^{page 194 note 1} An account of it will be found in Shewring's, W. H. articles, ‘Prose-Rhythm and the Comparative Method’, C.Q. xxiv. 164, xxv. 12Google Scholar.

^{page 194 note 2} Cf. also Skimina, S., État actuel des étudessur le rhythme de la prose grecque, i. 35Google Scholar (Bulletin de I'Académie polonaise, Suppl. 3), and Kalinka, E. in Bursian's Jahresbericht, cclvi. 49Google Scholar.

^{page 195 note 1} Handbook of Greek Prose Rhythm, p. 221. Cf. Quintilian ix. 4. 93: ‘neque enim ego ignoro in fine pro longa accipi breuem… aures tamen consulens meas intellego multum referre uerene longa sit quae claudit an pro longa', i.e. Quintilian's ear disagreed with the school rule.

^{page 195 note 2} I am not convinced that it was the ancient practice to scan as long a diphthong or long vowel before another vowel, cf. Billig, L., fournal of Philology, 1920, p. 232Google Scholar, and Wilamowitz, U. v., Hermes, 1900, p. 36Google Scholar.

^{page 196 note 1} Dr. de Groot's figure for form –∪––∪∽(2·4) is above the average according to my statistics, and the high rates in aqua an ignis(3·0), de musica (3·6), and frag. Tyrw. I (4·0), all doubtful or spurious works, may have some significance. I possess figures for some other forms also; the only remarkable feature presented by them is that whereas the ratio of –––∪∪∼ to ∪∪∼ in the sentence was was found by Dr. de Groot to be 5:9, the former is usually more frequent in the clausula, the only notable exception being aqua an ignis with a ratio of 1:5.

^{page 196 note 2} In counting I found the distribution of this particularly erratic within single works.

^{page 196 ntoe 3} Alexander II is not a sequel to Alexander I, but another and probably later handling of the same theme. As h is uneven in style, varying from the grave to the rhetorical, I suspect it may be an early work, revised in maturer years.

^{page 196 note 4} Cf. Wilamowitz, U. v., Hermes, xxxv. 26Google Scholar.

^{page 196 note 5} I am not convinced by the arguments Lattanzi, G. M., Riv.fil. 1932, p. 332Google Scholar, who maintains that de latenter vivendo is a late work.

^{page 196 note 6} Cf. H. Dessau, Prosopographia Imperii Romani, s.v. Plutarchus.

^{page 198 note 1} C.Q. 1929, p. 15. Dr. Fotheringham informed me that the astronomical data taken from Ginzel needed some modification in the light of more recent knowledge, but that the changes did not essentially affect the argument.

^{page 198 note 2} This was remarked on so far as concerns de liberis educandis by F. Bock, Pk.W. 1922, col. 70. In that article he notices the high frequency of –∪––∽ in what he calls ‘the six youthful works’, my columns 1–5 and 8. A count in the three ‘Pythian dialogues’, de exilio and de tranquillitate animi gave low figures for the sum of forms 1–3, and made him think that there was a tendency with advancing years to be less careful with clausulae. My statistics give no support to this idea.

^{page 198 note 3} De vitando aere alieno, by Volkmann, R.. Plutarchs Leben und Sckriften, p. 180Google Scholar, and Pohlenz, M., G.G.A. 1918, p. 328Google Scholar; de amore prolis by Schmid, W. in Christ-Schmid, , Griech. Lit. Gesch. II. i⁶, p. 506Google Scholar; Amatorius by Hirzel, R., Der Dialog, ii. 171Google Scholar, and Cichorius, C., Römische Studien, p. 406Google Scholar. I omit lesser names.

^{page 198 note 4} I have noticed 956 A∽736 A, 956 B∽995 D, 956 D∽968 A, 9S7 A∽98 c, 957 D∽1129 D, 958D∽frag. incert. 114. A glance at the new Teubner edition will show that correspondences between genuine works and the spurious de liberis educandis or Consolatio ad Apollonium are not infrequent.

^{page 199 note 1} Ph.W. 1922, col.70. The implication should be noticed that Plutarch himself made the excerpt.

^{page 199 note 2} By this phrase he means my forms 1–3; my figures show 36 per cent, instead of 39 percent. because I have excluded corrupt clausulae.I have not applied de Groot's rhythmical test of counting the occurrence throughout the sentence of ∪∪, ∪–∪ ∪––∪, ––, –∪–, –∪∪–, …to more than 500 syllables. This trial suggested that to persevere would reveal a distribution within or nearly within known Plutarchean limits.

^{page 199 note 3} Unfortunately Wyttenbach's index is not so accurate that statements about occurrence of words can be made confidently. I hope that all above are true. They have been very kindly checked by Mr. W. C. Helmbold, of Trinity College, Hertford, U.S.A., so far as he was able to do so, from the index to Plutarch which he has begun.

^{page 199 note 4} This outdoes in absurdity anything in the second sophistic. Since the four elements are represented in the universe by the aether, the atmosphere, the land, and the sea, it was considered striking to call sea an element: Artemon (Seneca, , Suas. i IIGoogle Scholarwonders whether it is πρεσβὺτατον στοιχεῖον we read in Apuleius of a man buried at sea, (Mel. iv. 11) iacet noster Lamachus elemento toto sepultus; and Polemo, describing the death of Cynaegirus at Marathon, has κυνα⋯γειρος μ⋯ν ἄνευ χερ⋯ν ναυμαχ⋯ν ⋯ εἰρ δ⋯ ἄνευ κυναιγε⋯ρου διώκουσα, κα⋯ ν⋯κrho;ος ɛἶς ἄμφω τ⋯ στοιχɛῖα πγηρώσας ⋯αυτο⋯ τοῖς μ⋯λεσσιν ἔκειτο, γ⋯ καὸ θαλ⋯σση μεμερισμ⋯νος (C. II). But in all these passages the sea as an element is a substitute for water, not an addition to it.

^{page 200 note 1} It occurs in the collection of moral maxims made by the monk Maximus (Migne, vol. 91, p.882) and runs as follows: π⋯σι δ' ὡς ἔπος εἰπεῖν διαμαχητ⋯ον ⋯μῖν πρ⋯ς τ⋯ν ὔπνον ὡς πλε⋯ονα χρ⋯νον το⋯ ζ⋯ν δι⋯ τ⋯ν ⋯γρ⋯γορσιν μηταλαμβ⋯νειν (sic) ⋯ λ⋯ρ ὔπνος ⋯σπερ τɛλώνης τἤμισυν ἢμῖν το⋯ β⋯ου συνδιαιρεῖται χρ⋯νον. The second half of this is an adaptation of a saying by Aristo, also quoted in aqua an ignis, and that is the sole reason for ascribing the fragment to Plutarch, who in fact took, as one would expect, a different view of the necessity for sleep: ⋯ναληπτ⋯ον ⋯αυτο⋯ς μ⋯θ' ὔπνου φθονο⋯ντας τῷ σώματι κτλ. de sanitate 136 A.

^{page 200 note 2} κα⋯ τ⋯ν φαυλ⋯ν ἔκαστος ⋯σἔπος⋯ μ⋯λλον ⋯ρνεῖται, τοσοὺτῳ μ⋯λλον ⋯νδὺεται καθε⋯ρλνυσιν ⋯αυτ⋯ν εἰς τ⋯ν κακ⋯αν ⋯μ⋯λει τ⋯ν πενομ⋯νων οἱ πρσποιοǺμενοι πλουτεῖν ἔτι μ⋯λλον τ⋯νονται δι⋯ τ⋯ν ⋯λαζ⋯νειαν ⋯ δ⋯ προκ⋯πτων ⋯ληθ⋯ς κτλ.

^{page 200 note 3} The author was so pleased with it that he tolerated the hiatus ἢνευ ἴσταται. According to Kolfhaus, O., Plutarcki de communibus notitiis liber etc. p. 13Google Scholar, aqua an ignis has more frequent hiatus than any other work of the Moralia withthe exception of an vitiosiias, but since he seriously exaggerates the frequency of hiatus in this latter work, I do not know what reliance is to be placed on his figures. The freguency of hiatus in aqua an ignis caused Benseler, G. E., De hiatu, p. 521Google Scholar, to accept Wyttenbach's theory of abridgement.

^{page 201 note 1} It is also interesting to compare the feeble way in which the anecdote about Diogenes is told in 956 B with the version in 995 D.

^{page 201 note 2} One might compare the statement at 957 D: δ⋯λον λ⋯ρ ὡ [τ⋯] οὗ στεροὺμενον οὐκ ἔστι, το⋯ το το⋯ ɛἶναι κα⋯ τ⋯ν αἰτ⋯αν παρ⋯σχηκεν ⋯τ' ἦν.

^{page 202 note 1} Cf. Wegehaupt, J.B.ph.W. 1913, col. 1316Google Scholar.

^{page 202 note 2} A word not found elsewhere in Plutarch,either in the positive or in the superlative.

^{page 202 note 3} The misplacing of the δ⋯ and the lack of adscript iotas will have led to the other changes.

^{page 202 note 4} Wilhelm's, F. article on our fragment, Rh. Mus. lxxiii. 466Google Scholar, ransacks contemporary litera ture for parallels.