¹ Publi Vergili Maronis Aeneidos Liber Quartus (1935), p. 392. I have not seen Braumüller, , Ueber Tropen und Figuren in Vergils Aeneis I (1877)Google Scholar, cited there.

² A partial analogy is provided by the extension of axis in the sense ‘axle’ to mean the whole ‘chariot, wagon, etc’ (TLL s.v. axis, II. 1636. 66), but the synecdoche here is far more natural.

³ Kaufmann art. s.v. axis, ἄξων, RE 2. 2632. The ancients are not more informative: Serv. Dan. ad Aen. 4. 482, axem: nunc pro caelo; non enim in axe sunt stellae.

⁴ Aeneidea III (1881), 417Google Scholar.

⁵ Virgil obviously selects the head as the support to maintain naturalism, but he has the sanction of one of the earliest of the descriptions of Atlas, Hes. Theog. 746 f. (head and hands).

⁶ Conington-Nettleship, , P. Vergili Maronis Opera III ² (1875), 96Google Scholar.

⁷ TLL s.v. orbis, IX (2), 913. 52 ff.

⁸ 1. 844.

⁹ QNat. 7. 1. 6.

¹⁰ TLL IX (2), 914. 23, ‘de sphaeris mundi’: note esp. Cic. Rep. 6. 17, novem tibi orbibus vel potius globis conexa sunt omnia; Varr. At. carm. fr. 14. 1 f. Morel, vidit et aetherio mundum torquerier axe \ et septem aeternis sonitum dare vocibus orbes. The use of orbis in poetry to mean caelum is as early as Cicero (TLL ibid. 913. 83), but there is no example (apart possibly from Aen. 8. 137) where the plural might be taken in this way. It is worth noting one other Ciceronian example of the poetic use of orbes in an astronomical context, referring not to the spheres, but to the circuit (ecliptic, tropics, etc.) projected on to the sphere of the universe: Arat. 296 f. hos [orbes] aequo spatio devinctos sustinet axis, / per medios summo caeli de vertice tranans. The language of line 296 is similar to that of Aen. 8. 137, but the concept is not as picturesque as that of the spheres, which, for that reason, should be understood in the Virgilian passage.

¹¹ It may be relevant that in a euhemerist interpretation of Atlas he is regarded as the inventor of the σπαιρικ⋯ς λ⋯γος (Diod. Sic. 3. 60. 2).

¹² On the Farnese Atlas: Thiele, G., Antike Himmelsbilder (1898), pp. 19 ffGoogle Scholar. Boll, , S.-Ber. Akad. München (1899), pp. 120 ff.Google Scholar, dates the group to the Augustan period. On the introduction of the sphere of the heavens in representations of Atlas: Roscher i. 710; the Farnese Atlas in the Pergamene tradition: Wernicke, , RE 2. 2132Google Scholar.

¹³ For earlier and later parallels see Pease ad loc.

¹⁴ Cited by Macrob. Sat. 6. 1. 9 as a parallel to Aen. 4. 482, 6. 797, but without any indication as to the context of the Ennian line.

¹⁵ For views on both sides see Wigodsky, M., Vergil and Early Latin Poetry (1972), p. 43 n., 203Google Scholar. Jocelyn, H. D., CQ n.s. 14 (1964), 295CrossRefGoogle Scholar, argues for Atlas, but his arguments are not cogent: he states that ‘no Roman poet of the archaic or classical period seems to picture Jupiter as standing apart from the caelum/mundus and making it rotate’, but ignores Lucr. 5. 1209 f. (cited below); and I do not believe that in Virg. Aen. 9. 93 ‘Virgil identifies Jupiter with the caelum/mundus on which the sidera are fixed’; is the Magna Mater speaking to an astronomical globe? For further considerations bearing on the problem see below and note 19. I here collect some passages in which turning the universe is the attribute of supreme divine power: Lucr. 5. 1209 f., ne quae forte deum nobis inmensa potestas/sit, vario motu quae Candida sidera versat (here the action is analysed into the several motions imparted to the heavenly bodies; the passage is possibly an echo – and deliberate criticism? – of Enn. ann. 29 V (note, in the immediate context, line 1205, stellisque micantibus aetherafixum, and cp. the Ennian stellis… aptum); Cic. nat. deor. 3. 93, fac [menlem divinam] esse distent am, caelum versantem terram tuentem maria moderantem this might also be an echo of the Ennian line – Cicero's language is poetic at this point); Virg. Aen. 4. 269 [Jupiter] caelum et terras qui numine torquet (torquet may also have a figurative sense here, being a favourite Ciceronian expression for ‘turn, direct‘ (LS s.v. torqueo HA), and the comment of Serv. Dan. ad loc. is pertinent: utrum quia mundus volubilis est ? an ‘torquet’ regit, sustinet? – I would answer both; cf. Aen. 9.93 [Jupiter] torquet qui sidera mundi, and compare Ov. Met. 2. 71, [caelum] sideraque alta trahit celerique volumine torquet); and compare Ov. Met. 2. 71, [caelum] sideraque alta trahit celerique volumine torquet); Apul. Met. 11. 25, tu rotas orbem,…regis mundum; Claud. Stil. 1. 63, volventem sidera Mithram. Examples in Greek mythological cosmologies: Kern Orph. Fr. no. 236, 1 f.; Plato Politic. 269^e; Rep. 617^c; Hippolyt. Elench. 5. 8. 34 (see Plato Cratyl. 408^c). Virgil also uses the idea of control by turning in the following passages: Eel. 9. 5, fors omnia versat; Aen. 3. 376, volvitque vices, is vertitur ordo; 7. 100 f., (the Trojans' descendants) omnia sub pedibus… vertique regique videbunt.

¹⁶ On Varro Atacinus and his influence on Virgil see Wigodsky op. cit.; Lenz s.v. Terentius 88, RE 2. 9, 699. In the context of the motif of turning the heavens note Morel FPL inc. fr. 29, luppiter omnipotens, caeli qui sidera torques, / ore tuo dicenda loquar, which Maass sought to claim for Varro, attaching it specifically to the context of fr. 14. Fragment 14 may also be relevant to the association of Atlas and the orbes in Aen. 8. 137, for in lines 1–2 we find in close juxtaposition axis and orbes (indeed the axis of the mundus is also logically the axis of the orbes). For his two descriptions of Atlas Virgil then echoes two distinct aspects of the Varronian passage.

¹⁷ Tièche, E., ‘Atlas als Personifikation der Weltachse’, MH 2 (1945), 65–86Google Scholar; Buffière, F., Les Mythes d'Homère et la Pensée grecque (1956), pp. 579 ffGoogle Scholar. In addition to Tièche's collection of examples it occurs to me to wonder whether the personification may not be alluded to at Arat. Phaen. 22 f., ἄξων αἰ⋯ν ἄρηρεν, ἔχει δ' ⋯τ⋯λαντον ⋯π⋯ντη / μεσςηγὐς γαῖαν, περ⋯ δ' οὐραν⋯ν αὐτ⋯ν ⋯γινεῖ. Might the rare use of the Empedoclean phrase ⋯τ⋯λαντον Emped. B. 17. 19 DK) contain an etymological pun on the name Atlas (Atlas as the cause of the equipoise of the earth)? The active role ascribed to the axis in the Aratean lines (⋯γινι) is unusual and may point in the same direction. This sort of wordplay would not be unique in Aratus: see Kidd, D. A., CQ n.s. 31 (1981), 355CrossRefGoogle Scholar.

¹⁸ Wernicke, art. cit., adduces passages to show the currency of the idea that Atlas turns the world-axis on his shoulders, but they do not in fact prove what he wants. Apart from the Virgilian passages he refers to Arist. mot. an. 699^a 27 ff., οἱ delta;⋯ μυθικ⋯ς τ⋯ν 'ατλαντα ποιο⋯ντες ⋯π⋯ τ⋯ς γ⋯sigmav; ἔχοντα τοὐς π⋯δας δ⋯ξαιεν ἄν ⋯π⋯ διανο⋯ας εἰρηκ⋯ναι τ⋯ν μ⋯θον, ὡς το⋯τον ὥσπερ δι⋯μετρον ⋯ντα κα⋯ στρ⋯ποντα τ⋯ν οὐραν⋯ν περ⋯ τοὐσ π⋯λους το⋯το δ'ἄν συμβα⋯νοι κατ⋯ λ⋯γον δι⋯ τ⋯ τ⋯ν γ⋯ν μ⋯νειν and σ. Hes. Theog. 509, ⋯π' ἄξον⋯ τινι λ⋯γουσι στρ⋯πεσθαι τ⋯ν π⋯λον. In both passages there is a clear distinction between myth (support on shoulders) and allegory (axis). A closer parallel for the conflation of two irreconcilable conceptions of Atlas is found in Aesch.P. V. 348 ff., ὃς πρ⋯ς ⋯σπ⋯ρους τ⋯πους / ἕστηκε κ⋯ον' οὐρανο⋯ τε κα⋯ χθονός / ⋯μοις ⋯ρε⋯δων, ἄχθος οὐκ εὐ⋯γκαλον (combining the Homeric notion of the κ⋯ονες guarded (ἔχει) by Atlas (Od. 1. 53) with the more common notion of Atlas as directly supporting the heavens). Hermann, G. (ppusc. 7. 254)Google Scholar took the Aeschylean passage as a poetic allusion to the axis-personification: contra Tièche art. cit., pp. 65 f., 85 f.

¹⁹ A much later relief shows Mithras framed by a zodiac band, which he grasps and turns with one hand (Mithras, ‘volvens sidera’, Claud., Stil. 1. 63)Google Scholar, illustr. in L'Orange, H., Studies on the Iconography of Cosmic Kingship in the Ancient World (1953), p. 32Google Scholar. Lucr. 5. 915 might be used as an argument for referring Enn. arm. 29 V to Atlas rather than to Jupiter, in which case one would naturally conclude that Ennius also refers to the personification. For a later passage where juxtaposition of the mythological and the scientific may imply awareness of the allegory: Avien.3. 101 ff., hie modus est orbis Gadir locus, hie tumet Atlas / arduus, hie duro torquetur cardine caelum, / hie circumfusis vestitur nubibus axis (the last line echoes another description of Atlas in Virgil, , Aen. 4. 248Google Scholar, Atlantis, cinctum adsidue cuinubibus atris; duro in line 102 appears to allude to the same etymology of Atlas as Virg. Aen. 4. 247).

²⁰ Cf. also Val. Fl. 3. 731; Arnob. nat. 2. 58.

²¹ E.g. Varr. At. carm. fr. 14. 1 (cited above); Cic. Acad. 2. 123; Manil. 1. 443 f., Arctos… / axem quae mundi stridentem pondere torquent (here it is the axis which is turned: cf. German. 227).

²² Maass, E., Aratea (1892), p. 270, n. 39Google Scholar, suggests that the immediate model may have been Arat. Phaen. 23, [ἄξων]περ⋯ δ' οὐραν⋯ν αὐτ⋯ν ⋯γινεῖ. I suggest above (n. 17) that the Atlas-axis personification may be present in the Aratean passage.

²³ E.g. Plato, Phaedo 99^c (possibly here combined with the axis-personification, Tièche, art. cit. pp. 78 f.); Aristot. cael. 284^a 18; Cornut. Theol. Gr. 26 (Cleanthes).

²⁴ In this connection it is interesting to compare the identification of the functions of a supreme power and the axis in Persian tradition, where the Great King is described as ‘the cynosure, axis, and pole of the world’ (Herzfeld, E., Iran in the Ancient East (1941), p. 320)Google Scholar.

²⁵ The earliest example in this sense (Accius trag. 566) antedates considerably the earliest example in the sense ‘axis’ (Varr. At. carm. fr. 14. 1). I would judge this to be an accident of survival.

²¹ See Maass, , op. cit. pp. 123 ff.Google Scholar; Kaufmann, art. cit. col. 2631; Tièche, art. cit. pp. 71 ff.

²⁷ There is other evidence to suggest that there is an easy shift between the sense of a word denoting the point or tip of something and the sense denoting the whole length of the object: (i)cardo, which is extended to mean a surveyor's line (and even to denote the three-dimensional sky, or part of the sky, TLL III. 445, 13 ff.); (ii) umbilicus, which, while properly referring to one of the bosses at either end of the stick round which the book-roll is wound, sometimes is better taken as referring to the whole stick (e.g. Hor. Epod. 14. 8); (iii) some instances of axis = ‘axle’ suggest that the sense is that of one of theends of the axle, if not actually, by transference, the ‘wheel’ itself (e.g. Lucan 8. 200).

²⁸ S.v.axis ¹ 5 b.

²⁹ On Vitruvius' dates: Fensterbusch, C., Vitruv: Zehn Bücher über Architektur (1964), pp. 4 if.Google Scholar: all, apart from the preface to bk. 1, probably completed before the end of the 20s B.C.

³⁰ See Fensterbusch, op. cit. p. 556 n. 347Google Scholar. Fensterbusch understands the phrase in all three places to mean the ‘south pole’; Reinhardt, K., Poseidonios (1921), p. 81Google Scholar, takes it to mean ‘der Durchmesser des Himmelsaquators im Schnittpunkte des Meridians, als eine zweite Achse', but admits that he thinks that Vitruvius has muddled his source.

³¹ S.v. axis II Ba (Luc. 7. 422, 8. 175; Vitr. 6. 1).

³² The Thesaurus groups the sense ‘pole’ under the heading ‘axle’, sc. Plaustri sideris septentrionalis, II. 1637. 20, which would rule out the reference of axis to the south pole; none of the quotations supports this bizarre notion. (The author of the article is perhaps misled by Servius ad Virg. Aen. 2. 512.) The definition of OLD s.v. axis ¹ 4 is at least ambiguous: ‘The extremity of the axis, the celestial north pole’; and none of the examples refers to the south pole.

³³ Axis in the sense ‘south pole’: Manilius 1. 375, 577, 589, 613, 624; Lucan 7. 422 (see Housman ad loc.), 9. 542. In some cases decision between the senses ‘pole’ and ‘part of the sky’ is impossible. ‘Pole’ may be the sense at Ovid P. 4. 10. 43, at Notus, adverso tepidum qui spiral ab axe: Notus is opposed to Boreas, which arises from the region of the circumpolar stars (for the idea that the winds blow from, or from the immediate vicinity of, the pole, cf. Varro Men. 271, ventiquefrigido se ab axe eruperant / phreneticiseptentrionumfilii; Lucr. 6. 720; (poss.) Ovid, Ib. 34; Manil. 4. 591; Sen. nat. 5. 16. 6; Plin. nat. 4. 89).

³⁴ It is frequent in Columella. OLD's second example, Quint. Curt. 7. 3. 7, is also open to doubt; the translation ‘pole’ is quite possible; it occurs, in any case, in a passage of deliberately heightened style. The first certain examples ofaxis = ‘part of the sky’ known to me are Ovidian: A A 2. 94;Met. 4. 214.

³⁵ I can compare only Eur.Or. 1685 ἄστρων π⋯λον.

³⁶ Cp. Ov. Tr. 1.2. 46, quantus ab aetherio personal axe fragor.

³⁷ E.g. Manil. 1. 454; Ov. Met. 6. 175.

³⁸ Cf. Lucan 5. 632, arduus axis / intonuit motaque poli campage laborant;, the possibility remains open that both here and in the Virgilian passage the sense is in fact ‘axis’; on sources of thunder cp. Aen. 1. 90, intonuere poli.

³⁹ It corresponds to οὐραν⋯ς in Θ. 68, ἦμος δ' 'η⋯λιος μ⋯σον οὐραν⋯ν ⋯μπιβεβ⋯κει and to caeli orbem in Aen. 8. 97, sol medium caeli conscenderat igneus orbem.

⁴⁰ Polus = ‘sky’ in the Aeneid: 2. 251; 1. 398, 608; 3. 586, 589; 4. 7; 5. 721; 9. 21; 11. 588.

⁴¹ Noted by Maass, , op. cit. p. 126 n. 18Google Scholar, ‘De Atlante caelum sustentante usurpari π⋯λος solet’, with examples.

⁴² So Tieche, art. cit. p. 72.

⁴³ The following is based on Deferrari, R. J. et al. A Concordance of Ovid (1939)Google Scholar.

⁴⁴ 6. 106.

⁴⁵ This is supported by the frequent use of gelidus axis in the poetry of exile: parallel expressions show that phrases like gelido…sub axe (P. 2. 10. 48; 4. 14. 62; 4. 15. 36; Tr. 2. 190; 5.2^b. 20) mean ‘under the north pole’, not ‘under a northern sky’: cf. e.g. P. 2. 10.45, sub cardine mundi; 4. 10. 39 ff.; Tr. 3. 10. Ovid exaggerates in his self-pity.

⁴⁶ 2. 94.

⁴⁷ Met. 4. 214.

⁴⁸ See note 38 above.

⁴⁹ Virg. Georg. 3. 223; Aen. 7. 288; Ovid Met. 6. 64.