The problem of the size of the Persian army can be illuminated by considering the size of the Persian fleet, since there must have been a proportion between the two forces.
Herodotos reports that the fleet consisted of 1207 triremes and 3000 lesser fighting ships and supply ships (VII 89, 184). The figure of 1207 triremes is itemized by specifying the number of ships contributed by the several subjects and allies of the Persian Empire (VIII 89-95). Nobody has succeeded in proving that any of these partial figures is questionable; the contributions made by the Greek subjects of Persia corresponds to what we know to have been their naval strength in other episodes of Greek history. Herodotos' figures are confirmed by several other sources. The historian Diodoros (XI 3) states that the triremes were 1200 at the time of the muster at Doriskos; the orator Lysias (II 27) mentions an initial force of 1200 triremes, whereas the orator Isokrates mentions 1300 triremes at the beginning of the campaign (VII 49) and 1200 on the eve of the battle of Salamis (IV 93); Plato (Laws, III 699 B) speaking in general terms refers to "one thousand ships and more." In order to find a trace of disagreement it is necessary to refer to the narrative of the historian Ktesias, as summarized by the Byzantine writers of the ninth century, Photios; in this text the figure of the triremes is given as 1000, but the text contains such an accumulation of obviously wrong information that either Ktesias or Photios must be dismissed as totally unreliable.
The most important datum is provided by Aischylos who fought at Salamis. In the tragedy The Persians he describes the Persian fleet as consisting of 1000 triremes plus 207 ships "of exceptional speed" (341-343). One may ask whether the distinction was introduced for the sake of variety in poetic diction or in order to refer to ships different from triremes; in any case the 207 ships appear to be militarily as important as the triremes. Herodotos asserts that the Persian triremes were faster and lighter than the Greek ones (VII 10,60). This would indicate that the Persians preferred triremes with light shells and had 207 that were of this type. The 207 ships included the 7 ships of the commanders for which speed was particularly desirable. But many scholars interpret the lines of Aischylos as meaning that the Persian triremes were 800: the poet would have mentioned 1000 ships and then added that 207 were "of exceptional speed," meaning that they were some sort of lesser cruisers. But this is a most forced interpretation of the text. The poet aimed at the dramatic effect of a crescendo of numbers, whereas by mentioning 1000 ships and then deducting 207 light cruisers from the figure he would have achieved an anticlimactic effect. It is true that there is a scholion to the lines of Aischylos that states that the 207 ships must be reckoned as part of the 1000 triremes, but this scholion also states that 207 ships were choice triremes. Even accepting the opinion of the grammarian who wrote this scholion, the conclusion would be that the Persian triremes were 1000 and not 800. In my opinion what Aischylos wants to indicate is that the Persians could afford to build triremes of special timbers that made for lighter shells. This agrees with Herodotos' intimation that Persian triremes in general were faster than the Greek ones. Herodotos states also that at the time of the battle of Salamis the people of Aigina kept their slower triremes to guard the island and sent their 30 fastest ones to meet the Persians (VIII 46).
Grundy, writing in 1901, agreed that the number of Persian triremes must have been around 1207. (74)
The following year Munro tried to apply to the fleet the method of analysis developed by Gobineau for the army. He recognized that the texts indicate that the standard strength of the Persian navy was 600 triremes; but he observed that in the report of the muster at Doriskos there are mentioned four admirals, two of whom are in command of 200 triremes each, so that he concluded that the total strength was 800 triremes. (75)
Tarn accepted that 600 triremes was the normal strength of the Persian navy, but tried to prove that this was also the strength in the campaign of 480 B.C. (76)
He remarked properly that the entire Persian fleet was not present at Doriskos, since Herodotos mentions that 100 triremes, the triremes of the Pontic Greeks (the inhabitants of the area of the Bosphoros and the Dardanelles), had been kept at the Hellespont or Dardanelles in order to guard the bridges against possible enemy raids (VII 95). From this Tarn concluded that there must have been five admirals of whom one was at the Dardanelles. He divided a total of 600 triremes into five squadrons of 120, but there is no evidence to the effect that the Persian navy operated by units of 120.
In describing the muster at Doriskos, Herodotos (VII 97) reports that a brother of the King commanded the 200 triremes provided by the Egyptians and that another brother commanded the 200 triremes provided by the Ionians, the Dorians of Asia Minor, and the Karians, whereas two other admirals commanded the rest. This would indicate that there were four squadrons of 200 triremes each. It is striking that Herodotos does not mention the commander of the strongest contingent, the 300 triremes of the Phoenicians. This suggests that the 300 Phoenician triremes were still tied together to form a bridge at the Dardanelles, while the 100 triremes contributed by the Greeks of that area were guarding them. Two admirals with 400 triremes were at the Hellespont, while four admirals were mustering their 800 triremes at Doriskos. It can be inferred that one of the two bridges formed at the Hellespont was left standing for the service of the supplies and for possible reinforcements, as long as the muster at Doriskos had not proved that the army was fit to start operations. Herodotos is in error when he assumes that the two bridges were left in position all through the year (VII 17). In another part he admits that the bridges were no longer there when Xerxes came to the Dardanelles in his retreat, so that the army had to be ferried across (VIII 130). The Persians could not afford to keep the Egyptian and Phoenician contingents immobilized as bridges, and, furthermore, in the course of months the ships would almost certaintly have been destroyed by storms.
It must be concluded that for the expedition of 480 B.C. the normal strength of the Persian fleet in time of war was doubled; the six admirals who usually commanded 100 ships each were put in charge of 200. This would explain the odd figure of 1207 triremes. There were seven extra triremes, six for the admirals and one for the King. In the army, too, the entire infantry was under the command of six corps generals, except for the 10,000 Immortals that formed a separate unit. Under the six corps generals there were thirty division generals who normally commanded a myriad or 10,000 men each, but on this occasion commanded 20,000.
Tarn continued his forced argument by adding that 600 triremes was only the "paper strength" of the Persian navy and that this number was never filled, with the result that at the battle of Salamis the Persian force was inferior to the Greek one, for which nobody questions in a significant way the total of 380 triremes mentioned by Herodotos (VIII 48, 82).
A large number of scholars have preferred the figure of 800 triremes, because they feel that they can justify it by the aforementioned questionable interpretation of Aischylos. Eduard Meyer, without submitting any argument, asserted that the figure of 1000 ships mentioned by Aischylos included the transport ships; the Persian fighting strength would have been between 400 and 500 warships, including warships of lesser size than triremes. (77)
Among the more recent writers Wilcken grants that the Persian ships were 1,000 out of which 207 were fast going, (78)
Berve reduces the total figure to 700 warships, and Giannelli estimates the total number of ships at 1,000 of which 207 were triremes. (79)
According to this last writer the Persians had fewer triremes than the Athenians alone possessed. But the majority of scholars agree that since the Greek fleet was outnumbered in the battle of Salamis, it must have engaged about 600 Persian triremes.
The Persian navy suffered substantial losses before Salamis because of storms and because of engagements. Herdotos specifies that the destruction caused by storms was high, and nobody questions him on this point, since the fleet had to follow the army along the coast for five months. The fleet was so large that it was not always possible to find a good shelter for all its units. Herodotos' declaration (VIII 66) that the losses were made up and that he is inclined to believe that replacements kept the fleet at full strength, is dismissed by Macan as "a fresh extravagance." But it is reasonable to assume that the Persian navy operated as any rational military organization in which forces are divided between first line contingents and reserves to be used as replacements.
Marg grants that the Persian triremes were 1207 at the beginning of the campaign, but twists the interpretation of the text of Herodotos (VIII 66) to meen that the losses due to storms and battles were made up only as far as crews were concerned, not for ships.
The reason for denying that the losses of the Persian fleet were made up by replacements is that critical historians feel compelled, for reasons that I shall explain, to reduce to a minimum the number of the Persian triremes that reached Attika on the eve of the battle of Salamis. Munro, who had put forward a solid argument for conluding that the Persian triremes mustered at Doriskos were 800, twenty years later gave in to the general tendency of scholarship. Without submitting any new arguments he reduced the initial strength to 600 triremes, of which 250 would have been destroyed because of storms, so that only 350 reached Attika. (80)
Some sort of inference about the original size of the Persian fleet can be drawn from the information that 674 triremes and penteconters were tied together as pontoons for the bridges across the sea at the Hellespont. Many of these ships were wrecked by a storm before the crossing of the army started. Since a storm could have made even a total wreck of 674 ships held together by cables and by a causeway, it follows that the Persians could afford to risk such a force. Since Herodotos reports that one bridge was built by the Phoenicians and the other by the Egyptians and also that their contributions to the fleet were 300 and 200 triremes respectively, it would follow that these two national groups used all their triremes for the bridges of 360 and 314 "triremes and penteconters". The bridges had to be formed with the biggest ships available.
The Great Persian War, p. 95.
J. A. R. Munro, "Some Observations on the Persian Wars," Journal of Hellenic Studies XXII (1902), pp. 299f.
W. W. Tarn, "The Fleet of Xerxes," Journal of Hellenic Studies 28 (1908), pp. 202ff.
Eduard Meyer, Geschichte des Alterthums, Vol. III (Stuttgart, 1901), pp. 375f.
Ulrich Wilcken, Griechische Geschichte ninth ed. (Munich, 1962), p. 140.
Giulio Giannelli, Trattato di storia greca 4th edition (Rome, 1961), p. 212.
J. A. R. Munro, "The Deliverance of Greece," in The Cambridge Ancient History, Vol. IV (1926).