¹ As Diels, , Doxographi Graeci 25 ff.Google Scholar maintained.

² Apollonius Rhodius iv 775–7: Iris interrupts Hephaestus at his work: As Lloyd says (op. cit. 314 n.), ‘it may be that the word (πρηστῆρες) still means “whirlwinds” and is only used poetically and metaphorically to refer to the bellows of Hephaestus”. (Homer uses φῦσαι in a similar context: Il. xviii 372, 412, 470. Note that Hephaestus' bellows are automata (ib. 468 ff.), which would make the metaphor easier in Apollonius.)

³ R. de Courcy Ward, quoted by Shaw, W. N., Manual of meteorology ii 356.Google Scholar See plate VI 1.

⁴ Lane, F. W., The elements rage (David & Charles, Newton Abbot, 1966) 60Google Scholar; see plate VI 2. Cf. Lucretius vi 423–50, Pliny, NH ii 132, 134Google Scholar, partly quoted below. (Some passages in other authors, which say that πρηστήρ or τυφῶν occurs, or may occur, over the sea, or that it wrecks ships, must also be referring to waterspouts: Theophrastus Vent. 53 and Arabic fragment §§52–6 [see Bergsträsser, G. ‘Neue meteorologische Fragmente des Theophrast’, SB Heidelberg phil.-hist. Kl. 9 (1918)Google Scholar Abh. 9]; Epicurus, Ep. ii 105Google Scholar; Seneca, NQ v 13.3.Google Scholar Elsewhere we usually cannot tell whether waterspouts, tornadoes or both are meant. Lucretius vi 448–50 says that the prester is more often seen on sea than on land, which accords with modern records, cf. note 6.)

⁵ Ward loc. cit.; Lane op. cit. 35 ff.

⁶ It is clear from modern records of tornadoes and waterspouts in the Mediterranean area that the ancients would have been familiar with them: the Mediterranean is one of the seas where waterspouts are most frequent (Gordon, A. H., ‘Waterspouts’ in Marine observer xxi [1951] 58).Google Scholar I have found no published record of tornadoes occurring in Greece; but Wegener, A., Wind- und Wasserhosen in Europa 59 ff.Google Scholar mentions several phenomena of this type occurring inland in Italy, and also one near Smyrna which, though described in his source (Gilbert, L. W., Annalen der Physik 73 [1823] 109Google Scholar) as a ‘Wasserhose’, is said to have destroyed houses and trees and so must have been over land for part of its course. (A waterspout passing over land becomes a tornado, Lane op. cit. 58.) (Wegener op. cit. vii says he has not considered the literature of the Balkan countries, so he may have overlooked records of Greek tornadoes. I have been informed by Professor E. G. Mario-poulos, Professor of Meteorology at the University of Athens, that tornadoes do occur, though rarely, in Greece, at least ‘in the western coasts of the country and the southern coasts of Crete’).

According to Gordon (op. cit. 90), waterspouts are rarely accompanied by lightning; but Wegener (op. cit. 104) says that ‘von 153 Beschreibungen [of tornadoes and waterspouts in Europe], die Angaben über die Witterung enthalten, berichten 92 von Gewittern’, and quotes (p. 15) a description of waterspouts in the northern Adriatic which descended from a cloud ‘der … unaufhörlich Blitz und Donner aussandte’. Evidently the ancients would have had opportunities of seeing these phenomena accompanied by lightning, and these would be what Aristotle calls πρηστῆρες.

⁷ στυλοειδῶς Usener, cf. Lucretius vi 433; ἀλλοειδῶς libri.

⁸ LSJ mentions also certain biological senses of πρηστήρ; but these seem of no possible relevance io Anaximander. At Euripides fr. 384 seems to be used metaphorically of the waterspout: there will be columns of blood from the eyes.

⁹ Ibid.; Aëtius ii 20.1, 24.2, 25.1, 29.1 (DK 12A21 and 22).

¹⁰ Pour l'histoire de la science hellène (2nd ed., 1930) 92.

¹¹ See Aëtius iii 3.1–2 (DK 12A23, 13A17). The wind presumably ignites as it emerges, at least when it strikes and sets fire to some object, though Aëtius does not say this. See Kahn, C. H., Anaximander 102.Google Scholar

¹² So already at Hesiod Th. 846, Heraclitus fr. 31 (cf. Aëtius iii 3.9 [DK 22A14]).

¹³ A possible difficulty here is the tornado's well-known effect of lifting things from the ground (Aristophanes, Lys. 973 ff.Google Scholar, Aristotle Mete. 371a14 ff., Lane, op. cit. 38 ff.; etc.); i.e., a tornado is more likely to suck things up than to blow them down through its funnel.

¹⁴ Cf. Rossmann, F., ‘Über Wasserhosen auf dem Mittelmeer’ in Deutsche hydrographische Zeitschrift xiv (1961) 63 ffCrossRefGoogle Scholar: he describes waterspouts in which a funnel descended from a cloud, and another rose towards it from the water below, but the two never met.

A tornado with its funnel not reaching the ground is shown in plate VI 1.

¹⁵ Aëtiusii 20.1, 21.1, 25.1 (DK 12A21, 12A22).

¹⁶ I know only the instances already quoted. (For accounts of πρηστῆρες besides those cited above, see Aëtius iii 3; Stoics ap. D.L. vii 154; [Aristotle] Mu. 395a23 ff.)

I should like to thank Dr G. E. R. Lloyd and Professor F. H. Sandbach for helpful comments on earlier drafts of this note; the Meteorological Office for assistance in finding information on waterspouts and tornadoes in the Mediterranean area; Professor E. G. Mariopoulos for the information quoted in note 6; and Mr F. W. Lane for making available the photographs reproduced as plate VI.