Harmodius and Aristogeiton; perhaps we have here a further clue as to the enmity of the Peisistratidae, who were as antagonistic to the earlier members of the Philaid family as were the Alcmaeonidae to the later: for the reference to the tyrannicides would hardly seem apposite unless their memory was cherished by Miltiades.

The arrival of the Plataeans raises the problem of the exact numbers of the Greek army. Nepos says that the total was 10,000 including the Plataeans-horum adventu decem milia armatorum completa sunt. Pausanias and Plutarch give the same numbers, but Justin makes it 11,000, giving 10,000 Athenians 1000 Plataeans. The order of the narrative of Nepos makes the Plataeans arrive at Athens and after they had joined forces Athenienses copias ex urbe eduxerunt, locoque idoneo castra fecerunt. Herodotus, with what seems like disregard for the more probable chronological order, times the arrival of the Plataeans after the Athenians had taken up their position-Αθηναίοισι τεταγμένοισι ἐν τεμένει Ἡρακλέος ἐπῆλθον βοηθέοντες Πλαταιέες πανδημεί. Nepos seems here undoubtedly to follow the better tradition, however preferable it may be from the melodramatic point of view to time their arrival at a critical moment.

His account is still further strengthened by topographical arguments, for there is no direct communication between Plataea and Marathon. The road from Plataea to Athens is straightforward enough, but the only way to reach Marathon without first going to Athens would be by going first to Thebes and thence to Eleon and Oropus and across the spurs of Mt. Parnes to Aphidna and so to the northern end of the plain of Marathon, and this would be a considerably longer march and a much more difficult road than if they went direct to Athens and then on to Marathon by the coast route. Of course Herodotus may have meant that the Plataeans went to Athens and, finding that Miltiades had already set out, followed him later: but he does not say that they went to Athens, and it seems highly probable that he thought that they came direct from Plataea to Marathon.

The most striking divergencies between the two accounts, however, appear in the subsequent narrative of the actual fight. The Greeks, says Herodotus, were ἐν τεμένει Ηρακλέος. Now this τέμενος is usually thought to be half-way up the mountain side in the valley of Avlona, where tradition and probability alike would place it. It was from this eminence, about a mile from the Persian camp according to traditional accounts, that the Greek forces descended at the double, gaining impetus as they went—δρόμῳ ἵεντο ἐς τοὺς βαρβάρους-says Herodotus.

But Nepos gives an entirely different version. After their departure from Athens the Greeks go straight and camp loco idoneo. Then without the delay of several days which Herodotus states took place (probably in order that he could find time for his mis-timed debate of the generals)

postero die sub montis radicibus acie regione instructa nova arte1) vi summa proelium comiserunt.

In this sentence no less than three new pieces of evidence appear. The first is the immediate action of the Greeks without the delay postulated by the confused chronology of Herodotus?). The second is that, whether their camp was on the level or on the hill, they began the battle not from the slopes as Herodotus implies, but from the foot of the mountains on the plain. The third is the assertion of the 'new strategy'-'nova ars'-used by the Greeks. In regard to the latter point both Herodotus and Nepos seemed to have gathered from the sources they used that there was some striking new feature about the battle: Herodotus, however, is at a loss to say precisely what this nova ars was: in despair he applies it to the δρόμῳ tactics-πρῶτοι τῶν ἡμεῖς ἴδμεν he says, δρόμῳ ἐς πολεμίους ἐχρήσαντο—the πρῶτοι τῶν ἡμεῖς ἴδμεν implying the novelty' of the manœuvre.

But, on reflection, to advance at the double upon one's enemies could hardly have been an innovation at the time of Marathon, and it seems foolish to give such prominence to so ordinary a matter. Incidentally, the difficulty critics have felt in understanding how a hoplite force could traverse a mile at the double and have strength enough left at the end to fight seems lessened if we accept this statement of Nepos, according to which the Greek camp was not on the mountain side, but well down on the plain, for then the Greeks would have considerably less distance to cover3).

On the face of it Herodotus seems to have missed the point and to have lost some valuable piece of evidence.

Nepos, however, seems to supply this lost evidence and to give quite a reasonable explanation of this nova ars: namque arbores, he says, multis locis erant rarae and so proelium commiserunt hoc consilio ut et montium tegerentur altitudine et arborum tractu equitatus hostium impediretur ne multitudine clauderetur.

1) Stange's edition reads non apertissima for nova arte, which appears in the first printed text. Whichever reading has better authority, nova ars is a reading justified by the remark of Herodotus πρῶτοι τῶν ἡμεῖς ἴδμεν.

2) To account for the five or six days' delay by adducing as a reason the waiting of Miltiades for the division of the Persian forces, and the waiting of the Persian forces for a signal from Athens, explains merely how the time was spent, not why it was spent. The delay was not necessary for either side: the Alcmaeonid intriguing might equally well have taken place in the preceding week.

3) Herodotus, of course, says that the distance between the rival armies was eight stades (ch. 112), but this assumes that the Greeks were in the Heracleium, which is not assumed by Nepos.

Now whichever historian may seem on the whole to be more reliable. it appears fairly obvious that Nepos gives at least a coherent account of the battle, implying that it was skilfully conducted and a triumph of tactics. All we learn from Herodotus is that it was a wild forlorn hope of men inspired with almost supernatural courage. On the face of it, Cornelius Nepos is preferable, and he at least would have no purpose to serve by elaborating the Herodotean version and improving it with touches of military realism. Herodotus, on the other hand, as will be subsequently discussed, drew from sources which were bitterly antagonistic to Miltiades and Philaid traditions, and which would go to no little trouble to deprive Miltiades even of the semblance of military ability. The whole account of Nepos, in fact, gives an absolutely different version of the battle. Four very important points are emphasised: points which completely alter the traditional conception of the battle. Firstly, there is the assertion that the plain of Marathon near the Greek camp was covered with trees scattered here and there fairly frequently; in the second place the mountain at the rear of the Greeks was of some strategic advantage; thirdly, this first fact-namely, the occurence of trees on the plain-was made use of to hinder the movements of the enemy's cavalry; fourthly, the mere mention of cavalry goes far to dispose of many difficulties which had been brought about by their absence in the Herodotean version.

In regard to the first point, it is generally believed that the plain of Marathon has been from time immemorial not only treeless but actually a marsh in some places. Thus Grote, quoting Finlay, says, 'the uninterrupted flatness of the plain is hardly relieved by a single tree'1). Now however true this is to-day we need definite proof of the assertion that it was equally true in 490 B. C. before we can dispute the very uncompromising statement of Nepos. And not only is the proof for such a dogmatic statement as Grote's not to be found, but there is every likelihood that in the 5th century B. C. the opposite was the case; for it has been shown) that Greece has suffered from a steady loss of vegetation: its surface is becoming every day less covered with trees, and evidence is now forthcoming of the greater prevalence of forest zones in earlier times particularly in northern Greece. But as a matter of fact all that Cornelius Nepos says is that the plain was covered with 'trees at intervals', and this is a description that applies admirably to the typical Greek 'scrubland' or 'bush'.

1) IV p. 274.

2) Zimmern, The Greek Commonwealth p. 41. Even in the last thirty years the area of forest land is said to have been reduced by one half. Cf. Myres, Greek Lands and Greek People.

Now the statement about the strategical value of the mountain is curious and rather difficult to understand. Presumably the meaning of ut montium tegerentur altitudine is that the mountain prevented them from being enfiladed or cut off in the rear, for if they were drawn up sub montis radicibus it would be almost impossible to surprise them in the rear or flank. But the safety of this position would apparently only apply as long as they remained there, for as soon as they left the radices montium they would obviously be unguarded on both rear and flank. But by far the most illuminating evidence is the statement that the Greeks deliberately took up a position where they could make the most of the natural features in order to avoid the dangers of hostile cavalry'); it implies not only that Miltiades was a thoroughly capable tactician, but also that the Persian cavalry, at least at the commencement of the fight. was a force to be reckoned with.

The actual topography of the plain of Marathon at the present day fully bears out these arguments and verifies the account of Nepos 2). The valley of Avlona is bounded on the south by Mt. Agrieliki and Mt. Aphorismos, and on the north by Mt. Kotroni. All these three mountains are wooded and the two mountains on the south side are covered with what is, for all practical purposes, a forest which reaches right down to the level plain. The actual plain is for the most part cultivated, and so denuded of trees, but in the few places where it is not cultivated there are several large trees, and it seems quite possible that in ancient times the whole of the western end of the valley was well wooded. It seems certain too that there were trees at fairly frequent intervals still further down the valley and well on to the main part of the plain, for the whole of the southern part of the plains along the coast towards the site of the ancient Probalinthus is covered with trees right down to the shore. The only part of the plain that was undoubtedly not wooded in ancient times is the part in the immediate neighbourhood of the Soros, between the Great and the Little Marshes.

Now Herodotus, after asserting that Marathon was ἐπιτηδειότατον χωρίον τῆς Ἀττικῆς ἐνιππεῦσαι nevertheless makes no further mention of the Persian cavalry at all: it has consequently been assumed that it never even landed from the Persian transports but was conveyed round to Phalerum as soon as the shield signal was given from Athens. But

1) Arborum tractu means that they scattered tree trunks and branches over the open spaces: they would hardly be likely to do this unless there were trees available on the spot.

2) I am indebted to Professor Lehmann-Haupt for the suggestion that the topography of Marathon supports the accout of Nepos, and I have since been able to verify it in detail myself.

this statement of Nepos shows that the cavalry must at least have been landed, or the Greeks would hardly have taken such elaborate precautions. The statement that the country was dotted with trees seems directly opposed to that which makes it ἐπιτηδειότατον ἐνιππεῦσαι. But perhaps a way out of the whole difficulty may be found by assuming from the evidence of the nature of the actual site that the plain of Marathon near the sea was admirably suited for cavalry action, and that the Greeks, seeing this, remained in their safe position (which commanded the coast road to Athens and so was adequate to prevent Persian advance) until the Persians re-embarked their cavalry and sent it round to Phalerum. Then it was that they descended doou upon the Persians.

Now the unsupported evidence of Cornelius Nepos for this completely different view of the outlines of the battle of Marathon is not likely to be accepted without some hesitation. But when in addition to the support of topographical evidence the narrative of Herodotus implicitly, and perhaps unconsciously, agrees with it, its truth is rendered much more likely, however little value we attach to the version of Herodotus. Such implicit agreement is found in the text: for Herodotus says that the Greeks pressed on at the double', οὔτε ἵππου ὑπαρχούσης και OCTE TOžeruátor1). Certainly no authority asserts that there was any Greek force of cavalry, and it seems redundant to remark on the absence of it; but considering that Nepos had declared that the nature of the land was unsuited to cavalry and that the Greeks had taken advantage of this fact, the remark of Herodotus acquires a new significance. By itself it seems superfluous to say that the Greeks had not got either horsemen or archers: there were innumerable things that they had not got as well. But Herodotus has got such an inveterate habit of inserting incidental and omitting essential points that it may well seem that he stated a fact and forgot to mention the reasons for it: the fact being the absence of cavalry and archers, the reason, as Nepos says, being the incapacity to use them-which would apply a fortiori to the Greeks if they assumed it in regard to the enemy.

Undoubtedly the Persian cavalry was reembarked not so much for its supposed utility on the plains of Phalerum but because of the discovery of the deceptive nature of the place which had been specially selected for its manœuvres. There is no necessity to represent the plans of the

1) It seems hard to understand the absence of archers, considering the discoveries of large numbers of Greek bronze arrowheads on the field. See Hauvette in Nouvelles Archives des Missions scientifiques vol. II p. 326-335. They might possibly have belonged to the Greeks of Karystus, who were pressed into the Persian service (ch. 99).