^{page 159 note a} Liddell and Scott⁸ were still unaware of this: according to them, o ι is found in Homer, ‘also in Aesch., and in late prose, as Lucian.’ But the 9^th edition gives an adequate idea of the true occurrence of o ι.

^{page 159 note b} 10 of these 12 occur in reported speech or tion. thought.

^{page 159 note c} In many of these I agree with Dyroff, who, however, is considerably more sweeping in his demands for uniformity, although he lacks the two supports on which I rely—comparison with Herodotus and a careful grammatical classification.

^{page 160 note a} One would naturally suppose that this applied only to cases where the noun is expressed, and that where the pronoun depends on an article alone there would be no objection to the non-reflexive pronoun following it, since no other position is available. But this also seems to have been avoided; for in the only instance which I find in Thucydides, 1, 31³ ὃπως μ⋯ σφίσι πρ ⋯ς τᾣ κερκυραίων ναυτικᾣ κα⋯ τ⋯ αὐτŵν προσ γενόμενον ⋯μπόδιον γ⋯νηται, half the MSS. offer 'Αττικ⋯ν instead of αὐτᾣν. Many editors have therefore done well to accept it: it gives the same sense, and the corruption is understand able.

^{page 160 note b} Madvig (Griech. Synt. § 10 Anm. 4) knows of no exception to the rule of position; and neither Kühner-Gerth (II i. p. 620§ 464. 4 and p. 564=§ 455. 5 Anm. 3) nor Jelf² (II p. 320=§ 652. 3) can adduce any prose example of αυτού, αυτŵν immediately between noun and article, where it cannot refer to the main subject.

^{page 161 note a} So also Dyroff (I.c. II 5).

^{page 161 note b} ⋯αυτᾣ or αύτᾣ, offered by MSS. other than C, is clearly a corruption by attraction to the case of the preceding word.

^{page 160 note c} It is well known that in the sense ‘wing of an army’ the contracted forms α⋯ρως κ⋯ρᾳ are used the gen. and dat. singular; but I can find no second example of the nom. or acc. plural in either the longer or the shorter form. ⋯κάτερον κ⋯ρας was a convenient expression, and the sense which Thucydides has to express in 5, 7^I1 would naturally be unlikely to occur often. Liddell and Scott on the forms of κ⋯ρας are unsatisfactory.

^{page 162 note a} Hude's total is only 121, not 122; for in 8, 56⁴ ναûς ᾐεlον ⋯âν (sc. αὐτοὺς) βασιλ⋯α ποιεîσθαι κα⋯ παραπλεîν τ⋯ν ⋯ â υ τ ο û γ⋯ň, he has adopted the reading εαντ⋯ν, which relegates this passage to category P (Prolative Infinitive).

^{page 162 note b} Strangely enough, all but three occur in the first 5 books. Perhaps Hude became more random between his edition of VI-VIII in 1890 and his Teubner text of 1898/1901, in which, however, books VI-VIII are avowedly little more than a reprint from the earlier edition.

^{page 162 note c} It requires some little thought to see that αὐτ⋯ν can refer only to the Eleians (the subject of ⋯φι⋯ναι), and not to the Lacedaemonians.

^{page 162 note d} It is often very difficult to decide whether a genitive should be held partitive or possessive. I have classed cases like ⋯ποΙκους ἔπεμψαν ⋯αυτ⋯ν (2, 70³) and ⋯στρ⋯τευσαν ⋯αυτ⋯ν χιλίοις ⋯πλίταις (1, 113¹) as partitive; but a passage like 5, 84¹ shows how slight the distinction can sometimes be: ⋯στρ⋯τευσαν ναυσ⋯ν ⋯αυτ⋯ν (possessive ?) μ⋯ν τρι⋯κοντα…, κα⋯ ⋯πλίταις ⋯αυτ⋯ν (partitive?) μ⋯ν μ⋯νδιακοσίοις κα⋯ χιλίοις κτ⋯.

^{page 162 note e} I have omitted 2, 100⁵ οὐδε⋯ς ὑπ⋯μενεν ăνδρας ἱππέας τε ⋯γαθοὺς και τεθωρακισμένους, ὐπ⋯ δ⋯ πλ⋯θους περικλῃ⋯μενοι α ὑ τ ο; ⋯; ς; πολαπλασίῳ; τῷ; ⋯μΙλῳ; ⋯ς; κίνδυνον καθίστασαν, ὣστε τέλος; ⋯συχίαν; ⋯γον. The reflexive is possible indeed, but exceedingly awkward at such a distance from its governing verb, and the demonstrative gives excellent sense. For the subject of καθίστασαν is deduced from οὐδείς, the verb περικλῃ⋯μενοι is transitive, as already the Scholiast saw (so 7, 52²: cf. κυκλοûσθαι and the change of subject in the consecutive clause (⋯γον) causes no difficulty.

^{page 163 note a} Counting 2, 39² οôτε γ⋯ρ Αακεδαιμ⋯νιοι καθ' ⋯ α υ τ ο ὺ ς, μεθ' ⋯π⋯ντων δ⋯ ⋯ς τ⋯ν γ⋯ν ⋯μ⋯ν στρατεύσιν. Most MSS. have ⋯κ⋯στους. ⋯αυτοὺς is read on inference from Valla's version, and I have found traces of it in one or two young MSS. Nevertheless, ⋯κ⋯στους is tenable, if Αακεδαιμόνιοι be taken as equivalent to Πελοποννήσιοι in 4, 81 ol 'Σπαρτιâται υ⋯ν … εὐθὺς ⋯βοήθουν…, τ⋯νδ⋯ ᾰλλων Αακεδαιμονίων βραδυτέρα ⋯γ⋯γνετο ⋯ ἔφοδος, where Αακ. is shown to mean Πελοποννήσιοι by the reference to the same fact in 14⁵ oi Πελοποννήσιοι, κα⋯ ⋯π⋯ πάντων ἤδη βεβοηθηκότες, κτ⋯.

^{page 163 note b} In Attic inscriptions of the second half of the fifth century they are all but universal (Meisterhans § 37a).

^{page 164 note a} κατ' ὦν ἔδησαν α ὐ τ ⋯ ν καταλιπόντες 4 157¹. But against the 2 Herodotean exceptions should be held an exactly similar passage in Thucydides, where the form of the pronoun is unambiguous: καταλιπόντες… ἕνα σ φ ⋯ ν α ὐ τ ⋯ ν 6, 50².

^{page 164 note b} Tucker regards the sentence as anacoluthic, and Αακεδαιμονίους as left without construction, That would require eir⋯π' αὐτοîς.

^{page 165 note a} I have omitted 5, 47⁷ ⋯ πόλις … τ⋯ν ⋯γεμονίαν ⋯χ⋯τω, ⋯ταν ⋯ν τῇ α ὐ τ ⋯ ς ⋯ πόλεμος ᾖ, not only because this occurs in a quoted treaty, but because even so αὐτ⋯ς is a conjecture (of Duker) for αὐτῇ of the MSS. Strangely enough Thucydides himself has no instance of a genitive singular in this category.

^{page 165 note b} For oratio obliqua itself see N, below.

^{page 165 note c} Perhaps here the pronoun refers less to the subject of ποιήσονται than to the singular subject of ἔφη, so that the example would fall under class II, below, pp. 172 f.

^{page 166 note a} Since clauses which we distinguish as temporal or causal are introduced in Greek by the conjunctions– ⋯πεί, ⋯πιδ⋯, ὡς and even ŏτε–it is sometimes hard to assign examples with precision to one class or the other.

^{page 166 note b} E.g. Shilleto on I, 30³ (⋯στρατοπεδεύοντο … φυλακ⋯ς ἔνεκα … τ⋯ν πόλεων ὂ σ α ι σφΙσι φίλιαι ἤσαν) writes: ‘The reflexive pronoun states their feelings on the subject: “the states which they felt were still friendly to them.” The ductility of the Greek indicative in orat. obliq. is well known.’ On the same principle 115⁵ (cited above) would mean ‘the officials who they felt were still amongst them!’ Dyroff also (II 9–13) believes that subjective force should not be sought for.

^{page 165 note c} οἰ⋯μενοι τ⋯ν Βουλ⋯ν … οὐκ ἂλλα ψηφιεîσθαι ᾖ ἅ σ φ ί σ ι προδιαγν⋯ντες παραινοûσιν σφίσι refers to the subject of ψηφιεîσθαι, not of οἰόμενοι. Editors have only found difficulty in this and proposed σφεîς, or ⋯ν σφίσι, or taken σφίσι with προδιαγνόντες as full reflexive, because they did not know that Thucydides uses semi-reflexives in simple relative clauses.

^{page 166 note d} ⋯γεμ⋯νες γενόμενοι τ⋯ν τε Ἰώνων κα⋯ ŏσοι ⋯π⋯ σ φ ⋯ ν ἧσαν ξύμμαχοι. I have treated σφ⋯ν as semi-reflexive referring to the Athenians (so Schol., Portus, Bredow, Haacke, Poppo-Stahl, Jowett, Marchant) rather than full reflexive ‘of their own accord’ (Göller, Arnold).

^{page 167 note a} The non-reflexive is not parallel with ⋯αυτοû, but in a dependent participial clause.

^{page 167 note b} The count includes both pronouns in 4, 128² (⋯φοβήθησαν, τ⋯ς τροπ⋯ς α ὐ τ ο î ς γενομένης σ φ ⋯ ν ⋯π⋯ τοû μετεώρον), on which, like Poppo-Stahl, I can do no better than quote Arnold: ‘αὐτοîς Υενομένης is exactly equivalent to ἰδ⋯ντες γενομένην’ and ‘thus αὐτοîς being virtually the subject of the proposition, σφ⋯ν properly follows as referring to it.’

^{page 168 note a} The count includes αὐτοîς in 7, 39³ Ἀρίστων … κυβερνήτης … πείθει τοὺς σφετέρους τοû ναυτικοû ἄρχοντας πέμψαντας ὡς τοὺς ⋯ν τῇ πόλει ⋯πιμελομένους, κελεύειν … ὅσα τις ἔχει ⋯δώδιμα, πάντας ⋯κεîσε φέροντας ⋯ναγκάσαι πωλεîν, ὅπως α ὐ τ ο î ς (αὐτοὺς vulgate) ⋯κβιβάσαντες τοὺς να⋯τας εὐθὺς παρ⋯ τ⋯ς ναûς ⋯ριστοποιήσονται. I take αὐτοîς as ethic referring to the Syracusan commanders, the subject of κελε⋯ειν, on which verb the final clause depends: ⋯κβιβάσαντες then refers to their ship-captains and officers, among whom the speaker Ariston includes himself. Alternatives are (1) to read αὐτοû, after A. Portus–a conjecture already found m. rec. in Camb. Nn. 3, 18–(so Poppo-Stahl, Hude etc.): (2) to refer αὐτοîς to πάντας (Marchant) or to τοὑς ⋯πιμελομένους (Göller, Arnold): but the interest of the hucksters or the aediles in feeding the sailors is not so relevant as that of their commanders who wished them to go into action again; (3) refer αὐτοîρ to ⋯δώδιμα, as abl. instr. with ⋯ριστοποιήσονται (Benedict)!

^{page 169 note a} This category largely coincides with Dyroff's prepositional phrases, präpositionale Ausdruche; for the pronoun is usually governed by a preposition. But this is not always so; see 1, 68², cited below, p. 170.

^{page 169 note b} Hude calls αὑτοû a ‘conjecture’ of Krüger. To designate such modifications of printing ‘conjectures’ is a custom widespread but none the less absurd.

^{page 169 note c} For 4, 86⁴παρέχεται λίμην ⋯ Πόντος 7hellip; ου πολλῷ τεῳ ⋯λ⋯σσω ⋯ ω υ τ ο û properly belongs under A (2).

^{page 170 note a} Not counting αὐτοîς 3, 101²: for by that point I think the subject of the sentence has widened and is rather the Peloponnesians than the Amphissaeans.

^{page 170 note b} For ⋯ξ αὑτ⋯ν Stobaeus is reported to read ⋯αυτ⋯ν which I have found also in several MSS. of Thucydides. It may be right, and ⋯ξ αὑτ⋯ν, which occurs some lines below in 38⁶.

^{page 170 note c} So, according to the MS. text: λ⋯γοντος ὡς α ὐ τ ῷ ⋯π⋯ τ⋯ν Νίσαιαν τῇ ⋯ α υ τ ο û μóνῃ στρατιᾷ οὐκ ἠθ⋯λησαν οἱ Ἀθηναîοι ξυμβαλεîν. If, however, as is probable, some participle such as βοηθήσαντι should be inserted governing τῃ στρατιᾷ, then the second pronoun would fall under class A (1).

^{page 170 note d} I have counted neither 6, 15² (στρατηγ⋯σαί τε ⋯πιθυμ⋯ν κα⋯ ⋯λπίζωυ Σικελίαν δι' α ὐ τ ο û λήψεσθαι), where all editors except Bauer rightly regard αὐτοû as neuter, nor 6, 77² (οἰόμεθα οὐ κα⋯ ⋯ς α ὐ τ ό ν τινα ⋯ξειν τ⋯ δειν⋯ν, πρ⋯ δ⋯ α ὐ τ ο û μ⋯λλον τ⋯ν πάσχοντα δυστυχεîν), where main verb and dependent clause are in different persons.

^{page 171 note a} Referring ⋯κείνου to Alcibiades, with Göller and Poppo-Stahl. If it be referred to Endius, either ‘the Peloponnesians’ or the like must be supplied as subject of ⋯ποστ⋯σαι (so Jowett and Tucker)–which seems to me unnatural–or ⋯ποστ⋯σαι emended with Krüger to ⋯ποστ⋯σαι.

^{page 171 note b} Taking ⋯ς αὐτοὺς with νεωτεριεîν, not with ὐποπτενάντων.

^{page 171 note c} In case any still hanker after ‘MS. authority’ in the matter of these pronouns, to whom therefore my proposals may appear daringly radical, I would mention that in 8, 50² the Vatican MS. B has αὐτῷ, and that in general the readings proposed are often already extant in one or more of the capital MSS.

^{page 172 note a} Not counting 5, 47⁵ ⋯πλα μ⋯ ⋯⋯ν ἔχοντας δίίέναɩ … δɩ⋯ τ⋯ς γ⋯ς σϕετέρας αὐτ⋯ν, which is in a treaty. The heavy reflexion and repeated article is characteristic of treaty style (see above, p. 161); when Thucydides refers to the same treaty-clause in his own person he says simply δɩ⋯ τ⋯ς ⋯αυτ⋯ν (56s). Moreover, Thuc. himself uses σϕέτερος αὐτ⋯ν only in direct reflexion (A 2, pp. 160 f.).

^{page 172 note b} In 4, 29³ (⋯νόμιζε σϕίσɩ μ⋯ν τ⋯ς ⋯κείνων ⋯μαρτίας … οὐκ ἃν ⋯μοίως δ⋯λα εἶναɩ, τοû δ⋯ α ὐ τ ⋯ ν στρατοπέδου καταϕαν⋯ ἄν εἶναɩ πάντα … λανθάνεɩν τε ἄν τò ⋯αυτ⋯ν στρατόπεδον διαϕθεɩρόμενον perhaps the intervening particle weakens an argument from position (cf. I.c. 217^a), but the parallel with the certain reflexive following and the fact that the pronoun stands in contrast (see above, p. 171) all but prove that τοû δ ⋯αυτ⋯ν is the truth. On the other hand Bekker was right in deleting αύτ⋯ν in 7, 48³: ἔϕη οὐ τούς αὐτούς ψηϕɩεȋσθαɩ περì σϕ⋯ν [αὐτ⋯ν]. For there is not there any contrast, and the double form σϕ⋯ν αὐτ⋯ν is highly unlikely in this construction; see pp. 163 f.

^{page 173 note a} Counting 8, 46³: οὐκ είκός (ἔϕη Ἀλκιβιάδης εἶναι, Λακεδαιμονίους ⋯πò μ⋯ν σ ϕ ⋯ ν τ⋯ν Ἑλλήνων έλευθεροûν τοùς Ἒλληνας, ⋯πò δ ⋯κείνων κτ⋯.σϕ⋯ν is probably not a full reflexive, but refers to those among whom the speaker Alcibiades numbers himself, whether Greeks generally or Athenians in particular.

^{page 174 note a} The last sentence of my previous article is therefore found to be untrue.

^{page 174 note b} It may be convenient to list the alterations which have been proposed in the course of this study, so far as they affect the texts of Hude and Jones.

(A)Substitute reflexive for non-reflexive (1) in both Hude and Jones: 1, 501; 1042; 112²; 115⁵; 121⁵; 143²; 2, II⁸; 15²; 16²; 34²; 34²; 81¹; 98¹; 3, 83³; 4, 8⁸; 5, 49⁵ (p. 162); 3, 60 (p. 167): 3, 91² (p. 169): 1, 10³; 50²; 68²; 137⁴; 2,65¹; 4, 27³; 83⁴; 6, 16⁵ (p. 170): 2, 81⁴; 7, 8² (p. 171): 4, 132² (p. 172): 4, 29³ (p. 172, n. a).

(2)In Hude alone: 2, 68³; 6, 78⁴; 7, 67²; 8, 72² (p. 162); 3, 37⁴ (p. 170); 8, 85² (p. 171).

(3)In Jones alone: 3, 65³ (p. 162); 1, 100³ 2, 27¹; 68⁹; 3, 92⁵ (p. 164): 5, 18² (p. 164); 2, 2⁴(p. 172).

(B)Substitute non-reflexive for reflexive (1) both Hude and Jones: 5, 43² (p. 167): 2, 95² (p. 167): 1,132⁵: 8, 85³ (p. 170): 8, 50² (p. 171).

(2)In Hude alone: 8, 47² (p. 169); 8, 47² (P172).

(3)In Jones alone: 8, 76³ (p. 167); 8, 85³(p. 169).

(C)Delete αὐτοȋς after σϕίσιν in Hude, 2, 65¹² (p. 165).