¹ I am indebted to Professor Adcock for his help and criticism.

² Partsch, J., Die Insel Korfu, Petermanns Mitteilungen Ergänzungsband, xix. (1887), p. 39 and 43Google Scholar, with geological map.

³ This accords with Strabo vii. 7, 5, who, describing the coast from north to south, mentions Buthrotum and then the Sybota islands ; sailing south one sights Cape Lefkimo and not Cape Bianco. The latter should be identified with Ptolemy's Cape Amphipagos, cf. Leake, , Travels in North Greece (1835) I 94Google Scholar, rather than Gomme, , Historical Commentary on Thucydides (1945) 183 n. 1Google Scholar, who places it on the west coast of Corcyra.

⁴ Thuc. I. 50, 3, ; 52, 1; and 54, 1, . Later authors mention only the islands Sybota, under which the anchorages both west and east of St. Nicolas Isle may have been comprised. When Leake lay at Port Mourzo, it was uninhabited. Thuc. III. 76: Alcidas anchored at mainland Sybota; he was then free to choose the best anchorage in the Channel.

⁵ Insets of these harbours will be found in Admiralty Chart 206 (1912).

⁶ de Bosset, C. P., Proceedings in Parga (1819), 46Google Scholar.

⁷ . In both cases the meaning of ἀντι- appears to be ‘opposite’ in the geographical sense. In the bay west of Cape Lefkimo there is excellent anchorage, according to the Mediterranean Pilot.

⁸ (I. 46, 3).

⁹ Jowett translates ‘Cheimerion is a harbour’ and adds a footnote ‘or “here there is a harbour”’; I prefer the former translation, which is supported by the analogous sentence at 50, 3, . I assume it was inhabited from Thucydides' silence, since in the case of mainland Sybota he states it to be uninhabited.

¹⁰ CQ xxxiii. (1939) 52, and Early Ionian Historians (1939) 50Google Scholar.

¹¹ NG iii 7 n. 1.

¹² Islets between Sybota and Cape Cheimerion are Megalo and Mikro Mourtemeno, Peramos and Katsonisi.

¹³ For this stretch of coast two other names are handed down. Scylax gives the corrupt ΕΛΕΔ for the harbour at the mouth of the Acheron. Today it is named Phanari after the name of the region and Splantza after the village of twenty hovels; in antiquity it may equally have had two names, Glykys Limen and that of Scylax, the latter being associated either with the fort on the south side or with the swamps which cut the harbour off from the interior. Ptolemy III. 14, 5 and Plutarch Antony 62 give Torone or Toryne, which is generally identified with Parga and came into use in the Roman period.

¹⁴ JGS. xviii 139 f.Google Scholar

¹⁵ Leake, NG III 6; Bursian, , Geographie Griech. (1862) I 28Google Scholar; Gomme, op. cit. 180.

¹⁶ A caique skipper with whom I sailed described the harbour as adequate in a storm (). Walking from Splantza to Parga I found no people near Agios Ioannes, except one shepherd a mile inland where a small dell afforded grazing.

¹⁷ CQ. xxxiv. (1940) 146 f.

¹⁸ The chronology of Gomme, op. cit. 197, is followed throughout.

¹⁹ Supplies could be provided by the area of Arpitza (a village of 600 families) and the hinterland plain of Margariti; as the post lay at the mouth of the South Channel, communication by sea with Actium was not endangered by the Corcyraeans. The two posts may have been linked for signalling by beacons, as in 427 B.C. when Alcidas lay at mainland Sybota and was warned of an Athenian fleet sailing from Leucas (III. 80, 2).

²⁰ I. 45, 3, , cf. 53, 4 may refer to the territory on the mainland which Corcyra held (III. 85, 2); the vagueness of the phrase, which is probably quoted from the original order, may be intended to cover Epidamnus also.

²¹ The evidence for the month is in IG I 2 295 = GHI. 55, cf. Gomme loc. cit.

²² Gomme, op. cit. 20, following Köster; under sail a trireme could cover up to ninety miles. Miles are here and below to be understood as sea miles. The direct course was probably that usually taken; in 427 B.C. Alcidas followed the coast from mainland Sybota to the isthmus of Leucas, in order to avoid the Athenian fleet which would take the direct route (III. 81, 1).

²³ To man her fleet Corinth also hired rowers who could be trained into good crews; for her Sicilian expedition Athens raised rowers from her empire to man die fleet which is so highly praised by Thucydides.

²⁴ It may be assumed that the Corinthians knew of the land base at Leukimme either from information or by observation from Cheimerion (cf. Plate IIa and b).

³⁵ In the map I have put the Corinthian line -at three miles in length. The nature of the battle indicates that both sides formed a single line for boarding tactics (the Athenians alone acting as a mobile reserve); for 150 ships a line of three miles allows forty yards per ship. Taking the dimensions of the trireme at 110 feet long and fifty feet wide from oar-tip to oar-tip, the interval of clear water between ship and ship sailing ahead would be twenty-four yards and between ship and ship turning broadside for boarding would be four yards. As boarding tactics only were employed in this battle, this seems to be a reasonable estimate; even so two ships turning broadside simultaneously would require skilful handling, for it was important to save one's own oars and timbers and to damage the enemy's. During the Peloponnesian War the tactics of breaking the line and ramming required a closer order in the line for defence; thus one cannot apply to Sybota deductions drawn, for instance, from the battle of Arginusae. On the Corcyraean side the Athenian squadron might occupy more sea room, in order to keep its freedom of manoeuvre; even so the Corcyraean ships would be spaced more widely than the Corinthian, probably between forty-five and fifty yards per vessel on the average. Thucydides I. 50, 2, , suggests that the line was unusually long. Cf. Rodgers, Greek and Roman Naval Warfare (1937) 47 and 187 (Arginusae)Google Scholar, and Köster, , Antike Seewesen (1923) 137 f.Google Scholar

²⁶ I. 49, 6 , ; the Corinthians presumably held with their ninety ships the left centre as well as the left wing. At 48, 4 the order of battle is given from right to left by national squadrons and I take it that the twenty-one ships of Elis Leucas and Anactorion occupied the right centre or part of it.

²⁷ Gomme, op cit. 194, and others have doubted the accuracy of the figure seventy. Thucydides' narrative suggests that the Corcyraean losses were very high; he mentions the great superiority of the Corinthians (49, 6) the Corcyraeans' effort to muster ships in offering battle in the evening (50, 4), the withdrawal of the Corinthians who feared the second Athenian squadron numbered more than twenty (51, 1), and the considerations which dissuaded the Corinthians from accepting battle on the next day (52, 2). This establishes a presumption that Thucydides did give a high figure for the Corcyraean losses, and, if we accept his figure thirty for the Corinthian losses, seventy for the Corcyraean losses need not be suspected; as the Corinthians killed survivors in the water (50, 1), the 1050 prisoners taken by them (55, 1) lend some support to the figure seventy, If we allow that the figure seventy was written by Thucydides and is not a MS error, is there any probability that he was correct ? He qualifies the numbers of losses with περί and μάλιστα, but the general trustworthiness of his account suggests that these numbers should be respected. In the Peloponnesian War losses in naval engagements were not so high in proportion to the size of the fleet as those of Corcyra at Sybota; but in the naval victory over Aegina c. 458 B.C. the Aeginetan losses were seventy ships (I. 105, 2) whereas their fleet was probably smaller than that of Corcyra (I. 50, 2. Thuc. notes Sybota was the greatest battle to date between Greek and Greek). It seems likely that the archaic tactics employed at Sybota caused higher loss in ships than the manoeuvring and ramming of the Peloponnesian War; all ships engaged, the fighting was bitter, and after charging one another disengagement was difficult (49, 3). Thus withdrawal was less easy than it later became.

²⁸ ἃς καταδύσειαν, Sunk in the sense of waterlogged; when a ship was carried by boarding, it was presumably holed to put it out of action. The towing in of waterlogged hulls must have been a slow process, which helps to explain the interval between the dawn engagement and the action which broke off in the evening.

²⁹ Allowing one and a half hours for the pursuit (8 miles), a half hour for pillaging the camp which would be on the beach, and one hour to reach the Sybota group (6 miles); for speeds cf. Köster, , Klio Beiheft 32 (1934) 87Google Scholar, who estimates that the Corinthian fleet in the second action against Phormio covered three to four miles in half-an-hour.

³⁰ Thucydides uses πλώιμος in a specialised sense at 29, 3, , where the meaning is ‘fit for manoeuvre in action’; here ‘the ships remaining’ appear to be those not fit for action by normal standards, but used now in emergency. At 52, 1 when the Athenian second squadron had arrived the Corcyraeans only employed the ships that were fit for action . Jowett translates ‘any others which they had in their docks’; as the docks would be at Corcyra town, it is doubtful if there was time to bring ships thence to Leukimme (fifteen miles); it seems more likely that ships not up to battle standard in formation at open sea were posted at Leukimme as a possible reserve.

³¹ Köster, , Antike Seewesen 138, n. 3Google Scholar, holds that at a distance of five miles low-lying land can be escried by a look-out with his eye five metres above sea level. This applies to the present case; a hull on an open horizon is as easy to pick up as low-lying land, and since Köster estimates the storm deck of a trireme at 2·20 m. high the Aphlaston (stern poop) is seen from reliefs to be at least 5 m. above sea level (cf. illustration at p. 141).

³² Cf. n. 30 above. Assuming that the forty ships which survived the main action were fit for action, these with the thirty Athenian ships make up a total of seventy.

³³ Mediterranean Pilot III. 132Google Scholar (current) and 509 weather at Corfu:

The N.W. wind may be excluded from the reckoning for the South Channel, which is sheltered in that quarter by the island of Corfu; the incidence of southerly to northerly winds is 18 to 18 in August and 30 to 12 in September. As the battle occurred at the turn of the month, the predominant wind then was southerly in the South Channel.

³⁴ Beloch, , Griech, Gesch. (1916) II. 2 222 f.Google Scholar rejects the figures of Thucydides for the fleets and their losses; his view has been well refuted by Gomme, op. cit. 191 f. Beloch also considers Thucydides' narrative to be a conflation of two accounts from different sources; in one of these Cheimerion was the base from which the first action was delivered, and in the other mainland Sybota was the base. In his argument Beloch misrepresents Thucydides as stating that huts were erected and stores shipped to mainland Sybota. Beloch is here carried away by his quest for doublets, and we need not follow him.

Gomme takes Agios Ioannes as Cheimerion harbour and camp, Cape Varlam or Trophale as Cheimerion cape, and Arilla Bay as mainland Sybota (180 f. and 186); finding it difficult to fit into this picture the pursuit and return of the victorious Corcyraean squadron, he argues that the pursuit was in fact to mainland Sybota and not Cheimerion, and concludes ‘this is the principal error of Thucydides’ (195). He considers Agios Ioannes ‘excellently placed for a fleet that was to hinder the Kerkyraians operating further south’ in 434 B.C. (181); in my opinion it is unsuitable either for observation or for forays by a small squadron capable of relatively small speeds. He finds the Corcyraean camp at Sybota islands improbable (194 and 182), as the landing place is small and the islets too rocky to provide sleeping-quarters; but on a campaign comfort and convenience are not primary considerations. ‘There seems to be no reason why the Kerkyraians should have chosen them (the Sybota islands) as an advanced base instead of waiting for the Corinthians to attack them off Leukimme. The farther they could make the Corinthians row before battle, the better for them; and if the enemy passed Sybota, sailing along the coast to threaten the town of Kerkyra, they could easily keep pace with them along the eastern shores of the island’ (182). My view on the strategy of the Corcyraeans is given above; the Corcyraean land force could not keep pace on land with a fleet sailing direct to Corcyra town. In differing with Pearson (179) Gomme does not take Strabo into account, although he refers to Strabo subsequently, and so concludes that Thucydides ‘did not fully understand’ his informant, ‘perhaps a Corinthian.’ The camp of the Corcyraean land force is put by Gomme west of Cape Lefkimo (183 and 187); I can see no advantage in putting it west rather than south of the actual point, and it raises difficulties for Thucydides' description of the camps as ‘opposite’ in 434 B.C. and for the sighting of the second Athenian squadron in action 2. At 50, 1–3 Gomme detects ‘a detail which came to Thucydides' notice after he had written his main account of the battle, and so from another source,’ and questions whether ἐς τὴν γῆν refers to Sybota or Corcyra near Leukimme (185). I have given my reason in the text for the Corinthian order of actions and attribute it to one and the same source, presumably Corinthian; in its immediate context I consider ἐς τὴν γῆν can only mean Corcyraean territory, that is Leukimme.

I also express my gratitude to the Oxford Philological Society and to Professor Gomme who have given me the benefit of their criticism.