^{page 76 note 1} I have simplified the interrelation of , B, and H. In 1952 Mr. D. M. Lewis kindly communicated to me the results of his own examination of H, from which it appears that the text of the first hand of H is very close to B throughout; passages before 6. 92. 5 where H is right and B wrong are the work of correctors. [See now Mme de Romilly's edition, i (1953), pp. xxvii-xxix.]

^{page 76 note 2} I have been fortified in this conclusion by discussion with Mr. Lewis. In classing G among the recentiores I am not suggesting that it does not contain much which, though of diverse origin, is also of great value; in this particular it resembles several of the Thucy-didean recentiores; but I do suggest that G is irrelevant to the problem discussed in this paper. [See now de Romilly, pp. xxiv–xxv.]

^{page 76 note 3} Bartoletti explained those passages in which M is right against as well as a few errors shared by EM against by supposing that β was collated with a manuscript independent of Θ; that is to say, M adopted most of the readings from this source, γ rejected nearly all of them, and δ rejected those that remained, thereby reverting to something very like the text of β before collation. M. Bertrand Hemmerdinger, who generously allowed me to read his unpublished thesis on the history of the text of Thucydides, put forward a radically different view of the interrelation of ABEFM; we disagreed on so much that I should like to say how grateful I am to him for goading me out of too complacent an acceptance of the currently adopted stemma of Thucydides.

^{page 77 note 1} In Bartoletti it means ‘the latest common ancestor of C and G’.

^{page 77 note 2} Mr. Powell uses it to mean the lost manuscript which I call Ω.

^{page 77 note 3} See also his review of Bartoletti in Gnomon xv. 281.

^{page 78 note 1} This is not obvious from the O.C.T. apparatus, and reference should be made to the 1898/1901 edition of Hude. No doubt Hude's apparatus contains inaccuracies, but even if so high a proportion of it as 1 per cent, is wrong the argument will not be substantially affected;

^{page 78 note 2} Cf. 8. 19. 4 E 8. 20. 2 CM:

^{page 78 note 3} For the absence of the pronoun in a similar genitive absolute cf. 4. 3. I, 16. 2, 21. 3, 41. 4, 47. 1, 94. 2, 135. 1.

^{page 78 note 4} viz. 1. 4. 1, 9. 2, 12. 3, 90. 1, 107. 4, 120. 5, 127. 2, 136. 4 bis, 137. 1, 2. 19. 2, 34. 6, 94. 1, 95. 1, 4. 29. 1, 69. 3, 135. 1, 5. 31. 1, 37. 3, 89. 6. 17. 2, 103. 4, 7. 40. 3.

^{page 79 note 1} The early part of book i is missing in C, the deficiency being made good by pages written in a later hand.

^{page 79 note 2} Mr. Powell denied that is a Greek geographical term, but I do not understand why it is objectionable. is the adjective of , and we should expect to be used like etc. At 8. 13 we find where the context does not provide adequate grounds for choice; but in 3. 94. 2 is clearly the city, not the island, and a place on the island would naturally be described as

^{page 79 note 3} This phenomenon receives a limited discussion in Kühner-Gerth, ii, p. 429, and Goodwin, Moods and Tenses, §§ 502, 555. Treatment of it in Schwyzer's Griechische Grammatik is scattered under different headings.

^{page 80 note 1} in the index nominum of I.G. xii. I, referring to 732. 7 (Camirus) seems to be an error; ad loc. (a name found several times on Camirus) is read, as the facsimile demands.

^{page 80 note 2} are vv.ll. in Polyb. 5. 8. 1, and see Burnet's apparatus on PI. Phd. 112 e.

^{page 81 note 1} Cf. Wilamowitz, , Sitzb. Preuss. Akad. 1921, p. 318Google Scholar

^{page 82 note 1} Bast's explanation (in Schaefer's Greg. Cor., p. 931) of is tortuous, and fails to take account of the fact that and are common words beginning and ending alike, so that something more than visual confusion of a single ligature is involved.

^{page 82 note 2} e.g. Dio 56. 33. 3 and fr. 28. Philo, de Gig. 61 PLond. Lit. 138. 9 .

^{page 82 note 3} The Scholia which appear in M, but are absent from the first hands in C and F and altogether from A, B, and E, abound in readings independent of Θ.

^{page 83 note 1} In the nature of the case I cannot adduce instances of errors present in E and known to have occurred in texts independent of Θ. Mr. Powell showed that 4. 15. 1 is irrelevant in this connexion, and that Hude's treatment of 2 ad loc. is misleading; E ^Gc must have read E corrected or corrupted it further to and the reading of Σ^Gc goes back ultimately to