I am most grateful to Mr G. T. Griffith and Mr W. G. Forrest for helpful comments on drafts of this article; and for the comments which were made when I read part of it to branches of the Classical Association at Exeter and Edinburgh in autumn 1966.

¹ Studies to which I refer in this article are abbreviated as follows:

Burn = Burn, A. R., Persia and the Greeks (London, 1962)Google Scholar; Caspari = Caspari, M. O. B., ‘Stray notes on the Persian Wars’, JHS xxxi (1911)Google Scholar; Curtius = Curtius, E., Griechische Geschichte (1888)Google Scholar; Delbrück = Delbrück, H., Geschichte der Kriegskunst i (Berlin, 1920)Google Scholar; Gomme = Gomme, A. W., More Essays in Greek History and Literature (Oxford, 1962)Google Scholar; Grundy = Grundy, G. B., The Great Persian War (London, 1901)Google Scholar; Hammond = Hammond, N. G. L., A History of Greece ² (Oxford, 1967)Google Scholar; Hignett = Hignett, C., Xerxes' Invasion of Greece (Oxford, 1963)Google Scholar; How = How, W. W., ‘Cornelius Nepos on Marathon and Paros’, JHS xxxiv (1914)Google Scholar; Kromayer 1 = Kromayer, J., ‘Drei Schlachten’ Abh. d. Sächs Akad. xxxiv (Leipzig, 1921)Google Scholar; Kromayer 2 = Kromayer, J., Antike Schlachtfelder iv (Berlin, 1924–1931)Google Scholar; Leake = Leake, W. M., ‘The Demi of Attica’, Transactions of the Royal Society of Literature i (London, 1829)Google Scholar; Macan = Macan, R. W., Herodotus ii (London, 1895)Google Scholar; Maurice = Maurice, F., ‘The Campaign of Marathon’, JHS lii (1932)Google Scholar; Meyer = Meyer, E., Geschichte des Altertums iv 1 (4th ed.Stuttgart, 1944)Google Scholar; Munro = Munro, J. A. R., ‘The Campaign of Marathon’, JHS xix (1899)Google Scholar; Myres = Myres, J. L., Herodotus: Father of History (Oxford, 1953)Google Scholar; Pritchett = Pritchett, W. K., ‘Marathon’, Univ. of California Publications in Classical Archaeology iv 2 (Berkeley, 1960)Google Scholar; Schachermeyr = Schachermeyr, F., ‘Marathon und die persische Politik’, Historuche Zeitschrift clxxii (1951)Google Scholar; Schliemann = Schliemann, H., ‘Das sogenannte Grab der 192 Athener in Marathon’, Zeitschr. f. Ethnologie xvi (1884)Google Scholar; Soteriades = Soteriades, G. in Πρακτικά τῆς ἀρχαιολογικῆς ἑταιείας 1935 (Athens, 1936)Google Scholar; Staës = Staës, B., ̒Ο ἐν Μαραθῶνι τύμβος, Ath. Mitt, xviii (Athens, 1893)Google Scholar; Whatley = Whatley, N., ‘On the possibility of reconstructing Marathon and other ancient battles’, JHS lxxxiv (1964).Google Scholar

² Kromayer 2.

³ Curtius ii⁶ 24 f.

⁴ Delbrück 63 ‘Riesenhaft steht die Gestalt des Feldherrn Miltiades an Eingang der Welt-Kriegsgeschichte; die vollendetste und seltenste Form der Schlachtenführung, die alle Kriegskunst bis auf den heutigen Tag hervorgebracht, die defensiv-offensive, tritt uns hier in den einfachen Linien des klassischen Kunstwerks.’

⁵ Munro, , CAH iv (1926) 245Google Scholar, keeping the cavalry and other troops at Eretria and then sailing for Phalerum; Grundy 184, ‘it seems almost certain that the cavalry had been already embarked’; Gomme 36, ‘they must … embark the cavalry first’; Schachermeyr 25, ‘daher entschloss sich Datis zur Einschiffung der Reiterei'; Pritchett 173, ‘Schachermeyr's interpretation is undoubtedly correct’; Burn 247, Datis ‘embarked some troops, including probably much of his comparatively small force of cavalry, destined for the dash up from Phaleron.’ Beloch, GG ii 2. 80Google Scholar simply denied the participation of cavalry in the expedition. Whatley 131, criticising such a hypothesis, remarked ‘mere theories are tending to be regarded as established truths’.

⁶ Meyer iv 1.311 ‘Datis entschloss sich zum Angriff und führte sein Heer in Schlachtordnung vor.’ De Sanctis, , Riv. di. Fil. liii (1925) 120.Google Scholar

⁷ Munro, , CAH iv (1926) 249Google Scholar: ‘The hand-to-hand fighting … is to be measured in minutes, not hours’; Delbrück 60 and 62.

⁸ Maurice 23 ‘The Soros existed before the battle’; Burn 254 ‘it may be therefore that for purposes of this burial an already ancient tumulus was dug up, re-used and perhaps enlarged; and it can not therefore be confidently said that the Soros shows exactly where “The Athenian dead lay thickest”.’

⁹ The same sort of theories' have bedevilled the Battle of Salamis, as I observed in my article on that battle in JHS lxxvi (1956) 39 f.

¹⁰ Macan 150.

¹¹ Leake 177. At the other extreme Meyer 311 n. 2 accepts as authentic tradition the account of Herodotus and a couple of remarks in Plutarch, sweeps all other literary evidence aside as modifications of the Herodotean tradition (‘alle andere Berichte haben keinen selbständigen Wert, sondern sind Modifikationen der bei Herodot erhaltenen Tradition’), and feels free to invent a Persian offensive which is not even a modification of the Herodotean tradition. It is anachronistic to suppose that all ancient writers except Herodotus, whether contemporary or subsequent, relied only on their powers of invention or modification in regard to this campaign; Gottlieb, G., Das Verhältnis der ausserherodoteischen Überlieferung zu Herodot (Bonn, 1963) 65Google Scholar, commenting on Meyer's view, writes ‘es ist aber m. E. einfach unmöglich, dass alle Einzelheiten, die Herodots Beschreibung ergänzen, erfunden sind’. And it is uncritical to treat alike Plutarch, for instance, and Clemens Alexandrinus, who made Miltiades undertake a night march in imitation of Moses, (Alex. Strom. i 162).Google Scholar

¹² Δελτ. Ἀρχ. 1890, 65–71; 123–132; and 1891, 34–67; and 97; Ath. Mitt. xviii (1893) 46–63.

¹³ Staës gives the height and diameter, Schliemann the circumference; Schliemann gave the height as 11 metres, which means either that his excavation docked its height by 2 metres or that Schliemann included the two metres which he excavated below the outside ground-level.

¹⁴ So Staës, Ath. Mitt. xviii (1893) 53Google Scholarἐπὶ τῆς ἐπιφανείας τοῦ στρώματος. Pritchett 141 puts the tray a metre lower (if I understand him correctly) perhaps misunderstanding Staës ibid. see also Δελτ. Ἀρχ. 1890, 65, 128 and 129. Staës hit water when he dug his small trench on the north side (Δελτ. Ἀρχ. 1890, 130); he may not have gone deep enough to hit the layer of ashes, or else the ashes were scattered to one side only of the pyre. The pyre, sometimes on a cairn of stones, is the centre of Bronze Age tumuli.

¹⁵ The method of constructing a mound is given in Iliad xxiii 255 f. and is illustrated in the tumuli of North Epirus; see my book Epirus 387 f. (Oxford, 1967).

¹⁶ Maurice 23–24; Pritchett 142 points out the impossibility of Maurice's suggestion.

¹⁷ Burn 254.

¹⁸ Staës 55.

¹⁹ Pritchett, 141, 142 n. 34, 154 n. 24 and 157 ‘the rise of three metres at the Soros’, which shows a failure to understand the effect of erosion; in the same way he says on p. 141 that ‘the Soros was originally a burial mound of earth at least twelve metres high’, which does not allow for the erosion of the top.

²⁰ Schliemann 85–88.

²¹ Soteriades 148 and Pritchett 140 n. 20.

²² Soteriades found a tholos tomb and also Geometric pottery in the plain; see JHS liv (1934) 189. The tholos tomb is about a kilometre west of the Mound; see Ergon 1958 (1959) 23 for the further excavation of it.

²³ This is not surprising in so great a mound. Staës did not find Mycenaean and Geometric sherds. Burn 254 n. 42 implies bad faith in Staës when he writes ‘the Greek excavators did not wish to find such evidence (scattered, in a disturbed tumulus) as Schliemann had reported’. But Staës, anticipating this modern tendency to question the good faith of excavators, obtained witnesses at the time. The fact is that Schliemann and Staës dug in different parts of the mound, and Schliemann's vast trench went down from the then top of the mound. There is no reason to suppose that they would have made similar finds.

²⁴ Leake, , Travels in Northern Greece ii 431Google Scholar and G. Finlay, quoted by Pritchett in his excellent note 20 on p. 140.

²⁵ Leake 172; Broneer, in Hesp. iv (1935) 114Google Scholar with fig. 4 top row and Salmony, A., Artibus Asiae xvii (Switzerland, 1954) 303 f.Google Scholar; I illustrate one more clearly in Epirus fig. 30B. Pritchett 159 discusses arrowheads but misses out the important find made by Leake.

²⁶ Forsdyke, E. J. in Proc. Soc. Antiq. xxxii (1919) 146 f.Google Scholar; Admiral Brock and General Meyrick had them from ‘a grave in the plain of Marathon’, no doubt the Mound. Forsdyke showed that ‘the Marathon arrowheads are not Greek but Oriental’. Herodotus vi 112.2 shows that the Greeks had no cover of arrows when they attacked.

²⁷ Kromayer 1, 10 estimated the effective range of archers against bronze-clad hoplites at 100 metres, and Delbrück 60 n. 1 at more than 100 metres. McLeod, W., ‘The range of the ancient bow’ in Phoenix xix (1965) 8Google Scholar makes the effective range of an archer ‘at least 160–175 metres but not as far as 350–450 metres’. He does not take into account the nature of the target. For instance his lower figure is based on the firing of incendiary arrows (his Ti = Hdt. viii 52.1), which needed no power of penetration, and the higher figure is based on unarmoured horses at that distance being out of range (his T2 = Hdt. ix 22–23), which is almost equivalent to the extreme range because a horse is easily stung into action by an arrow. The only deduction we can make from the evidence for the Persian Wars is that a Persian archer could shoot an incendiary arrow as far as at least 155 metres and not as far as 355 metres. The only piece of ancient evidence which concerns archery in battle gives 600 feet = 178 metres as the practice range for men training to fire in acie, and it relates to the fourth or fifth century A.D. (his T8 = Vegetius, , Epit. rei. mil. ii 23Google Scholar). At Marathon the archers no doubt held their first shots until the Greek bronze-clad hoplites came within a range at which an arrow could penetrate armour. I should be inclined to put this at less than the 178 metres of Vegetius and nearer to Kromayer's estimate. I take 150 metres here as a reasonable compromise. McLeod's suggestion that the Greeks at Marathon were under fire for 200 or even 300 yards (p. 13) is an exaggeration. Once the battle was joined, the bow played no part in the hand-to-hand fighting.

²⁸ Schachermeyr 27, having put the Greek position on Mount Agriliki, made a general statement that the heaviest losses in ancient battles occurred first ‘im Zurückweichen’ and so felt justified in putting these losses about a mile away from the place where the engagement began. This distance means losses in flight, such as occurred when the Macedonian cavalry pursued the defeated enemies of Philip and Alexander, but a hoplite set-to was different. Herodotus puts the long fight before the Greek centre broke and it was in this long fight that the thinner files in the Athenian centre were laid low. Gomme 33 assumes, pace Herodotus, that the Greek centre did not break. The casualty list, recording the tribe to which the dead belonged, showed that the tribes Leontis and Antiochis had suffered most, and Plu. Arist. 5 reported that their regiments held the centre; this can be taken as an intelligent inference by Plutarch's source, or as independent evidence that the centre suffered most casualties.

²⁹ Estimates vary from the 2,750 yards of Myers 210 and 2,300 yards of Maurice 23, who made the Greek centre two men deep, to ‘at least one kilo metre’ of Kromayer 1, 10. Pritchett 143 f. puts the line at 1,641 yards as a maximum.

³⁰ Soteriades 133; Pritchett 143 and 173, if I. understand him correctly, for his sketch-map does not show any positions at all; and now Vanderpool, E. in Hesperia xxxv (1966) 103.Google Scholar When the Persian line is put at right angles to the coast with its back along the line of the Charadra, the Persian break through the Greek centre carries the Persians towards Mount Agriliki, not towards the Schoeniá.

³¹ Leake 164; Milchhoefer, A., Text zu Curtius-Kaupert, Karten von Attika, iii 49Google Scholar; Radke in P-W s.v.; Soteriades, , Praktika 1935, 129Google Scholar; Philippson-Kirsten, , Die Griechischen Landschaften i 3.789 (1952)Google Scholar; McCredie, J. R., Hesperia Supplement xi (1966) 41.Google Scholar

³² Suidas s.v., ἐμπίς. Arist. HA 569b, mentioning midges in shaded marshy districts, says they occurred thus ἐν Μαραθῶνι i.e. either in the Great Marsh or in the small marsh or in both.

³³ Leake 168.

³⁴ Leake ibid. and Pritchett 155 did not notice this difference.

³⁵ Leake 162 mentions these inundations.

³⁶ Pritchett's plate 7 gives an excellent aerial view of the coast-line and the marsh. For Pylos see Gomme, A. W., Historical Commentary on Thucydides iii 483.Google Scholar

³⁷ This tent, like the tent of Xerxes, made a great impression on the Athenians; see Broneer, O., ‘The tent of Xerxes and the Greek Theater’ in Univ. of California Publications in Classical Archaeology i (1929) 305 f.Google Scholar and Thompson, D. B., ‘The Persian Spoils in Athens’ in The Aegean and the Near East ed. Weinberg, S. S.; (New York, 1956) 282 f.Google Scholar

³⁸ The last thing mentioned by Pausanias in i 32.7 is not connected with the battle, but it gives some idea of his itinerary. Having observed the spring Macaria, the marsh, the mangers of Artaphernes' horses, the marks of the tent and the outlet of the river from the marsh into the sea, he mentions the mountain and cave of Pan ‘a little further away from the plain’ Such a cave was reported on a hill to the west of the Marathóna, valley (Ergon 1958 [1959] 15 f.).Google Scholar

³⁹ Leake 173; see now Vanderpool, E., who has found fragments of the column, in Hesperia xxxv (1966) 93 f.CrossRefGoogle Scholar

⁴⁰ Raubitschek, A. E. in AJA xliv (1940) 53CrossRefGoogle Scholar; see Hampe, R. in Die Antike xv (1939) 170 fig. 1Google Scholar and Shefton, B. B., ‘The dedication of Callimachus’ in BSA xlv (1950) 145.Google Scholar A Nike probably was on the column on the Acropolis; there may have been one also on the column by the church of Misosporétissa.

⁴¹ Leake 162.

⁴² Leake's identification of Oenoe on p. 163 has been widely accepted; see references in Pritchett 149.

⁴³ Soteriades, in Praktika 1935, 132 f.Google Scholar

⁴⁴ Schachermeyr 17 and Pritchett 157.

⁴⁵ So Pritchett 156 reports from personal observation; and now Vanderpool, in Hesperia xxxv (1966) 103.Google Scholar

⁴⁶ Leake 165.

⁴⁷ Soteriades, in Praktika 1935, 124.Google Scholar

⁴⁸ Pritchett 154. He proceeded, like Soteriades, to ‘the elimination of the marsh’ in antiquity. I am not convinced by his arguments; for instance by one that the ‘black silt’ here was not marsh-silt but was silt ‘brought in by the sea’, a procedure contrary to the normal processes of nature.

⁴⁹ Leake 165.

⁵⁰ Pritchett 138; he is very critical of Soteriades (see his note 9). Vanderpool, E. goes further in AJA lxx (1966) 322Google Scholar and says that Soteriades' enclosure wall ‘is not even ancient’, but what of Karo's confirmation on the evidence of sherds? The wall is relatively unimportant, because the boundary stone shows there was a temenos of Athena hereabouts.

⁵¹ Praktika 1933, 42, reported by Payne, H. G. G. in JHS liv (1934) 189.Google Scholar

⁵² Soteriades, in Praktika 1933, 42Google Scholar held the stone was in situ (κατὰ χώραν) because the letters were clear.

⁵³ Soteriades, in Praktika 1934, 30Google Scholar

⁵⁴ Ibid. 1934, 38 Illustrated ibid., 1935, 87, fig. 1.

⁵⁵ Ibid. 1935, 90 For Athena's connection with Marathon see Od. vii 80.

⁵⁶ Soteriades, having found what he held to be the acropolis of Marathon and having decided the Heracleiim should be near it, set out to look for the temenos in the vicinity and found it in the expected place and not a trace at all of any other temenos (Praktika 1933, 34 He decided this was the temenos of Heracles and when he found a boundary-stone of the temenos of Athena cheek by jowl with it—ὁμόζνγον πρὸς αὐτό, p. 412—he failed to make the obvious deduction that the one and only temenos on the western side of the plain— p. 42—was that not of Heracles but of Athena. Payne and others accepted his decision without demur; but Wrede, in PW s.v. Marathon, made the cautionary remark ‘doch ist das vage Vermutung’. Vanderpool, E. mentions the boundary stone in AJA lxx (1966) 319CrossRefGoogle Scholar but makes no deduction. Little attention has been paid to Soteriades' work and least to the boundary stone of Athena's temenos by most of those who have studied the battle. Hignett 60 f. does not mention his work; he just places the temenos of Heracles on Mount Agriliki on his a priori view that the Athenians came by the coast road from Athens. Burn 243 refers to ‘a precinct sacred to Herakles, now identified at a Chapel of St Demetrios where there are ancient foundations, at the foot of the valley where the Vraná track comes in’; the last words are inaccurate as his own contourless map on p. 244 shows, since the Chapel is on the lower slopes of Mount Agrilíki and not at the foot of the valley and much less where the Vraná track comes in (i.e. from Athens).

⁵⁷ Pritchett 140 and n. 17. I find Vanderpool, 's identification unconvincing (AJA lxx [1966] 321)Google Scholar; having downgraded Soteriades' other discoveries, he upgrades his ‘farm-houses’.

⁵⁸ Pi. P. viii 79; the games are mentioned also in O. ix 89 and xiii 110. This is where Leake tentatively put the temenos of Heracles. The word μυχός is translated ‘vallis’ in this passage by Rumpel, I., Lexicon Pindaricum (Stuttgart, 1883)Google Scholar; it is used of Delphi by Pindar in P. v 68 and P. x 8. An inscription mentioning either the ‘Heracleia’, i.e. the games (Vanderpool, E. in Hesp. xi [1942] 329 f.CrossRefGoogle Scholar) or the Heracleum (Peek, in SEG x 2Google Scholar) has been dated to shortly after or shortly before 500 B.C. The stone was found in a vineyard ‘in a locality known as Valaria’ (Vanderpool, ibid.), which is marked on Leake's map = Plan 4 here. As the stone had been partly re-faced and trimmed at some unknown period probably for incorporation in a building, the place of its discovery does not give us the position of the stadium for the games or for the Heracleum but only a general indication that it was in the western part of the plain. Vanderpool, E. admits as much in AJA lxx (1966) 323.Google Scholar

⁵⁹ How, W. W. and Welk, J., Commentary on Herodotus ii 113.Google Scholar Macan 239 describes the two routes.

⁶⁰ See Macan 227 and Caspari 105 f.: ‘the oldest piece of evidence for the reconstruction of the battle of Marathon’.

⁶¹ So Amandry, P. ‘Sur les épigrammes de Marathon’ in ΘΕΩΡΙΑ Festschrift Schuchhardt (1960) 4.Google Scholar

⁶² Meritt, B. D., ‘Epigrams from the Battle of Marathon’ in The Aegean and the Near East (1956) 270 f.Google Scholar

⁶³ The word πεζός is used of men on foot with a latent or expressed contrast to men at sea, in a chariot or on horseback (for the first see Thuc. ii 94.2 and L-S-J⁹πεζός I). Here the contrast is expressed. They fought on land and at sea (as in the epigram commemorating the battle of the Eurymedon River; see my History of Greece ² 259 f.). The restoration proposed by Wilhelm, in Anz-Akad. Wien lxx (1933) 10.95Google Scholar has not been tenable since Meritt's publication in 1956; it was used by Schachermeyr 16 in 1951 and by Hignett 66 n. 4 in 1963 as an argument in favour of Persian cavalry being engaged in the Battle of Marathon.

⁶⁴ Amandry did not refer to the following passage which strongly supports his view: D. S. xi 33.3 Pritchett 160 f. attributes the epigram to the battle of Salamis but the emphatic position of πεζοί can hardly be reconciled with the small-scale attack on Psyttalia.

⁶⁵ In AJA xliv (1940) 58 f.; he follows H. L. Crosby in finding a reference in Aristophanes, Wasps 1079 f.Google Scholar to fighting at Phalerum; but see p. 50 below.

⁶⁶ See n. 61.

⁶⁷ Rivalry between the supporters of Cimon and the democracy was then very strong, and Salamis was the pride of the democratic group, (Arist. Pol. 1304 a 22 and Ath.Pol. xxv 1–2).

⁶⁸ Forrest, W. G., ‘Aristophanes' Acharnians’, in Phoenix xvii (1963) 10Google Scholar, thinks it ‘no doubt is an exaggeration’; but having visited the Royal Chelsea Hospital for Veterans I have no doubt that some Acharnians of 85 years and more had tales to tell of Marathon.

⁶⁹ We must remember that we are dealing with a single day of fighting and we must not be misled by Whatley's analogy of men's memories of protracted, complicated wars such as the First World War becoming blurred and confused.

⁷⁰ Modified by Beloch, GG ii 2.80Google Scholar and Maurice 16, who supposed there to have been no cavalry units in the expeditionary force, and by Munro, , CAH iv 243Google Scholar, who kept the cavalry at Eretria in the Marathon campaign. One might as reasonably deny today the presence of tanks at the Battle of the Somme in 1916 and hope not to be refuted by participants in that battle.

⁷¹ Rejected, for instance, by How, and Welk, Commentary ii 112Google Scholar and Hignett. 62 ‘a feat which no phalanx of citizen militia could have performed’, and modified by Delbrück 60 who reduces the δρόμος to half a stade and by How and Wells ibid. who reduce it to one stade (unnecessarily, if one reads Hdt. ix 59.1 and Th. iv 78.5). Delbrück 56 converts the attack over 8 stades into a pursuit over 8 stades.

⁷² Rejected, for instance, by Delbrück 65 who supposes a ‘längere Pause’ between the action and the pursuit; a view favoured by Hignett 72.

⁷³ Excerpted by Blakesley, rejected by Delbrück, converted by Macan 167 to before the action, by Wecklein, Über die Tradition der Perserkriege 38Google Scholar to before the Athenians left Athens for Marathon and by Curtius to the occasion of the (supposed) reembarkation of the cavalry.

⁷⁴ Rejected by How, and Wells, , Commentary ii 113Google Scholar on the ground that ‘the distance from Marathon to Athens is more than an army could march after a pitched battle’, and by Hignett 73, while the Persian voyage via Sunium is rejected by Schachermeyr 29 as ‘ganz unglaubwürdig’.

⁷⁵ The underlying reason is that many students of ancient history tend to regard ancient history as ancient, dependent on ‘oral traditions’ (e.g. Hignett 69) and written sources. For Herodotus and his audience Marathon was contemporary history; he asked the participants for the facts and the participants criticised his recitals at Athens; and even if he had misunderstood an informant, the participants would have corrected him when he gave his recitals.

⁷⁶ Hdt. vi 113.1 For this part of the fighting we may compare the hand-to-hand fighting in the battle of Pydna: Plutarch, Aem. 21.1 The victory at Pydna was gained in less than one hour (22.1). I imagine the Greeks defeated the Persian wings and centre within an hour or so. The analogy is with a bout in fencing rather than with trench warfare.

⁷⁷ Hdt. vi 113.2 as compared with 113.1 As T. J. Cadoux and W. G. Forrest have pointed out to me, the inference is not certain; but as the sentences follow one another, the inference is in my opinion the probable one.

⁷⁸ Arch.Zeit. xxiv (1866) 220f.; see now the reproduction in Hesperia xxxv (1966) pl. 35 by Vanderpool, who aptly remarks that ‘it might be studied with profit by those who believe that the Persian cavalry had been withdrawn before the battle’, p. 105, n. 26.

⁷⁹ See now Vanderpool, E. in Hesperia xxxv (1966) 102 n. 17Google Scholar; he believes Leake is correct in his identification, but he confuses this trophy with the memorial to Miltiades which are two separate things in Paus. i 32.4 and i 32.5.

⁸⁰ κατά means ‘about’ (L-S-J⁹s.v. B. vii 2), pace Burn 251 who thinks the number precise; some corpses in the lake and swamp may not have been recoverable.

⁸¹ is omitted by some manuscripts.

⁸² Including Paros which sent a trireme to Marathon (Hdt. vi 133.1).

⁸³ The Persian expedition was launched against Eretria and Athens (Hdt. vi 94.2) and was ordered by Darius to enslave the Eretrians (Hdt. vi 101.3) and presumably the Athenians, and it was prudent to secure the islands en route and the lines of supply; it was also advisable to land the horses as often as possible in order to keep them in fighting condition.

⁸⁴ Schachermeyr has propounded a theory that the Persians' slow advance was planned to be ‘a war of nerves’ on Athens and so timed that the landing at Marathon coincided with the phase of the moon during which, as Hippias knew, the Spartans would not leave Sparta. But the fall of Eretria by treachery was not predictable; indeed, if the 4,000 Athenians from Chalcis had entered Eretria as they intended, the city might have held out for weeks. His theory seems to me to be born of after-knowledge. Schachermeyr's view has been strongly attacked by Kraft, K., ‘Bermerkungen zu den Perserkriegen’, in Hermes xcii (1964) 153–158.Google Scholar

⁸⁵ Gomme 30 writes of these 4,000 men as follows. ‘So they go, to Athens, but we hear no more of them. There is one difficulty—what did they then do?’ I see no difficulty whatsoever; they served in the army at Marathon.

⁸⁶ Hdt. i 62 and 63.2 for mention of Peisistratus' sons.

⁸⁷ How and Welk ii 358 find ‘the reasons … inadequate. Nearness to Eretria would not com pensate the Persians for remoteness from Athens, and the plain of Athens … is more suitable for the operations of cavalry’. They fail to realise that Herodotus' reason is a single one combining two assets, namely suitable ground for cavalry and proximity to Eretria.

⁸⁸ Hignett 71. Meyer 306 envisages a smaller force still: ‘mehr als 20,000 Mann können es schwerlich gewesen sein, vielleicht beträchtlich weniger’.

⁸⁹ Personally I should opt for a figure of 30,000. The Persians had good information; for the battle of Plataea Sparta produced 10,000 hoplites, Megara 3,000, Corinth 5,000 while Athens and Plataea together produced 10,000 and Eretria may have had 2,000 or 3,000 in 490 B.C. Of course the Persians hoped to fight the Greek states separately, but they could not count on it at all.

⁹⁰ See my History of Greece ² 389 f.

⁹¹ The 600 triremes of Hdt. vi 95.2 is evidently a conventional figure for a Persian fleet (see How and Wells ii 103). The Athenians had little opportunity in this campaign of assessing the size of the Persian fleet.

⁹² Ampelius 15. 9 gives the number of Persian troops as 80,000; his source was Trogus Hht.Phil. prol. 2 (see ed. by O. Seel p. 57). Leake 190 f. calculated the total numbers at 177,000; he accepted the number 600 for the Persian warships.

⁹³ In the same way the Eretrians had asked Athens to send help in their defence (vi 100.1 αφίαι βοηθσὺς γενέσθαι); earlier still, the Athenians, faced by the same problem when Peisistratus landed at Marathon, had gone to the defence of Attica (i 62.2 βοηθέουσι ἐπ' αὐτόν); the Spartans were asked now to come to the defence of the Athenians (vi 106.2 σφίσι βοηθήσαι) and after the battle the Athenians went to the defence of the city (vi 116 ἐβοήθεον ἐς τὸ ἄστυ). The Persians had ravaged the territory of Carystus (Hdt. vi 99.2), and [D] lix 94 says that Datis landed with a great force in Attica and began to ravage (ἐπόρθει. I see no justification for Gomme's view, p. 32, that ‘the only purpose’ of the Athenian army was ‘to engage the army’.

⁹⁴ Burn 241 and Hignett 61 think that Miltiades' proposal was carried before the Persians landed and was meant to apply ‘wherever they might land’; but the citations of the decree have a specific and not a general purpose, and the decisions of the Athenian assembly were usually made to meet a specific crisis and at the last moment. Burn 257 sends Philippides to Sparta three days before the Persians land at Marathon; this runs counter to Herodotus’ sequence at vi 102–103.1 and at vi 105.

⁹⁵ The message from Athens would reach Plataea that evening. If the Plataeans set off at dawn on the next day, they could have reached Vraná during the following night. Heurtley, Pendlebury, Skeat and I walked from Athens to Thebes via Phyle in 1930, leaving Athens about 2 a.m. and reaching Thebes by 6 p.m. to catch the train for Athens. From Plataea to Vraná is slightly farther as the crow flies. For speed of movement and night marches in 424 B.C. see my article in BSA xlix (1954) 112 f.

⁹⁶ In 479 B.C. the lines of supply for the Greek army crossed Mount Cithaeron, which is more arduous going than over the foothills between Kefisiá and Vraná. Vanderpoo's proposal to put the Greek camp out in the open plain at the mercy of the Persian cavalry seems to be mistaken; his position of the Heracleum, is shown in Hesperia xxxv (1966) 104 fig. 3.Google Scholar

⁹⁷ Burn 250 suggests that some of the slaves were killed in the Athenian camp, but we are not told that the Persians captured the camp; the word of Pausanias ἐμαχέσαντο is more trustworthy. Helots fought at Thermopylae. ‘Pausanias’ assertion … that Miltiades liberated the slaves before the battle arouses suspicion’ in the mind of Hignett, 59 n. 7; but he does not advance any argument against the authenticity of the decree by Miltiades and the Athenian state.

⁹⁸ The cleruchs from Chalcis, presumably hoplites, numbered 4,000 (Hdt. vi 100.1); it is difficult to believe that Athens and Attica yielded only another 4,000 or so. At Plataea 8,000 Athenian hoplites fought; at the same time there were Athenian troops serving at Mycale.

⁹⁹ Philippides reached Sparta on the 9th of the lunar month (Hdt. vi 106.3), and the night of full moon was on the 15th. The Spartans left presumably early on the 16th and reached Athens on the 18th (Hdt. vi 120 cf. Isoc. Panegyr. 87). Since the battle occurred with a waning moon as the commemorative coin shows, it was after the 16th and before the 18th; that is, it occurred on the 17th. The Spartans then arrived the day after the battle (so Plato, Laws 698EGoogle Scholar), and went that day (Plato, Menex. 240Google Scholar) to see the Persian dead still unburied (Hdt. vi 120). As Philippides reached Sparta on the second day from Athens, the Persians landed on the 7th, the Athenians were at Marathon on the 8th, the Plataeans joined them on the 9th, and the battle was fought on the 17th. Burn 256, putting the battle on the day before the waning moon first appears, departs from the ancient evidence.

¹⁰⁰ The Macedonians were the first to integrate the two arms. Kromayer 1 19 f. thought that the Persian cavalry fought in the middle of the infantry line, but no one has followed him in this idea. The generals were mounted, Datis and Artaphernes appearing in the picture in the Poecile Stoa (Pliny, NH xxxv 57Google Scholar), and Mardonius on a grey at Plataea (Hdt. ix 63.1). Mardonius was in the infantry line with the 1,000 picked infantrymen (ix 63.1 resuming the of vii 40.2 and not the 1,000 picked ἱππόται of vii 40.2 and viii 113.2, as the description of the fighting shows: ix 63.2).

¹⁰¹ In 511/510 B.C. 1,000 Thessalian cavalry had defeated an army of Spartan hoplites in the plain of Phalerum (Hdt. v 63.2); this was a further warning to Miltiades.

¹⁰² Later in his narrative Herodotus rejected the attribution of the signal to the Alcmeonids (vi 121 f.), but he does not doubt that the signal was made.

¹⁰³ 216 n. 2.

¹⁰⁴ This seems to me to be the meaning of Herodotus; Hignett 73 seems to think that night intervenes between the battle and the Athenian return to Athens but there is no sign of this in Herodotus' straightforward μέν and δέ clauses.

¹⁰⁵ ii 113.

¹⁰⁶ 191; on 173 he had called it ‘a long day's march, and a hard journey’.

¹⁰⁷ Burn 243 n. 14 thinks my figure to be ‘an underestimate’. He questions the route via Vraná over the hills to Kefisiá and Athens, adding ‘one must wonder whether Col. Hammond whose … service in Greece was with guerilla resistance forces, would really choose to take an infantry division that way’. The answer to his submerged question is ‘yes, if they are Greeks’; Greek guerilla troops went fast, each man for himself, because they were accustomed to covering tough and hilly ground which is general in Greece, and hoplites must often have moved in the same way and not, as Burn 243 supposes, ‘in column’.

¹⁰⁸ A point accepted by Burn 251 n. 33 but doubted by A. H. Hodge in a letter to me. I make the following deductions from Hdt. vi 115, 121 and 124.2. The signal was thought to be a pre-arranged signal, indicating a change of situation at Athens and so leading the Persians to sail for Sunium; as men on the battlefield could not know the situation at Athens, the signal was not from the battlefield but from somewhere on the Athens' side of the Marathon plain. The Persians were out at sea, on course for or near Aegilia; their ships being lowish in the water, they could have seen a signal sent from south of the Marathon plain only if the signaller was on high ground. As the signaller and the Persians were some miles apart, a man simply lifting up a shield would not have been visible to the naked eye of a Persian. Herodotus uses two expressions for signalling, ἀναδέξαι ἀσπίδα here and ἀναδέξαι σημήιον of Xerxes on a ship (vii 128), and Xenophon likewise, ἆραι ἀσπίδα at Aegospotami, (HG ii 1.27)Google Scholar and ἀραι σημεῖον (Cyr. vii 1.23); just as a crow's nest at sea is not a crow's nest, so I suppose the ἀσπίς was not a hoplite's shield, a ὅπλον, but a shield-shaped σημεῖον (ἀσπίς is ‘a round, flat dish’ in a fourth-century writer, Aristopho). In 432 B.C. a pre-arranged signal, first raised and later pulled down at Potidaea, was seen by men some seven miles away at Olynthus (Th. i 63.2). The signalling object at Potidaea and at Marathon must have glistened in the sun to be visible for seven miles or so, and I imagine it was a round, flat polished disc in each case.

¹⁰⁹ See n. 81 above; the timing here, as in Herodotus, is counter to Burn's view.

¹¹⁰ I quote here the text of A. Monginot (Paris, 1882), who is the best commentator on Nepos. He keeps the reading of the codices ‘e regione’ and prefers the reading of M to those of D ‘nona’ or ‘nova partis summa’, A ‘nana partis summa’, B ‘non apertis summa’, and R ‘in parte montis summa’, whereas Roth, followed by Winstedt in OCT, alters ‘e regione’ to ‘regione’ and reads ‘non apertissuma’ as a correction of B's reading. Monginot's reading ‘stratae’ is a conjecture for the ‘rarae’ of the codices, and this conjecture is almost essential, if there is to be any sense in ‘nova arte’, ‘namque’ and ‘hoc Consilio ut … arborum tractu equitatus hostium impediretur’. Monginot notes the ugliness of the change of subject in ‘tegerentur’ and ‘impediretur’ as unusual in Nepos; the reason is probably that his original had a μέν and δέ here, being in Greek as I point out below on p. 51. Cleomenes felled trees and used them to impede the cavalry of Hippias in 510 B.C.: ‘Cleomenes Lacedaimonius adversus Hippiam Atheniensem, qui equitatu praevalebat, planitiem, in qua dimicaturus erat, arboribus prostratis impediit et inviam fecit equiti’, Frontinus ii 2.9.

¹¹¹ The cavalry were similarly placed before the infantry line at the Granicus river; see my History of Greece ² 604.

¹¹² As the plain was most suitable for cavalry (Hdt. vi 102) it was treeless; see also Hdt. v 63.4 where the Thessalians cleared the Phalerum plain for cavalry action.

¹¹³ As I mentioned in my account of the campaign in my History of Greece ² 215. See n. 110 above for a similar use of felled trees by Cleomenes.

¹¹⁴ We have no clue to the number of the Persian cavalry; the Thessalians had 1,000 cavalry when they defeated the Spartans in the plain of Phalerum in 511/10 (Hdt. v 63.3; Ath.Pol. xix 5) and I should be inclined to put the number of Persian cavalry at a higher figure. The Theban cavalry, which probably numbered considerably less than 1,000, was able to inflict heavy loss on 4,000 Megarian and Phliasian hoplites in the plain by the Asopus (Hdt. ix 69.2). Meyer's ‘wenige Hunderte’, p. 306, is not acceptable; Athens, Eretria and possible allies of the two could do better than ‘a few hundred’.

¹¹⁵ The Chancellor of the University of Bristol, the Duke of Beaufort, told me this is the case in his experience; I found it to be so with mules moving at night in occupied Greece during the war.

¹¹⁶ A more correct translation is ‘apart’, (as in Hdt. ix 32.2 ), that is ‘apart from the army’ and as the end of the Greek passage shows ‘away from their position’.

¹¹⁷ The primary meaning of ἀνέρχομαι is ‘go up’ and from the sea ‘go inland’, as on the Anabasis. There is no likelihood of the Greek phrase meaning ‘they climbed up the trees and signalled’ and anyone who has climbed a tree knows that it is impossible to signal from inside it to people a mile away either at night or in daylight, when the absence of cavalry would have been obvious without any signal; yet all students of the battle seem to have translated the passage thus until I wrote my account in my History of Greece 215. Burn 248 n. 23 accepts my translation, but he seems to go astray on the word ἀναχωρήσαντος which he translates ‘according to Byz. and modern usage, not “withdraw” as in Homer, but “leave”, “start”, “depart” as in a modern Greek timetable’. The passage, however, is not in Byz. or modern Greek. A more fruitful comparison is Thuc. v 75.1–2 where we have ἀπεχώρησεν and ἀναχωρήσαντος in successive sentences—as here but in reverse order ἀναχωρήσαντος and ἀποχώρησις. In Thucydides Pleistoanax came to Tegea and learning of the Spartan victory at Mantinea ‘withdrew’ (L-S-J⁹ἀποχωρεω 2 abs.); the victorious Spartans ‘going back (home) themselves (L-S-J ἀναχωρέω 1) and dismissing their allies' conducted the Carneian festival. Here Datis went back i.e. to his camp and his tent and Miltiades realised ‘the withdrawal’ of the cavalry Schachermeyr 22 stretches several points when he says the passage must mean ‘die Wiedereinschiffung der Armee’.

¹¹⁸ Once the infantry lines were interlocked the cavalry, like the archers, could not intervene effectively; as Hignett 69 puts it, ‘once the hoplites came to close quarters, this cavalry would be of no use’.

¹¹⁹ Macan 223 f. quotes the evidence for the festival. It was held on the 6th of Boedromion (Plu. De Glor.Ath. 7 Plutarch (Cam. 19) assumed this was the day of the battle, but he was mistaken as Herodotus put the arrival of Philippides at Sparta on the 9th of a month and the battle was some days after the 9th. Scholars have debated whether the battle was in the month of Boedromion or in the preceding month, Metageitnion; most favour Metageitnion. Burn 240 n. 10 and 257 concludes for Metageitnion, but on p. 256 retails the vow of Callimachus, which, as I argue below, is better explained if the battle was fought after the festival, i.e. in Boedromion.

¹²⁰ Seltman, C. T., Greek Coins 91Google Scholar; the coin is illustrated in my Hitory of Greece plate XIIa. Burn 256 ‘probably commemorating the date of the battle’.

¹²¹ I am very grateful to the Astronomer Royal, Sir Richard Woolley, who has given me the times of moonrise, moonset, sunrise and sunset for the September moon in 490 B.C. at a position of latitude N. 38° and longitude E. 24° the accuracy of the times of moonrise and moonset is probably about ten minutes, and the accuracy of the times of sunrise and sunset is probably less than five minutes. Full moon rose on September 9th in 490 B.C. The times are as follows, given in local mean time for a position of latitude N. 38° and longitude E. 24°:

On my interpretation the battle was joined on the morning of Sept. 11th, when sunrise was at 05.33, and this was the 17 th day of the lunar month in Attica.

¹²² The evidence of Xenophon, an Athenian born c. 430 B.C., that the vow was made by the state, is much to be preferred to the statements of the Scholiast to Arist. Eg. 657 and Aelian V.H. ii 25, which attribute the vow respectively to Callimachus and to Miltiades. At the time of the festival Athens was about to fight for her existence, and Artemis Agrotera was the war goddess (as PW i 907 puts it, ‘die kriegerische Artemis Agrotera’), as we know, for instance, from the procession of the ephebi in armour in her honour (Syll. ³ 717). It is therefore almost certain that the state festival on the 6th of Boedromion was the occasion on which the state made this vow. Apollo shared the cult with Artemis; he too was honoured for his share in the victory, as we know from the inscriptuon at Delphi (Tod, GHI no. 14).

¹²³ The sea level was then some five feet lower than today (see my article in JHS lxxvi [1956] 35 nn. 9 and 10) but there must have been considerable advancement of the coast as silt accumulated.

¹²⁴ So too Suidas s.v. διεξιφίσω and Schol. Arist. Eq. 778 ἐνωρμίσαντο and ἐνώρμησαν.

¹²⁵ For the ten yards see my article on the Battle of Salamis, in JHS lxxvi (1956) 42.Google Scholar Myres 207 calculates the space at sea for 600 ships and Maurice 20 n. 6 for 250 ships.

¹²⁶ The general attitude of scholars to Herodotus' account of the Persians sailing round Sunium and the Athenians' march back ‘post-haste’ is sheer incredulity. The Persians are too tired (Schachermeyr 29 ‘eine besiegte Truppe wäre für ein solches Unternehmen gar nicht zu brauchen gewesen’); so too are the Greeks (How and Wells ‘The distance … is more than an army could march after a pitched battle’). The number of miles from Marathon to Athens by sea increases with increasing incredulity from the 70 miles of Myres 211 and How and Wells ii 113 to the 90 miles of Grundy 191 and Hignett 73. Grundy put the voyage under the most favourable circumstances at 9 or 10 hours for 90 miles; Hignett criticises Grundy for this estimate. Caspari 104 says the fleet ‘may have required 20 hours or more’ for his 70 miles.

¹²⁷ For the chronology see my ‘Studies in Greek Chronology’ in Historia iv (1955) 406 f.

¹²⁸ As Sparta and Corinth did in the campaign of Plataea (Hdt. ix 28).

¹²⁹ Lindos Chronicle D 5 in Lindos ii i. 183 and 194, where Blinkenberg combats the scepticism of Beloch, , GG ii ²2.81 f.Google Scholar

¹³⁰ IG ii 1.471 line 26.

¹³¹ E.g. Philippides' speech at Sparta, Hippias' remarks about his tooth, and Miltiades' conversation with Callimachus (who was killed a few days later). As W. G. Forrest points out to me, Herodotus treats Marathon less fully and less continuously than the campaigns of 480 and 479 B.C.

¹³² Macan 164.

¹³³ See n. 5 above.

¹³⁴ Kromayer 1, 6 and Grundy 183.

¹³⁵ Cf. Hignett 60.

¹³⁶ Hignett 64. Gomme's paper is a typical example, beginning ‘everyone knows that Herodotus' narrative will not do’ and creating ‘difficulties’ such as the later movements of the 4,000 Athenians from Chalcis (see n. 85, above).

¹³⁷ No deeper meaning should be attached to this expression, since the story shows Miltiades to have had no powers superior to those of his colleagues. Macan 158 is mistaken in claiming that ‘this phrase can hardly be taken to mean that Miltiades was last and least of the ten strategi, but rather that he was chief and foremost’; he seems to shut his eyes to the equality which is obvious. See Dover, K. J. in JHS lxxx (1960) 70Google Scholar, who is of the opinion I am expressing but imparts a ‘concessive’ implication which I do not find.

¹³⁸ Technically, the division being 5 versus 5, it was a deadlock, but Herodotus may be right in suggesting that in a deadlock the proposal to engage would lapse and the Greeks would stay on the defensive.

¹³⁹ Macan 157, Callimachus ‘voted with the Strategi as one of themselves’, obscures the issue; Gomme 31, ‘president of the board of generals with a casting vote’, is even wider of the mark. The polemarch was not a Strategos and did not attend the generals' meeting; he was imported later with a vote, ψηφιδοφόρος. In early times he had been ὁμόψηφος by right, such a right as the Argives claimed when they suggested that the Argive king should be ὁμόψηφος with the two Spartan kings (Hdt. vii 149.2).

¹⁴⁰ There is of course no idea of purpose in the Greek it was a matter of fact, a coincidence of time, that the day of the attack happened to be Miltiades' own day of command by rotation.

¹⁴¹ I have not discussed the order in which the tribal contingents were drawn up because the evidence seems inconclusive and the solution does not affect the course of the battle; see Macan 216 f. Munro, , CAH iv 246Google Scholar, followed by Burn 250 n. 29, suggests that on the day of the battle the Greeks advanced in column and then deployed; it is much more likely that they bivouacked in line in position for immediate action. I believe the contingent of the tribe Aeantis was commanded not by the polemarch but by its taxiarch. The polemarch's position was determined at that time by the law, namely (Hdt. vi 111.1); here the meaning of ἔχειν, for which see L-S-J⁹ ἔχω A.3, is purely positional as in Th. iii 107.4, where Demosthenes, having been chosen positioned himself on the right wing, The regiment of the tribe Aeantis, to which the polemarch belonged, was on the right wing, probably because Marathon was a deme of the tribe (Plu. Mor. 628). Pritchett 145 f. discusses the position of the tribal contingents.

¹⁴² When the votes of the Strategi were 6 to 4, for instance, there was no point in invoking the polemarch. When Herodotus says τὸ παλαιόν, he is referring not to the context of the battle, which was for most of his audience a contemporary context, but to earlier times, i.e. to a period before the reform noted in Ath.Pol. xxii 2; he means that then the polemarch sat by right with the generals and was ὁμόψηφος. That Strategi existed before 501/0 is fairly clear from Hdt. i 59.4 and Ath.Pol. xxii 3; see Hignett, , History of the Athenian Constitution 169.Google Scholar

¹⁴³ For discussions of this matter see Macan 156 f. and Hignett, op. cit. 169 f. Hignett has confused the issue by saying that ‘in Herodotus the polemarch Kallimachos presides over the board of strategoi’ (the point of Hdt. vii 109.1 and 109.4 is precisely that Callimachus was not present, let alone presiding, when the generals voted) and by not realising the force of τὸ παλαιόν (see last note). His statement on p. 171 that Callimachus as polemarch ‘must in fact have been commander-in-chief’ at Marathon is counter to the evidence of the picture in the Poecile Stoa and of Herodotus’ account of the other instance of command in the 490's when Melanthius was sent to Ionia as ‘strategus’ and not as archon polemarchus (Hdt. v 97.3). Dover, K. J. in JHS lxxx (1960) 71Google Scholar thinks, like Hignett, ‘the generals were all subject to the polemarch’ in Herodotus' account. He refers to IG I² 609, the dedication by Callimachus, but Shefton, B. B. in BSA xliv (1950) 160 f.Google Scholar and Raubitschek's version in 164 show that one cannot rely on the restoration by Hiller.

¹⁴⁴ Ath.Pol. lxi 1 notes the change only from ‘one a tribe’ to election ‘from the whole community’; the mention of this change and of no other change suggests that election was by the people in and after 501/0 B.C. = Ath.Pol. xxii 2.

¹⁴⁵ See Macan 157 f. for this view. Oncken's suggestion, that the lot was used to decide which elected candidate went to which office, may serve as a compromise; but the fact remains that ‘the bean’ is used to express complete sortition unless it is qualified as in xxii 5 (cf. viii 1).

¹⁴⁶ These ideas were expressed by Lolling, H. G. ‘Zur Topographie von Marathon’, Ath.Mitt. i (1876) 90Google Scholar, and have been much repeated since then. He wrote of Nepos ‘Die Schilderung der näheren Umstände der Schlacht bei dem eben genannten Schriftsteller ist sehr rationalistisch gehalten und von Herodots Angaben wesentlich verschieden’. Macan 234 expresses this view: ‘the rationalist réchauffé (by Ephorus) of the Herodotean and other traditions has come out badly in cross-examination’.

¹⁴⁷ Wordsworth, C., Athens and Attica (1869) 39Google Scholar; C. E. Graves in his edition of the play tries to attach the details to Marathon; Raubitschek, A. E. in AJA xliv (1940) 58 f.Google Scholar and Crosby, H. L., Classical Studies presented to E. Capps 75Google Scholar apply them to Marathon and ‘events in Phalerum’.

¹⁴⁸ E. Meyer, iv 1.312 n.; answered by How, W. W. in JHS xxxix (1919) 49 f.Google Scholar, who regarded the use of Ephorus as ‘improbable’. Macan 205 exclaims: ‘surely Ephorus might have led to something better than this!’ I am sure he would have done, if he had been used by Justin for this campaign. The immediate source of Justin was not Trogus (see the parallel passages in Seel, O., Pompeii Trogi Fragmenta p. 57, and n. 79 above).Google Scholar The account in Justin was bowdlerised by the anonymous writers of cod. Laur. 66, 40, 104 and cod. Bamb. E III, 14, 95, quoted by Seel, loc. cit.

¹⁴⁹ Having decided to fight, Datis ‘in aciem peditum centum, equitum decem milia produxit’. ‘Producere’ is probably short for the common phrase ‘producere castris’ or ‘producere pro castris’, and a few lines later Nepos says that the defeated Persians ‘non castra sed naves peterent’. Datis used all his cavalry but not all his infantry; for a considerable force is likely to have been defending the camp.

¹⁵⁰ Some of these points are noted by Macan 207 and by How, in JHS xxxix (1919) 51Google Scholar, who regards them as ‘anachronisms which may with great probability be ascribed to Cornelius Nepos himself’.

¹⁵¹ Only one general was appointed to command the twenty ships which helped the Ionians (Hdt. v 97.3).

¹⁵² For instance in Ath.Pol. xli 2 and xxvii 5 the beginning of τὸ δεκάξειν.

¹⁵³ The Souda quotes a proverb which is on the same topic as the second of these. The style here is different. Zenobius v 29 gives this proverb with slight variations. On the other hand Photius gives the Souda's proverb verbatim. The paroemiographers evidently drew on two separate versions, one of which we know was that of Demon. Strabo viii 6. 16, C375 alluded to the version in a corrupt passage, which Meineke does not include in his text.

¹⁵⁴ This view was held by Crusius, O. in Rh. Mus. xl (1885) 316 f.Google Scholar; he based his argument on other fragments of Demon than that which I have quoted, and he did not make any connection between the entry in the Souda and the trees in Nepos.

¹⁵⁵ The differences are that in Ephorus Miltiades ravaged some islands, in Nepos he forced some to return to obedience and took others by storm; in Ephorus the fire is on Myconus, in Nepos in Asia Minor; in Ephorus the Parians think Datis is signalling by beacon and do not surrender to Miltiades, in Nepos both sides think the King's fleet is signalling, the Parians give up ideas of surrender and Miltiades burns his siege-engines. How exaggerates in saying Nepos', chapter ‘is translated from Ephorus’ (JHS xxxbc [1919] 50).Google Scholar

¹⁵⁶ Lamb, W. R. M., Clio Enthroned (Cambridge, 1914) 122Google Scholar for Democritus and 225 for Gorgias.

¹⁵⁷ The evidence is discussed by Cadoux, T. J. in JHS lxviii (1948) 117.Google Scholar

¹⁵⁸ Herodotus does not record the fate of Hippias. Justin ii 9 fin. says he was killed at Marathon. The Souda and Justin pass the same judgement on his death, namely that the gods of his fathers took vengeance on him, but this was probably a common place. Ctesias, Pers. 18Google Scholar reported the death of Datis in the battle.

¹⁵⁹ The Souda s.v. Polyzelus adds the superhuman size of the apparition, exaggerating the μέγαν of Hdt. vi 117.3.

¹⁶⁰ This agrees with Hdt. vi 106.1 δευτεραῖος, and the Souda s.v. Philippides expands to

¹⁶¹ This may refer in particular to Theopompus, who thought the actual battle differed from the eulogies all men made of it (FGrH 115 F 153 and to his successors.