¹ This article, a very delayed publication, is based on research I conducted in Athens in 1967, and though I have several times presented the material at various conferences and lectures, it has not previously appeared in print. To all those who have discussed it with me, as to the anonymous reviewer of this article, I offer my appreciative thanks; and in particular to the cheerful encouragement of Dr Luis A. Losada.

In addition to the customary abbreviations, the following will be used:

Hammond = Hammond, N. G. L., ‘The campaign and the battle of Marathon’, JHS 88 (1968), 13–57CrossRefGoogle Scholar.

Evans = Evans, J. A. S., ‘Herodotus and the battle of Marathon’, Histona, 42 (1993), 279–307Google Scholar.

Lazenby = Lazenby, J. F., The Defence of Greece 430-479 BC (Warminster, 1993)Google Scholar.

A very full bibliography is to be found in Lazenby, 263-71.

² Denied, for example, by Delbrück, H., Geschichte der Kriegskunst (Berlin, 1920) i. 73 n. 9Google Scholar. Maurice, F., ‘The campaign of Marathon’, JHS 52 (1932), 17CrossRefGoogle Scholar, found it ‘hard to swallow’, and to a writer with first-hand experience of intelligence work in Greece, Mackenzie, Compton, Marathon and Salamis (Edinburgh, 1934), 83Google Scholar, it was ‘a typical piece of popular mythopoeics’; so also Myres, J. L., Herodotus (Oxford, 1953), 208Google Scholar. More recendy, Lazenby, 73, in a careful and discerning analysis, thinks that ‘all in all, it seems unlikely that there was such a signal’. Evans, 303-4, takes a modified view: ‘It bears all the marks of a carefully cultivated political myth, even though its core was an actual sighting’. He sums up: ‘the shield signal remains a mystery’.

³ Hammond, 28.

⁴ Of these critics, however, I might suggest that their ferocity may not have been equalled by their accuracy. Soldiers' personal recollections of a battle fought long ago are not always reliable, much less their appreciation of the overall strategy involved. And, given the time lapse, Herodotus' auditors must have been in the very junior ranks on the day of Marathon, not privy to the debates and decisions in the generals' councils.

⁵ Cf. Hdt. iii. 80 and vi. 43 for his insistence on other points sometimes doubted by modern commentators (sec How, W. W. and Wells, J., A Commentary on Herodotus (Oxford, 1912)Google Scholar, ad loc.

⁶ Lazcnby, 72; Evans, 289.

⁷ Evans, 289. Hignett, C., Xerxes' Invasion of Greece (Oxford, 1963), 73Google Scholar, notes that ‘there is no evidence for the modern view that the signal was sent from the summit of Mount Pentelikon’, and the same is true of Agrieliki. Pendeli: How and Wells (n. 5), ii. 361; Reynolds, P. K. Baillic, ‘The shield signal at the battle of Marathon’, JHS 49 (1929), 100CrossRefGoogle Scholar. Agrieliki: e.g. Hudson, H. G., ‘The shield signal at Marathon’, Amer. Hist. Review, 42 (1917), 450Google Scholar. Aubrey de Selincourt, the World of Herodotus (London, 1962), 258, put the flashing shield ‘presumably’ on the Acropolis in Athens, which, with the whole bulk of Pendeli in between, is of course quite impossible (unless the signal is relayed). Actually, if the signal were sent from the top of Pendeli it could not have been from the summit, from which the area of the Persian camp cannot be seen. The only spot from which Athens and the area occupied by the Persians are both visible is one of the lesser peaks, Mavrovouni. I am grateful to Professor E. Vanderpool for bringing this to my attention and taking me there for a look around.

⁸ Listed by Merriam, A. C., ‘Telegraphing among the Ancients’, Papers of the Archaeological Institute of America, Classical Series 3 (1890), 10Google Scholar. He is chiefly concerned with tracing the course of the beacon chain in Aeschylus' Agamemnon but also includes a consideration of all forms of telegraphy. I should perhaps here comment on Professor Hammond's response (Hammond, 37 n. 108) to the above, communicated to him while he was writing his Marathon article. Strongly maintaining that there was a flash, he argues that when Herodotus mentions a shield it must have been not a real one but a ‘round, flat polished disc’, quoting in support the use of aspis as ‘a round, flat dish’ by the fourth-century writer Aristophon (as quoted in Athen. Deipn. xi. 472 D). There are two things wrong with this. First, the context makes it clear that the object concerned is 111 fact a kylix from which wine can be drunk; and one cannot drink wine from a flat plate. Second, the word is here used in a parody of its normal meaning, and is so identified by C. B. Gulick, the Loeb editor of Athenacus (v. 85 n. c). And we cannot reject a word, ‘shield’, that is clearly present in Herodotus, in order to save one, ‘flash’, that is not. ‘The ‘flash’ interpretation seems to have been started by Colonel Leake (so Evans, 288). It was still stated as fact by Hammond in CAH ² (1988), 512.

⁹ So Burn, A. R., Persia and the Greeks (London, 1962), 251Google Scholar; Pritchett, W. Kendrick, Marathon (Univ. of California Publ. in Classical Archaeology, 4; Berkeley, 1960), 171Google Scholar. Lazenby, 72–3, rightly insists on the ‘lift up’ meaning as opposed to ‘flash’.

¹⁰ Hammond, 37, id., CAH ², 512. The view is attacked in Hodge, A. Trevor and Losada, Luis A., ‘The time of the shield signal at Marathon’, AJA 74 (1970), 31–6CrossRefGoogle Scholar (esp. 35 n. 17). The inability of shields to flash is picked up by Evans, 289. Lazenby, 73 n. 65, notes that since there is no evidence the signal came from Pentelikon (quoting Hignett, n. 7 above), ‘modern attempts to time it, based on the position of the sun, are worthless’. I agree, but my own view is rather stronger: even if the shield was on Pendeli it could not, under any circumstances, send a flash visible at Marathon.

¹¹ Bailie Reynolds (n. 7), 102.

¹² So Lazenby, 73.

¹³ In WWII aircrew flying over Germany were issued, in their survival kit, with a small metal reflector for signalling. It was flat and about 4 cm square; for sighting, see below.

¹⁴ Most conveniently illustrated in Richter, G. M. A., Sculpture and Sculptors of the Greeks⁴ (Yale, 1970), 104Google Scholar. She dates it to the early 5th c.

¹⁵ Snodgrass, Anthony, Early Greek Armour and Weapons (Edinburgh, 1964), 61Google Scholar.

¹⁶ I have recently (1999) been again in contact, and he has confirmed that his comments, as I here print them, still stand.

¹⁷ The concave reflectors of flashlamps or car headlights, which may come to mind as a possible parallel, are essentially different, as, unlike the sun. the light source is located close within the are of the curved reflector, resulting in different angles ol incidence.

¹⁸ Merriam (n. 8, above), records that in tests run by the US Coast and Geodetic Survey, using ordinary mirrors, flashes could be transmitted over very long distances, up to 191 miles; and ‘for one hundred miles a mirror ol four and a half inches was found to answer very well’. This was in very clear air, and he does not say how the flashes were aligned on target.

¹⁹ A modern military heliograph is accordingly fitted with an elaborate sighting device (which I have seen in the Canadian Forces Communications and Electronics Museum in Vimy Barracks, near Kingston, Ontario). The airerew's signalling reflector (n. 13) was to be sighted, somewhat primitively, by aligning a hole bored in the centre of the plate with a short metal rod held out at arm's length. As for thrillers and films where the hero averts imminent peril by reflecting a signal flash from some handy object, such as the glass of his wristwatch, this proves only that the author has never himself tried it.

A further point arises out of an experience of Professor Vanderpool, partly written up by him but never published; I am grateful to him for telling me about it. He was crossing over from the mainland to Spetsai, and, it being winter, the ferry ran only on request. One taverna was still open, apparently run by a girl aged 13 or so, who transmitted the request. To do it, she took a large, bedroom-type mirror and, at the water's edge, held it at waist level before her and facing Spetsai. With a slow turning movement she then raised it above her head, edge-on to the target. This operation she repeated for some time, making no attempt at sighting; moreover, the sky was completely overcast with not a glint of sunlight. To his sceptical comment, ‘You don't imagine anybody's actually going to see that, do you?’, Professor Vanderpool got no reply whatever, until, after some minutes of this, the girl remarked laconically ‘They're coming’, and turning on her heel disappeared back inside the taverna, while, to his amazement, he saw a small boat leaving Spetsai which came across and picked him up.

Both Professor Vanderpool and I were at a complete loss as to how the thing could be done, particularly with no sun, so I tried an experiment to reproduce it, arranging for a mirror to be manipulated on a dull day across an Ontario lake (my acknowledgements are due to Mile Corinne Gardais of Chartres). At a range of 1 km I could see clearly a signal, flashing on and off as the mirror moved up and down. It did not have the bright intensity of a real flash of sunlight, but rather a kind of grey luminosity—because, as I then realized, what I was seeing was a reflection of a portion of the sky, which, even on a dull day, is a lot brighter than the darkness of the land mass in front of which the mirror was held up. The sighting problem is eliminated, for while, with a true heliograph, the mirror has to be aligned so that the observer sees reflected in it the sun, with this system any bit of sky anywhere will do. Naturally, this does not affect Marathon, where no ancient author speaks of anyone flashing anything.

²⁰ Hodge, A. Trevor, ‘Marathon: the Persians' voyage’, TAPA 104 (1975), 155–73Google Scholar; see also id., ‘Marathon to Phaleron’, JHS 95 (1975), 169-71. Fleet speeds: Casson, L., Ships and Seamanship in the Ancient World (Princeton, 1971), 282–6Google Scholar. This very low fleet speed has been questioned in view of the performance of the rebuilt trireme Olympias. So Shaw, J. T., in Gardiner, Robert (ed.), The Age of the Galley (London, 1995), 169Google Scholar, in a theoretical study of oar mechanics, estimates a top speed of ‘nearly 10 knots in a sprint lasting about 5 minutes’, with a regular cruising speed of 71/2-8 knots, assuming ‘a carefully selected, well led and well trained crew’. At Athens, ancient crews were apparently not selected at all. Everybody who could not buy his own armour lor national service in the army was liable to end up in the fleet, as conscripts; we do not know what happened in the Ionian cities that supplied most of the Persians' fleet, but it was probably the same. Training is a different matter, though in 494 (Battle of Lade; Hdt. vi. 11-12) the Ionian seamen refused a rigorous training programme and mutinied to stop it. In actual speed trials in 1990, the Olympias got up to around 8 knots in a sprint, while cruising speed was around 4-4[1/2] knots, but often less, under stress of wind and weather. Full account by Shaw, J. T., in The Trireme Project (Oxbow Monograph 31; Oxford, 1993Google Scholar: a report published by the Trireme Trust), 40-3.

Moreover, there are two reasons why the Olympias figures are inapplicable to the Persian fleet. First, fleet speeds, as noted by Casson, are always much slower than single ships sailing independently. There is the difficulty and delay of mass manoeuvring, and the invariable presence of one or two laggards that hold up everybody else. Second, the Olympias is a trireme, and its figures are invalid for the freighters and transports carrying the Persian army—one remembers that in WWII the Flower class corvettes built to escort the Atlantic convoys were engined for a speed of 16 knots, while even the fast convoys travelled at only 8 knots. I thus feel that Casson's figures for fleet speeds are still the most reliable, while my own reservations on the Marathon-Phaleron voyage in particular are set out in the articles noted above.

²¹ There may be a connection here with the Souda gloss Χωρὶς ἱππής (= ‘The cavalry are gone’—?). For the cavalry issue, analysis and discussion in Lazenby, 59-61.

²² J. Papademetriou, Praktika, 1956, 88.

²³ Papademetriou (n. 23) notes that in 1956 the beach was still heavily mined from WWII, just because it offered such fine landing-ground.

²⁴ The details of the topography may be studies on Curtius and Kaupert, Karten von Attika (1:25,000), Sheet 7 (Spata), on which my FIG. 6 is based. The accompanying text by Milchoefer (to Curtius and Kaupert, Heft III-VI, 5) describes the hills. The ‘Spuren einer Passbefestigung’, which I have reproduced on my map as ‘traces of walling’, have long since disappeared; it is also unlikely that they existed as early as 490, when Athens had not even built the Long Walls.

²⁵ Sailing at 4 knots. Casson (n. 20), 293.

²⁶ See also Gomme, A. W., Phoenix, 6 (1952), 83CrossRefGoogle Scholar; Pritchett (n. 9), 173.

²⁷ See Burn (n. 9), 248; Pritchett (n. 9), 170; Harrison, E. B., ‘The south frieze of the Nike temple and the Marathon painting’, AJA 76 (1972), 363CrossRefGoogle Scholar; Hammond's comments (30 n. 78) on the Brescia sarcophagus are perhaps weaker, since the sarcophagus shows only one Persian horseman. See also Evans, 294, n. 68; Lazenby, 59.