^{page 208 note b} Thucydidis (sic) I. with collation of the two Cambridge MSS. and the Aldine and Juntine editions by Richard Shilleto, Cambridge 1872.

^{page 208 note c} Demosthenis oratio in Midiam, cum annotations critica et exegetica. Curavit Philippus Buttmannus, Berolini 1823Google Scholar.

^{page 208 note d} Dyroff, also, in the work cited below, II 17 and 161, is inclined to attach too much weight to this criterion.

^{page 209 note a} I mention the Oxford editor not as being in all cases the first to write the text thus, but because his edition is so widely used at present in this country.

^{page 209 note b} The words underlined mean nothing, since, whenever the subject of the main sentence and of the subordinate clause are the same, the problem of indirect reflexion does not occur, and the reflexive must be used if the subject be expressed at all.

^{page 209 note c} A Grammar of the Greek Language, chiefly the German of Raphael Kühner, by Jelf, William Edward², Oxford 1851Google Scholar.

^{page 209 note d} This false rule (already repudiated expressly by Dyroff. 1. inf. cit. II. 160) derives ultimately from Krϋuger, K. W.: Griechische Sprachlehre 1845 I § 51. 5Google Scholar. Yet it is no worse than the statements still current in Grammars: for instance Thompson, F. E. (Syntax of Attic Greek 1907, p. 63 § 42)Google Scholar makes αυτο the normal indirect reflexive, but adds incidentally that the oblique cases of airbs are sometimes so used (!).

^{page 210 note a} οτοι, Πείτε ς ΑίθιοΠίην Πίκοντο, διδοṇι σ ϕας ωυτοủς τᾦ ΑιθιóΠων βασιλι. δσϕεας τᾦδ ντιδωρέεται. So all the MSS. But since σϕας ωυτοủς is an impossible combination, a corrector of P changed ωυτοủς αύτοúς, and on his ‘authority’ that is read in texts. It would, however, be natural to suppose that διδοṇι ωυτοủς alone was the original, and σϕας crept in from the line below: yet this will then be the only passage where the plural reflexive in the phrase διδóναι ωυτóν τνι is ωυτούς, and not σϕας αύτοúς: see below A2 (a).

^{page 210 note b} These are commonly called indirect reflexives, or in Germany simple pronouns; but since the former name presumes the answer to our inquiry, I have preferred, for reasons which will appear, to call them semi-reflexives.

^{page 211 note a} For example, he confuses prolative infinitive and accusative and infinitive under the single rubric “abhangige Infinitivstruktur,” and object clauses, final clauses and clauses of fear under the heading ‘Absichtsatzen.’

^{page 211 note b} The two dissertations of Arndt, C. E.: de ronominum reflexivorum usu apud Graecos observationes, Neubrandenburg 1836Google Scholar, and de pronominum simplicium constructione reflexiva apud Graecos dissertatio, ibid. 1840 were rare already in Dyroff's clauses, time, and to me totally inaccessible.

^{page 211 note c} The count includes 6, 23³ (where class domits the pronoun) and 6, 122² in a passage probably spurious.

^{page 212 note a} On Herodotus' tendency to stereotype turns and phrases see below p. 213.

^{page 212 note b} Stein believes that Herodotus could at will place the possessive reflexive outside an article agreeing with a noun, and adduces in proof six passages (Annotated Edition⁵ V 5, 7). Besides 6, 23³ and 9, 37², dealt with above, he cites 2, 26¹ (where τἠν διξοδον ω υτ ο is merely Stein's own conjecture for αὐτῷof the MSS.); 2, 107¹ (where the d-class has, not ὐτο, and where a reflexice makes no sense: Stein, like Hude and other editors, reads α ὐτο there himself I); 5, 5; φζεται ὑπ το οικηιοττον ωυτς (where οικηιττου is not a substantive); and 9, 31¹ (which is a false reference). Thus does Stein's evidence for this supposed Herodotean irregularity crumple to ntohing at a glance.

^{page 212 note c} He suggests, however, αὐτο (Cf. above p. 211): Krüger had already altered to ωυτν.

^{page 212 note d} The Omission of αὐτοὔτω following.

^{page 213 note a} Both times corrected by Stein, and Hude after him, to σφισιon which see below. Dyroff also (I 127) concurs with Stein here.

^{page 213 note b} And a peculiarity not confined to Herodotus. Thucydides uses the expression αἰεἰ ποτε only in reference to friendship or enmity, and uses it so at least 4 times.

^{page 213 note c} There is nothing to prevent ωυτοῖσι from having this sense of λλλοις but the only apparent case in Herodoius (3, 49¹) is dubitable: εσ λλλοισι διποροι ντες ωυτοῖσι (ωυτο ). Without ωυτοῖσι all would be in order: for εωυσι…öντεσ see the pasages adduced by me in Hermes LVIII 1933 p. 125Google Scholar, to which this may be added. It is possible, however, with Stein to assume a lacuna after ωυτοῖσι.

^{page 214 note a} This is characteristic of Stein. Likewise, having decided on the slightest evidence that Σὐριοι and only Σὐριοι, were the Cappadocians, σὐρο ι, and only σὐρο ι the Syrians (n. on I 6¹), he imposed his rule uniformly throughout without any regard for the MSS.

^{page 214 note b} Under this head A belongs also 1, 132² αυτῷ τῷ θὐοντι οὔ ο ι γγνεται ρâσθαι γαθ:, for the contruction there is ωυτῷ ρ;σθαι γαθ, and οι pleonastic.

^{page 215 note a} I have counted 1, 77⁴: τōν παρεōντα κα μαχεσμɛνον στρατν π ρσῃσι δς ν α ὐ τ ο ξεινικōς, πντα πες διεσκδασε, though I feel almost certain that Stein is right in regarding ατο as partitive, ‘the part of it, that is, which was merecenary.’ For the partitive genitive idiomatically placed within the relative clause see Shilleto's note on Thucydides 2, 45¹.

^{page 215 note b} In 9 cases some MSS. omit the pronoun altogether; in another (3, 135³) οι is the Aldine's certain emendation for.

^{page 215 note c} Twice (1, 29² 117⁵) some MSS. have σφισι.

^{page 215 note d} Kühner, however (II i p. 562), cites the first two passages, explaining that in such cases the relative clause is felt as a simple part of the main sentence.

^{page 216 note a} In one instance half the MSS. ofler the reflexive: ἠν δι' αὐτο (a Suid.: ωυτο⋯) κατλθωσι 5, 30³; but there ωυτο. which neither Stein nor Hude accepts, may well be due to αυτν in the line above.

^{page 216 note b} Once the MSS. vary between σφι and σφσι elsewhere σφι.

^{page 216 note c} μιν (3, 30³); οι(I,9¹); σφεας (4, 140⁴); σφι(7, 191¹). In 9, 18²). δεσαντες μἠ κα σ φ 㽷 γνηται τρμα, the stronger semi-reflexive formseems to show thatκαι preceding, which Hude omits on slender authority, is genuine; but Stein had no reason to ask for σφσι in 7, 191¹ also. In this construction occurs Herodotus' single instance of the genitive singular of the semi-reflexive: 3, 135³ δεσας μ ε Ὁ ༐ κπειρῷτο∆αρεῖος: it may be that at the back of his mind was the Homeric T 463 f., μν ντος ἤλυθεγοὺνων / εἴ πὠς ε ὑ πεφδοιτο λαβών κα ζων φespiv;ཷη.

^{page 217 note a} It ought not to be thought that the position of αὐτν between article and substantive suggests its emendation to ωτν: for HErodotus admits this position of the non-Reflexive pronoun, provided one or more particles also intervene; examples by Stein on VI 30¹.

^{page 217 note b} I have not counted either I, 141⁴, τοῖσι ἓδοξε πμπιν γγλους δεησομνους σ φ σ ι τιμ, ωρε ιν, where σφσι is Naber's conjecture for ἓωσι nor 7, 162², στερισκομνην τν 'ελλδα τς ωυτο συμμαχης εἴκαζε ὡς εἰ κτ., which occurs in a passage considered spurious; nor yet 6, 137², where van Herwerden's emendation σφι αὐτοླྀ for σφσι αὐτοῖσι seems almost certain.

^{page 217 note c} So I, 134² we find νομζοντες τοὐς κασττω οἰκοντασ π ω υ τ ν κακστους εἶναι followed immediately by ἦρχον … μῆδοι … τνἂγχιστα οἰκεōντων σ φ σ ι

^{page 218 note a} Van Herwerden apparently emended all instances of σφι in indirect questions to σφσιand left those occurring in indirect statements; but the distinction is an arbitrary one.

^{page 219 note a} It is probably worth noting that these 10 all occur in Books V and VI and the five introductory chapters of Book I.

^{page 219 note b} Ihave counted 1, 205¹: συνιεῖσα οὐκ α ὐ τ ν μ ι ν μνὡμενον λλ ττ μασσαγετων ρχν which I believe equivalent to οὐκ αὐτν μν μιν μνὠμενον, αὐτν μιν being the reflexive (see above under A2 p. 213), and μιν doing double duty, by an idiom common in Herodotus: see Stein's note on 1 137⁴ for a list which might be considerably amplified from other authors, Greek and Latin.

^{page 219 note c} It is per worth noting that in Books I-III the proportion of reflexives to semireflexives is 19: 19; but in Books IV–VI it is 11: 20, and in Books VI1-IX only 2: 11. It might seem, then, that Herodotus progressively became less inclined to use the reflexive in this construction.

^{page 220 note a} I have included 7,152³: πν σ β ουλōμεν᯶ι σφι εἶναι πρ τς παρεούσησ λύπης where σφσι is read on the authority of Plut. Mor. 863. But Plut. loses his evidential value for this point when we consider that(I)he is not there quoting exactly, since he reproduces the words as βουλōμενο σφσι προεῖναι τς π λ, and (2) that according to Wyttenbach's Lexicon of his complete works (leipzig, 1843 he limself uses σφσι, but apparently never σφι.

^{page 221 note a} In such an article as the present it was obviously undesirable to give more than a few of the hundreds of references involved; but I have a record of these myself and would gladly furish information of them on any particular point.