¹ The first reliable text was published, accompanied by a Latin translation, by Müller, C. in Geographi Graeci minores, I (Paris, 1853), 257–305Google Scholar. Then came Fabricius', B.Der Periplus des Erythräischen Meeres von einem Unbekannten (Leipzig, 1883)Google Scholar, a new text accompanied by a German translation; this for long was treated with a respect it did not deserve, for Fabricius' cavalier emending makes it thoroughly unreliable (cf., e.g., CQ 32 [1982], 182Google Scholar). The best text is that of Frisk, H., Le Périple de la Mer Érythrée, GṐteborgs Högskolas Årsskrift 33 (Göteborg, 1927)Google Scholar.

² Müller's text served as the basis of the translation by McCrindle, J. in his The Commerce and Navigation of the Erythraean Sea (Calcutta, 1879)Google Scholar. Subsequently Schoff, W. published a translation with extended commentary, The Periplus of the Erythraean Sea (New York, 1912)Google Scholar; this has been standard for over half a century – unfortunately so, since Schoff not only was careless but made wide use of Fabricius' text. A much better translation, based on Frisk's text, has recently appeared: Huntingford, G., The Periplus of the Erythraean Sea, Hakluyt Society, Second Series 151 (London, 1980)Google Scholar.

³ The source of the text is a single manuscript, Heidelberg Univ. Pal. Graec. 398, fol. 40v–54v, of the early 10th century; another, of the 14th–15th, in the British Museum (Add. Mss. 19391) is but a copy of the Heidelberg codex. I give the passage as it appears in the manuscript save for three obvious corrections: the insertion of πρ⋯τεροι before μικροτ⋯ροις (there are other equally plausible possibilities, e.g., ο⋯ π⋯λαι μ⋯ν); the deletion of τ⋯ν before κατ⋯ καιρ⋯ν (this deletion is one of a few corrections that the scribe of the London manuscript introduced in his copy; cf. Frisk 32); οἵ τε εἰς for the ms. οἱ δ⋯ εἰς before Σκυθ⋯αν. In Frisk's edition ὑψηλο⋯ has been erroneously transposed before ⋯κ τ⋯ς χώρας.

⁴ What follows has gained much from the invaluable help of my friend and colleague, N. Lewis.

⁵ That ⋯φ' οὖ used temporally denotes time prior to the main verb is as true of the Greek of the papyri (cf. Mayser, E., Grammatik der griechischen Papyri aus der Ptolemäerzeit, 11. 3 [Leipzig, 1934], pp. 77–9, 156Google Scholar) and of the New Testament (cf. Luke 13. 7, 13. 25–6, 24. 21; Rev. 16. 18) as it is of literary Greek (LSJ s.v. ⋯π⋯ II). I have found but one instance in which it denotes later time, Wilcken, U., Urkunden der Ptolemäerzeit (Berlin, 1927), 46. 3–7 (162–161 B.C.)Google Scholar. I take the second instance, a few lines below, causally, ‘because of which’; cf. note 24 below.

⁶ Reading, with all editors, ⋯ξευρηκ⋯τος for ⋯ξευρηκ⋯ναι.

⁷ On the nature of the monsoon winds and how they affected sailing to India, see Böker, R., RE Supplbd 9, s.v. ‘Monsunschiffahrt nach Indien’ (1962)Google Scholar; Casson, L., ‘Rome's trade with the East: the sea voyage to Africa and India’, TAP A 110 (1980), 21–36Google Scholar (= Ancient Trade and Society [Detroit, 1984], 182–98), esp. 31–5Google Scholar.

⁸ Cf. Pliny, , HN 6. 100Google Scholar (where he describes sailing ‘ab Syagro Arabiae promuntorio Patalen [at the mouth of the Indus] favonio, quem hippalum ibi vocant’) and 104 (‘vento hippalo navigant…ad primum emporium Indiae Muzirim’ [on the Malabar Coast]). The ‘ventus hippalus’ was actually more ‘africus’ or ‘austro-africus’ than it was ‘favonius’; Pliny manages to get the other monsoon wind wrong as well (see Casson, note 7 above, 33). Between November and April the wind blows as steadily from the northeast as it does from the southwest the rest of the year, and this made the homebound voyage just as easy and direct as the outbound. Indian and Arab skippers had been using these supremely convenient winds for centuries but somehow kept them a well-guarded trade secret. Hippalus' claim to fame was in being the first westerner to find out about them.

⁹ Thus Fraser, P., Ptolemaic Alexandria, I (Oxford, 1972), 183Google Scholar, in translating the passage simply omits ⋯φ' οὖ (‘For, when the Etesian winds are blowing with us, in the Indian Ocean etc.’). Possibly scribal negligence left something out after ⋯φ' οὖ just as happened after οἱ μ⋯ν a few lines above; omissions are common in the manuscript (8 = Frisk 3. 25, 10 = 4. 7, 33 = 11. 8, 45 = 15. 17, 54 = 18. 3, 59 = 20. 1). Or possibly ⋯φ' οὖ is a miswriting induced by the occurrence of the same sequence of letters a few lines below (-πλουν ⋯φ' οὖ).

¹⁰ See Sailing Directions for the West Coast of India (U.S. Defense Mapping Agency, Hydrographic Center Pub. 63, 5th ed. 1967, rev. 1976), Section 1–37, p. 55: ‘Southwest monsoon winds…may very well be the strongest and most persistent winds over any water area of the globe.’ Cf. also Casson (note 7 above), 24, 33–5. Greek and Roman merchantmen, with their massive construction and conservative rig, were better able to handle the southwest monsoon than the Arab and Indian craft of later centuries (Casson, 23).

¹¹ A few commentators were aware that the routes to India involved sailing with a favourable wind. Cf. Hourani, G., Arab Seafaring (Princeton, 1951), p. 27Google Scholar (‘ships could sail straight on before the southwest monsoon…and arrive off the Malabar Coast’); Thiel, J., Eudoxus of Cyzicus, Historische Studies uitgegeven vanwege het Instituut voor Geschiedenis der Rijksuniversiteit te Utrecht 23 (Groningen, 1966), p. 44Google Scholar (where he talks of the Greek skipper's ability ‘to hold a given course across the open sea, with a following seasonal wind as the only compass’). Most, however, as we shall soon see, were unaware, and even Hourani and Thiel failed to exploit their knowledge; cf. note 17 below.

¹¹ See E. Gardiner's full discussion, abundantly illustrated from vase-paintings, in JHS 25 (1905), 272–8Google Scholar; de Ridder, A. in DS s.v. ‘Lucta’ (1918) 1340Google Scholar and Jüthner, J. in RE s.v. ‘Pale’ (1949), 86–7Google Scholar add nothing. That twisting was involved is not only manifest in the ancient representations but made expressly clear by Plutarch, , De Curiositate 521 BGoogle Scholar: an Olympic victor, riding in a procession in his chariot, was unable to keep his eyes off a shapely courtesan among the spectators and kept turning his head around to catch sight of her (Παρεπιστρεφ⋯μενον); ‘Look at our big athlete,’ commented Diogenes, ‘being neck-locked (τραχηλιζ⋯μενον) by a little girl.’ For a neck-lock ending in submission, see Plutarch, , Apophtheg. Lacon. 243 DGoogle Scholar; for it ending in a toss, Gardiner, , loc. cit., 272Google Scholar and Athletics of the Ancient World (London, 1930), p. 189Google Scholar. That the neck-lock was a very powerful hold is attested by the metaphors in which it appears, e.g. Philo, , De Mutalione Nominum 81Google Scholar (cf. Harris, H., Greek Athletics and the Jews [Cardiff, 1976], p. 70)Google Scholar.

¹³ Cf., too, Bunbury, E., History of Ancient Geography, II (London, 1879), 470Google Scholar (‘if they are going to Limyrice [they] have to struggle longer’); Lamotte, E., ‘Les premières relations entre l'lnde et l'occident’, La Nouvelle Clio 5 (1953), 83–118, at 104Google Scholar (‘ceux qui font voile vers la Limyrikē…louvoient la plus part du temps’); Huntingford, p. 53 (‘those sailing to Limurikē turning the bows of the ship against the wind’).

¹⁴ Sailing vessels cannot travel into a head wind but only at an angle to it. They achieve forward progress by zigzagging, by sailing at as close an angle as they can get to the direction from which it is blowing – about 70° in the case of ancient square-riggers – first a zig of a certain distance to one side of it and then a zag to the other; each of these zigs and zags is called a tack in sailor's jargon, and the process of sailing such a course is tacking.

¹⁵ In referring to the wind as a ‘beam wind’, Schoff may have been following Bunbury, who for some reason called it that (note 13 above, p. 471). Bunbury, however, made it perfectly clear that a beam wind required no steering off course.

¹⁶ Cf. Casson, L., Ships and Seamanship in the Ancient World (Princeton, 1971), p. 224Google Scholar.

¹⁷ Cf. Warmington, E., The Commerce between the Roman Empire and India (Cambridge, 1928), p. 46Google Scholar: ‘throwing the ship's head off the wind with a constant pull on the rudder and a shift of the yard (thus sailing in an arc of a circle)’, which is more or less repeated in Cary, M. and Warmington, E., The Ancient Explorers (New York, 1929), pp. 76 and 223, n. 51Google Scholar. Rawlinson, H., Intercourse between India and the Western World (Cambridge, 1926 ²), p. 110Google Scholar, talks of ‘throwing the ship's head off the wind’; similarly, Charlesworth, M. in CQ 22 (1928), 96Google Scholar (‘those who are making for Limyrike…have to throw the ship's head more off the wind’). Schoff's translation is repeated by Hourani (note 11 above) p. 25 (who failed to see that it directly contradicted what he himself would say two pages further along; cf. note 11 above), by R. M. Wheeler (in Grimes, W., ed., Aspects of Archaeology: Essays Presented to O. G. S. Crawford [London, 1951], p. 368)Google Scholar, and by Miller, J. (The Spice Trade of the Roman Empire [Oxford, 1969], p. 145)Google Scholar. Thiel (note 11 above, p. 45) recognized that ‘Schof's translation… is incorrect’ and that ‘Fabricius’ term laviren is an exaggeration', yet fell into the same error himself: as he saw it, skippers did not tack, ‘they only sailed close to the wind’. Sailing ‘close to’ a southwest wind would have meant, for an ancient square-rigger, sailing southeast by south, a course that would have left it wandering in the reaches of the lower Indian Ocean.

¹⁸ See Huntingford's words cited in note 13 above.

¹⁹ Frisk thereby avoided the emending of his predecessors, e.g., Müller's suggestion that τραχηλ⋯ζοντες be changed to τραχηλ⋯ζονται or Fabricius' altering of ⋯ντ⋯χουσι to ⋯ντεχ⋯μενοι.

²⁰ Frisk, of course, was simply following any number of predecessors in assuming that the three-day sail was made against the wind: cf. Müller's translation (‘non ultra triduum contra nituntur’), McCrindle's (‘they are not retarded for more than three days’), and Bunbury's comment (note 13 above, p. 470: ‘contend with the wind for about three days, and thenceforward have a favourable wind’). Huntingford, too, takes it this way (‘hold out to the contrary for not more than three days’).

²¹ Almost all commentators and translators have been aware of this: Müller (who commented ‘per triduum oram legebant usque ad Syagrum promontorium’), Fabricius (who translated ‘hältsich nicht länger als drei Tage am Ufer’), Schoff (who translated ‘keep along shore not more than three days’), Lamotte (note 13 above, 104: ‘ne longent pas [les côtes d'Arabie] plus de trois jours’), Thiel (note 11 above, p. 46: ‘ships must hug the Arabian coast for the first three days’).

²² With a favourable wind ships averaged four to six knots on the open water, slightly less along a coast; see Casson (note 16 above), p. 288.

²³ A verb *παρεπιφ⋯ρω is not attested elsewhere but it is a compound that suits the context; cf. ⋯πιφ⋯ρεσθαι as used by Herodotus in his description of the Nile sea anchors (2. 96). Possible restorations are παρεπιφερ⋯μενοι (suggested by M. D. Reeve) or παρεπιφ⋯ρον<ται> πρ⋯ς ἴδιονδρ⋯μον <κα⋯>.

²⁴ For the translation of ⋯φ' οὖ, cf. Aristotle, , Ath. Pol. 18. 2Google Scholar; Demosthenes 10.20; Xenophon, , Anab. 5. 6. 30Google Scholar; Plutarch, , Sulla 29. 5Google Scholar, Demosthenes 22. 1.

²⁵ So far as I can tell, ὑψηλ⋯ς in this figurative sense ‘high (sc. on the open water)’ is unparalleled. In 33 (= Frisk 11. 12) ὕψος is used to mean ‘high sea’.

²⁸ δι⋯ τ⋯ς ἔξωθεν γ⋯ς: Müller emended to το⋯ and Frisk added the deletion of γ⋯ς. Perhaps τ⋯ς <θαλ⋯σσης>. As Müller aptly comments, ‘certe mira est verborum redundantia in istis’.