^{page 514 note 1} The authors were these (for the last seven he spared only a cursory glance): Isoc, Dem., Gorg., Antisth., Alcid., Antiph., Andoc, Lys., Isaeus, Demades, Lye, Din., Aeschin., Herod., Thuc, Xen., Theopomp., Polyb., Plut., Dionys., Diod., Jos., App., Arr., Dio Cass., Herodian, Aelian.

^{page 514 note 2} In the circumstances a small bibliography may be of service.

Plutarch: Sintenis, , De hiatu in Plutarchi vitis parallelis (Zerbst, 1845)Google Scholar; Schellens, , De hiatu in Plutarchi moralibus (Bonn, 1864)Google Scholar; Bemardakis, , Plutarchi moralia, i (Teubner, 1888), lxi–lxxGoogle Scholar; Ziegler, , P.–W., ‘Plutar–chos’ (1951), 932–5Google Scholar.

Polybius: Hultsch, , Philol. xiv (1859), 288–319Google Scholar; Brief, , Wie beeinfiufit die Vermeidung des Hiatus den Stil des Polybius? (Hradisch, 1907)Google Scholar.

‘Longinus’ and Onesander: Rohden, H. von, Commentationes in honorem F. Buecheleri et H. Useneri (Bonn, 1873), 68–94Google Scholar.

Aristotle: Blass, , Rh. Mus. xxx (1875), 481–505Google Scholar; Kaibel, Stil und Text der aristotelischen Άθηναίων Πολιτεία (1893), 9–16Google Scholar.

Diodorus: Kaelker, , Leipz. Stud, iii (1880), 303–20Google Scholar.

Galen: Marquardt, , Galeni scripta minora, i (Teubner, 1884), xlvii-lvGoogle Scholar.

Appian: Zerdik, , Quaestiones Appianeae (Kiel, 1886), 49–82Google Scholar.

Polemo: Schmid, , Alticismus, i (1887), 58–60Google Scholar; Jiittner, , Breslauerphil. Abh. viii (1898), 68–75Google Scholar.

Dio, Herodes, Lucian: Schmid, Ibid. 168, 198, 404.

Philo: Jessen, H., De elocutione Philonis Alexandrini (Hamburg 1889Google Scholar); Cumont, , De aeternitate mundi (1891), 20–22Google Scholar; Wendland, , Philos Schrift iiber die Vorsehung (1892), 116–17Google Scholar; Poland, , B.Ph.W. xiv (1894), 1009–11Google Scholar; Arnim, M., De Philonis Byzantii dicendi genere (Greifswald, 1912), 160–4Google Scholar.

Aristides: Schmid, , Atticismus, ii (1889), 248–53Google Scholar.

Aelian: Ibid., iii (1893), 291–6.

Josephus: Schmidt, W., Jahrb. für class. Phil. Supp. 20 (1894), 345–550passimGoogle Scholar; Auerbach, M., Archiwum Towarzystwa Naukowego we Lwowie, dzial 1, tom 1, zeszyt 4 (Lwów, 1924), 37–58Google Scholar.

Philostratus: Schmid, , Atticismus, iv (1896), 469–75.Google Scholar

Plato: Janell, Jahrb. für class. Phil. Supp. 26 (1901). 273–324Google Scholar.

Dionysius: Kallenberg, , Rh. Mus. lxii (1907), 9–32Google Scholar;lxvii (1912). 11–19.

^{page 515 note 1} Jakrb. fur class. Phil, lxxvii (1858), 165Google Scholar.

^{page 515 note 2} C.Q. xxix (1935), 52–7, 96–112Google Scholar.

^{page 515 note 3} When scholars between these dates show awareness of hiatus, acknowledgement is made in the appropriate place.

^{page 515 note 4} When Vilborg on p. 131 of his commentary on Achilles Tatius awakes from his sleep and asserts that ‘the dative gives … an unpleasant hiatus’, he is merely echoing Jackson 105.

^{page 515 note 5} e.g. Benseler ix–x, Schmid i. 58, Kaibel 10 n. 2.

^{page 515 note 6} Dionysius, for instance, diagnoses hiatus in (εύτυχουντα òρών and δέ őλον at Dem. 2. 22 (Den. 43).

^{page 515 note 7} Heibges, S., De clausulis Charitoneis (Halle, 1911), 54–6Google Scholar. Cf. also Jüttner (see p. 514 n. 2) 69 n. 1, Dewing, H. B., ‘Hiatus in the accentual clausulae of Byzantine Greek prose’, A.J.P. xxxii (1911), 188–204Google Scholar. ούδέ εε and ουοε εν (Hel. 5. 21. 4. 4, Longus 2. 16. 1. 5; 2. 19. 2. 1) are presumably not elided.

^{page 515 note 8} Kühner-Blass i, § 53. 5 E.

^{page 515 note 9} Blass, Über die Aussprache des Griechischen ³ (1888), § 17, Schwyzer, p. 195Google Scholar.

^{page 515 note 10} Lucius, A., De crasi et aphaeresi (Strasburg, 1885)Google Scholar; Kühner–BIass i, § 54, § 51, § 52.

^{page 515 note 11} Kühner–Blass i, §48. 3.

^{page 516 note 1} See however p. 528 n. 1.

^{page 516 note 2} Not, of course, before δέ itself but before the group of words it belongs to.

^{page 516 note 3} Heibges rightly infers Chariton' articulation from the clausulae (27–41), and it is reasonable to transfer his conclusions to writers whose rhythm is not so regular.

On pauses cf. also Hultsch (see p. 514 n. 2), 298–304 and Fraenkel, Kolon und Satz, ii with Nachträge (Kleine Beiträge, i. 93–139).

^{page 516 note 4} Cf. Fraenkel (see n. 3), 114.

^{page 516 note 5} Neither of these pauses is mandatory: the point about enclitics and rhythm that is made below in connection with έφη also applies to the vocative (cf. Heibges 40–1, Fraenkel, , Sitzungsberichte der bayerischen Aka-demie, 1965 [2], 71-3)Google Scholar.

^{page 517 note 1} Pauses after adverbial clauses parenthetically placed (e.g. Ach. Tat. 2. 31. 2. 4) should perhaps have been included here rather than in (a).

^{page 517 note 2} Hiatus overdetermined in other ways (e.g. coi dAAd) is ignored unless both classes to which it may belong are small.

^{page 517 note 3} Hel.4. 6.1.3; 5.20.7.5; 10. 9.5. 8,Ach. Tat. 2. 21. 4. 4; 5. 13. 5. 3, Xen. 3. 3. 3. 3; 3. 5. 7. 1, Ninus A V. 12–13.

^{page 517 note 4} Chariton (Heibges 38–9).

^{page 518 note 1} In Bekker, ' edition (Teubner, 1855) Heliodorus occupies 309 1/2 pagesGoogle Scholar; in Hercher, ' Erotici scriptores Graeci (Teubner, 1858–1859)Google Scholar Achilles Tatius occupies 176 1/2, Chariton 154 1/2, Longus 85 1/2, and Xenophon 71.

^{page 518 note 2} Gartner, H., Antike und Abendland, xv (1969), 48Google Scholar, mentions the divergences among the extant witnesses to the text of Chariton and Achilles Tatius and goes on to say ‘wir werden gut daran tun, mit ahnlicher Verwilderung auch bei den Aithiopika … zu rechnen, wo vorlaufig keine entsprechende Kontrolle moglich ist’. The state of the text gives no ground for such nervousness.

^{page 518 note 3} ‘From the opening of the Aethiopica to their ever–receding close, final is treated for all purposes as a homophone of έ’, Jackson 54–5.

^{page 518 note 4} πλήν εĩ τι με διαλέληθενas Dionysius says in a similar context (Comp. 23). The same reservation attaches to all the other figures in this article.

^{page 518 note 5} Cf. the accentuation of.πρòπαλαι

^{page 518 note 6} τήε δ^ ύετεραίαε Jackson, unnecessarily.

^{page 519 note 1} Rattenbury' disregard of hiatus is one of the three main shortcomings of his text, which is on the whole, as Jackson 112 hoped it would be, ‘a genuine and important service to Greek letters’. The others are his reverence towards the ms. C even at its most irresponsible (e.g. 1. 11. 5. 8 , 1. 22. 6. 4 , 5. 12. 1.9 and the infantile conjectures of Lumb' that disfigure the apparatus (e.g. 10. 31. 2. 2 and occasionally the text (5. 12. 3. 3 , 8. 11. 2. 5 ).

While Rattenbury' text is under discussion, attention may as well be drawn to a few unusual lapses of judgement: neither οδεν at 7. 19. 6. 5 nor νòον at 10. 9. 6. n has the remotest chance of being right, and the same is true of his conjectures at 1. 12. 4. 7; J. 21. 7. 8; 8. g. 15. 9, and 9. 7. 1. 4, two of which are nothing short of grotesque. In all these passages except 1. 12. 4. 7 the reading best attested is faultless.

Unnecessary conjectures are also printed at i. 17.3.4 (=5.22. 2. 7); 5. 14. 1. 4 (cf. 1.28.2.7; 1.31.2. 6); 8. 5.10.458. 13. 1.6; 9. 15. 5. 1, and 10. 9. 4. 4 (cf. Xen. Eph. 1. 12. 4. 2).

The wrong variant is chosen at 1. 8. 1. 4; 3 7 5 8; 4 14 1 2 (read 4 17 5 5; 5 24 5 5; 6. 13 3 3; 7 1 4 7; 9. 24. 8. 8 (read ), and 10. 9. 5. 1.

These lists do not purport to be exhaustive.

^{page 519 note 2} This hiatus and others in the list are defensible, as it turns out. The corresponding lists for the other authors likewise contain some defensible conjectures.

^{page 520 note 1} It comes in 8. 9. 3 at the end of the following sentence: . Rare though glosses are in Heliodorus, the relative clause is surely a gloss designed to furnish with an object ( is easily enough supplied, as with three of the other infinitives in the sentence, despite the intervention of ). If the common word for cup, , was in the mind of whoever jotted down the words, he will have given no more thought to the gender of the pronoun than editors have done since.

^{page 521 note 1} The other work attributed to an Achilles (Maass, E., Commentariorum in Aratum reliquiae, 1898, 27–85Google Scholar) is of so different a nature that nothing is proved by its admission of hiatus. On the style and the author' date see Rohde, 471.

Mr. James O'ullivan, who is compiling an index to Achilles Tatius, very kindly read this section of the article and made a number of helpful comments.

^{page 521 note 2} Though Vilborg draws up his text on absurd and incoherent principles (see the last paragraph on p. lxxxv), his reporting of the manuscripts, even if he makes mistakes (Russo, , Gnomon, xxx [1958], 585–6Google Scholar), is so much easier to apprehend than Jacobs' that his edition is indispensable.

^{page 522 note 1} [αύτ] Jackson, unnecessarily.

^{page 522 note 2} The other instance of ώεπερ after a vowel has been put in 5 (c) because no participle accompanies it.

^{page 522 note 3} Vilborg' remarks about Θ on p. xxxi derive from Dorrie, notwithstanding that Müller, B. A. in a valuable review (Phil. Woch. Ivii [1937], 925–7Google Scholar) had shown him to be guilty of misinterpreting Jacobs' introduction.

^{page 523 note 1} άποχρήται owes its rescue from the apparatus to Mr. O'ullivan. The blame for printing άποχρήεθαι attaches not to Cobet,who knew only the reading άποχρήεθαι, but to Vilborg.

^{page 523 note 2} μέν τοί<νυν> O'ullivan.

^{page 524 note 1} Colon, P., inv. 901, published by Henrichs in Zeitschriftfiir Pap. undEpigr. ii (1968), 211–26.Google Scholar

^{page 524 note 2} See Appendix.

^{page 524 note 3} This also appears to be the reading of the new papyrus: .

^{page 525 note 1} Cf. Fraenkel (seep. 516 n. 3), 103–11.

^{page 525 note 2} An unpublished papyrus that Mr. P. J. Parsons kindly allows me to quote reads

^{page 525 note 3} Perry has an interesting note on the manuscript in The Ancient Romances (1967), 344–5Google Scholar, due mainly to Professor Aubrey Diller. Its text of Xenophon is evidently not unique after all, because the British Museum possesses a copy written in the sixteenth century.

^{page 525 note 4} The reference in Lesley, ' Geschichte der griechischen Literatur² (1963), 926Google Scholar to a text and translation by F. Zimmermann (i960) actually leads to an essay on the architectural history of Dresden. In a footnote to a paper on Chariton written in 1959 and published in 1961, Zimmermann expressed the hope that his text would appear in the course of 1962, but the world is still waiting for it.

^{page 526 note 1} But at 1. 14. 7. I read or

^{page 526 note 2} Jackson.

^{page 526 note 3} Jackson.

^{page 528 note 1} This instance apart, Chariton uses ώεπερ after a consonant both in the middle (19) and at the beginning (7) of a sentence, after καί (7), and after τι (1). καθάπερ follows a long vowel (2), a consonant (6. 9. 4. 5, 7. 1. 2. 3), and a short vowel both in the middle (3) and at the beginning (3) of a sentence. This instance is therefore highly suspect (read and so is ), at 7.3.2.2, the more so because does occur (4.4.1.3).

Heliodorus’ only obvious rule in the use of (107) and (38) is not to put after a long vowel. Achilles has after a long vowel in 3 places out of 42 and uses only 3 times. For Longus’ practice see p. 530 n. 1. Iamblichus has (p. 7. 14), (p. 7. 18) (fr. 96) Ninus has Meliochus has

^{page 528 note 2} Cf. Hercher cited on p. 515 above.

^{page 528 note 3} Proc. Camb. Phil. Soc. cxciv (1968), 65Google Scholar.

^{page 530 note 1} This conjecture is beyond question illicit, for Longus is as sensitive as anyone to the difference between and . he uses after a consonant (25), καί (8), δέ (4), and άλλά (2), καθάπεμ after other short vowels (9) and all long vowels (10). The exceptions are 1. 32. 4. 5 would have been perverse when καθάπεμ has just preceded in a balancing phrase), 3. 12. 4. 1 (to avoid an excess of sigmas?), 4. 25. 2. 6 νύκτα ώεπερ (too many short syllables in νύκτα καθάπερ ίκέτηε?).

Norlind, W., ‘Nar levde Longos?’, Eranos, xxiv (1926), 189–91Google Scholar, appears to have been too sweeping and in part misguided on this point (p. 190); indeed, his remarks about hiatus in Longus are altogether too superficial.

^{page 530 note 2} e.g. 1.15. 1. 2

It is worth mentioning here that καιτοι stands before a consonant (3), καίτοιγε before a vowel (4). Incidentally, καίτοιγε should not be printed as two words because Longus does not use γε, a point noticed as long ago as 1834 by Struve, De exitu versuum in Nonni Panopolitani carminibus, n. 9 (not quite accurate).

^{page 531 note 1} He is called by Suidas, just as the Xenophon who wrote . is called (Suidas’ third novelist of the name, who wrote , presumably did come from Antioch).

Among the Xenophons listed at Diog. 2. 59 the fifth is someone μυθώδη τερατείανπεπραγματευμένοε. Menagius in his commentary (London, 1664) asserted that this Xenophon was the θαυματοποιòε.6c whose apprentice Cratisthenes of Phlius was able to conjure up fire and perform other bewildering tricks (Athenaeus 19 e; neither gentleman is registered in P.–W.). Wickert in P.–W. ‘Xenophon’ 2089 (1967) suggests no identification but assumes that he was a ‘mythologischer Schriftsteller’, an assumption evidently shared by Jacoby, who gave him a place in the first volume of F.G.H. (no. 24). Menagius’ view has little to commend it, Wickert' nothing; for πεπραγματευμένοε denotes authorship, and μυθώδηε is no epithet to use of a mythological compilation.

One of the works that Photius read with his friends was to τά ύπέρ Θούλην ãπιετα by Antonius Diogenes. Here are a few phrases from his summary of it (Bibliotheca, no. 166):

No one has ever called Antonius Diogenes a mythological writer or taken him for a conjuror: he is usually ranked with the novelists.

It may be inferred, then, that the Xenophon in question was a novelist of some sort; but unless all fiction was alike to Diogenes (or to Demetrius of Magnesia, if Diogenes derives the notice from him), it is improbable that the μυθώδηε τερατεία was the extant Ephesiaca, because Ephesiaca bears a much closer resemblance to the novels of Heliodorus, Achilles Tatius, and Iamblichus, than to Antonius Diogenes’ fabrications, and nowhere in his summaries of those novels (Bibliotheca, nos. 73, 87, 94) does Photius employ similar language. If in spite of this Diogenes’ novelist is one of the three in Suidas, he could be any of them.

After this note was written it turned out that Rohde had anticipated it: ‘man könnte… unter der μυθώδηε τερατεία eine, wie es dem Demetrius scheinen mochte, schamlos erlogene (und doch als wahr erzählte) abenteuerliche Geschichte verstehen, einen Roman, nach unserer Ausdrucksweise’ (p. 346 n. 1). Nevertheless, confirmation from Photius seemed welcome, and when Pauly gives currency to error the truth can bear reiteration.

^{page 532 note 1} Rohde 401, Bürger, K., Hermes, xxvii (1892), 36–67Google Scholar. Admittedly there are people who remain unconvinced (e.g. Rattenbury, , Gnomon, xxii [1950], 75Google Scholar); but they ought to demolish Büger' arguments.

It has also been maintained that the novel had been worked over for religious purposes before it fell into the epitomator' hands: Kerenyi, , Die griechisch–orientalische Roman–literatur in religionsgeschichtlicher Beleuchtung (Tübingen, 1927), p. 232 n. 11Google Scholar; against, Zimmermann, F., Würzburger Jahrbücher, iv (1949–1950), 252–86Google Scholar; in support of Kerenyi, but ignorant of Zimmermann, Merkelbach, , Roman und Mysierium (Munich, 1962), pp. 91–113Google Scholar; against Merkelbach, Gärtner, P.–W. ‘Xenophon von Ephesos’, 2072–80.

^{page 532 note 2} Cf. Zimmermann 253 ‘daß manaus dem ungleichmäßig erhaltenen Zustand unseres Textes von Sprache und Stil des eigentlichen Verfassers nur ein äußerst unvollkommenes Bild gewinnen kann, liegt auf der Hand’. Castiglioni was not so despondent: ‘Seno–fonte non esclude compiutamente l'iato, ma cerca, quanto gli e possibile, di evitarlo’ (Boll. Fil. Class, xxix [1922–1923], 205Google Scholar). Gärtner talks even less questioningly about Xenophon' ‘Gleichgültigkeit gegenüber selbst schweren Hiaten’.

^{page 532 note 3} μειρακίου is superfluous and should perhaps be deleted.

^{page 532 note 4} The others are őτι 4 times, δή twice, πρò, που, and τινι, once each, one on either side of έφη at 3. 2. 1. 1, and one before a pause at 3. 3. 5. 3.

^{page 532 note 5} The clausulae marked are and the more obvious of their resolutions. When resolution obscures the shape of a clausula, commitment may seem arbitrary, but there is often something to go by, e.g. the accent or the way the words divide. If there is a choice of acceptable clausulae, Heibges prefers the one ‘quae maximum praebeat ambitum’ (16).

The prosody of phrases like is too doubtful for account to be taken of them in the earlier stages of a rhythmical analysis.

^{page 532 note 6} The parentage of this clausula is immaterial.

^{page 533 note 1} In Attic the prosody is but Heibges has made it probable that in Chariton the lighter combinations of mute and liquid are ambivalent (57–9). This is the justification for arbitrary procedure below, e.g. but

^{page 533 note 2} The heroine is either or (1. 12. 2. 7); the former produces hiatus in a number of places, the latter inferior clausulae. Klaffenbach prefers (Zimmermann, p. 265 n. 3).

^{page 533 note 3} Hemsterhuys: F.

^{page 533 note 4} Jackson, p. 96.

^{page 533 note 5} 5 This should be the order at 1. 10. 10. 7 as well, instead of

^{page 533 note 6} F, edd. omnes; fort, etiam ἴccuc delendum.

^{page 534 note 1} Habrich, has taken over Hinck' collations (Polemonis declamationes [Teubner, 1873], viii–ix, 45–57)Google Scholar as though no further work had been done on the manuscripts; but a much fuller account of them, albeit hard to follow, was given by Naechster, M., De Pollucis et Phrynichi controversiis (Leipzig, 1908), 47–57Google Scholar. Properly appraised, so Naechster argued, they offer no support at all for the ascription of fr. 101 Habrich to Iam-blichus. Naechster also discussed at length (57–9), but too confidently, Iamblichus’ debt to Xenophon (not the novelist) in fr. 1 Habrich.

^{page 534 note 2} A structural fault lies bare on the surface at this point, and no one but Rohde appears to have been disconcerted by it. The piece falls into the following sections: p. 5. 3 (the king' chariot and apparel are described) … , p. 7. 1 … , p. 7. 2 (and more about their armour) … , p. 7. 10 (and the horses are described). The absurdity of was not lost on Rohde, who proposed There are two objections to this: (1) why are the summarily dismissed? (2) who are the people riding on Nisaean horses? A better solution would be to delete or at least obelize it. The one objection to this is that on a literal interpretation only those of the who are on foot would have their armour described. This anomaly cannot be irond out, however, unless were to mean ‘others’ of the ; and what would they be doing on horseback? Read therefore, until someone has a brighter idea,

^{page 535 note 1} Heibges' dissertation does not confirm a similar qualification added by Wilamowitz, in the 3rd edition of Die griechische Literatur des Altertums (1912Google Scholar) to his remarks about Chariton' rhythm: ‘so finden wir die Rhythmen… besonders stark bei Chariton wenn er direkte Reden einfuhrt’ 226, ‘…in den Reden noch ganz mit den hellenistischen Rhythmen verziert’ 258.

Achilles Tatius, however, would repay study in this respect: see for instance 3. 10. 2–6, 6. 21. 2–3, 7. 7. 2–6.

^{page 535 note 2} On this see Heibges, 92–3.

^{page 535 note 3} Dr. Stephanie West has very kindly pointed out that this piece may well come from a novel. Incidentally, it is quite strongly rhythmical.

^{page 535 note 4} The enormous labour that Zimmermann expended on annotation may appear to confer on his supplements more than the usual authority. In fact it was almost all wasted, partly because too little survives of the texts, partly because his command of Greek and his feeling for congruity are equally unsure. The following passages will illustrate all three of these criticisms: no. 1 AI 8 no. 9. 28 Zimmermann' Muse is nodding …έκ

If Zimmermann had confined these fantasies to the apparatus, they would have been less of a nuisance. As it is, not only the text but the index too is full of them.

Zimmermann prided himself on being the first to offer ‘eine Rezension im eigentlichen Sinn’ (Phil. Woch. liv ]1931], 193Google Scholar) of Lavagnini, ' Eroticorum fragmenta papyracea (Teubner, 1922Google Scholar). What ‘eine Rezension im eigenltichen Sinn’ would have made of his own compilation is one of the more tantalizing secrets that history was never permitted to disclose. Miiller, B. A.' review in Phil. Woch. lviii [1938], 561–8Google Scholar was charitable to the point of sycophancy: ‘in mustergültiger Weise herausgegeben’, 563; ‘die Text-gestaltung dieses Bandes, in deren Dienst eine ausgezeichnete sprachliche Schulung von einer heute nicht äaufigen Abgeschlos-senheit steht, ist nach jeder Richtung sehr vorziiglich’, 565; ‘überall fühlt man hier in Kenntnis, Erkenntnis und Behandlung des Sprachlichen ein hohes Mass sprachlicher Meisterschaft’, 566.

^{page 536 note 1} Cf. Vitelli, , ‘L'iato nel romanzo di Nino’, S.I.F.C. ii (1894), 297–8Google Scholar; Schmid iv. 471 (deficient).

^{page 536 note 2} As the same name occurs in the new fragments of Lollianus’ Phoenicica (Hen–richs, A., Die Phoinikika des Lollianos, Cologne, 1970Google Scholar), and their style is equally unpretentious, 2620 doubtless belongs to the same work.

^{page 537 note 1} Zimmermann must be commended for taking the trouble to state in his introduction to each piece whether or not hiatus is avoided in it; but unfortunately he does not seem to know what hiatus is. At any rate, ‘Hiat nicht vermieden’ does not apply to this piece on the evidence available.

^{page 537 note 2} 'Hiat offenbar nicht vermieden Zimmermann; but see n. i above.

^{page 537 note 3} See now West, S. R., ‘The Greek version of the legend of Tefnut’, J.E.A. lv (1969), 161–83Google Scholar.

^{page 538 note 1} References where none are given can be found on pp. x–xiii of Vilborg' edition.

^{page 539 note 1} The content of the lacuna suspected by Jacobs before 3. 3 is supplied by 1. 5. 3 των ãμα τήν θέαν, as Russo 400 points out; and the connection between Dionysus των ãμα τήν θέαν in the digression and Dionysus in 3. 3 (Dörrie 88, Russo 401) is dispensable.

^{page 539 note 2} Jacobs, to whom they appeal, put Leucippe' name in his supplement.