mentions that the decks of both fleets were crowded with heavy infantry archers and javelin-men, "for their naval engagements were still of the old clumsy sort." Possibly this last sentence gives us a clue to the explanation of the apparent discrepancy. The Athenians were, as we know, expert tacticians at sea, and adopted the method of ramming hostile ships, instead of lying alongside and leaving the fighting to the troops on board. They may, however, have been forced to revert to the latter method, in order to provide for cases where ramming could not be used; as, for instance, in narrow harbours crowded with shipping, like that of Syracuse. It is perfectly certain that the Phoenician ships which formed the most important part of the Persian fleet at Salamis carried fighting-decks. We have seen already (p. 28) that they used such decks in the time of Sennacherib, and we have the distinct authority of Herodotus for the statement that they were also employed in the Persian War; for, he relates that Xerxes returned to Asia in a Phoenician ship, and that great danger arose during a storm, the vessel having been top-heavy owing to the deck being crowded with Persian nobles who returned with the king. [Illustration: FIG. 11.--Galley showing deck and superstructure. About 600 B.C. _From an Etruscan imitation of a Greek vase._] Fig. 11, which represents a bireme, taken from an ancient Etruscan imitation of a Greek vase of about 600 B.C., clearly shows soldiers fighting, both on the deck proper and on a raised, or flying, forecastle. In addition to the triremes, of which not a single illustration of earlier date than the Christian era is known to be in existence, both Greeks and Persians, during the wars in the early part of the fifth century B.C., used fifty-oared ships called penteconters, in which the oars were supposed to have been arranged in one tier. About a century and a half after the battle of Salamis, in 330 B.C., the Athenians commenced to build ships with four banks, and five years later they advanced to five banks. This is proved by the extant inventories of the Athenian dockyards. According to Diodoros, they were in use in the Syracusan fleet in 398 B.C. Diodoros, however, died nearly 350 years after this epoch, and his account must, therefore, be received with caution. The evidence in favour of the existence of galleys having more than five superimposed banks of oars is very slight. Alexander the Great is said by most of his biographers to have used ships with five banks of oars; but Quintus Curtius states that, in 323 B.C., the Macedonian king built a fleet of seven-banked galleys on the Euphrates. Quintus Curtius is supposed by the best authorities to have lived five centuries after the time of Alexander, and therefore his account of these ships cannot be accepted without question.