(sumpsit) exercuit (1.44. 3) following Virgil's: laeva malorum exercet poenas (6. 542), quia Romanis Germanisque idem conducere (putabam), et pacem quam bellum (malebam) probabam (1. 58. 2), cunctos adloquio et cura sibi (conciliabat) et proelio (con)firmabat (1. 71. 5), sed Maroboduum regis nomen invisum apud populares (reddebat), Arminium pro libertate bellantem favor (sequebatur) habebat (2.44. 3). Thus far we have been dealing with zeugma, as though it had to do only with the verb. But the present and past participles of the verb are often used as adjectives, as in: umentes oculos et pallida bracchia tendens (Ov. Met. 14. 734), or: arte laboratae vestes ostroque superbo (pictae) (Aen. 1. 639), an example of zeugma with the past participle. We have further examples of this in: pallam signis (variatam) auroque rigentem (1. 648), tumulum, viridi quem caespite inanem (exstructum) et geminas, causam lacrimis (exstructas), sacraverat aras (3. 304–5), ipse Quirinali lituo (praeditus), parvaque sedebat succinctus trabea, laevaque ancile gerebat Picus (7. 187-8), quem pellis ahenis in plumam squamis (composita et) auro conserta tegebat (11. 771). In: videre rates... inter opacum adlabi nemus et tacitos (remiges) incumbere remis (8. 108) from rates adlabi we must imply remiges (= remigantes) with incumbere. We have already noticed how in poetry the noun is often substituted for the verb, and that in the prose of Tacitus this substitution is common. So in: fidem atque pericula polliceantur (Ann. 2. 40. 3) we should expect: se fidos fore atque audaces polliceantur. In: mederetur fessis, neu mortem in isdem laboribus, sed finem tam exercitae militiae neque inopem requiem orabant (1. 35. 2) mortem, finem, requiem are used as co-ordinate with mederetur. So in: praeda famaque onusti (12. 28. 1) and: ordinem agminis disiecti per iram ac tenebras (Hist. 3. 22. 2) we have examples of our first variety of zeugma not dependent on inflexion; and in: Inachus Acrisiusque patres mediaeque Mycenae (patria) (Aen. 7. 372) one dependent on likeness of meaning and derivation. In: diversis animorum motibus pavebant terrebantque (Ann. 1. 25. 2), where for metuendo et minando Tacitus substitutes a general term: diversis animorum motibus, we have a good example of our last class favoured by Tacitus. In: vides quae maxima credis Esse mala, exiguum censum turpemque repulsam for the general term labore we must substitute: dolore et periclo to restore the fourfold union, balancing them with: exiguum censum turpemque repulsam. Here animi dolore must be related to exiguum censum, and is fully described in vv. 45-6, while capitis periclo is the risk run to avoid a disgraceful defeat in an election (turpem repulsam), having nothing to do with loss of life in Horace's day, and referring merely to the loss of caput or civic status consequent on a conviction for bribery. Dolore and periclo both appear as readings for labore in the manuscripts of Acron, whose note: vis etiam mori propter honorem, seems to have misled some of our editors. XXXVII HYPALLAGE To: nube candentes umeros amictus augur Apollo (Od. 1. 2. 31) Acron's note is: nube candentes. Melius candenti nube, quam candentes umeros, amictus. Candidis nubibus velatus, scilicet qui videri possis; irati enim di nequeunt conspici. Hauthal wishes, with the usual wisdom of the emendator, to change irati to velati; but it is easy to believe that when the gods, like Tennyson's Hera, grow angry, they withdraw from human view 'into the golden cloud'. Bentley thinks that Acron wishes to change Horace's text here from candentes to candenti, and pleads that the resulting hiatus makes this impossible. But from Acron's note it is perfectly plain that candentes was the reading he found in his manuscripts, as it is in the best we have. As regards the meaning, it is true that the gods wrap themselves in clouds to hide from mortal view, if we may believe the ancients. Homer represents this in: vepéλy eiλvμévos wμovs (Il. 5. 186), and Virgil in Venus obscuro faciem circumdata nimbo (Aen. 12. 416). But that, as the gods were hidden from mortals obscuro nimbo, so they could be revealed to them candenti nube, the contradiction revolts Bentley. Hoc tamen inepte incommodeque, ut nihil supra. So he turns resolutely from the candenti nube of Acron to the candentes umeros of the Sungod, and seeks a parallel in other shining objects that the Roman poets present, the candidos lacertos of Tibullus's Pholoe, and the candidos umeros of Horace's Lydia. Shall we join in his refrain: hoc tam apte quam nihil supra? For we may feel that Lydia's shoulder-polish was a little different from the radiance that made the shoulders of the favouring god visible to his worshippers. And it is to show us the source of this radiance that Acron draws our attention to the shining cloud, which as it enwraps a modern saint or angel, our worshippers to-day, strangely enough, term a nimbus. Acron feels that from candentes umeros we must imply candenti with nube; for from the shining cloud comes the light that reveals to mortals the beneficent deity, who when angered veils himself in the nimbus or storm-cloud. So the threefold: nube candentes umeros goes back to a fourfold: nube candenti candentes umeros; the epithet candens belongs rather to the nubes, the source of the radiance, than to the umeri, which the radiance lights up. This seems the meaning of Acron's melius; candenti would be better in prose, and it is prose he is writing in his notes. The epithet candens is transferred by the poet from nubes to umeri; and by this reduction of a fourfold union to three we get the figure of the transferred epithet, the most usual form of hypallage. Of the nature and origin of this transference Bentley has no idea, nor does he always recognize it when he meets it. In: iam tibi lividos distinguet Auctumnus racemos purpureo varius colore (Od. 2. 5. 10-12) he rejects the manuscript reading just given, as well as Lambinus's emendation to: purpureus vario, and gives us in purpureo varios the plain prose to which his nature directs him. Here the poet gives Auctumnus a purple complexion, because he changes the grapes to purple, just as with the poet atra mors passes to pallida mors because of the pallor of the dead. It is the fourfold union: varius Auctumnus racemos varios, with the omission of varius in its apter and more obvious application. And yet Bentley understood why Horace in: inaequali tonsore (Ep. 1. 1. 94) calls a barber 'uneven': inaequalis est tonsor, qui tam prave capillum secat, ut una parte brevior sit, altera longior. But again in timet... miles sagittas et celerem fugam Parthi (Od. 2. 13. 17) against all his manuscripts he wants to change celerem to reducem. The soldier has no dread of the swift flight of the Parthian; it is the feigned flight he dreads and the swift arrows with which, as he turns back, he transfixes the soldier. Celerem is the epithet transferred from sagittas to its concomitant fugam; and it is because the feigned flight of the Parthian is swift in these speeding shafts that the soldier dreads it. On the next page I find his note on: pugnas et exactos tyrannos densum umeris bibit aure volgus (2. 13. 31-2), where following, as he imagines, Acron's note: bibit, avide audit, he changes umeris to avida. Densum volgus, he says, is sufficient without further qualification, but we want an epithet with aure. So he writes avida aure, failing to see that, when Acron makes bibit equivalent to avide audit, he shows how Horace by his striking oxymoron: bibit aure really supplies this lack. When Ovid writes: densum trabibus nemus (Met. 14. 360) it is short for: nemus densum densis trabibus; and Horace writes here: densum umeris volgus for: densum densis umeris volgus 'a crowd packed with thronging shoulders'. Let us begin with an example where the fourfold union has not been reduced to three. We read: illum absens absentem auditque videtque (Aen. 4. 83), where, though absens could be omitted and is usually omitted in prose, its expression adds greatly to the poetic force of the verse. The usual prose construction is that shunned in verse; and the omission of absentem would have made the line weaker still. But in pulverulenta fuga Rutuli dant terga per agros (12. 463), where fuga dant terga is for: conversi fugiunt, of the repeated terms we have that retained which is the less obvious and goes with the word which gives the verse its distinctively poetic character. In: procella velum (adversum) adversa ferit (1. 103), serpens quem obliquum rota (obliqua) transiit (5. 274), mixtoque insania (mixta) luctu (10. 871), furens (media) mediisque in milibus ardet (1. 491), of the several pairs the terms omitted are so clearly implied in those expressed that their omission would not be felt, were it not that the term omitted in each case is the more obvious, and that which would have found expression in prose. In: saxum (spumans) spumantia contra litora (5. 124) the very omission of spumans tends to call up in the mind of the reader the picture Stevenson gives us in the Merry Men; and the same seems true of the omissions in: postes auro spoliisque superbi (2. 504), or: Troiae renascens alite lugubri fortuna tristi clade iterabitur (Od. 3. 3. 62), or: miramur, facilis ut (lyram) premat arte manus (Prop. 2. 1. 10). ... The old grammarians, starting from the most puzzling form of hypallage, defined it as: mutua casuum permutatio, as we have it in: dare classibus Austros (Aen. 3. 61) for: dare classes Austris, the obvious prose form. The form of hypallage of which we have just cited some examples they styled hypallage imperfecta; for hypallage perfecta they required that both cases should be exchanged with each other, as in: et cum frigida mors anima seduxerit artus (4. 385) for: animam seduxerit artubus. Let us cite first a few simpler cases of such double hypallages; for double they seem to be. In: Ibant obscuri sola sub nocte per umbram (6. 268), Servius says we have a hypallage for: sub obscura nocte soli ibant. This seems to be the result of a union by distribution of two threefold clauses reduced from ibant soli sub sola nocte, and ibant obscuri sub obscura nocte, in each of which the more obvious term of the |