112 HISTORY OF GREECE. Hardly was the fleet mustered at Salamis, and the Athenian population removed, when Xerxes and his host overran the de- but the items which he gives amount, when summed up, only to three hun- dred and sixty-six. There seems no way of reconciEng this discrepancy except by some violent change, which we are not warranted in making. Ktesias represents that the numbers of the Persian war-ships at Salamis were above one thousand, those of the Greeks seven hundred (Persica, c. 26). The Athenian orator in Thucydides (i, 74) calls the total of the Grecian fleet at Salamis " nearly four hundred ships, and the Athenian contingent somewhat less than two parts of this total (vavg fiev ye eg rag TETpanoaiag 67.iyu) kXucaovg ruv 6vo fioipHv)." The Scholiast, with Poppo and most of the commentators on this pas- sage, treat tuv 6vo jioipuv as meaning unquestionably tvx> parts out of three : and if this be the sense, I should agree with Dr. Arnold in considering the assertion as a mere exaggeration of the orator, not at all carrying the au thority of Thucydides himself. But I cannot think that we are here driven to such a necessity ; for the construction of Didot and GoUer, though Dr. .fVmold pronounces it " a most imdoubted error," appears to me perfectly admissible. They maintain that al 6vo fiotpal does not of necessity mean two parts out of three: in Thucydid. i, 10, we find Kairot. HeXonovvfjaov tuv TTEVTE Tuf 6vo fioipuc VEfiovTat, whcrc the words mean tivo parts out of five Now in the passage before us, we have vavc fisv ye eg rug TerpaKoaiag d?iiy<f> i/.aaaovg tuv 6vo fioipCiv : and Didot and Goller contend, that in the word TE-paKoaiag is implied a quaternary di'ision of the whole number, — foui hundreds or hundredth parts: so that the whole meaning would be — "To the aggregate four hundreds of ships we contiibuted something less than two" The word Terpaiioaiag, equivalent to reaaapag tKarovraSag, naturally includes the general idea of reaaapag fioipug : and this would bring the passage into ex^ict analogy with the one cited above, — tuv Tvevre rag dvo fioipar. "With every respect to the judgment of Dr. Arnold on an author whom he had so long studied, I cannot enter into the grounds on which he has pronounced this interpretation of Didot and Goller to be "an un- doubted error." It has the advantage of bringing the assertion of the orator in Thucydides into harmony with Herodotus, who states the Athe- nians to have furnished one hundred and eighty ships at Salamis. Wherever such harmony can be secured by an admissible construction of existing words, it is an unquestionable advantage, and ought to count as a reason in the case, if there be a doubt between two admissible construc- tions. But on the other hand, I protest against altering numerical state- ments in one author, simply in order to bring him into accordance with another, and without some substantive ground in the text itself. Thus, for example, in this very passage of Thucydides, Bloomfield and Poppo propose to alter reTpaKoaiag into rpcaKoaiag, in order that Thucydides may be in harmony with JEschylus and other authors, though not with Herod-
