr. 172, fol. 19r: Helena 610-68. Conv. Soppr. 172, fol. 19v: Helena 669-724.

IV (a) Laur. 32.2, fol. 14or: Iphigenia Taurica 826-40. (h) Pal. 287, fol. 127r: Iphigenia Taurica 826-30. (c) Pal. 287, fol. 121r: Iphigenia Taurica 177-90. (d) Pal. 287, fol. 144r: Iphigenia Aulidensis 1275-91. V (a) (h) (c) (d) (e)

Laur. 32.2, fol. 152r: Iphigenia Aulidensis 1274-88. Laur. 32.2, fol. 69r: Supplices 169, 171, 173. Laur. 32.2, fol. 106v: Helena 73-99 (recent photograph). The same, Helena 93 and 95, before June 1960. Conv. Soppr. 172, fol. 14v: Helena 95f.

VI Cod. Laur. 32.2, fol. 135r: Iphigenia Taurica 115---71. VII VIII IX

Cod. Laur. 32.2, fol. 135r: Iphigenia Taurica 172-241. Cod. Laur. 32.2, fol. 11or: Helena 622---75. Cod. Laur. 32.2, fol. 110v: Helena 676---702.

X Cod. Laur. 32.2, fol. 192r: Electra 1-38. XI

Cod. Laur. 32.2, fol. 192r: Electra 37---76.

XII

Cod. Laur. 32.2, fol. 192v: Electra 77-113.

XIII

Cod. Laur. 32.'.?-,fol. 192v: Electra 113-84.

XIV

Cod. Laur. 32.2, fol. 193r: Electra 185-228.

XV XVI

Cod. Laur. 32.2, fol. 193r: Electra 227-60. Pap. Oxy. 2336: Helena 63off.

Xl

PREFACE The dedication of this hook to A. Turyn amounts to a plain statement of fact: but for him, it would not he there. He obliged me by sending me his fundamental work on The By:r_antine Manuscript Traditionof the Tragediesof Eu;f1!!4es ~nviting me to test and pursue his results. For 6 better or worse I fell in' with the flattering request, but the task could not have been completed without Turyn's constant encouragement and help. He answered countless questions, lent me photographs and checked many details in manuscripts which at the time were inaccessible to me. His hook will he found quoted on almost every one of the following pages; it is in fact the starting-point and basis of these disquisitions. I had set out, and for a long time persisted, in the expectation of confirming one of Turyn's more revolutionary theses and have been genuinely sorry in the end to find it controverted by irresistible evidence. But this, I suppose, is how by steps we move towards the / truth-in the rhythm described by Hegel. The reader will not find in the I following a mere reassertion of the views of W ecklein and P. Maas. The progress of the actual inquiry is reflected in the successive chapters of this hook (and hurried critics are advised that a casual reference, prompted by the index, to one point or the other may lead them on to statements which actually are developed, qualified or even reversed with the progress of the a~gun;:ient).t_ In addition to my paramount indebtedness to Turyn, I am obliged to many others. P. Canart has been good enough on my behalf to examine two problematical passages in cod. Pal. Gr. 287; in papyrological matters I have been allowed to invoke the expert judgement of C. H. Roh_~rt_s and E. G. Turner (the latter moreover allowed me to present the'gist· of this hook to a meeting of the London Classical Institute in N ovemher 1960). I am not a specialist in the fields represented by these experts; if nonetheless I have ventured, in a number of details, to differ from authoritative views, this has been_,·.,the outcome of ..•careful and .• • ,.J •• J ,, , •. ,•; '.' prolonged consideration. He would he a rash that fancied to he able to settle controversial points of reading in manuscripts or papyri by a quick glance at a photograph or even at the original documents. The present disquisitions started with the study of microfilms by ,. ·"l . • • •

•.• .., • •

~

man

xiii

PREFACE •

J

PREFACE

••

means of projectors kindly 1f;~1:'°bythe British Academy and the University of Manchester; to the generosity of my university moreover I owe it that I have been able twice to study the original two manu.·scripts which form my central subject (I have not seen the orlginalof the Roil!;,mp.,

SELECT BIBLIOGRAPHY Full bibliographies are given by Koster and Turyn. The following is a short list of publications frequently referred to in this book.

R. Aubreton, Demetrius Triclinius et les recensions mearbales de Sophocle, Paris 1949 (cited 'Aubreton'). A. Dain, Les manuscrits, Paris 1949 (cited 'A. Dain'). K. Holzinger, 'Kritische Bernerkungen zu den spatbyzantinischen Aristophanesscholien ', in Charisteria A. Rz.ach• •. , Heidelberg r 930, pp. 58 ff. J.Irigoin, Histoire du texte de Pindare, Paris 1952 (cited 'Irigoin'). -- Les scholies metriques de Pindare, Paris 1958 (cited 'Irigoin, Les scholies.. .'). W. J.W. Koster, Autour d'un manuscrit d' Aristophane ecrit par Demetrius Triclinius, Groningen 1957 (cited 'Koster, Autour ... '). A. Pertusi, 'Selezione teatrale e scelta erudita nella tradizione de! testo di Euripide ', in Dioniso xrx (1956), pp. ur ff. and 195 ff.; ibid. xx (1957), pp. r8 ff. -- 'La scoperta di Euripide nel primo umanesimo ', in Italia medioevalee umanistica ,III (1960), pp. IOI ff. K. M. Setton, 'The Byzantine background to the Italian Renaissance', in Proceedings of the American Philosophical Society roo (1956), pp. r ff. H. Weir Smyth, 'The commentary on Aeschylus' Prometheus in the Codex Neapolitanus', in Harvard Studies in Classical Philology XXXII (r92r), PP· I ff. A. Turyn, The Manuscript Tradition of the Tragedies of Aeschylus, New York 1943 (cited 'Turyn, Aeschylus'). -- 'The Sophocles recension of Manuel Moschopulus', in Transactions and Proceedings of the American Philological Association LXXX (1949), pp. 94 ff. (cited 'Turyn, Mosch.'). --· Studies in the Manuscript Tradition of the Tragedies of Sophocles, Urbana 1952 ( cited 'Turyn, Studies • .• ' or 'Turyn, Sophocles'). -The Byz.antine M.:'"~ :

;,!•,.

•. ··, •

,

INTRODUCTION

i students of the text. Turyn has taught us that the 'interpolator'

of codex L-which, for the plays in question, will prove to be the source · of P-was not an unknown and irrelevant Italian humanist but the / outstanding Byzantine scholar Deme_t_rius-:r::r_iclini11si it is th~refore · essential to ascertain the extent and quality of his interference with the text. Meticulous collation is thus required, not only here hut also in future editions· its results while far from dramatic, will at last provide a reliable basis for the criticism of the text, and appease the critics' conscience. It is true that outstanding scholars, like Porson, Hermann and Elmsley, have been able without this basis to re:over the original wording in hundreds of places (and particularly also m cases where no collation, however exact, could have helped them); it is equally true that they have often heen misled by faulty information about the acr_ual tradition. But beyond this critical use it is hoR:? ~at the presentation and evaluation of a considerable amount of dreary detail may serve to clothe the dry skeletons of sigla, variants and stemma~ with some flesh of historical reality. As Pasquali urged long ago, this should benefit also the speciat.,ts in the field of textual criticism. . . Our two codices contain also those plays which ar~,-!rat~srmtted,ipother medieval manuscripts; oddly enough, codex p...ifrowned upon, with good reason, by modern critics-is the fullest of all: in fact,. the only complete one. The presence of these plays in Lan? ~-the creation, that is in the fourteenth century, of a new corpus Euripideum-cal.ls for an ex;lanation in the context of the scholarship of the_age.. . The inquiry so far outlined does nothing to~rds illummatmg the early history of the texts so miraculously emergmg towards the end of the Middle Ages.. This however has become feasible thanks to a papyrus (TT,i.e. S">:JCY· 2 336) containin~ a:1outsta~ding passage fro~ the ilelena. The-comparison of its Hellemstic text with that of the medieval manuscripts proves instructive, and a final sketch of the whole course of · tfi.etradition can thereupon he attempted. The researches of Turyn are basic to the following disquisitions. We shall not repeat those results of his which everybody grate~ully accepts, nor duplicate his exhaustive descriptio~s of ~e _manuscripts; but start at the point where his views have met with ohJecttons: namely, with the question of the relationship between the two MSS. L and P. !

CHAPTER

I

THE RELATIONSHIP BETWEEN LAND P 1. THE

STATE

OF THE

QUESTION

The two manuscripts are closely related. Is one of them a direct copy of the other, or do they derive, independently, from some common -ancestor, or has still another relation between them to he acknowledged? The pr_ohlemhas been discussed for nearly a century;t with, at least, two indubitable results: (a) L cannot be a copy of P, for too often Lis right where P is wrong; and (h) no single answer can apply equally to all plays. In the 'alphabetic plays' the similarity of the two is such as to , invite the assumption that P was copied from L; on the other hand, P : has Tro. which is lacking in L, and it also has Ba. nearly complete while L breaks off at v. 755. In the 'Byzantine triad' (Hee., dr.,Phoen.) there i~.~nf1!1ount of diverg~;_~~;-wlthp often giving the better text, which f6i-bi~s its straight derivatiopJi;om L. * As for the remaining 'annotated play~', the similarity of the ~ording in L and P-frequently opposing jointly the other witnesses-is striking indeed, hut a n,umher of instances (limited, it is true, but marked) where p a;oids f~uft; 'of L, § here too sta~d in the way of straight derivation. In consequence, views : differ regarding the relation of L and P in the annotated plays,, but · most scholars today believe that ? copied the alphabetic plays from L. ·. This view is supported by two different types of argumen!. I (1) Lis very often right, in these plays, where P is wrong; hut P is I1 very rarely right against L, and never in a really impressive manner. This point-the absence of an unambiguous 'separative error' (Trennt Cf. Turyn, p. 265.

p.

§ E.g.Med.336;(?48, XXX.

II5I;Alc.463,

:I: Cf. Turyn, pp. 298 ff. 473, ro7r;Turyn,p.285;A.M.Dale,Alcestis,

,- E.g. 'P is not a mere copy of L, but has independent value' (D. L. Page, Medea, p. xlvii); 'at least partially independent of L' (E. R. Dodds, Bacchaer, p. Ii); 'it seems more probable that for the select plays P had another source closely akin to L, but expert opinion is far from united on this point' (A. M. Dale, Alcestis, p. xxxi).

XX ZE

THE

RELATIONSHIP

BETWEEN

LAND

P

fehler) from L-has been urged especially by P. Maas.t He who refuses to accept this argument as cogent will have (i) to assume that L, in copying the common ancestor, committed no, or next to no, faults, and (ii) to try and trace overlooked instances where, after all, P is right against L. . ,c>:'.~,_.,,-,. :, >i.ssil>lype a flaw·i~ the paper? On running my hand over it I felt no i:irievenneJs. Finally, on the date just mentioned, I placed the page under the quartz lamp. The peculiarity of its colour stood out still more markedly; but what its cause was, the lamp could not tell. Hence I invoked the help of the Assistant Librarian, Dr Anna Lenzuni. An experienced palaeographer, she too failed to find an analogy for the mark in question anywhere in the handwriting of this ~crip~- In the end, she, ?¾n her hand over the pla_ct-and the 'colon' stuckto hei; finger. Th~ 'lieafof the lamp had loosen~d it. It was a tiny piece of'si:ra~-a residue from the production prq P ') was to be maintained. And yet, even W ecklein's own apparatus criticus contained a number of agreements of P and l against L; many more appeared ili-a~~is~ the corrector l being labelled L corr or L 1 or L2 or L3 '(thus throughout I.A.) or, vice versa, all manus correctrices being lumped together as l (I.T.). The resulting haze blurred facts which fatally threatened Wecklein's basic assumptions. There is no need here to retell the further history of the problem; .the reader not familiar with it may read it in Turyn's epoch-making book. \Vith_regard to this problem, Turyn's main merit is threefold. First, he stressec! tlie coincidence of land P-imidmissible on Wecklein's theory -in an impressive number ofi11sta:nc;es.,~econdly, by the brilliant disi') covery that the corrector l is none other than Demetrius Triclinius he changed the basis of the whole problem, placing it in the definite, historical frame which for so long it had lacked. Finally, by the tentati".:'_e hint that P n:i~yhave been written earlier than commonly assumed ('may bectated even as early as c. 1340 A.D.'),t he provided an essential element for a new assessment of all the facts involved. If we may judge by the observations presented in the preceding chapter, Turyn did not quite achieve this complete reassessment; nor is this in the least surprising. The extent of his new observations, and their itpplications, were so revolutionary that no one at the time could have fully assimilated them. In fact, though, Turyn had in his hands all the keys for the solution of the age-old and confused problem. If none the less he arrived at a result which-unless the present writer is quite mistaken-cannot stand, this is due to the fact that he r~i:;ihed, without by his questioning, one traditional tenet which actually had been own discoveries.· As long as P was supposed to have been written a long time after L, and the corrector l to have worked on L still later again, the ackno~l~dgement of fe~~i:~~d~~is~d by l recurdng in the original writi!1g of P would indeed have been fatal to the assumption that P was copied from L. Turyn had invalidated all these premisses; and yet he retained the conclusion. 'Of course, P did not know L.,!ricl' ::I:starting from this (no longer axiomatic) presupposition J1e

upset

t Turyn, p. 269. Ibid. n. 268 he even compared the writing of P with that of a :i: Thus p. 272, n. 272, and similarly throughout. MS. dated A.D. 1323. 2

17

ZE

WORK

OF

TRICLINIUS

INLAND

EFFECTS

UPON

P

rejected the derivation of P from L and traced their common features to ,i"common ancestor from which he supposed both P and L, and also· Triclinius, to have derived them. This hypothesis was an indispensable · step which gave a fresh impetus to the stagnating discussion; hut it was not final. The traditional refusal to acknowledge any impact upon P of the innovations due to the corrector l is quite understandable. Indeed the date is no longer a serious obstacle ( one ·merely has to bring oneself to move the writing of P hack by about another twenty or twenty-five ' years beyond the earl:y.A~re which Turyn, in his book, considered admissible). But th~_piiusa:1'of a few pages of the apparatus criticus will sh?w, time and again, readings of l (i.e. Triclinius) not shared by_P. .· l!ow_gien is one to account for the fact that other l readings-many more,. ~1l f~~!r,than the apparatus criticus suggests-do appear in P? Mere wHim of the scribe P cannot in general account for his accepting or rejecting the l readings-seeing that Triclinius often so effaced the original reading of L as to make it irrecoverable, except with the help of P. As already noted it was in places·of this kind that even Wecklein arid his_ adherents conceded the value of P. How then could P have repudiated, together with many other l readings, even those which had completely ousted the original ones? How could he, in these places, he in agreement with the original wording in L and yet reproduce, elsewhere, a considerable number of Triclinian alterations? Turyn's answer-that both P and Triclinius derived these readings from a common ancestor (whose wording L had missed), while those readings which are found interpolated only in L are Triclinius' own and never came heffethe eye~ of the scribe P-was perfectly legitimate; though not compelling, it' provided a reasonable answer to the questions raised. If now we feel compelled to abandon his solution, these questions still demand to he answered. If P actually copied L, not excluding at least part of Triclinius' alterations, the reliability and value of the whole LP tradition-for these plays, the only tradition which survives-is called into question. What authority has the primitive text of L? What was the source of Triclinius' alterations? If, at least occasionally, P copied them where L is no longer legible, are we, in some : places, left with nothing but his free inventions? In short, are there ' places where actually we have been taking Triclinius for Euripides? 18

THE

PROBLEM

These questions are serious enough to justify the most wary and sceptical approach and a punctilious inquiry which may strain the reader's patience. We proceed to examine the dealings ofTriclinius with L, and their relation to P. And we shall, for the time being, disregard the results of the preceding chapter, in so far as they may have yielded hints for the solution of the problems here arising. In carrying out this inquiry independently we may hope to gain a useful check upon our results. 2. METRICA (a) Triclinian changes of the division of cola recurring in P In the Avertissement at the beginning of vol. III of the Bude Euripides (p. ii) one reads: 'Particulierement dans les chceurs, il [the corrector l] donne en marge les noms des metres tels qu'il les concevait, il a souvent modifie la division anterieure des cola, et meme corrige le texte pour le mettre en concordance avec sa theorie. Aucune de ces notes metriques et ... aucune des corrections du texte qui y correspondent n'a ete reprise par la premiere main de P. On reconnah done ici surement !'intervention d'une main recente, posterieure a la confection de P.' On the notes metriques something has been said in the first chapter (abqve, pp. 6 ff.). As to the colon divisions in lyrical passages, it is true that, generally speaking, they are identical in P and L to a most impressive degree. Particularly in parts which Triclinius left, in this respect, unaltered, one may compare page after page without finding a single difference, and this identity (which anyhow is not matter of course) extends even to places where a word is broken by the end of a colon;t so in I.T. 190 !3cxc;v-e!!c.:iv, 399 8ovcxl!K6Xi\ocx, 835 o:yi'i:z'.

jpap:-~:-Ha;iis 3sl

t P and B separate this colon only by a punctuation mark and/or enlarged space. This happens sporadically in all MSS.-a first step towards the obliteration of original divisions. :f:B breaks the word one syllable later. § L r75v, P 83r, B r3§v. , Cf. below, pp. 37f.. II Here B wrongly adds;ywa1Kwv. tt Here, though, P reflects a temporary alteration by Tricliniiis, cf. below, PP· 37f. :f::f: L though had a division also (like pap.), after !3po-rois-quite possibly a survival of the Alexandrian arrangement. It was removed by Triclinius (and hence (?) is absent from P).

32

ME TRI CA

P

checked against P (Pal. 287, 195v) only. The division of verses is generally the same in both, but P repeatedly puts on one line two cola which in the papyrus have separate lines. In y. I 158 a c;mniabeforeei\o:l3ev still indicates-the old division, and in 1183 ff. it is unmistakable because the changes of speakers are indicated by wide spaces into which the ruhricator (as usual) has written their names; but P squeezes the whole of w. I 185 f. (veos • •• crnCXA6Tplxo:) on to one line without indicating any division, while the papyrus breaks the line afterµ6c,xos (the next is mutilated). A basic identity and its gradual, superficial' obliteration are here evident. There is comparatively copious papyrus evidence for various lyrics in the Phoenissae; it is however, for various reasons, not very instructive. Phoen. 175-8t is set out, in _p~p- Ox. 1i77 (of the Augustan age) exactly as in Wecklein and Murray; that is, by dochmiac dimeters (divisions after 175 o:EAiov,176q,Eyyos, 1770-wq,povo:).Thus the verses stand in P and M, except that both these join ,rwi\01s to the third colon. L and B on the other hand offer a completely different and obscure layout-and in both it is the same. Of Antigone's next stanza (182 ff.) the same papyrus preserves only the very beginning. Here it is noteworthy that p !/ scripts. Vitell!J:i~ld that P was copied from a copy of L into whicli thC>se hee_n introduced in which P differs from L; the n?.~ahle \ fact being that all of them occur in one or other of the extant manu' scripts, whence they could affect that supposed intermediate. I venture to present a few chance observations which seem to support Vitelli's view. They suggest a relation, in P, to the metrical efforts of Triclini.µ~ a_!l:~lC>gous to that ohserye1__in.~e. alp~abetic plays. - Alte~atiori.s, by Triciinius, of the colon divisions in L as a rule do not recur°I~ P. For example, in Hipp. 362 ff. Triclinius changed the division by dochmiacs which L * (159r) had identically with Par. 2713 (89v) and (except in the first verse) Marc. 471 (14or) by transferring 363 .vpavvov, 364 \ ing the changed word order. :i: It is hard to imagine that this could have come about independently. Triclinius put beside the beginning of this dochmiac passage ( 1251), in his usual big letters, the note &v,10"1Ta:O"T1Kcx and, moreover, beside vv. 1273 and 1279, the small notes &v,10"1Ta:O"TtKo: 6' and againav.10'1TO:O"TIK6pµ1yycxs.The MS. L cannot have lain before P in this altered state. 201t aiCYXwovc;'(with colon-end after the second syllable) L This reading makes the following T)µo:sunderstandable. Tr. changed it into -vas with black ink; P has aiCYXwcxs. 218 [3iovL *: [3ioTovTr. (black) = P. The ductus and colour of this alteration are as in 146. Tr. made exactly the same easy conjecture (cf. Hee. 1034) in S. El. 225 (Aubreton, p. 167), and for the same reason. It yielded a metrical equivalent for 199ovoµcx§-at the price of the intrusion of an unsuitable word. Biows and [3toTa:are, with Euripides, mere synonyms of [3ios,and the connotation of this word is so non-specific as to make the question 'which life have you not borne?' meaningless. Biov-the one really transmitted reading-is corrupt and the cure for it still to seek., Note that Triclinius did make some changes for metrical reasons already during his earlier spell of work on L. In the same verse there is another trace of Triclinius which however with a different does not recur in P. He wrote cxsover the end of ETA'l')S, ink (namely, the brown of the bulk of the metrical alterations) and a different ductus (namely, in carefully shaped and fairly large letters). It

*.*

t Jn parenthesi: Murrayis right on (189)µv(x)epe1s II63, joining the following, original divisions by dashes. Noting the faulty responsion he put i\ehm (not Tt i\elmt) after &ei\lo1sII64. Subsequently he endeavoured to mend it. He deleted i\el1re1and added svbefore crvµq>opa:is (not before a6i\fo1s)II64, while deleting Toov(actually only Too)in II 50; also, at the very end, he replaced aii\ivots by ii\lo1s.The gain from so much interference is disappointing; but since it all was part of his later effort, in brown ink, it did not affect P. One may note that Tr. could preface a personal conjecture, as here-and also 263 and 1615-by yp, i.e. ypa:,r-reov. This compendium then, with him, does not guarantee that he is recording an authoritative variant reading (cf. above, p. 42, on Y. 164).

47

WORK

OF

TRICLINIUS

IN

L

from L) there was a manuscript in which v. 1423 (omitted owing to the identical beginning of 1424) had been re-inserted in the wrong place. This more distant ancestor was presumably written down the columnsdifferently; that is, from Land, presumably, the direct model of L. This is the only instance I have been able to find where figures indicating a change of order were written (or, rather, copied) by L; otherwise, the bulk of them appear to be due to Triclinius (who here added opa in the margin). 1448 AV1Tp6:. 6q,ei11.c.u L *: AV1Tp6: y' 6q,. Tr. (black) = P. One more instance of the misapplication ofTriclinius' panaceat (cf. v. 888). 6e L *: 6' Tr. (black) = P. 1495 oIµa L (n6r): 6 superscr. Tr.:ol6µa P (26v). This is something of a test case, and a difficult one. On consideration, I believe that Turyn (p. 251) has diagnosed it correctly. It is admittedly difficult to dogmatize about the origin of a single added letter. The 6 here however differs from the products of the original scribe both in style and colour. Its shape is typically Triclinian,* and by its dark brown colour it stands out from the original text. In view of the wording of P one would have expected it to look black, and it is in fact possible to demonstrate that it is on a level with really black additions such as are found both before and after. The colour of this 6' is the same as in the marginal variant yp &pµa added by Triclinius and differs markedly from the reddish-brown of the additional TpoxaiKcx,ic.uv1K6:, etc. underneath the main notexop1aµ!31K6:ad v. 1451. That reddish-brown in turn compares with the colour of the ink used by Triclinius in his detailed work throughout this ode. Here then we have the familiar distinction of earlier and later strata, and the one letter under discussion belongs to the former. This conclusion is confirmed by the fact that in the course of his later work on this passage Triclinius deleted the whole colon 1495; the alteration of one word in it must then be earlier. One would here infer that when Triclinius went over the work of L for the first time, he compared some other manuscript. The useless t After prolonged examination I believe that Wecklein ad loc. is correct. Differently from Murray, he notes '7\V1Tp6:L, Av1tp6: y' !G'. This then is one more instance (cf. above on l.T. 166 and He!. 218) which ought to have made him reconsider his theory. :J:Nicolaus Triclines uses the same; but there is no trace to show that he had a hand in this part of L. Cf. below, p. 52, n. t.

48

HEL.]

[HEL.

=-}

EFFECTS

UPON

P

alteration ol6µa may or may not have been of his invention; but the variant &pµa, though equally useless, t appears to be drawn from somewhere-seeing that as a conjecture it could not commend itself either metrically or palaeographically or by any specious improvement of the ·meaning. By origin it may have been a (misleading) gloss on irrrnov olµa. 1532 (L 116r) Tr. (definitely black) = P: see ad 1413. 1535 i;/

_o?e

t I am not impressed by its tentative defence in J. Jackson, Margin. Scaen. 136'. The Dioscuri are horsemen, not charioteers. The Homerism oIµcx••. ieµevo1is as correct as 6:y&vas .•. 6pcxµeivI.A. 1455. :J:One does not lightly qontradict an authority like Vitelli (ap. Murray), but K(cxi)s.l. does appear-differently from the ending of the preceding word-to have been added by Tr., who therefore is naturally supposed also to have erased y'. Since P is here = L *,this diagnosis anyhow imposes itself (unless the main result of the present disquisition be rejected). All modern texts then are here, consciously or otherwise, based upon his alteration of L *. Is it a conjecture-or did he draw it from some tradition? § 78, 140, 287, 592, 893, 924, 1448, 1535, 1615. II 774, 1413, 1532. , 405 f., 902, 915. tt Another typical endeavour of Tr.-about which more later-is in correcting L's typical form 6veiv to 6voiv; we noted 571 (Tr' = P) and 731 (Tr'; 6veiv P); cf. on Suppl. 33, below, p. 64. 4

49

ZE

WORK

OF

TRICLINIUS

IN

L

[HEL.

• during his_~_rste,ffort, Tridipius corrected scribal slips of L * (79, 105, 10~245~ 542, 871, 921, 962, 1406) and it pO..os(277) to the end of the colon beginning with 1rposyevet6:6os, added the ending-o:i to the apostrophized verb CXVToµ' ( 279, retracing the neighbouring letters) and finally effected, in v. 282, one hexameter (as in the following two cola in L *) by changing M to evand deleting the colon after it. None of these changes reappears in P. The ink which he used for his final and most conspicuous: effort looks markedly red in most instances (o-rp(oq>11) at the very beginning, though, is dark brown). In the margin he noted 6CXKTvAtKoi e~o:µETpot;in a rasura which, one may infer, effaced the standard description xop10:µ[31K6: of Tr 2 (and it is of some interest that on consideration it appeared to him inapplicable here). t Accordingly he now joined, by dashes and enlarged letters, the remaining divisions which L * had been careful to indicate within the hexameters. Besides he added to o-oicn (v. 284) the final -v which p was to copy. As usual, none of these very marked contributions reappears in P. :i: 317 (cf. above, p. 5). The tell-tale addition µeyo:v was written by Tr 1 with black ink, in small, elegant letters, above q>o:vAov. Absurd though it is, it looks like a perfectly serious variant and may well come from some older manuscript. It is deleted or, rather, lightly scored over with black lines which may he Tr 1 or Tr 2 • Even if the addition lay before Pin this final form, the scribe could easily suppose it to have been smudged accidentally-for so itlooks to this .day. At any rate, P * put it into his text. K(o:i) &rrepoppc.v6ovo-'L * (for K/ = Ko:i cf. 318): TI 344 TITEKovo-o: TEKOV0-0: crvx' VTI.Tr 1 (black-brown): TITEKOV0-0: x' U1T.P*. P* copied ur 1 , skippingov after-o-o: (thus arriving by mistake at the correct text). Triclinius interpolated o-v, for' il a juge que, clans la erase, la voyelle qui en resulte pouvait etre breve'; indeed he considered this to he 'the normal scansion'.§ L * had given the correct text-in his usual scriptio plena; Triclinius, as usual, effected crasis. 365-80 (L 70v, P 87v). In this song, Triclinius found nothing to

alter at his first, but much at his second effort. Then, he indicated its beginning by enlarging the initial I 365 and defined its rhythm as io:µ[31Koi Ko:ixop10:µ[31Kol (marg.). Accordingly he noted io:µ[3(os)at the beginning of 365,369,373,375,377 and 379 (these verses actually make iambic trimeters in L, their division being as in the editions of e.g. Wecklein, Murray, Gregoire)t and altered two places which, in L *, fell oov) short of this metrical understanding. In 373 he wrote the sign over \\ (ov) so as to make Ko:Aoov out of Ko:Aov(before &yo:Aµo: ), :i:a1,1din 375 he furnished Ti µ6TIToAi0..1ov 371 to q>lAo:v, so as to secure a choriambus to correspond with Ko:iµeyaAa 367 (though thereby he spoiled the syllabic correspondence of the next metre). P reflects none of these alterations-not one is by Tr 1 • 396 **OOTtL *: oi6' cmTr 1 (black)' P: L * seems to have had o-6:q>o: Ei66Tt (anticipating Musgrave). t Wilamowitz, Comment. Metr. I (1895) = Verskunst, p. 155 proposed a dif-

baffling wording in L (in consequence I have time and again felt tempted to read 6:µq,lTwoov IKe--rav (eµ') 6:Acrrav-and been deterred by the need to insert oiKT10-a1 t Cf. above, p. 10, n. :i:. the precarious form eµE). :I: Verse 279 indeed stands on one line in P, with colon after To o-ov. P had omitted these words, and p mended the fault imperfectly. § W. J. W. Koster, Autour d'un manuscrit ... (1957), pp. 219 f.

ferent one in the second strophe. :I: Tr. dealt in the same manner, and for the same reason, with A. Eum. 306. § P has the same reading, not followed by either space or punctuation. L * often leaves spaces out of mere whim. , Here and in the next instance, the 'black' stands out as a very dark brown, markedly different from the brown of Tr3 and the grey-brown of Tr2.

~(

71

WORK

OF TRICLINIUS

IN L

SUP PL.]

fsuPPL.

408 Kai o L *: x>ooTr 1 (black) P.t 43oeio-i L* (71r) P: -iv Tr3 (brown); 435 eo-n L *: -iv Tr 1 (black) P; · 430 and 435. In identical additions, the difference of colour in L coincides with difference of text in P. 4 36 T**e> L *: Tave> Tr 1 (black) P: superscr. TO:CXUTO: Tr 1 • 448 11p1vovL *: -ov Tr 1 (black) P. The accent (which Tr. altered) shows that the ambiguous last letter in L * was meant to be read as v. 451 oos LP: superscr. os Tr 2 (grey). 458 -1TEµl.(Je L *: -ev Tr 1 (black) P. 478 o-q>piyoOvT, (?) L *: -ooVT, Tr 1 (black) P. 1 486 6veiv L *: 01 superscr. Tr (black): 6voiv P. 496 ovr' &v LP: add.y' Tr 3 (red), &'Jµoo-e L *: -o-ev 498 mii\iJo-ivL * (sic) P: Tr 3 (red) made 1Jinto ai. Tr 1 (black) P. 520 (L 71 v). Tr 2 (grey) wrote mpio-o-cxto the left and right of this verse. Why? 521 npcxyµaT' L *: -ae, Tr 1 (black) P. 525 av6P****TaS L *: av6poKµfj-msTr 1 (black) P: Tr. seems (in ras.) to have corrected some slip of L *. 574 ow L (72r) P: add. o-, Tr3 (brown). 577 eu6aiµova LP: ei supra -va Tr 3 ( or Tr 2 ?). The tiny compendium resulted in a thick, dark blob the colour of which yields no certain indication (cf. ad 611). If this is a conjecture (which seems likely), it is among Tr.'s best (inspired by v. 3, cf. Amir. 420?). 589 TE P: L can be read as either TE or ye (differently 858). 598-633 (L 72r, P 89v). Triclinius never perceived the strophic form of this passage. This may explain why there are far fewer alterations of cola in it than in others. On his first reading he corrected 619 KiAavAOS 43 5 (for q,1M6EAq,os L *P) and (e1)e1;\10-o-6µevos(to gain room for this expansion Tr3 altered the preceding word to 1= as ~O frequently passed from Tr 1 to P, are also absent from P's text , o{ .lj_a_.Such are in vv. 55, 138, 475 and 653 (quoted above, p. 113) and also 257 cpepwvL *P: j,tp~£_~_a_!e instances of P agreeing, against L *, .with Triclinius; for example 77 60-io1sL *: 60-£010-1 Tr. = P; /I'_.'1' 200 ocdµoo-1 L *: -CYIV Tr. = P; 401 Eµoi L *: Eµo1ye Tr. = P; 501 0µµ0:0-1L*: -o-1vTr. = P; 567 xopru*** L *: xopevo-wv Tr. = P; 659 cpev~6µe6o: L *: -ovµe6o: Tr. = P. If it is held that L and P are copies of one and the same model, one will in instances like these assume that the copyist L blundered and Triclinius restored the wording of the model. This interpretation would he perfectly natural in instances like the last two, hut not in the others. For the addition of a final-v to mend the metre, or at the end of a verse, is one of Triclinius' favourite devices. He introduced it also in vv. 97,369, 394 where P has not got it; could he, somehow, he implicated where it recurs in P, i.e. in vv. 200 and 501? The same suspicion arises with regard to the added ye in v. 401 (add. Tr. also in vv. 490 and 655-we have found it in countless other places) and the lengthened ending in v. 77 which Triclinius appears to have introduced in order to satisfy a mistaken notion of metrical correspondence. These are, so far, mere hints at faint possibilities. There are others, and more striking ones. One of them is in 13 yvo:s L: yvio:s Tr. = P. The mis-spelling of this word, :i:no doubt on the model of yviov, seems to have set in early, seeing that in Phoen. t I am less certain than Dodds that the emendation was 'doubtless conjectural'. Tr. knew his Aristophanes, cum scholiis. Cf. above, p. rn8, on El. 435 ff. :j: Cf. Valckenaer ad Phoen. 648.

II5

WORK

OF

TRICLINIUS

IN

L

648 it is reported from all manuscripts; but it is characteristic of the later Byzantine period. In Aesch. Prom. 369 and 708 it is in the recentiores, in opposition to the vetustiores, and Moschopoulos quotes it, in his Lexicon, from Soph. Ant. 569 where, as far as information is available, all manuscripts agree in giving the correct form (as also in O.C. 58). Triclinius was interested in the spelling of the word since the metre was affected by it. He noted ad Soph. Ant. 569t yum· 610:To µhpov evTei:>yume~el3ATJ6Ti TO L ei 8i:µETo:TOVi !3ovAe1 yp6:q,e1v,fo-rw c;o1cxvTi Ko1vf\sovMal3f\s: he regarded the diphthong as authentic; where the metre indicated a short syllable, he accounted for this by declaring the diphthong' ambivalent'; when faced with the correct form he assumed the iota to have been 'thrown out' for metrical reasons. His·practice confirms this. In Hee. 454 he conjectured TO:Syvlas (sic) for m5la.:t The original text of L always has the correct spelling-an achievement which must be ascribed to its source (or sources); namely, in Ale. 590 (with B only) and 687 (alone), Andr. 1045 (with H only), Med. 479, Held. 839, El. 79, He!. 3 and 89.§ Triclinius changed the two instances in He!., writing Ko1viJover the first., He did just the same in Ba. 13. One is tempted to suspect that he did so too in the manuscript from which P was copied; both here and perhaps also in the three other places where P, in this detail, differs from L; namely, Ale. 590 and 687 and Andr. 1045. / P_givt:sthe text of Ba. complete in three neighbouring passages where / ~~ J:i~c:l. riotaple lacunae which were subsequently filled in by Triclinius.11 They are (L 79v-8or; P 19or) 524 3evs om. L *: add. Tr.; hab. P;) 525 Ta:6' cxval3oiJc;o:s Tr. in Ta:8' cxvo:!306:c;o:s P; 552 evcxµiMa1c;1v cxv6:yiya colon sign. This argument may seem flim§y:' ·The following observations will bear out its cogency. The division of the lyrical cola in Ba. shows differences between L and P which could not have come about if both had been copied from one and the same manuscript. In the course of the present inquiry we have had ample evidence to show that in this respect P imitates its model with complete exactness (except that now and then P allots two lines to two cola which L, indicating the division by : , had crowded into one (half-) line. It was this consistency which made significant the agreement of P with certain divisions which Tridinius had introduced in L. t Also in Ba. the two manuscripts agree largely in their division of the lyrical verses; but there are significant exceptions. I do not pretend that the following survey is complete. :I: 399 ... ovxi q>Epe1 IIµcx1voµEVc.uv L (78v): ... ovxi q>Ellps1 µ. P (189r). 419 ... Eip11Jlvcxv Kovp. L (79r): ... sip11vriv· 11Kovp. P (189r). 427 L * (79r) CYOq>CXV o·crrrE)(_ElV ,rpcmiocx q>pEVCX TE 11irsptcrcrwv ,rcxpo: q>C.UTOOV II: p ( 189v) (j. 8'crrrE)(_E1V IIircxp' O:CY1Tl0CX q>pEVCX TE1TEptCYCYWV 1T.q>.IINote that Triclinius added, in L, : after crrrE)(_Etv but left the division after TE unaltered. Cf. below, w. 540 f. ••• o:pCYEVCX ... !3mh VT)O\JV: II &vcxq,cxvw ... ooII: 526 L (79v) Ye' cY:i P (19or) Ye' oo... o:pcrsvcx ... !30:61IIv118w• o:vcxq,cxvw ... ooII- Triclinius made expcrsvcxinto o:p:crsvcx(which comes to the same). to EVcroi; L divides it 531. P has one long verse from CYTECXVT)VTEVCYE x06v1os o:yp1c.u,rov11TEpcxsov q>WTCX !3p6Ts1ovII EA.Ol'TO, - ,r(ooe) · o:v6o:: mxp1v Ci) µ'ETIEVEVO"EV: - Ci) ,i\.fjµov: II ,i\.ccµoov,i\.ccµoov· 006' rnei\.o:cr' o:iy&rnw: eh' O:V'TE6ooK' ei6ooi\.ov(we) creeevKi\.voo:II

,a BeKcrra µei\aepa:1raeea 1rcceea::

µ&rep oi fyw· - ,i q>1Je:II OVKEO"'Tl µCX'TT}p. 6:yx6v1ov 6e j3p6xov · St' EµEKo:'TE6T]eO:'TO 6vcryo:µoe o:icrxvvo:vII 00µ01· 6vyo:Tp6cr 6'epµ16vT}crEO"'Tl j3iocr; ayo:µoe CX'TEKVOe &-reKvoe& ,r6cr1e Ko:'TO:O"'TEVEl II yccµov ayo:µov o:icrxvvo:: er,,rav Ko:'T'&Kpo:crcrooµ' Eµov ,repcr(o:e) ,rccp1e. II 'TCC6E K0:1O"Efowi\.ecreµvp1cc60:e'TE xo:i\.KEOTIA.OOV 6o:vo:oov· 11 EµE6e 1To:Tpi6oearro, Ko:KOTIOTµov 6:po:io:vej3o:i\.i\.e eeocr arro ,r6i\.eoe o:TIOTE cre6ev· 11 OTl µei\.e6po:i\.exeo::T'ei\.mov ov i\.movcr' ETI'o:icrxpoie yccµo1e:

Tr' > P 691 owµ' 696 µei\.a6pa Tr3 679a-b e6f\-x' linea coniunx. et notis ( :) ante TO:et post Ko:Kwv, · add. trim. iamb. effecit (n.b.: fuerat fort. colon (:) post 0-01,quod Tr3 lit. T 681 mevevo-'·rescripta delere et Pin spatio angustiss. neglegere poterant) 1 688 ECfT!V 689 I/\ TATATTPOCO'l'IC 6 ]MC1>0HN 6 ]OY.6.IOC/\EKTPA/\H.t.ACTE ]MTTA.t.WNKOPOI 7 64 ]OlzYNOMAIMONEC s---9 ]/\BICANEMECETEMATAN s ]N 9 ]N s ]NrE/\AYNEl0EOC IO ]KPEICCW

4

3-4 3-4

IO 10-II

II II +2 II +2

]A00N ]Ar ArEN 6:)TTOCEI ]MWCONA[ 2 ]ANTYXAC ]TA.6.~NEYXOMA[ ] ]NOYXOMENT/\HMWN[·].t.OY

] II

10 II

]TICTENOMENOY.t.A/\rW ]XOMENEXOMENONEMENON ]IACTTO/\ YETHMO/\EIN

658

~[

660

I)[

663 664 664a 665 666 667 668 669 670 670a 671 672 673 674

~Er[

N

Q[

·Q[ O[ TT[

[ [ O◊[

M[

~[ 8A[2]J:.(

-

KA['

~A[

Notes:

THE

HELENA

PAPYRUS

634 H.6.0NH and 636 TTPOCO't'IC (the second), and also 0 in 636a EME]M0HN.Problematical, though, arethedotonthe left of the letter and the short vertical above it. My suggestion that these might be remainders of the final N of the abbreviated name of the speaker (MEN(E/\AOC)) did not commend itself to Professor Turner. 2. RECONSTRUCTION

OF THE

RECONSTRUCTION

TT

PAPYRUS

To reconstruct the fragmentary papyrus is not to reconstruct the original text; it does mean recovering evidence for it older than L by 1400 years or more-a large though preliminary endeavour. This endeavour is greatly helped by our previous examination of the passage in L and P (above, pp. 21 1 ff.), for we are thereby enabled to confront the papyrus with the evidence of the majuscule MS. E. The majority of the required supplements thereby become matter of course. In putting the two witnesses side by side we assume that in E, as in TT, the lyrics were set off from dialogue verses by efo6ecnsand that each lyrical colon had a line to itself. As to the wording, it is hard to sayand unimportant-whether superficial faults in L (such as 631 &p~oµo:1, 634 xeipo:s, 666AEKTpov,671 rnEAo:ae)go back to E or originated on the way to L. Although the latter is more likely, I have retained them for E,t together with the punctuation, which is hardly older than the age of Eustathius, and the crasis in 630 K&ycband 644 K&µewhich may have been introduced by Triclinius (above, pp. 49, 129). The division of cola in E as ascertained by the elimination of Triclinian alterations in L affords welcome confirmation of many easy restorations; in some problematical passages it proves essential. In presenting now what to me seems the correct reconstruction of the papyrus I am in a few places anticipating the results of their discussion in the comment subjoined to it. I have not however as yet supplied the paragraphoi indicating the speakers in vv. 630--51, because this difficult point is bound up with the later evaluation of the restored evidence (see pp. 222 f.).:i:

OF THE

PAPYRUS

In principle of course all restoration is hypothetical; alternatives are not always out of the question and the scribe who blundered at the end of v. 634 could have blundered (careful though he is) in the parts broken off as well (e.g. v. 634 el3CXAi\ov ?) ; practically, though, the comparison with E allows of certainty on all except one crucial point (635). Some minor details however remain open. TI apostrophizes 6e and ye before vowels (vv. 642, 647, 649); even so the data are insufficient to show whether or no scriptioplena was used in 631 0160:and in 630 Ko:1 eyw and 644 Ko:1µe; t nor can we decide whether o:p~oµo:1or -wµo:1was written in 631 and6ve1v or6vo1v in 647. Where as little is preserved as in 663 ff., the range of uncertainty must grow; we cannot tell whether ec;01c;oµevor -µ0:1was written in 664 a, :f:whether the word order in 666 was as in L, and whether in 669 the particle (which?)§ was omitted as it is in L. I have marked these points of uncertainty in the supplements in the same manner as in the preserved parts; the remaining amount of trustworthy evidence is enough for comfort. 3. COMMENT

ON THE

RECONSTRUCTION

OF TT

When, as here pp. 222 f., E and IT are seen side by side, one is immediately struck by the fact that, almost throughout, the layout of verses and cola is identical in both. Differences there are-in four places where, significantly, the wording of L has always been recognized to be corrupt (vv. 634, 640 f., 650, 670). This general identity is evidence of a common archetype, for the divisions are anything but a matter of course, as may be seen e.g. in vv. 635-6a (which implies, in TTas well as in L, that Menelaus begins at 636a), in v. 637 ending with Te, in separate lines being allotted to one iambic metre in v. 648 (cf. v. 641-1a), to one trochee in v. 636a and to one dochmiac in vv. 664a, 667 and 671. The common archetype could only be the 'Alexandrian edition' reasonably fathered on Aristophanes of Byzantium. This conclusion is not novelt

t Except 665 Toi which Triclinius restored for TI L *. yy. 658-62, of which no more than two fragmentary letters remain. Ed. princ. shows that these fit the arrangement of verses in E without any difficulty. Cf. below, p. 248.

exceed the space available in v. 644. Lenting's conjecture fo-oio-oµevseems necessary to me. The middle form is incorrect (see Andr. 757 eio-oi,mst.6yov; cf. Ba. 650, Suppl. 600, Ion 1002, I.A. 97), the change from -µextto -µev trivial, and the alteration between the 1st pers. sing. and plur. typically Euripidean (649). § Barnes's yap is widely accepted; I should prefer Wecklein's Kcxi(Denniston, p. 312) or 617or oOV.

220

221

:I: I have refrained from including

KCXI eµs would

:j:

THE

HELENA

PAPYRUS

TT

He!. 630--51 and 663-74 sec. E

RECONSTRUCTION

OF

THE

PAPYRUS

He!. 630--51 and 663-74 sec. TT

ere. TIOA/1.0VS 6'ev µeerwi\6yovs E)(WV, 630 [µsv.] KCX)'W OVKol6' 01TOIOV TIPWTOV a:p~oµo:tTavw • 631 yey,i6o:· KpaTi 6'6p6iovs e6sipo:s 632 [ei\.] cxvETITepwKo: · Ko:i6a1VKO:S of the others . appears as v.l. in V. Does not all this suggest that, in each case, both variants were in one common archetype? To posit one archetype furnished with copious and substantial marginal variants: this in fact seems to me the conclusion demanded by the facts upon which Pertusi based his different theory. He was himself by no means unaware of this possibility;§ hut he regarded it as inadmissible in view of a number of other, substantial variants which can he shown to antedate the Middle Ages. These we now turn to examine. The 'shibboleth' (Wilamowitz) is Med. 1078 6pav µeMw L and all ( 12) quotations: ToAµ-fJ crwcet. codd.' Tot.µ-fJcrwno doubt is an ancient variant and not merely a Byzantine slip. Even so, it is not beyond possibility that it stood in the text of the archetype and the alternative in its margin. If so, the ancestor of L adopted the marginal readingll while the others followed the text. Med. 840176vnv6ovs hab. pap. Antin. 23 et LPv: om. cet.tt The occurrence of the genuine word in the papyrus does not prove its omission to he an ancient feature. It could have been omitted, by t Cf. Turyn, pp. 328 and 334. :(: C. Collard reminds me of the odd parallel in Suppl. 225 66µovs L: q,6j3ovsP. There however the fault was caused by q,l:l\ovsfollowing; it is one of the mental slips typical of the scribe P (above, p. 137). § Dioniso, l.c. p. u6. 1 Not mentioned by Pertusi, but see Turyn, pp. 286 f. (Cet.) codd. includes the Gnomologium Vatopedianum (ed. Longman, C!.Q.r!.N.S. IX (1959), p. 139). 11 Here again the wording of P or, rather, its direct model TI had been adapted to the current texts. tt Cf. Turyn, pp. 284 and 314.

MAIN

ST AGES

OF

THE

THE

TRANSMISSION

chance, in a copy of the archetype which became the ancestor of the other manuscripts. If this assumption is rejected, one may suppose that the word was omitted in the text of the archetype, presumably in imitating a faulty model, and was added in its margin, presumably from another model. The ancestors of L and of the other manuscripts would in this case have proceeded as suggested in the previous instance. Med. 1099 ecropoopap. Strassburg and LPv: opoo BVA.t As before, the correct reading in the old papyrus does not prove the (easy) fault in the discordant manuscripts to be old. It is without difficulty traced to their hyparchetype, where it could have been caused by the verbum simplex used in the paraphrase (schol. BV). Med. 1283 ywaiK(a) evpap. Harris 38 and LP: ywa1Koovevpap. Strassburg:!:and HBV A.§ Here the papyri make it highly probable that both readings are ancient (it is not absolutely certain, for the genitive, after µiav, is so natural a corruption that it could conceivably have arisen spontaneously again in the Middle Ages). Anyhow the possibility is not excluded that one archetype had the genitive in the text and the accusative in the margin. Med. 1295 Toicnv pap. Harris 38 and LP: Toio-6e y' HBVA. This I should consider an instance of progressive and typical corruption, from the original Towi6' (Canter), via the (archetypal) LP reading corrected to Toicr6e, to that of the other manuscripts. Among the many parallels quoted in Page's instructive note, Hipp. 1393 exhibits all the steps just posited. This passage then cannot strengthen Pertusi's case.1 Thus even these ancient and more or less substantial variants do not enforce the assumption that LP descend from another archetype than the rest of the Byzantine witnesses; it is possible to hold that they merely reinforce the case for a common archetype furnished with substantial marginal variants. 11 It is indeed hard to visualize, in investigat Turyn, p. 248 n. and p. 284. :t: The papyrus is not preserved beyond yvvatKu>v, but there is no reason to assume that it omitted the preposition. No extant witness does. § Cf. above, p. 37, and Turyn (as note t). , Equally inconclusive are (quoted by Pertusi) Andr. 639 and Rhes. 90. In Rhes. 66, VO contain a mere itacistic slip, and the variants in Hipp. 406 and 458 are too trite to allow of safe conclusions. II In Ale. 1025 one would have to reckon with an alternative wording for practically the whole verse.

/

TEXT

IN

THE

MIDDLE

AGES

tions of this kind, a set of facts which would exclude the latter, and enforce the former conclusion. The most telling indication of the former would come from the presence in one branch, and absence in the other, of corruptions unambiguously traceable to majuscule writing. Such have in fact been listed at the beginning of the present disquisition (above, p. 264); but even these are not, after all, decisive. It has been observed (above, pp. 183 ff.) that in the alphabetic plays a number offaulty transcriptions had been corrected in an ancestor-no doubt the original transcript-and that both faults and corrections continued to be transmitted as though they were veritable variant readings. The same could legitimately be assumed to have happened also in the transmission of the Selection, and thus even these 'majuscule variants' could be traced to one medieval archetype. To proclaim this possible alternative as the correct interpretation of the facts could be held to amount to an illegitimate exploitation of a legitimate principle in view of the large amount and, in some instances, the gravity of the differences between the two branches of the tradition. It is therefore imperative that, in conclusion, we should look out for , arguments which would render the assumption of one Byzantine arche. type probable or even certain. It has already been observed that elementary means of proof exist but are scanty and fail to yield complete certainty. There are others, less elementary but, I believe, cogent. We may stress, first, a point made previously. The assumption that the vast scholarly labour of transliterating these ten plays had been '; carried out twice is in itself so improbable· that it would require the most cogent proof to carry assent. It has meantime become clear that the arguments in favour of it, though in part impressive, are not really cogent; moreover, if the L text had really come from a separate source, one would have anticipated a far greater number of alternative transliterations and other significant variants than we have been able to trace. t The assumption that one book of the Pindaric hymns had been transliterated repeatedly is not exposed to this reservation. In fact, the vast majority of the variants separating the two branches of the tradition are of a trifling kind, such as would naturally arise among the descendants t Why, for example, is the spurious verse Andr. 7 in all our manuscripts, even though the scholia state that it was 'added by the actors' and pap. Ox. 449 (3rd cent. A.D.) actually omits it?

MAIN

STAGES

OF

THE

TRANSMISSION

of one ancestor, and far from intimating the characteristics of two different 'recensions' (as Pertusi asserted). This fact is in itself a powerful argument against their separate origin. It is finally excluded by the following points of detail: Med. 140Tov H: o cet. (s.l. H). I cannot visualize a credible theory to account for· the occurrence of the correct reading in H only and the fault in all others, if LP are supposed to descend from a separate archetype. Andr: 1037 6:y6povs Musgrave: 6:yopcxia:x.6povs LPM*v: 6:yopcxs a:x.6povsM< rr (hiat B): 6:yopcxsA: a:x.6povsV. The primary cause of this muddle appears to have been y miscopied x and producing a:x.6povs -and this mistake by itself suggests a model written in early minuscule. The various attested corruptions could have resulted from o:yop written above the faulty word by way of correction. Other explanations are not excluded; but ifLP reproduced a separate transliterated archetype, they could not have been drawn into this medley. Med. 741 EAE~CXS fV Myois L*P: EAE~CXS c';:, ywcxi BVALtr. Page's diagnosis (following Sigonius, Valckenaer and Elmsley) is convincing. E.6.ElzACwas wrongly transliterated; LP preserve the wording of the archetype; the others, an unavailing cure of the same. It passes belief that two independent majuscule manuscripts should, previous to transliteration, have exhibited the same abstruse fault. Hipp. 228t OE07TOlV' a/\!CXS VB2 A0Ur: OE07TOlVCX 6icxsMB* L*P. An error, evidently, in the original transliteration(/\ > b.) and its correction. Sharing the fault with the other most respectable witnesses, the L text proclaims its derivation from the common archetype. It would be wild here to posit two independent archetypes, both infected by the same fault b.lAC.* Other reasons apart, where, in this case, could the correction have come from? It is not a conjecture a Byzantine critic would light upon. But it could have been, and was, put right by the corrector of the one archetypal transcript.§ Med. IIOKCXJ