^{page 59 note 1} The version is already in print (from a Wolfen büttel manuscript, Gudian. Gr. 15) in the apparatus to Dindorf's edition of the scholia.

^{page 61 note 1} Since is represented not by one manuscript but by two, only one of them could be an ancestor of (unless one were an apograph of the other). But it would in any case be fantastic to suppose that a single Egyptian manuscript should have become, before it was consigned to a provincial rubbish-tip, the ancestor of the whole medieval tradition.

^{page 61 note 2} The author seems to be strict in his avoidance of hiatus. For instances after prepositives (where it is venial) see p. 62, n. 1; apart from these I find it, in the papyrus texts, in two passages only. One is P.S.I. 1286. A. i. 16f.Google Scholar (Rhes.) here I observe (a) that the first hiatus could be made venial by reading (b) that is an odd verb and the wrong tense (the author seems not to use the historic present), (c) that the medieval tradition has its verb after (and so may avoid hiatus): (for the verb Harl. has, corruptly, V leaves a blank). The other is P. Oxy. 2455.Google Scholar 85 (Skiron) : I should expect corruption, but the text is too fragmentary for speculation. (I do not include P. Oxy. 2455. 215 (Phaethon) for Turner reports ‘the lower parts of two verticals: v would also be possible’; perhapstherefore, space permitting, but the context is fragmentary and the sense may be quite different.)

^{page 62 note 1} Only the strictest practice eschews hiatus after or the article. But our author seems to be chary of it; and he seems also (what tallies) to have it predominantly before proper names. I find the following instances, or possible instances, in the papyrus texts (all in P. Oxy. 2455): 118 123 (?) (proper name ? one expects 228 (conceivably 254 (why the intrusive ? we have been told already that she is his sister; therefore pointless with but in place instead of Turner's articulation is uncertain at fr. 115 and fr. 109 similarly fr. 54

^{page 66 note 1} Turner prints twelve, but I can add three more: see section VI.

^{page 66 note 2} Or see section VI.

^{page 66 note 3} I say ‘is not preserved’: it must be remembered that the beginnings of B (up to Hek. 522) and V (up to Hek. 31: Turyn, op. cit. go) are later replacements.

^{page 66 note 4} Line 2 (Murray) A, Mosch. (against (against (against

^{page 66 note 5} The epitome will thus have occupied, without the heading, 36 lines of the papyrus. If the first and last lines are on a level (Turner, , p. 33) a column of text will have con tained 35 lines (columns which include a heading, whose three lines occupy the space of five or six lines of text, will have had 32 or 33 lines in all); Turner’s estimate (loc. cit.) of ‘between 43 and 45 lines’ must there fore be reduced, and calculations depending on it must be revised.Google Scholar

^{page 68 note 1} Turyn's dating of A; I accept it here ad hominem.

^{page 68 note 2} Turyn never formulates this explicitly as a principle, but he appears (with rare exceptions) to adhere to it in practice.

^{page 69 note 1} have taken the readings of the recensions from Turyn if he cites them, from manuscripts if he does not: Moschopoulos from the Oxford manuscripts X and Xa; Thomas from the Cambridge manuscripts Z and Zd (Cambridge University Library, bound together as Nn. 3. 14; Zd does not include Phoinissai). For the readings of other manuscripts I follow Prinz-Wecklein, but have verified those of MBHLP from the facsimiles.

^{page 69 note 2} Where he does mention a reading, it will be found that he deals with the evidence very differently: at Or. 61 the agreement of Moschopoulos with the papyrus is ‘certainly fortuitous’ (it could be, of course, but why ‘certainly’ ?); in the Life he supposes that A, in copying Moschopoulos’ text, ‘felt the lack of an object after and left a blank space for the missing word to be filled in later on’ (no suggestion that there might be another, and more credible, interpretation of the facts); at Ph. 1597 he does not report the readings of L and P; at Hek. 734 he merely states that and are alike ‘interpolations’.

^{page 69 note 3} See Longman, G. A., C.Q.N.S. ix (1959), 129 ff.CrossRefGoogle Scholar