In this provocative and intelligent book, Wallinga (henceforth W.) argues for the substantial accuracy of the fifth-century tradition concerning the naval campaign of 480 BC. His argument is sophisticated, and illuminating even for those who are inclined to place less confidence in the sources than does he.1
In his first chapter W. argues that the emergent power of Persia was the driving force for naval innovation during the later part of the sixth century BC, reprising his argument in Ships and sea power before the great Persian war that the trireme was originally developed in Phoenicia, and that the first great battle fleets employing this ship were those used for Cambyses’ invasion of Egypt in 525 BC. His basic point, that the use of sea power appears vastly better developed in the territory of the Persian King by 500 BC than it was anywhere in Greece, seems unimpeachable — the tradition in Thucydides that the trireme was first developed in the west is late and unreliable.
Perhaps less unimpeachable is W.’s reconstruction of the decade after Marathon. In W’s view, the tradition that Darius began the preparations for a second invasion after the failure at Marathon is false. The Persians, who would not, on his reading, have regarded Marathon as a disaster, became concerned about the situation in Greece only after the creation of a new Athenian navy in the late 480s (p. 21; 25-26). Although Themistocles’ naval bill was directed at Aegina, a Persian would have seen the construction of a modern battle fleet as a threat to Persian domination of the Aegean (p. 30). For this view to be sustained we need to assume that the Persians regarded the expedition of 490 as a substantial success, and that Herodotus’ statement that Darius planned a second expedition when he got the news of Marathon is false. On a priori grounds the notion that Herodotus could overstate the Persian reaction to Marathon is not unreasonable, but in this case the internal consistency of Herodotus’ chronology — taking account as it does of both regime change and revolt in the Persian Empire — is not self-evidently absurd. Nor is there other evidence to set against what Herodotus has to say (which I think there is in a case where I would take issue with W.’s support for the tradition).
In the second chapter W. looks at the question of Persian numbers. Noting that there is a good deal of internal consistency in the numbers that Herodotus gives, region by region, from the sixth into the fifth centuries, he argues that Ionia and Phoenicia floated roughly three hundred triremes each c. 500 BC (p. 35). The loss of the Aegean squadron to the rebels at the outset of the Ionian revolt in 500/499 BC caused Darius to double the size of the Phoenician fleet (p. 36). This calculation, when the recovery of the Ionian fleet is factored in, brings W. within striking distance of Herodotus’ 1207 triremes as a number representing the total naval power of the Persian fleet. W. supports this calculation through reference to Aeschylus’ statement that there were a thousand Persian ships at Salamis, even after ships had been lost on the way (p. 44). Perhaps the most intriguing suggestion that W. has to make is that numbers of ships do not automatically translate into numbers of sailors. He argues that the Persian ships sailed with skeleton crews that were only made up to full strength when battle approached (p. 42-43 with p. 42 n. 35). Given that extensive training was needed for effective coordination of a crew under battle conditions, I doubt that, if ships did indeed come in this way, they were front line vessels. This does not mean, of course, that ships could not have been brought as spares, to be crewed with survivors from ships too damaged by storm or enemy action to be usable again.
In chapters four through nine W. reconstructs the battlescape of Salamis, and offers his reconstruction of the decisive battle. His arguments about the topography are first rate (the battle was fought by the Greeks facing north in a line from kra Kinsoura to the shallows off Pharmakoussa). So too is his analysis of the Persian plan, which was simply to engage the Greek fleet in a battle of annihilation while troops were landed on Salamis to slaughter the civilian population (the point of the Persian occupation of Psyttaleia was to facilitate this attack, as argued in chapter six). Themistocles recognized the Persian plan from his observation of maneuvers on the afternoon before the battle itself (chapter five). Seeing the efficiency of the Persian maneuvers, he wanted to weaken the Persian attack force by convincing the king to divide his strength between two squadrons in the mistaken belief that the Greeks were trying to flee (p. 78-81). The result of the message was that Xerxes made the classic error of dividing his force in the face of the enemy.
In chapters seven and eight W. reviews the tactical capabilities of the Greek and Persian fleets, showing that the Persians should have been better in an open battle, which is why Themistocles sought to weaken the Persian attack by getting Xerxes to divide his forces. The key to the action on the day of the battle was that the Persians were not able to make use of their superior ability. Instead of confronting the Greeks in an unbroken line, the Persian squadron fell into confusion, allowing the Greeks to encircle and destroy it. The proximate cause of the Persian confusion was an unexpected maneuver by the Corinthian squadron, which had broken ranks to occupy the straight between Pharmakoussa and Amphiale (chapter nine). Confronted with the evident flight of a portion of the Greek fleet the leading Persian ships had to alter their course, and in so doing began to unravel an operation that depended on extreme precision if it was to work (p. 136; 151). The minimal Greek plan —hold fast like a hoplite phalanx — proved the more practical. This reconstruction has the particular advantage of making the best sense out of Aeschylus’ narrative in the Persae, and should, to my mind, be accepted.
When it comes to understanding the course of the actual battle, W. is correct to see that the crucial point is that the Persian fleet had to pass through a strait that is 1200 meters wide at its narrowest point to engage an enemy that could deploy on a 3600 meter front (the maximum possible space for the Greek deployment). Where disagreement is possible is in the estimate of the number of ships that could do this. The issue is important because it gives us an idea of the scale of military operations in the early 5th century and might act as a control for other figures that appear in Herodotus.
There are there are three possible approaches to the numbers question:
1. To try and reconcile the numbers given in our texts with the space
2. To estimate the number of ships on the from figures generated on the basis of calculations supported by what is known of ancient tactics
3. To estimate the number of ships on the basis of comparative data.
W. adopts the first course, making astute use of Aeschylus’ statement that 207 ships ( Persae l. 343; 366 with W. p. 57) formed in three rows ( stoichoi) upon entering the bay (thereby allowing 69 ships per rank, which would fit through the 1200 meters at the entrance, assuming that each ship occupied roughly 24 meters). The rest of Xerxes’ thousand ships remained outside the bay, as, indeed, Herodotus says that they did. As for the Greeks, he notes that, although Herodotus says that there were 378 ships (Hdt. 8. 82), Aeschylus says that 300 were engaged in the battle, suggesting that all the ships were not manned. Although it is possible that he is correct, these numbers, especially those for the Greeks within the bay, depend upon the calculation that a trireme in line of battle occupied only 17 meters (11 meters for the ship and its oars, with 3 meters clearance beyond the oars); a figure that W. admits is very tight (p. 59).2 A further problem is simply that if Xerxes lost only about a quarter of his fleet and still would have had a vast superiority in numbers after the battle, why did he give up?
While it is not unreasonable to object to assertions on the part of many historians that the Persian fleet was much smaller than attested in the ancient sources, as W. does, he has, to my mind, used the wrong variables to calculate the number of ships engaged in the battle. He is not solely to blame for this. The figure of six meters that he uses for the interval between triremes in line is derived from J.S. Morrison, J.F. Coates and B. Rankov, The Athenian Trireme 2nd ed. (Cambridge, 2000), 59 (who share his view of the size of the Persian fleet). In order to defend the traditional numbers, he (and did they) calculates the available space in the bay of Salamis on the basis of an equation that begins with the figure of 11 meters for the width of a trireme and its oars, with three meters on either side, allowing a maximum of seventeen meters per ship. I doubt that this is realistic. The distance between ships is less than the 6.5 meters allowed in modern rowing competitions where the boats are expected to cover relatively short distances in a straight line under, ideally, absolutely calm conditions.3 The distance is also substantially less that the 20-yard minimum between ships engaged in refueling operations in the contemporary US Navy, a useful figure since it reflects what is considered safe only under abnormal conditions. The reason for this is simply that it is very difficult to exercise any control over the lateral drift of even modern vessels. In a wind there is no way to keep a ship from sliding to leeward, and the slower the ship, the more acute the problem (the authors of The Athenian Trireme note the problem of drift on p. 258).4 As indeed happened at, for instance, Phormio’s first battle in the gulf of Corinth, triremes drawn up without sufficient intervals get blown into each other.
How then can we arrive at figures that may be more accurate than those generated with the aid of The Athenian Trireme ? As I suggested above, there are two possible methods, which, though yielding results that differ by a factor of 50%, suggest that the battle was on a very much smaller scale than W. allows. In this calculating the space between triremes on the basis of ancient tactics, the key is the diekplous, or rowing through the line, which was only possible if there was an interval between ships that would allow a third ship to pass easily between them. On this basis, and it should be immediately obvious that there is not much to go on here beyond the demonstrated width of the Olympias, I suspect that triremes in battle formation allowed around 25 meters between oar tips, and thus that each ship occupied roughly 36 meters. This would give a maximum length of the Persian line as 100 triremes, meaning that the Persians might have attacked with about half their available ships (assuming for the sake of argument that the number 207 represents the number of battle-worthy triremes) while the others were held in reserve or formed part of the blocking force sent around the island.
The second line of approach to the numbers question is offered by accounts of battles between Christian and Turkish fleets in the sixteenth century. We know, for instance, that Venetian war galleys were roughly 19 meters from oar tip to oar tip (W.L. Rodgers, Naval warfare under oars 4th to 16th centuries: a study of strategy and ship design [Annapolis, 1939]: 230), or roughly one third greater than the Olympias. At the battle of Lepanto in 1571, the center of the Christian fleet occupied 2000 meters with 62 galleys in line (32.25 meters per galley allowing a twelve meter area between each). If the Greek and Persian fleets used intervals such as those used at Lepanto, then the Greek line could not accommodate more than 145 ships (allowing for the smaller beam of the trireme), while the Persians would enter the bay in three lines of roughly 48 ships. It is also important to note that there was one encounter in which very narrow gaps are reported for a defensive fleet in line. This took place at Modon in 1572, where about two hundred Turkish galleys were moored on a 2700-meter front with 14-meter gaps from mainmast to mainmast, or almost exactly the formation that W. supposes for the Greeks at Salamis. The problem is that the Turkish ships were anchored with additional stern lines running to the shore to stabilize their batteries, a “far closer order than it was possible for the Christians to maintain while under way” (Rogers, Naval warfare, 225) and the Christian fleet declined to attack.
While we obviously cannot claim to know the exact numbers at Salamis, study of the action in terms of ground-scale can help us far better, I think, than the texts of Herodotus and Aeschylus. In the limited space available at the bay of Salamis, we have no reason to think that ships were deployed for battle (whatever the approach formation) in more than a single line, and thus, I suspect, the actual numbers in each fleet were closer to 150 than 200 ships, though I would allow that the remarkably precise number 207 may represent the totality of he Persian battle fleet, and that perhaps a third or more were lost, which would have left the Persians at potential disadvantage if the victorious Greeks tried to press the matter. Even allowing that there were between three and four hundred ships in the region of Salamis on the fatal afternoon, this is still by far the largest naval encounter that took place in the fifth century.
I have taken much space over the issue of space because W. himself has deployed the argument in what strikes me as a potentially significant way, and because I feel that he has picked upon issues that should enable us to understand what happened. Even if we disagree on the scale of the naval operations in 480, I feel that I am in essential agreement with his account of the course of the action. W.’s is a highly intelligent, learned book, from which any reader will learn a great deal.
1. I am grateful to Professor Michael Flower for comments on an earlier draft, and to Dr. Adam Kemezis for reading repeated drafts and improving each one. Errors that remain are obviously my fault.
2. I admit to some confusion about his argument here. In his reconstruction of the battle W. argues that the attack force consisted of 207 ships (p. 132), but on p. 60 he writes “conversely, the file/line of 207 ships cannot be combined with a hypothetical plan of attack that would bring the Persian ships in line abreast into [S]alamis Strait, which is roughly 1650m wide over much of its length i.e. 8m per ship for 207. Even for 207 ships in double line abreast the strait is no doubt too narrow (and why the odd number?). For lines of 69 ships there would be more than enough room across the strait (some 24 meters per ship), but such lines would never have been called stoichoi, nor would three of them operate in combination.” But if we are to get 207 ships into the strait, then we must allow that they were in successive lines abreast, and, since this is the number that he wants, I suspect that there may be an issue with the editing here.
3. I am indebted to Kate Bosher on the University of Michigan for this point.
4. I am indebted to Commander Steve Roper, USN retired, for advice on this point.