in the satires; and some of the later poets ventured on palatia and Vaticanus. But to shorten an essentially long Italian syllable like the a of Apulia or Appenninus would be portentous in classical times. One might explain the corruption in two ways: either Apuliae was a marginal gloss, and, as so often happens, has supplanted the word it was meant to explain, as for instance limina Dauniae' which Ritter and others have adopted: or else the copyist's eye was caught by the Apulo of the preceding verse, and he wrote down Apuliae for some other feminine ending in iae or ae: Bentley suggests 'Nutricis (or, Altricis) extra limina sedulae'. Mr. Yonge and, quite independently of him, one in whose judgment and taste I have the greatest confidence, have proposed limina villulae'. To this I feel the following objection: diminutives, used to such excess in the language of the people, in the comic poets, in Catullus and others, almost disappeared from the higher poetry of the Augustan and later ages: most of all perhaps in the Aeneid and similar poems, and next to them in the odes of Horace. Though puera had vanished from the language, Virgil may be said in the Aeneid to have rejected puella, as it occurs only twice in the same conventional phrase 'pueri innuptaeque puellae', borrowed probably from an older poet: villarum is found in the eclogues; but villa never appears in the Aeneid. Now though Horace in his odes does use villa and often puella, I would not venture to introduce on mere conjecture the double diminutive villula, any more than puellula or ocellulus or the like. It is said he has parmula: but the motive of that ode is a sportive self-abasement, and my poor shield' has its point. Again, though that might be used to defend villula, if found in all the MSS., it can hardly justify its introduction on mere conjecture.

In connexion with the above I will now speak of another puzzling passage: III 24 3 Caementis licet occupes Tyrrhenum omne tuis et mare Apulicum' or 'Ponticum' or 'publicum' of MSS. If Apulicum, which is scarcely Latin, be older than the other words, then it too may be a marginal

gloss and have superseded 'et superum mare' or the like. But both Ponticum and publicum seem to have higher authority in their favour than it has as Ponticum, adopted by Ritter, appears as much out of place as the Poenus in the Bosporus, I have with much unwillingness accepted Lachmann's Terrenum omne tuis et mare publicum', for Tyrrhenum sounds to me genuine, and terrenum to my ear has not a Horatian ring.

Carm. III 29 6 Aefulae, not Aesulae of all editions: the ƒ is found in some of the best MSS. of Horace, in the best of the scholiasts, as well as of Livy, as shewn by Huebner in the Hermes I p. 426; who completes the proof by citing three inscriptions, one of them Greek, in which the gentile names Aefolanus, Aefulanus, Aipouλavós occur.

.

me

Epod. v 87 Venena (magnum) fas nefasque, non valent Convertere humanam vicem': I have thus arranged this obscure passage, with one point of difference as Lambinus has done the meaning is venena (id quod magnum est) fas nefasque valent convertere, humanam vicem non valent'. Lambinus has shewn with his usual brilliancy that this is a good Latin construction, citing among other passages Cic. ad Att. x 14 istum, qui filium Brundisium de pace misit, legatum iri non arbitror' i. e. istum arbitror legatum iri, me non arbitror. But magnum I do not think can be joined with 'fas nefasque'; I have therefore made it parenthetical, where it seems to me to have much force. Venena can upset the natural order of right and wrong, as shewn by what precedes: comp. Virgil's 'fas versum atque nefas', Ovid's 'fasque nefasque Confusura ruit: but they cannot change the retribution due to mortals according to their deeds; which you therefore will incur; for Diris agam vos, cet.': for this use of vicem comp. 'fors et Debita iura vicesque superbae Te maneant ipsum'. The explanations of Orelli and others I cannot even understand; nor have any conjectures the least probability, neither Bentley's, nor Haupt's modification 'Venena maga non fas cet.'; nor Keller's 'humana invicem '.

I

Let me touch concisely on a few more passages: Carm. 1 1 13 I am unable to decide between 'dimoveas' and 'demoveas'. 4 8 though I have retained the ordinary reading of editions here, I believe that MS. authority properly interpreted indicates that Horace wrote not writ but uissit (i.e. visit, as Bentley and before him Rutgersius read): the MSS. of Plotius, who quotes this passage in his book de metris, appear to have ussit or iussit; clearly therefore he did not find in his Horace urit, and he is of course a much older authority than any existing MS. the old Bernese fails us in this ode, nor do the scholiasts either of them quote or explain this verse. Several of the oldest of Keller's Paris MSS. have uisit; one of the oldest Swiss and one of the oldest Paris MSS. have writ written on an erasure, that is some older reading. Oddly enough a Bodleian MS. cited by Bentley has uissit. It seems to me then more than probable that Horace wrote uissit, a common way of spelling visit in his day: comp. the well known passage in Quintilian 1 7 20 'quid quod Ciceronis temporibus paulumque infra, fere quotiens s littera media vocalium longarum vel subiecta longis esset, geminabatur? ut caussae cassus divissiones: quomodo et ipsum et Vergilium quoque scripsisse manus eorum docent': and to illustrate this we find in inscriptions of that date causa and caussa side by side over and over again. Ribbeck's proleg. p. 445 will shew how often Virgil's MSS. especially P, the best on this head, retain traces of the ss: thus among 50 other cases we find in geor. I 167 provissa, and, a very suggestive instance for us, Aen. v 637 iussa, i.e. uissa, for visa. By Quintilian's time the usage in this as in most other points of orthography had settled itself, and we find only visus, misit, but missus, though chance might have ruled it missit and misus: comp. the next words of Quintilian 'atqui paulum superiores etiam illud quod nos gemina dicimus iussi, una dixerunt'. When therefore uissit became obsolete, some traditions rightly kept uisit; others altered the unknown form to iussit or ussit; and then the unmeaning perfect was changed to urit. None of the passages quoted

to support this use of it, least of all the Greek pay, which = accendit, seems at all applicable; but Venus dancing in the moonshine, while her husband is away visiting the stithies of the Cyclops, is a beautiful picture. I 7 27 I leave 'auspice Teucro' of most MSS.: the old Bernese fails us here: though auspice Teucri' of many MSS. seems certainly to have been read by the Acron scholiast: Bentley's 'Phoebo', accepted by many, is also specious. 20 10 Tum for Tu is not I

n.

think necessary. be right. II 2 2 Lambinus' Abditae' would render this obscure stanza somewhat less dark. 13 23 though I still think 'discretas' more poetical, I ought I fear to have read discriptas', which has far higher authority on its side: see Buecheler's learned discussion in the Rhein. Mus. vol. 13 f. p. 598-604, though I do not admit that the sense requires the latter. 17 14 whether Horace wrote Gyas, Gyges, or gigas I cannot decide. 20 6 and 7 my mode of arranging this passage, which is the scholiast's, seems to me more poetical and to give a better sense, than those of most of the editors. III 20 7 tibi praeda cedat Maior an illi', i.e. tibi potius cedat an illi: this reading of all MSS. has I think been most successfully supported by Macleane against Peerlkamp's 'Maior an illa', accepted by Orelli and Haupt: he cites an exact parallel from Cic. ad Q. fratr. 1 1 13 'maioraque praeferant fasces illi ac secures dignitatis insignia quam potestatis': the way in which Latin writers sometimes use adjectives for adverbs is very striking. IV 1 22 'lyra,

26 9 Pimplea for Pimplei can hardly

fistula, Berecyntia'; here and in many other places, such as Epist. 1 2 32 hominem', 6 31 'putes', the readings of the 'antiq. Blandin.' alone or with few other MSS. are more elegant and to be preferred to those of all the other MSS. together; but on the other hand the 'terras' of this MS. in Epist. 1 3 4 for the turres' of all others is a corruption; and many such like might be pointed out in the odes.

III 27 5 Rumpit', 15 'vetat': though Bentley has proved the first and Lambinus the second of these indicatives to be

necessary, they will surprise perhaps many of my readers, as all recent English editors without exception have 'Rumpat' and vetet' and so have the best known German commentators, Orelli, Ritter and others. Who Galatea was, or what Horace meant by entering into such curious details of augury in the mingled pathos and sportiveness of these first stanzas, we do not know; but as explained by Lambinus and Bentley the sequence of the argument is clear enough. May the godless be guided on their way by the parra, the canis, etc.: such omens are fatal for a traveller: a serpent too darting athwart your path is cause for breaking off a journey begun. Here 'rumpat', as Bentley shews, has no proper sense, though Macleane quietly observes 'Bentley appears to have mistaken the meaning of the passage'. Though 'rumpat' has got into most MSS. by attraction of the preceding subjunctives, yet properly considered the indicative has most authority on its side: several of the best of Keller's Paris MSS. have it; and, what is more decisive, the best codices of both the Acron and Porphyrion scholia attest it, though the inferior MSS. conceal this by senseless interpolation and are unskilfully followed even by the latest editor of these scholia, Hauthal. I, the poet continues, a careful observer for one for whom I have fears, will summon from the east the raven and see whether its cry is propitious. Yes, all is well: it has uttered its cry on the right hand: you may journey with good fortune. wherever you list. And there both the woodpecker and the crow by appearing on the left shew that they too do not forbid you to set out. vetat is a conjecture of Lambinus, or else of the Vatican MS. in which he says he found it; yet it is undoubtedly true, the vetet having come from the attraction of the preceding subjunctives, a very common source of error in Horace's MSS. Otherwise the passage itself has no meaning, and is in direct contradiction to other authorities: Cicero, de divin. 1 83, asks 'cur a dextra corvus, a sinistra cornix faciat ratum?'; and Plautus asin. 260 says 'Picus et cornix ab laeva, corvus porro (parra) ab dextera Consuadent'; and