'saepius vera lectio in margine exstat.' In some cases the variants are still legible in F, but many have been wholly or partially erased, and, though we know that they existed, we do not know (except by inference) what they were. It is desirable, therefore, to begin by proving that I. has in fact a marked liking (often shared by II.) for F's variants, and that its choice and rejection of them is largely, at least, independent of p. The scribe of sometimes follows the variant, where I. follows the text; sometimes the text, where I. follows the variant. At other times & and Iagree in ignoring the variant or in ignoring the text. It will be enough at this point to quote a few passages where I. has adopted variants ignored by 4, but still visible in F, and recorded by Helm. I give them briefly, using Fv as a symbol for 'original variant in F': I. II., etc., will not imply the agreement of all members of the Class. I. 22 (20, 6) fortiter F & 11., firmiter Fv I. III.: I. 9 (33, 2) adtraxerat F 4, adstrinxerat Fv I. II. III. : II. 24 (45, 5) quassato F & IV., quassanti Fv I. II. III. : IV. 1 (74, 16) leuigatos F & II. III., leuatos Fv I.: VI. 20 (143, 13) prospicua F & II., uel propitia Fv I., propitia III.: we shall meet other instances of I.'s slavish retention of such veľ's: VII. 27 (175, 21) deiecto F II. III., derelicto Fv I. VIII. 26 (198, 2) choraula FII. III., ceraula Fv I.: IX. 9 (209, 11) curriculi F 4, curuli Fv I., circuli III., phrase omitted II. (B3): XI. 9 (273, 7) facuum lumine F II., facti luminis Fv, ficti luminis I., factiuum luminis III.: XI. 11 (274, 22) huius F II., cuius Fv I. III.: XI. 22 (283, 25) competeret F & II. III., comperiret Fv I. These instances suffice to prove that in passages where F shows an illegible variant, ignored by 4, while I. shows a novel reading, it is not unlikely that I. has preserved the reading of F's variant.

:

I now pass to a selected list of passages where I. throws new light on F's original variants: here, as before, p's readings have not been hitherto recorded, unless an editor's name is attached to them:

+I. 17 (16, 1) [miser Helm] miser F, in margin res: ve is by the first hand, but s (which has been crossed out) is later. I think that a later hand has tried to change miser to mirer. [Hild. seems to record this variant, though his language is obscure : he writes of F 'quamquam in textu miser habet, ab eadem manu correctus est in mire.' Subsequent editors are silent.] miser : mire A1 E a: miser L1 V2 II. III. IV. I think that mire is defensible, but I prefer miser. +I. 22 (20, 15) [foribus Helm]. foribus F, but a word ending in; (=us) was written over foribus by the first hand, which also placed a row of dots under it: the word fortiter earlier in this chapter (20, 6) was altered to firmiter in exactly the same way: there firm, written above, has been partially erased, and the dots below almost completely (see Helm, ad loc.: he there read the letters, but missed the dots). edibus : edibus I. III. (but V7 has also the variant foribus in text): foribus II. (eight MSS. checked: V6 has also variant edibus in text): edibus L5 of IV., foribus V5 of IV. Presumably F's scribe changed foribus to edibus, and I fancy that he was right. II. 14 (36, 19) [etiam arisnotus unicus frater meus Helm]. F has two dots over the a of arisnotus (the scribe's commonest method of calling attention to a marginal note), and in the margin, by the first hand, nom ē (=nomen est). This was doubtless explanatory, but I. and II. treated it as an omitted phrase, and inserted it: it has influenced all MSS., except 4, which shows no trace of it. etiam nom ē arisnotus A1 L1: etiam nomine arisnatus E a (with arisuatus) : cui nomen est arisnotus II. (most MSS. checked) III. (V7 N2 8 B2) IV. (L5 V5). This well illustrates both the strength of F's influence, and the extraordinary faithfulness of Ai to F. III. 1 (52, 8) [uesperni Helm]. uesperni F 4 (but 4, as Helm says, has ti added by a later hand). But Helm has not noticed the erasure in F of an original variant over uesperni. Other MSS.: uel uespertini A1 E a: uesperni II.: uespertini LI III. Probably I. here preserve F's variant, though in X. 35 (266, 8) they have altered F p's uespernae to uespertinae without any such justification. +III. 2 (53, 21) [Helm pducunt (sed ras. supra et infra p, ut videatur fuisse p) : pducunt.'] Helm's explanation of the erasure below p seems to be right, but it cannot explain an

erasure above p. That erasure exists, and is almost certainly the erasure of the characteristic of F's first hand. No doubt it refers to a short erasure visible in the margin, which we may guess to have been a variant p, inserted by the first hand. Other MSS.: perducunt AI E a II. III. Both readings are defensible. +V. 17 (116, 17) [colubrum Helm] colubrum F 4, but in F there is an erasure over the second u of something by the first hand, which Michaelis reports as i erased: but it may have been a non-Beneventan a, such as F often uses when pressed for space (cf. Lowe, op. cit., p. 133). colubram AI E a: colubrum II. III. Neither coluber nor colubra occurs elsewhere in Apuleius, but he may well have chosen the feminine form, which is often used generically. +VI. 19 (143, 7) [Helm 'ad recolens alqd. in mg. add. quod iam non dispicitur; recalcans non cognoui: recalcans Oudendorp, cf. 211, 8']. The phrase is recolens priora uestigia: the passage (211, 8) to which Helm refers is IX. 11 mea recalcans uestigia. Oudendorp read recalcans, but not as a conjecture: it was the traditional reading. F's illegible marginal note cannot be recalcans, but it may well be e.g. calc: recolens : recalcans AI LI E a: recolens II. III. I should print recalcans. †VI. 29 (151, 13) [suspiritus Helm]. suspiritus F 6, but in F over the , and in the margin, by the first hand, vat;. Other MSS.: suspiratus A1 E a III.: suspiritus II. For suspiratus cf. Ovid, Met. XIV. 129. Apuleius may well have chosen this uncommon form. +VIII. 5 (180, 1) [Helm 'dente copulso (4) in mg. ead. m. add. dente èpulsu.'] F's variant is undoubtedly by the first hand: is used both in text and margin: but I think Helm has read it too confidently. Nothing seemed to me certain, except that the first word ended with the projecting Beneventan abbreviation for final m, while the second began with cp: this at once suggests dentem compulsum,' but it is merely a suggestion. dente compulso : dentium compulsu A1 V2 E a: dente compulso II. III. IV. It can scarcely be doubted that I. has preserved F's variant, and that this variant is the better reading: the passage describes a wild boar at bay... incendio feritatis ardescens dentium compulsu, quem primum insiliat, cunctabunda rimatur. The noun compulsus is very rare, but Probus quotes it (in the ablative) and it is used by Avienus and Marius Victorinus. Apuleius loves verbal nouns in -tus and -sus in the Apologia alone he uses eight which have no earlier authority (see Butler and Owen, Apulei Apologia, 1914, p. xlix.) VIII. 12 (186, 12) [Helm iam futuras vdVl. cod. Dorv. secutus ']. In F the passage runs oculi isti . . . qui quodam modo futuras tenebras auspicantes uenientes poenas antecedunt. Helm's note is inadequate. Not merely has (as Vliet almost correctly states) quodam modo tam iã facturas (u by the first hand), but in F itself, as no one has observed, tam ia by the first hand is still clearly visible written above futuras, though partially erased (Professor Rostagno kindly confirmed this for me). Other MSS. : quodam modo tamen iam futuras A1 a IV.: quodam modo futuras B3 V4 of II.: quodam modo iam futuras E of I., V6 N3 of II., III. It seems clear that and I. rightly understood the intention of F's original scribe: he meant to add a phrase which he had accidentally dropped, not to substitute one phrase for another. Obviously we ought to print qui quodam modo tamen iam futuras. A striking and instructive contrast with this passage occurs in the same chapter (186, 14 and 15), where & has, without justification, expanded F's absit ut simili to absit ut uelim simili. This false reading has not affected I., but it has been adopted by the whole of III. and IV. and by V4 V6 N3 (but not by B3) of II. It would be difficult to find a neater illustration of the merits of I., and of its general independence of 4. +VIII. 13 (187, 12): this is a very interesting case: it is, I think, the only passage where a reading of d's, not an easy correction of F's, is yet so obviously excellent that all editors accept it. It is not in 's manner: and it haunted me until I happened on the truth. [Helm 'caput. ò̟, capulu.'] On my first inspection of F, in 1910, I noticed nothing that threw light on the matter: caput is original and untouched: but when I re-examined F ten years later, my eye was caught by an erasure in the margin,

26

opposite the line which ends with the words preceding caput (Sed Charite | caput). I found that on close inspection cap. . .~ was distinguishable, also a tall erasure corresponding in position to the l of capulum: and further that the characteristic was erased over the initial c of the erased word (there is, however, no such mark over caput or over any other word in the text). I measured the erasure carefully, and found that it was the precise length of capul in F's script. It seems impossible to doubt that an original F variant is the source of p's capulum. Other MSS.: capulum caput AI E S, capulum capit a: caput II. (P V6 O N1 V4 N3 L6 B3) and III. (L3 N2 V7 B2 8): capulum L1 V2 of I., IV. (L5 V3 V5 L4). The classification is unusually complete and clear cut: Li and V2 are impure members of their class. I.'s original reading was clearly capulum caput: IV. copies p's capulum, and the rest have caput. It is, I think, probable that I. have borrowed capulum from : such borrowings, resulting in double readings, indisputably occur: for instance, VII. 12 (163, 17) os parti mortui F, oms parati morti : omnes partim parati mortui A1 L1, omnes partim parati morti E a But it is possible that the double reading was taken direct from F, or that caput has been borrowed from II. †VIII. 13 (187, 23) [Helm 'pflauit (4)'] 4 has pflauit, but F has p-flauit, with an erasure over fl. The dash - is later. F's erasure might be of a, or of e, or even of a small af or ef. Other MSS.: pafflauit Ai LI, pefflauere a: pflauit E II. III. It seems likely that F had pflauit with a variant intended to signify afflauit or efflauit. Few editors have relished perflauit. Hild. printed perefflauit, Vliet and Gaselee (after Pricaeus) proflauit, Helm (as his own emendation) efflauit. Helm's conjecture seems to be supported by the evidence of F and I. †VIII. 18 (191, 5) [quid miseros homines . . . inuaditis atque obteritis? Helm]. This is the reading of F and 4, but in F two lines of writing, apparently by the first hand, are erased in the margin exactly opposite the line containing inuaditis atque obteritis. The erasure is absolutely illegible. Other MSS.: inuaditis ac prosternitis atque obruitis A1 E, inuaditis ac perteritis atque obruitis a, inuaditis ac perterretis atque obruitis S* (teste Mod.), inuaditis perteritis atque obruitis S* (teste Oud.). All others follow F's text. It seems not unlikely that I.'s reading (probably as given by AI and E) was an original F variant. It should, however, be pointed out that I. certainly contains some interpolated words and phrases, which seem to have no justification in F: for instance, IV. 17 (87, 15) specus roridos et fontes amoenos F 4, specus roridos frigidos et fontes amoenos Ai Li (and B2 of III., L5 of IV.), specus roridos et colles frigidos et fontes amoenos E a: X. 15 (248, 13) iam totos ad me dirigunt animos F 4, iam totos dirigunt ad me omnes animos A1 a, iam totos ad me omnes dirigunt oculos atque animos E. All these seem to be, ultimately, instances of glosses or variants interpolated beside the word glossed. In other cases the gloss seemed to have driven out the word glossed: e.g. VIII. 28 (199, 20) paruam F 4, modicam A1, IX. 5 (206, 18) lucerna F&E a, lanterna A1. In one case, at least, the intruder has failed to shake off the uel which introduced it: VI. 19 (142, 17) telam struentes Fa, telam listruentes E, telam listrientes AI (no doubt corruptions of uel instruentes). Sometimes the true reading has been restored as a variant, and stands, with an apologetic uel, beside the usurper: e.g. X. 15 (248, 5) muscas F, mures uel muscas Ar E a S*: sometimes the true restored reading has triumphed, but still retains the tell-tale uel; e.g. II. 28 (48, 21) obuersus incrementa F 4, obuersus uel incrementa A1 LI E a, and B3 of II. But I must return to Class I. in its better aspects. +VIII. 22 (194, 14) [uxori suae Helm]. uxori suae F, but in F is erased over the u of uxori, and there is an illegible erasure in the margin. luxurie sue A1 L1 E a So, uxori suae II. III. I think that if we change sue to sua, the new reading is most attractive, and I have little doubt that luxurie sue, or, very likely, sua, once stood in F's margin. The passage deals with a slave, whose unfaithfulness had driven his wife to child-murder and suicide: seruulum qui causam tanti sceleris (uxori suae or luxurie sua) praestiterat. For luxurie compare VI. 11 (136, 12), where Venus confines the scalded Cupid to his bedroom, ne petulanti luxurie uulnus

..

grauaret. XI. 4 (269, 4) [Helm 'ad uirgule in mg. add. **gule (fuit uir aut un)]. uirgule (with no note): ungule A1, ? E, a: uirgule II. III. +XI. 15 (277, 22) [Helm ⚫ innouandi (p, al. m. superscr. ti) parv. ras. supra ua']. All editors print inouanti, which they ascribe to Aldus or Colvius. Helm has overlooked the fact that, besides the small erasure over ua (which we may guess to have been ), there is in F an erased marginal variant. inouanti A1 E a: innouandi II. III. Probably in(n)ouanti comes from F's margin.

I hope that I have killed the doctrine that of all the transcripts of F only one is of importance, and that is 4.' The man who would decipher F needs all the help that he can get.

οὗτοι ἀπόβλητ ̓ ἐστὶ θεῶν ἐρικυδέα δῶρα.

D. S. ROBertson.

THREE NOTES ON APPIAN.

I.

APPIAN, B.C. i. 16. 3, τὸ κράσπεδον τοῦ ἱματίου ἐς τὴν κεφαλὴν περιεσύρατο, εἴτε τῷ παρασήμῳ τοῦ σχήματος πλέονάς οἱ συντρέχειν ἐπισπώμενος.

These words occur in Appian's account of the riot which led to the death of Tiberius Sempronius Gracchus in 133 B.C. The tribunician elections had been adjourned from the previous day, and Gracchus, who irregularly sought re-election, had with his supporters taken possession of the temple of Jupiter on the Capitol. The assembly broke up in disorder amid wild rumours that Gracchus had deposed all his colleagues or had declared himself tribune for the following year without election or had actually demanded the diadem. The Senate meanwhile had been in session in the temple of Fides, and upon receipt of the news from the Capitol Publius Cornelius Scipio Nasica Serapio urged the consul, Publius Mucius Scaevola, to crush the tyrant. When he declared that he would not put any citizen to death without trial, Nasica, calling loudly upon those who desired the safety of their country to follow him, mounted the Capitol at the head of a considerable body of Senators and led the attack on Gracchus and his partisans.

Such is in outline the story told, with some discrepancies1 and in varying degrees of fulness, by Appian, B.C. i. 14-16, Plutarch, Tib. Gracchus, 16-19, Velleius Paterculus, ii. 3, Valerius Maximus, iii. 2. 17, Cicero, Tusc. iv. 23. 51, Livy, Ep. lviii, Auctor ad Herennium, iv. 55. 68, and other authors referred to in Greenidge and Clay, Sources for Roman History, 7 sqq. I am not here concerned with the larger historical problems raised by these narratives, but only with a single action on the part of one of the leading actors in the tragedy.

Speaking of Scipio Nasica, Valerius Maximus uses the phrase 'deinde laeuam manum parte togae circumdedit,' and Velleius refers to 'circumdata laeuo brachio togae lacinia.' On the other hand, Appian and Plutarch both speak of a different act-Appian in the words above quoted, Plutarch in the very similar phrase rò κράσπεδον τοῦ ἱματίου θέμενος ἐπὶ τῆς κεφαλῆς ἐχώρει (Tib. Gr. 19. 3). The author ad Herennium may be thinking of either of these versions, or of neither, when he describes Nasica as standing on the Capitol contorta toga.' The two Latin historians give no explanation of the action they describe, but apparently regard its aim as that of freeing the legs from the hampering folds of the toga, and perhaps

1 For these see especially E. Kornemann, Zur Geschichte der Gracchenzeit, 3 sqq.

also of providing a means of defence which might partly compensate for the lack of shields. To Eduard Meyer this was an obvious and sufficient reason, and he maintained that Appian puzzles himself needlessly to explain this very natural procedure: in order to be able to dash up the Capitol, the legs must be free, and at the same time the toga wound round the head serves as a protection' (Untersuchungen zur Geschichte der Gracchen, 95, note 2). But the Greek historians did not think thus. In Plutarch's narrative Nasica alone treats his toga in this way, while the Senators who follow him wrap their togas round their hands (τῇ χειρὶ τὴν τήβεννον Tepieλíĝas, Tib. Gr. 19. 3). In Appian, again, none of the three suggested explanations of Nasica's act refers either to mobility or to protection. E. Kornemann, who calls attention to the prominence of Nasica's pontificate in the story told by Appian and Plutarch and its neglect by the Latin historians, seems to me to be right in combating Meyer's criticism1 and in concluding that 'diese Massnahme . . . ist nichts anderes als die Herstellung der Priestertracht' (Zur Geschichte der Gracchenzeit, 5).

But is Kornemann equally right in maintaining (loc. cit.) that for Nasica's action Appian vainly sought an explanation"? What did Appian mean in speaking of Nasica as possibly τῷ παρασήμῳ τοῦ σχήματος πλέονάς οἱ συντρέχειν ἐπισπώμενος ?

It is not a question of historical fact which here confronts us, for we are dealing with the meaning of Appian's explanation, not with its validity; still less is it a problem of textual criticism, for the MSS. are unanimous. It is simply a question of translation and interpretation.

Candidus' rendering, quoted by Mendelssohn ad loc., is 'siue hoc gestu plures excitaturus siue ad cursum futurus aptior,' which suggests either that he had a different text before him or that he was baffled by the phrase. Schweighäuser renders siue quod miro illo habitu plures ad sese sequendum adlicere uellet,' but does not touch upon the difficulty in his notes. Horace White is content with this view, translating 'either to induce a greater number to go with him by the singularity of his appearance.'2 E. F. M. Benecke, on the other hand, has either to get more to follow him by displaying the badge of his rank.' Finally, Strachan-Davidson comments thus upon the phrase:

'Benecke's interpretation, "by displaying the badge of his rank” (i.e. the toga praetexta which Nasica would wear as pontiff), is tempting, but I can find no such sense for oxμa: it seems best to follow Schweighäuser and translate "by the strangeness of his appearance."'

4

In this view I have never been able to acquiesce. That Scipio Nasica,3 a member of one of the proudest families of the Roman aristocracy, who had been consul five years earlier and was now pontifex maximus, should have sought to swell the number of his followers by the bizarrerie of his appearance seems to me not only to run counter to all historical probability but also, and this is more important for our present purpose, to be inconsistent with the tone of Appian's narrative, which goes on to tell how the people είξαν ὡς κατ ̓ ἀξίωσιν ἀνδρὶ ἀρίστῳ. Scipio's aim was to ús secure not only a large crowd but the active support of all who, whether Senators or

1 Meyer's omission of the note in question from the second edition of his essay (Kleine Schriften, 412) is a tacit admission of the force of Kornemann's polemic.

2 The Roman History of Appian, London, 1889. No change has been made in the revised version published in the Loeb Library, 1913.

3 See F. Münzer's article in Pauly-Wissowa, R.E. IV. 1501 sqq., s.v. Cornelius, No. 354, and

the genealogical table, ibid. 1429 sq.

♦ E. Meyer (Untersuchungen, 95, note 1) declared that Appian was mistaken in regarding him as already pontifex maximus; but Münzer (op. cit. 1503) and Kornemann (op. cit. 4, note 3) have defended Appian with a cogency which has convinced Meyer himself (Kleine Schriften, 412, note 1).