^{page 204 note 1} De Aristotelis dicendi ratione (on particles) and Ueber den Sprachgebrauch des Aristoteles (on prepositions) Berlin, 1868Google Scholar; E.'s work deals with all the prepositions and is valuable, though in some respects incomplete and open to criticism

^{page 204 note 2} This would not apply to the lost dialogues, which were apparently of a more literary nature, and in the surviving fragments of which there are e.g. no instances of the later uses of prepositions illustrated below; similarly the more careful attention to style and diction in the Ath. Pol. is shown by Kaibel, , Stil und Text der Ath. Pol., esp. pp. 81 ff. and 102 ff.Google Scholar; see also pp. 37 ff. where K. gives instances of deliberat ‘atticizing’ from the Ath. Pol., and remarks ‘die Politeia aber will attiseh sein und ist es auch, nicht des Stoffes oder der Athener wegen sondern weil sie ein Litteraturbuch ist’; cf. Diels, ' review in Gött. Gel. Anzeig. 1894, pp. 293–307Google Scholar. Even in other works of A. there may be some conscious effort to conform to Attic standards.

^{page 204 note 3} E.g. Thumb, , Die griechische Sprache im Zeitalter des Hellenismus, p. 205Google Scholar; Wilamowitz, , Geschichte der griech. Sprache, p. 48Google Scholar; Kaibel, , op. cit. p. 37Google Scholar.

^{page 204 note 4} See Powelll, and. Baber, New Chapters in Greek Literature, Series II, pp. 114–116Google Scholar, where about 340 B.C. is suggested as a likely date.

^{page 204 note 5} See Jannaris, , Hist. Gr. Grammar §§ 1491–3Google ScholarRadermacher, , Neutestamentliche Grammatih, pp. 107–112, 115Google Scholar; Blass-Debrunner, Neutest. Gramm. (ed. 6) § 203Google Scholar; Moulton, , Proleg. to N. T. Grammar, pp. 106, 7Google Scholar; Robertson, , Hist. Grammar of the N.T., p. 451(d)Google Scholar; Wackernagel, , Vorlesungen über Syntax II, pp. 207–8Google Scholar; Thumb, , Handbook of Modern Greek Vernacular (English trans, of ed. 2) § 158Google Scholar.

^{page 205 note 1} Thumb, , op. cit. §§ 161, 162, 164Google Scholar.

^{page 205 note 2} Radermacher, , op. cit. pp. 108, 116Google Scholar; Blass-Debrunner, , op. cit. § 185Google Scholar; Krebs, , Die Präpp. bei Polybius, pp. 144 ffGoogle Scholar.

^{page 205 note 3} It is to be regretted that there is no book dealing with Greek prepositions on the scale of Goodwin's Greek Moods and Tenses or Dennis-ton's Greek Particles. In the absence of any such work (apart from the valuable but naturally not exhaustive sections in Kühner-Gerth) I have mainly consulted and given some references (by author's name only) to the following special dissertations: Helbing, Die Präpp. bei Herodot und andern Historikern; Sobolewski, De Praep. usu Aristophaneo; Lutz, Die Prpp: bei den en Rednern; Lina. De praep. usu Platonico quatit. selectae; Eucken, Ueber den Sprachgebrauch Aristoteles; Hagfors, De praep. in Arist. Pol. et Ath. Pol.usu; Regard, Contrib. à I'étude des prép dans le N.T.; Johannessohn, Der Gebrauch der Päpp. in der Septuaginta; Krebs, Die Päpp. bei Polybius; Rossberg, De praep. Grace. in chartis Aegyptiis Ptol. aetat. usu; Kuhrung, De praep. Grace. in chartis Aegyptiis quaest. selectae.

^{page 205 note 4} Cf. a rare use of sub c. acc. quicquid sub et Borean hominum sumus, Luc. VII 364.

^{page 205 note 5} Also in a few references in the tragedians to gods of the underworld, e.g. , Aesch.Eum. 952Google Scholar ảθανάτοɩς τοȋς θ' ύπò ϒαίαν, Eur. Hec. 144, Alc. 899.

^{page 205 note 6} In K.-G. (I 525) this passage is included under a category ‘Annäherung,’ with Soph. El. 720 κεȋνος δ ὑπ αὐτήν ⋯σχάτην στήλην ἒΧων ‘unmittelbar unter der Saule hin.’ Here motion is implied, but not in Isocr. (examples of prep, c. acc. after κεˆɩσθαɩ in pregnant sense, as pass, of τίθημɩ, do not offer a close parallel to the present passage). In sense (a) above the idea of coming near to, as well as under something higher is present; so perhaps here ‘nearness’ alone may be expressed; in so far as the idea of extension is present this use is also akin to ('c) des above. Possibly the sense ‘under’ (already present in the verb) may be due to the idea that. Euboea lies close to and under the control of Athens; cf. a few lines later κρατοȗντες (Euboea) μâλλον ᾓ της ήμετέρας αὐτων. Examples of ὑπ⋯ c. acc. denoting subjection in classical writers are given below; cf. an inscription of the Christian era from Selymbria (A.E.M. VIII, 1884, p. 212 no. 29) where ὑπ⋯ Νακολίαν means ‘in the territorium of.’

^{page 206 note 1} In , Xen.Symp. VIGoogle Scholar 4 the MS reading is ὑπ⋯ τ⋯ν αὐλ⋯ν δɩαλέϒωμαɩ, and it seems unnecessary to accept (with Marchant in the O.T.) Cobet's correction του αὐλου.

^{page 206 note 2} In Phaedrus 268A Εωμεν δή τά ϒε σμɩκρά. ταντα δέπ ανϒας μαλλον ίωμεν motion (metaphorical) is implied, ‘holding up to the light.’ Cf. Eur. Hec. 1154 νπαὐϒας τούσδε λεύσσουσαɩπέπλους.

^{page 206 note 3} In Il. III 371–2 αϒΧε δέ μɩν πολύκεστος ίμ⋯ς ⋯παλήν ύπ⋯ δεɩρήν δς οί ύπ ⋯νθερεωνος ⋯χεύς τέτατο τρυϕαλείης, there seems to be no real distinction between acc. and gen. with ύπъ. In Hdt. cf. VI 137 Χώρην τήν ύπο τον ϒμησσον εονσαν and, in the same chapter, κατοɩκονμένους τούς πελασϒούς ύπο εɩναɩ θεω the ace. is used to denote position under, though the idea of extension is hardly present. It is a familiar fact that some instances of the characteristic usages of post-classical Greek can frequently be found in earlier Greek, and that poetic and Ionic characteristics sometimes disappear in literary Attic prose and appear again in the Koine. The use of ὑπъ C. ace. offers some slight confirmation of this; of. the use of πρɩν 1) in Ionic and again in Aristotle and the Koine. But on the whole the point is perhaps better illustrated by vocabulary than by syntax.

^{page 206 note 4} Cf. Blass-Debrunner § 232 ‘das alte lokale ύπъ τɩνος und ύπό τɩνɩ ist in ύπь τɩ aufgegangen.’

^{page 206 note 5} Dittenberger, , Sylloge II 826E₃₀Google Scholar.

^{page 206 note 6} Ibid. I 495₁₃₀. Contrast Dem. LI 22 ύπ⋯ κήρυκος πωλου σɩ.

^{page 206 note 7} From Lutz, Lina and Krebs. The quency of the gen. throughout is of course mainly due to its use to denote the agent; in Polyb. it is only used in a local sense in the phrase ύπ⋯ ϒ⋯ς (Krebs, p. 47).

^{page 207 note 1} Cf. Monro, , H.G. § 218Google Scholar; Kühner-Gerth II 1, p. 488.

^{page 207 note 2} Stein and Bekker emend to Μέμϕɩɹος πόλεως, the gen. being the normal usage, and perhaps with special reference to II 12 ὃρος τουτο τ⋯ ὐπ⋯ρ Μέμϕɩος ἔχον.

^{page 207 note 3} , Pind., Pyth. II. 80Google Scholar has ⋯βάπϒɩστ⋯ς είμν ϕελλ⋯ς ὢςπ⋯ρ ὑ ἒρκος ἃλμας.

^{page 207 note 4} Cf. with Od. VII 135 (above) τ⋯ν δ ὑπ⋯ρ Od. XVII 575; no real distinction can have been felt.

^{page 208 note 1} Krebs, p. 40.

^{page 208 note 2} Blass-Debrunner § 230.

^{page 208 note 3} Collitz-Bechtel, 4110.

^{page 208 note 4} Krebs, pp. 40, 42.

^{page 208 note 5} J.H.S. XXII, p. 369, No. 143A.

^{page 208 note 6} Homer, , Il. VI 524Google Scholar ὑπ⋯ρ σέθεν αἲσχε ακούω is doubtful, since some idea of interest may be involved; so Monro, (H.G. 189)Google Scholar ‘on your account.’

^{page 208 note 7} Kühner-Gerth (I § 435e), and still more L. and S., seem to underestimate the extent of this use of ὐπ⋯ρ in the Orators. On the other hand of the passages cited by , L. and , S., in Hdt. II 123Google Scholar Hude (with better MSS) reads ύπЬ, and in Soph, . O.T. 1444Google Scholar the meaning is ‘on behalf of.’ There is also a certain amount of MS variation between ύπέρ and περɩ; cf. Wyse on Isaeus III 55. 8.

^{page 208 note 8} This use of ὑπ⋯ρ is also fairly common in some of the Atticists (Schmid, III 290, IV 466, 630); it was not perhaps regarded as un-Attic, though it does not occur in Attic inscriptions till after 300 B.C. (Meisterhans, 222).

^{page 208 note 9} E.g. in business letters with the sense ‘with reference to’; see Robertson, , The use of ὐπέρ in business documents in the Papyri. Expositor Ser. VIII 19, 1920, 321–7Google Scholar.

^{page 209 note 1} In the classical period περɩ c. gen. is of course used in certain contexts in much the same sense as ὑπέρ, e.g. in μάχεσθμɩ περɩ; in Greek περɩ is more widely used in this sense (often with a v.l. ὑπέρ). e.g. in the T., N.Matt. xxvi. 28Google Scholar αɩμά μον…τ⋯ περɩ πολλων ⋯κυνόμενον[D ύπέρ]; cf B.-D. § 229 n. 1.

^{page 209 note 2} See Grégoire, H., Recueil des inscriptions grecques chrétiennes d'Asie Mineure, Fasc. 1, p. 4 No. 2Google Scholar; cf. No. 2; cf. Nos. 15, 40, and Nachmanson, E., Syntahtische Inschriftenstudien, Eranos IX, p. 74Google Scholar. The existence of this form does not seem to be noticed in any grammar or lexicon. Both ὑπ⋯ρ εὑΧης and περί ενχης are common at the beginning of ex voto inscriptions.

^{page 209 note 3} At any rate in prose; in Aeschylus, Ag. 448, 1453 δɩαί c. gen. may have a causal sense.

^{page 209 note 4} The cause seems to be thought of as an instrument; cf. Aristotle, , De Part. An. 641b 16Google Scholar τον ούρανον ϒεϒενησθμɩ ύπο τοɩαύτης τοɩαύτης αίτɩας…καɩ εϊναɩδɩά τοɩαύτην αίϒίαν.

^{page 210 note 1} In the classical period ςɩά c. gen. was sometimes used in practically the same sense as an instrumental dative; this use becomes more common in Aristotle (Eucken p. 18), and still more common in the post-classical period; cf. Humbert, , La Disparition du Datif en Grec, pp. 119–124Google Scholar.

^{page 210 note 2} Cf. [, Xen.] Cyneg. 13Google Scholar. 13 βελτίους ϒίϒνονταɩ καί σοϕώτеροɩ δɩʼ οὒ δɩδάζομεν ‘they become and wiser, and we will give the reason’. Authorship and date are, however, disputed (Christ-Schmidd, I, p. 516.)

^{page 210 note 3} With this causal use of δɩά c. gen. cf. the use of per in such expressions as per timorem, with pronouns, e.g. per hoc ne… caperentur (Livy) and, in later Latin, more freely with nouns, e.g. Amm. 14. 7. 5 per multas causas; see Stolz-Schmalz, , Lat.-Grammatik⁵, p. 521Google Scholar, and cf. Glotta IV, pp. 278–9.

^{page 210 note 4} Krebs, p. 68. The Atticists are careful observe the distinction (Schmid, , Atticismus IV 627)Google Scholar.

^{page 211 note 1} Cf. Propertius IV 3. 20 et struxit querulas rauca per ossa tubas. This use is particularly common in technical works on medicine and cookery, to denote the ingredients of prescrip- tions etc.; numerous examples are given in the Thesaurus of Stephanus s.v. δɩά

^{page 211 note 2} From Lutz, Lina, Regard and Krebs.

^{page 211 note 3} So also in such phrases in the Orators as δίκη προς τον πολέμαρχον ‘motion to’ is apparently implied.

^{page 212 note 1} See Rudberg, G., Ad usum circumscribenttm praep. Graecarum adnotationes, Eranos XIX, pp. 173 ff.Google Scholar, where the use of irepl and jcord in periphrasis in post-classical Greek is very fully illustrated and a number of examples from Plato are given, but no distinction is made between easier and more extreme forms of periphrasis. For similar uses of ⋯π⋯, ⋯κ and παρά see K.-G. I, p. 336, Anmerk. 8.

^{page 212 note 2} περί c. gen. is rarely so used. England (on Laws 685c 2) makes the distinction that περί c. gen. represents an objective, περί c. ace. (generally) a possessive gen., but there are so many exceptions to this that the distinction does not seem worth making.

^{page 212 note 3} England (in the Index to his edition s.v. περί) refers to sixteen examples of περί c. acc. so used and nine of περί c.gen., but some of these, esp. those of the gen., may be otherwise explained.

^{page 212 note 4} Eg. in το⋯σ περι δυσιαν λ⋯λουσ, possibly (as was suggested by Mr. J. D. Denniston) the λòγοι are personified as a sort of cortege of L.; the personification of λ⋯λοσ is of course characteristic of Plato, eg.⋯λαζ⋯νεσ λ⋯γοι Phaedo 92D, Rep. 560c, and for a strong instance see Rep. 503A. So possibly Laws 775B 4 τὦν περι τ⋯σ Mο⋯σασ ν⋯μων, where euphony may have some influence; in Laws 668B 10 τ⋯ περ⋯ αὐτ⋯ν (τ⋯νμουσικ⋯ν) ποι⋯ματα ‘its productions’ there seems to be no special reason, except perhaps considerations of rhythm or avoidance of hiatus; cf. Wyse, on Isaeus I 36. 1Google Scholar.

^{page 213 note 1} Here, however, the relationship may be thought of as local; cf. Rhet. 1360b 25 τ⋯ περ⋯ ψυχ⋯ν (⋯γαθ⋯) και τ⋯ ⋯ν σὡματι, E.E. 1218b 33 γ⋯ ⋯ν τῇ ψυχῇ ⋯γαθ⋯. In the geographical sense περ⋯ c. acc. is often used by A. in the same context and the same sense as ⋯ν c. dat.; cf. Pol. 1315b δευτ⋯ρα δ⋯ περι κóρινθον ⋯ τ⋯ν K. {τυραννισ) with 1310b 26 πειδων μ⋯ν περι” αργοσ … και, and again, ibid. 1341a 33 και γαρ εν αρ but in ps.-Ar. Mirab. Auscult. 838a 5 εν τη κυμη τη περι την the acc. with περι replaces the usual gen.

^{page 213 note 2} Here also εν (or επι c. gen.) might be used to denote pleasure residing in the exercise of a faculty; cf. 1175b 30 αι εν αυαισ. περι and κατ⋯ are in fact used in A. to express a very wide range of relationships between nouns, and in many cases either could be used (or replaced by a gen. or another prep.) without alteration of meaning; cf. E.N. 1175b και γαρ εν αρ ενεργειαν, where it would be possible to make a distinction in translation, but the variety is in the expression, not the thought.

^{page 213 note 3} Cf. Phys. 189b 32 τα περι εκαστον ιοια and De An. 421b 19 το … μη αισθανεσθαι ιδιον, instead of the usual gen. with ιδιοσ, and on this pleonastic use of επι see Vahlen, , Philologische Schriften, I 334–6Google Scholar.

^{page 213 note 4} Cf. Xen. Hell. V 4. 2 την περι, possibly influenced by αισθανεσθαι ‘ the conduct of A.’4, in the previous line.

^{page 213 note 5} I have not attempted in all cases to distin- guish between easier and more extreme forms of periphrasis; they all in varying degrees illustrate the same tendency. The total number of examples would not be great in comparison with the bulk of A. is mistaken in stating that this usage is rare in A.; he quotes only four exx. from genuine works of a prepositional phrase equivalent to a gen., and adds that there are several passages where such phrases are equivalent to an adjective. (Similarly from the Orators R. cites only Dem. II 27, apparently regarding it as a unique example).

^{page 213 note 6} Except XVIII αισθανεσθαι… (see below on the phrase 01 irepl nva). XXVIII 1 αισθανεσθαι ‘ things went well enough with the state ’ is not purely periphras- tic; so in the other three examples, which are instances of the common formula for summing up, e.g. Ill 5 rb. iitv oiv [irepl} ris apxfo TOVTOP etxc TOV Tpbicov.

^{page 214 note 1} Kuhring, pp. 12–15; Rossberg, pp. 11 ff., 34 ff.

^{page 214 note 2} Further examples are given in Krebs, p. 104.

^{page 214 note 3} Robertson (Historical Grammar of N.T., p. 620) compares the vernacular ‘ you all’ for a single person, in the Southern States of America.

^{page 214 note 4} See Schweighaiiser's Index s.v. irepl, and Krebs, pp. 5 and 113.

^{page 214 note 5} 0! irepl TWO. is also used to denote two friends or lovers, e.g. ol ir. ‘αισθανεσθαι,’ A. and Melanio*. Other examples are given in Kiihner, Ice. cit.

^{page 214 note 6} The more familiar use is of course also found in Arist., and in Pol. 1314b 23 αισθανεσθαι (αισθανεσθαι… /αισθανεσθαι}αισθανεσθαι, the man himself is clearly contrasted with ol irepl airbv.

^{page 215 note 1} See Blass-Debrunner§§ 203, 215,216; Rader- macber, pp. 26, 55; Monlton, Prolegomena to N.T., p. 159; Wackernagel, of. cit., II, p. 158; Robertson, Hist. Gr. of N.T., pp. 636–7; Jannaris §§ 1491–2; Hatzidakis, Einleitung in in neugriech. Grammatik, p. 213; Thumb, Handbk. § 158; Schmid, Atticismus IV, 625.

^{page 216 note 1} On the language of Hyperides see U. Pohle, Die Sprache des Redners Hypireides in ihren Bezit- hungen air Koine, and D. Gromska, De Sermone Hyperidis.

^{page 216 note 2} See Stein on Hdt. I 67, and cf. Indogerm. Forseh. XXXI, p. 448 ff. lus 0$ is found in an inscription from Olbia, 230 B.C. TOO uipovt las oS i ffpwt 0 2Wias {Ditt. I 496), in the N.T. ace. to some MSS (Blass-Debrunner § 383), and the LXX (Kings II 23. 10). In later Greek it becomes more common, and survives into mod. Greek in the contracted forms (Jannaris § 1785, Thumb § 275).

^{page 217 note 1} Blass-Debrunner § 216; Johannessohn, pp. 304–5; Krebs, p. 19. I wish to thank Mr. J. D. Denniston and Prof. A. Cameron for advice and criticism on several points.