* I am indebted for help with this study to various people and bodies. Some will be mentioned at the appropriate place; here I mention the British School of Archaeology at Athens and its staff; Dr R. A. Donkin of Jesus College and the Department of Geography, Cambridge, for help with the maps; Constantine and Sibylle Mano for help and hospitality on Skiathos; my sister Elizabeth Bowen who with Richard, Heather and Matthew Clay, Marie-Catherine Gauthier, and David Livermore made stimulating company in August 1992 on board a yacht hired to work the waters between Khalkis and the Peneios mouth; Professor W. K. Pritchett for very helpful comments on an earlier draft; and Dr Paul Millett, who with CQ's referee has brought it into acceptable form and focus. I am grateful for financial help to the Faculty of Classics and to Jesus College, Cambridge, and especially to the Clay Foundation.

¹ N. Whatley, JHS 84 (1964), 119–39.

² J. F. Lazenby, The Defence of Greece (Warminster, 1993)

³ See chs. 2 and 17 of T. Shaw (ed.), The Trireme Project (Oxbow monograph 31, Oxford, 1993).

⁴ F. Maurice, JHS 50 (1930), 210–35. Maurice used his experience of campaigning in Gallipoli in the First World War to set an upper limit to the forces which could have passed through the area in seven days; water supply was the critical factor. A similar approach is adopted by T. Cuyler Young, ‘480–479 B.C––A Persian Perspective’, Iranica Antigua15 (1980), 213–19; he considers the size of both army and navy in terms of supplying it with food and water. G. L. Cawkwell (Xenophon, The Persian Expedition[Harmondsworth, 1972], pp. 36–8) analyses the battle of Cunaxa in order to establish a likely size for Persian armies.

⁵ The essential points about food, water, and land are made in discussion of notable relevant texts in J. S. Morrison and J. F. Coates, The Athenian Trireme (Cambridge, 1986), pp. 95ff

⁶ I am grateful for help with this and the next paragraph to Professor T. H.Van Andel of the Department of Earth Sciences, Cambridge

⁷ See Ilhan Kayan, Studia Troica(Mainz, 1991), p. 90 for a useful diagram. I am grateful for this reference to Dr Sturt Manning of the Department of Classics, Reading. For Thermopylai see J. G. Kraft, G. Rapp Jr, G. J. Szemler, C. Tsiavos, and E. W. Kase, Journal of Field Archaeology14 (1987), 181–98. But the great complexity of the larger scene, together with the chance of local anomalies, can be well seen in N. C. Flemming, ‘Holocene eustatic changes and coastal tectonics in the north-east Mediterranean: implications for models of crustal consumption’, Philosophical Transactions of the Royal Society of London1362 (1978). For a particular local feature, see F. J. Frost, Plutarch's Themistocles(Princeton, 1980), pp. 110–11, rebutting a suggestion of A. J. Podlecki, Life of Themistocles(Montreal, 1975), p. 176 that the coastline has receded at Artemision; he says ‘a small headland, about eight kilometres to the south-west, now an island, was indisputably connected to the mainland in antiquity’. In correspondence he tells me ‘There are sherds and tile fragments on the bottom between the little headland...and the shore but more definitely there is an Asklepieion with footings more than 2 m deep near the shore in Skopelos harbour.’ While we were sailing in the area in August 1992 an underwater site of the second millennium B.C. was being investigated by the south shore of Peristeri off Halonnesos.

⁸ A. Dascalakis (Problkmes historiques autour de la bataille des Thermopyles [Paris, 1962], p. 123, n. 1), summarizing earlier work of E. G. Mariopoulos (Etude sur le climat de la Grec [Paris, 1925] and of V. Eginitis [Annuaire Scientifique d' Athenes 1946–7]), says 'le climat de la Grece n'a subi aucun changement e t...les conditions climatologiques d'aujourd'hui sont celles de l‘ancienne Grece’. V. McGeehan-Livritzis, citing more recent work (Journal of Nautical Archaeology 17 [1988], 237–56) quotes ‘the present climate is similar to that of the 3rd millennium B.C., so our wind data are reliable’, and ‘sea currents will be essentially the same’. H. E. Wright (Antiquity 42 [1968], 123–7) reports virtually no change in the pollen record in the last 3,000 years.

⁹ Murray, W. M. (‘Do modern winds equal ancient winds?’, Mediterranean Historical Review 2.2 [1987], 139–67) says that the results of his study ‘fully support the view that the winds of classical antiquity were essentially the same as they are today’. See also W. T. Loomis in His tor ia 39 (1990), 489–490, n. 18.Google Scholar

¹⁰ See R. Heikell, Greek Waters Pilot (Huntingdon, 19904), pp. 205 and 246

¹¹ The most recent edition (1993) has gone coloured, with loss of many details from the previously engraved editions. The chart is to be read in conjunction with the Mediterranean Pilot vol. IV, now in its tenth edition (cited hereafter as MPIV).Google Scholar

¹² Herodotus' report of the ebb and flow in the gulf of Malis (7.198.1) exposes a different phenomenon, caused by wind and barometric pressure operating on enclosed and shallow waters.

¹³ The drier the ship, the faster through the water; dryness of ship is basic to winning a seafight. The training and experience of the crew matter of course, as said by Lazenby (n. 2), pp.139 and 250, and see Shaw (n. 3), ch. 17; but the evidence of Th. 7.12.3–4 is prime: see Morrison and Coates (n. 5), pp. 64, n. 4, 152–3, and 230–3.

¹⁴ If they were south of it, they would be trying to escape ‘best sailing’ ships by working upstream, possibly upwind, and in the direction of the enemy base, which would be odd; downstream and downwind they would get ever nearer their own base and might even evade capture long enough to deter their pursuers from risking further chase. The suggestion (J. A. R. Munro, JHS 22 [1902], 308 and J. F. Lazenby Hermes 92 [1964], 275) that the pursuers were returning from a trip to Skyros seems unlikely: it ignores the risk to the Persians in going so far forward before they had cleared any enemy out of Skiathos or its neighbouring islands. From the Peneios delta to Skiathos is over 60 miles, and Skyros is 60 miles beyond that. Herodotus' words 6piJ.wfji.fvos fK @4pixr s napefiaXe...Wii Zxiadov (‘based on Therma they worked the coast in a line to Skiathos’) imply exactly the sort of Persian interest to be expected. H. Horhager (‘Zu den Flottenoperationen am Kap Artemision’, Chiron 3 [1973], 43–59) overlooks their need to check the coast of Magnesia for anchorages and obstacles. A small Greek presence at sea so far north is not, at this stage of the campaign, surprising: some spy work was needed.

¹⁵ Herodotus' words in 7.183.1 appear to place the Greek fleet at Artemision at this stage, but I think no more than an advance force is to be understood, preparing the encampment: such a force might well have retired as described.

¹⁶ I accept the proposal of J. N. O'Sullivan, CQ 27 (1977), 92–4, to read for . The subject of πθηκαν in 183.2 is by implication the same three Persian ships.

¹⁷ The word is very rare, and unique both in Herodotus and as an adverb. I take it to mean a full day's worth of travel, from before dawn to before dusk. It seems to have been standard practice to start before dawn (see below and nn. 46 and 47), and it was clearly sensible to anchor and disembark with daylight still remaining for the last ships in.

¹⁸ It is often said (first in Diodorus 11.12.5–6) that the fleet anchored at several beaches, both on the first night under Pelion (note, however, the discomfort of W. K. Pritchett, AJA 67 [1963], 1–6 in suggesting that Koulouri below the village of Keramidhi was the fleet's headquarters) and at the second anchorage on the south coast of Magnesia or even partly in Akhaia Phthiotis (see P. W. Wallace, GRBS Monograph 10 [1984], 305–10); but the notion gets no support from Herodotus (or from Euripides: O.KTO.V aXifievov 77ijAt'ou, Ale. 595–6) and overlooks the question of contact, control, and security within the fleet as a whole. Herodotus' only mention of separate anchorages is at 7.123.3, in a very different context. The beaches under Pelion are in any case too far from Therma for a reasonable day's voyage for the whole fleet and none can properly be called alyiaXos (see n. 24).

¹⁹ See Heikell (n. 10), p. 223

²⁰ The canal through the neck of Athos was allegedly built for two-way traffic (7.24), like Nekos' canal in Egypt (2.158.1). Investigation of the canal is in hand; soundings suggest a top width of 30 m, just enough (see below) for two triremes to pass with care if the canal was full to the brim. See B. J. Isserlin (to whom I am grateful for correspondence on the matter) et al. in BSA 86 (1991), 83–91, 89 (1994), 277–84 and 91 (1996), 329–40.

²¹ L. Casson, TAPA 82 (1951), 136–48, reckons that two and a half knots was the best speed possible for a sailing vessel working against a moderate wind and current; in favourable conditions four knots was reached in coastal waters and up to six on the open sea. The word OXKOLS (see 7.25.2 and 191.1) indicates that towing was a recognized practice, but we do not know what speed could be achieved by triremes doing it.

²² Over 10 km measured on a car odometer. The beach is backed by a parallel road the whole way. Pritchett (n. 18) estimated it from a map at 6 km.

²³ Two of the streams had water in them in spring 1995; these are the two streams marked on the map in D. Miiller, Topographischer Bildkommentar zu den Historien Herodots (Tubingen,1987), p. 360. More important than rivers are springs and wells, rivers being very seasonal in flow in Greece and often dry in summer, whereas springs and wells may serve the year round.

²⁴ Herodotus’ word αἱγαγ⋯ς for beach is worth note. He uses it elsewhere only of Doriskos, Argilos, Artemision, and Mykale. I have not seen Mykale; the others are like Ayi⋯kambos, long, gently shelving, and unobstructed. Pritchett (n. 18) has a good discussion of the word.

²⁵ The battlefield of Plataiai is a good example of features carefully named then but now causing much discussion or resisting identification altogether. See A. R. Burn, Persia and the Greeks (London, 1984), p. 522.

²⁶ at 7.179 means ‘using Therma as their base’ as the present participle shows; departure of the fleet as a whole is marked by the aorist

²⁷ Connecting the two with re xal is in Herodotean idiom; so too would be a change of preposition to mark the change of focus. After Kal I suspect that h has been lost: see 1.2.2, 1.27.3, 2.7.1, 8.7.1, 8.66.1, and perhaps 7.123.3; also 7.188.1.

²⁸ Lazenby (n. 2), p. 221

²⁹ See Kirsten, E. editing Philippson, A., Die Griechische Landschaften vol. l. i (Frankfurt am Main, 1950), p.161, n. 1.Google Scholar

³⁰ History of Herodotus vol. 4(London,1880), p.149.

³¹ J. E. Powell, Lexicon to Herodotus (Cambridge, 1937), s.v. I add to his list in 7.25.2, both of which he lists under their adjective7.25.2, both of which he lists under their adjectives.

³² For Attica as ansee E. Hel. 1673. See Appendix

³³ There is an unfortunate extra distraction in it. Nowadays it is the name for what Herodotus at 7.193.1 calls simplyThe earliest Royal Navy surveyor, Cdr. R.Copeland in 1831, labels that headland Cape St George, after a nearby church; subsequent surveys (by Cdr. Brookes in 1846 and Lt. Drew in 1862) say Cape Sepias, which has established itself. I am grateful to the Hydrographer of the Navy for permission to inspect these surveys in the archives at Taunton

³⁴ See Pritchett, W K., Studies in Ancient Greek Topography vol.2(Berkeley,1969), p.18, n.29, andMuller(n. 23), p. 286.Google Scholar

³⁵ The beach at Phaleron, which he doubtless knew well, was apparently adequate for the fleet which came together after the losses off Pelion, Euboia, and Artemision; it is, or rather, seems to have been before encroachments this century, about two miles long. If Herodotus is essentially right that the Persian fleet was not much reduced in numbers for Salamis (not because its losses had been made good, as he says at 8.66, but because rather few had been lost), then at Phaleron too, for anything much more than 300 ships, a Mediterranean moor was needed.

³⁶ See his article in The Mariner's Mirror 74 (1988), 283ff. For the probable origin of the phrase see J. S. Morrison and R. T. Williams, Greek Oared Ships (Cambridge, 1968), p. 56n. See also Shaw (n. 3), ch. 18.

³⁷ By Cape Sepias here and henceforward I mean the cape referred to by Herodotus asSee n. 36.

³⁸ See Th. 2.84.3 for poles on board triremes.

³⁹ They could not be hauled up the beach unless they were lying stern to, because of the rams at the prow. For prow to shore see 7.100.3, where the abnormality of the mooring (for a royal inspection) is described with care.

⁴⁰ What it means at 4.152.4 is a further question; probably each griffin's head in the upper row was set over the gap between two in the lower row.

⁴¹ Dascalakis (n. 8), p.123, n. 2; Horhager (n. 14), pp. 49–50

⁴² Burn (n. 25), p. 389

⁴³ This and all further such numerical references in this paragraph are to the pages of Weather in the Mediterranean vol.1, Air Ministry, Meteorological Office (London,1962), 2 vols (2nd edn). Volume 2 has climatic tables: those for Skiros (Skyros) are the most relevant to this study. I am grateful to ProfessorChorley, R. J., lately of the Department of Geography, Cambridge, for his help in preparing this paragraphGoogle Scholar

⁴⁴ Dascalakis (n. 8), esp. pp. 120–7

⁴⁵ As Dascalakis shows ([n. 8], p. 127), Herodotus' ‘thunder from Pelion’ is not an indication of wind direction; it marks rather how those there were impressed by it. Thunder is associated with mountains; Pelion was likely to be prominent in Persian consciousness

⁴⁶ See R. W Wallace, TAPA 119 (1989), 201–7, for a good study of the word. Although he does not cite this instance, it fits his analysis perfectly.

⁴⁷ About 3.30 a.m., I suggest (experto crede and see Horhager [n. 14], p. 50). They would want to use the cool of the dawn and to conserve themselves for what might be a difficult end to the day.

⁴⁸ (in the area of Sepias itself): some MSS have eg. For nepi, probably lectio difficilior, see 8.16.2; these two instances of are the only two in Herodotus or, apparently, anywhere. If is is preferred, the argument is hardly affected: the change of verb is more important.

⁴⁹ N. Georgiadhis, Thessaiia, Volos (1894, 2nd edn), pp. 19 and 142.

⁵⁰ Pritchett (n 18).

⁵¹ See Pritchett (n. 18), plate 2, fig. 2, and Miiller (n. 23), pp. 330–1. Miiller includes for good measure a picture of a peasant oven.

⁵² The wordused of the ships wrecked on the great beach, may add a little strength to the identification offered. Herodotus repeats the word in 190 for the objects obtained by Ameinokles' beachcombing; he uses it nowhere else (Plutarch noted it: see Moralia 864C). The repetition suggests that Ayiokambos was the place of that beachcombing.

⁵³ For the inscription see A. M. Woodward, Liverpool Annals of Archaeology and Anthropology vol. 3 (1910), p. 158. This series, long defunct, may be found in the Liverpool City Record Office, and I am grateful to the archivist, Miss J. Smith, for a photocopy of the relevant pages. For the coordinates see R. Stilwell (ed.), The Princeton Encyclopedia of Classical Sites (Princeton, 1976). Its articles on the sites at issue in this study are mostly cautious.

⁵⁴ This inscription is cited by A. J. B. Wace, ‘The topography of Pelion and Magnesia’, JHS 26 (1906), 145, n. 10, referring to ApXaioXoyiK6v AeXriov 5(1889)92. W. W. Tarn in JHS2& (1908), 210, says firmly ‘Meliboia is Thanatu; epigraphic evidence fortunately renders this certain.’ He should have waited two years for Woodward. I am grateful to CQ's referee for alerting me to this inscription

⁵⁵ Lazenby ([n. 2], p. 126) calls 400 Herodotus' estimate, and appears happy to accept the figure anyway, as others have. Burn ([n. 25], p. 390) observes more cautiously, and in my view more soundly, ‘The estimate was exaggerated in proportion to the original exaggeration of the size of the fleet.’

⁵⁶ The error of Sandokes (7.194–5) deserves note. He must have lost contact with the preceding squadron (perhaps he was charged with picking up any survivors on the Magnesian coast; perhaps he had damaged ships to nurse), but why should he have thought that ships halfway across the straits to the south were part of the Persian fleet unless the original plan had been to go to Euboia if possible?

⁵⁷ Pritchett (n. 34), ch. 2, has a useful map; see too plates 9 and 10.

⁵⁸ H. G. Lolling, AMS (1883), 7–23 and 200–10

⁵⁹ Pritchett [n. 34, ch. 2 expresses some worry about the identification: Lolling's road to the south of the site now runs to die north of it. But Dr E. B. French tells me in correspondence, mentioning her father's map of 1895 which has the road to the south still, that it has been common for roads to migrate closer to the shore since then.

⁶⁰ 5 vols. and maps, (Athens, 1987: survey conducted between October 1979 and March 1981). I am grateful to Messrs Pergaliotis, Drosos, and Banos for permission to study it in their office in Athens

⁶¹ In August 1980 47,000 cubic metres flowed in the Xeropotamos, the main river of the plain, as measured under the bridge on the east side of Istiaia

⁶² s suggests that it was not significantly inhabited, and so could have been prepared in fair safety even before Xerxes entered Thessaly.

⁶³ Tim Severin also seems to have used it (The Jason Voyage [London, 1985]): though neither text (p. 80) nor map (p. 58) mentions Platania by name, the map makes it plain.

⁶⁴ The case is thoroughly presented by Pritchett (see n. 59). I am grateful for the help of Professor C. J. Tuplin of Liverpool University in correspondence on the topic

⁶⁵ For example, Burn (n. 25), p. 391 and map 340.

⁶⁶ See n. 33.

⁶⁷ See also E. Fabricius, Gnomon 2 (1926) 11–15, reviewing F. Stahlin, Das hellenische Thessalien (Stuttgart, 1924). Fabricius is rightly cited by Pritchett (n. 34), vol. 5, p. 95, n. 5.

⁶⁸ I am grateful for the information that does exist to Mr Apostolos Dhimoulas of the Geological section in the Ministry of Agriculture's office in V 610 s.

⁶⁹ Maurice (n. 4), p. 221.

⁷⁰ Severin (n. 63), p. 181

⁷¹ See for instance J. F. Coates, S. K. Platis and T. Shaw, The Trireme Trials 1988 (Chippenham, 1990), p. 43. My Greek informant, Mr Xypolitidhis, may have been referring to the voyage mentioned there, but I omitted to ask.

⁷² Shaw (n. 3), p. 108

⁷³ I have a photograph taken in London in 1993 when Olympias graced the Thames which shows a 2 litre bottle in position beside a zygian porthole.

⁷⁴ See Maurice (n. 4), p. 221, who seems to assume a lesser need on the part of orientals.

⁷⁵ This bluff does not appear on the otherwise useful map of Fabricius (see n. 67).

⁷⁶ I am grateful to Mr A. Potamianos for a trip on his yacht Thelginos (built in Glasgow in 1926) specially to check these depths in July 1993; also to Stuart Mann, Thasos Economou, and Dimitri Mellios who helped me to do it.

⁷⁷ See Heikell (n. 10), p. 227.

⁷⁸ Ibid., p. 205.

⁷⁹ See E. Koster, Studien zur Geschichte des antiken Seewesens, Klio Beiheft 32 (Leipzig, 1934), 73; Lazenby (n. 2), p. 130 says ‘perhaps’.

⁸⁰ See MP IV 10.87: ‘Drift seaweed, wood and other debris may be encountered in the strait [between Skiathos and Magnesia] during NE winds’. See also ibid. 10.79 and Heikell (n. 10), p.

⁸¹ Of several photographs in F. Welsh, Building the Trireme (London, 1988), that on p. 205 shows the inboard position of thalamian oars most clearly, but that on p. 275 is more relevant because it shows the ship without its burden of crew aboard.

⁸² Nearly the same point was made by Koster (n. 79), pp. 73f. We may reckon that on the second evening when the Greeks in their raid disabled some Cilician ships, the Cilicians were on patrol duty

⁸³ Hignett, C.,Xerxes' Invasion of Greece(Oxford,1963), app. VI, following K. J. Beloch, Griechische Geschichte2 II.2 (Berlin, 1931), pp.87–90Google Scholar

⁸⁴ Pace Lazenby (n. 2), pp. 174–5.

⁸⁵ Th. 1.48.1 and 3.49.3 show that triremes could carry food in a particular situation; but such moments seem to be very rare

⁸⁶ 86 See 7.184.2

⁸⁷ Pammon is otherwise recorded only as the name of a son of Priam (Iliad 24.250). It may not be Greek (I am grateful to Dr D. J. Thompson for this observation), which would bear out what Thucydides says (1.98.2) about the Skyrians. That Pammon was only one of several who helped the invaders is marked byPowell's (n. 31) classifications of are not satisfactory forsatisfactory: for preceding a proper name and modifying it see 2.49.3, 2.50.1, 5.120, and 8.136.2.

⁸⁸ See MPIV 10.50 and 51.

⁸⁹ See MPIV 10.16.

⁹⁰ See MPIV 7.522–5.

⁹¹ The Greek ships at Pogon (8.42.1) and those presumed to be on Aigina (46.1) which did not join Eurybiades at Artemision were not posted to prevent a circumnavigation, nor could the Athenian reserves have done that job had they stayed by Attica, as appointed in the Trozen inscription.

⁹² The total length of voyage is not a problem in itself: the famous non-stop voyage from the Piraeus to Mytilene (Thuc. 3.49) was longer. See Morrison and Coates (n. 5), pp. 95–6. But the weariness of these crews would be a huge problem during the final miles under Kaphereus and Geraistos.

⁹³ H. J. Mason and M. B. Wallace, Hesperia 41 (1972), 12810.

⁹⁴ See Morrison and Coates (n. 5), pp. 103–4.

⁹⁵ The facts are set out clearly in W. W. How and J. Wells, A Commentary on Herodotus, 2 vols (Oxford, 1928), vol. 2, pp.372–3.

⁹⁶ Herodotus points explicitly to only two occasions when triremes were at sea in the night, the night before Salamis (8.70.1) and here; implicitly there is the Greek plan, unfulfilled, to meet the ships circumnavigating Euboia. Sailing at night, especially in a trireme, would not be undertaken lightly.

⁹⁷ For Tpiraios see Lazenby (n. 2), p. 123. But the force of KO.1 SIJ is relevant, and the phrase needs to be considered as a whole. Powell (n. 31) records the instances of KU SIJ under both KCU and S77. Unfortunately, his two lists do not tally, his count is incomplete (he misses 5.67.5), and his classification unsatisfactory. In total (but omitting the disputed 7.10)91) there are 51 instances, of which 39 are both initial and connective in sense, and a 40th, 4.11.2, probably is (see H. Stein, Herodotos [Berlin, 1868], ad loc.). That at 7.14, introducing a question in direct speech, stands on its own. The other ten (4.102.1, 7.196, 8.94.3,9.6,781,11.2,48.2,66.3, 89.1,102.1: the number of instances from Book 9 is noteworthy) are either non-connective or ambiguous, to adopt the cautious initial classification of J. D. Denniston {Greek Particles2 [Oxford, 1954], p. 248). 7.196 is best paralleled by 9.6 and 102.1. Denniston says (p. 252) of this KO.1 817 that it ‘frequently approximates in sense (particularly in the historians) to 7(817, though it is always more vivid and dramatic in tone’. I observe that in Herodotus this ical 817 quite often occurs in direct or quoted speech, with a sort of challenge: Xerxes was there within—yes!—two days! I think that Herodotus found no peg on which to hang Xerxes' arrival in Malis and used rpiratos as it was used to him. If the journey had been planned to take four days, however, then the diaries would tally and surprise at Xerxes' speed would make sense.

⁹⁸ Curiously, a route via Skyros is scouted by N. G. L. Hammond (CAHIV2, 550), but in a different context.

⁹⁹ MP 10.30 calls it Ormos Aragma, Arazo Road, and Ormos Araxo; Admiralty Chart 2048 uses the two latter names. The bay is guarded by several small islands, whose negotiation in the dark would be perilous. I estimate that up to sixty triremes could lie side by side at the beach's north-west end; there are reefs further east. There is abundant fresh water at the sea's edge a little further east along the beach, as marked on the chart

¹⁰⁰ Did they have Pammon to thank for it all? If so, Herodotus might have treated him as he did Ephialtes, the traitor at Thermopylai (7.213–4); but perhaps Poseidon and Boreas took him.

¹⁰¹ See Wace (n. 54), p. 146.