The theory of the comparative method has been discussed in a previous article. I pass on to the results obtained—the actual history of prose-rhythm in the practice of particular authors. I give below the figures of normal frequency on which my statements are based. For Greek prose, as I have already said, Thucydides may be considered a practically unmetrical author, since nearly all his clausulae occur with about the same frequency as might be expected from the natural proportion of long and short syllables in the Greek language, and since there is little difference between his sentence-metre and his clausula-metre. (A slight difference between the two is natural, because there are some words, such as the article, which can scarcely be used to close a sentence.) Thus the theoretical frequency of — ∪ — is 14·19 per cent.; its frequency in Thucydides' clausula-metre is 14·2, in his sentence-metre 14·4. But there are two cases where the difference in these proportions seems too great to be due altogether to chance. Compared with the theoretical calculation and with the sentence-metre, —∪∪—— occurs in the clausula considerably more frequently (6·1 per cent, as against 2·6 per cent.) and —∪—∪ considerably less (3·7 per cent, as against 5·1 per cent.). The first form seems to be sought by Thucydides; the second seems to be avoided (doubtless as suggesting an iambic trimeter ending). For these two forms, therefore, I have considered as normal the average percentages of 2,000 cases in the sentence-metre of Thucydides and Xenophon (1,000 each), for the rest the percentages of 2,000 cases in the clausula-metre of Thucydides. For normal frequency in Latin metrical prose I have used de Groot's figures, based on 2,000 cases from nineteenth-century Latin translations of Gregory and Athanasius.