§81. Nescio qui: Goer. is quite wrong in saying that nescio quis implies contempt, while nescio qui does not, cf. Div. in qu. Caec. 47, where nescio qui would contradict his rule. It is as difficult to define the uses of the two expressions as to define those of aliquis and aliqui, on which see 61 n. In Paradoxa 12 the best MSS. have si qui and si quis almost in the same line with identically the same meaning Dav. quotes Solinus and Plin. N.H. VII. 21, to show that the man mentioned here was called Strabo—a misnomer surely. Octingenta: so the best MSS., not octoginta, which however agrees better with Pliny. Quod abesset: "whatever might be 1800 stadia distant," aberat would have implied that Cic. had some particular thing in mind, cf. Madv. Gram. 364, obs. 1. Acrius: οξυτερον, Lamb. without need read acutius as Goer. did in 69. Illos pisces: so some MSS., but the best have ullos, whence Klotz conj. multos, Orelli multos illos, omitting pisces. For the allusion to the fish, cf. Acad. Post. fragm. 13. Videntur: n. on 25. Amplius: cf. 19 non video cur quaerat amplius. Desideramus: Halm, failing to understand the passage, follows Christ in reading desiderant (i.e. pisces). To paraphrase the sense is this "But say my opponents, the Stoics and Antiocheans, we desire no better senses than we have." Well you are like the mole, which does not yearn for the light because it does not know what light is. Of course all the ancients thought the mole blind. A glance will show the insipidity of the sense given by Halm's reading. Quererer cum deo: would enter into an altercation with the god. The phrase, like λοιδορεσθαι τινι as opposed to λοιδορειν τινα implies mutual recrimination, cf. Pro Deiotaro 9 querellae cum Deiotaro. The reading tam quererer for the tamen quaereretur of the MSS. is due to Manut. Navem: Sextus often uses the same illustration, as in P.H. I. 107, A.M. VII. 414. Non tu verum testem, etc.: cf. 105. For the om. of te before habere, which has strangely troubled edd. and induced them to alter the text, see n. on I. 6.

§82. Quid ego: Bait. has sed quid after Ernesti. Nave: so the best MSS., not navi, cf. Madv. Gram. 42. Duodeviginti: so in 128. Goer. and Roeper qu. by Halm wished to read duodetriginta. The reff. of Goer. at least do not prove his point that the ancients commonly estimated the sun at 28 times the size of the earth. Quasi pedalis: cf. D.F. I. 20 pedalis fortasse. For quasi = circiter cf. note on 74. Madv. on D.F. I. 20 quotes Diog. Laert. X. 91, who preserves the very words of Epicurus, in which however no mention of a foot occurs, also Lucr. V. 590, who copies Epicurus, and Seneca Quaest. Nat. I. 3, 10 (solem sapientes viri pedalem esse contenderunt). Madv. points out from Plut. De Plac. Phil. II. 21, p. 890 E, that Heraclitus asserted the sun to be a foot wide, he does not however quote Stob. Phys. I. 24, 1 ‛ηλιον μεγεθος εχειν ευρος ποδος ανθρωπειου, which is affirmed to be the opinion of Heraclitus and Hecataeus. Ne maiorem quidem: so the MSS., but Goer. and Orelli read nec for ne, incurring the reprehension of Madv. D.F. p. 814, ed 2. Nihil aut non multum: so in D.F. V. 59, the correction of Orelli, therefore, aut non multum mentiantur aut nihil, is rash. Semel: see 79. Qui ne nunc quidem: sc. mentiri sensus putat. Halm prints quin, and is followed by Baiter, neither has observed that quin ne ... quidem is bad Latin (see M.D.F. V. 56). Nor can quin ne go together even without quidem, cf. Krebs and Allgayer, Antibarbarus ed. 4 on quin.

§83. In parvo lis sit: Durand's em. for the in parvulis sitis of the MSS., which Goer. alone defends. Quattuor capita: these were given in 40 by Lucullus, cf. also 77. Epicurus: as above in 19, 79 etc.

§84. Geminum: cf. 56. Nota: cf. 58 and the speech of Lucullus passim. Ne sit ... potest: cf. 80 quasi quaeratur quid sit, non quid videatur. Si ipse erit for ipse apparently = is ipse cf. M.D.F. II. 93.

§85. Quod non est: = qu. n. e. id quod esse videtur. Sui generis: cf. 50, 54, 56. Nullum esse pilum, etc.: a strong expression of this belief is found in Seneca Ep.. 113, 13, qu. R. and P. 380. Note the word Stoicum; Lucullus is of course not Stoic, but Antiochean. Nihil interest: the same opinion is expressed in 40, where see my note. Visa res: Halm writes res a re, it is not necessary, however, either in Gk. or Lat. to express both of two related things when a word is inserted like differat here, which shows that they are related. Cf. the elliptic constructions in Gk. with ‛ομοιον, μεταξυ, μεσος, and such words. Eodem caelo atque: a difficult passage. MSS. have aqua, an error easy, as Halm notes, to a scribe who understood caelum to be the heaven, and not γλυφειον, a graving tool. Faber and other old edd. defend the MSS. reading, adducing passages to show that sky and water were important in the making of statues. For aqua Orelli conj. acu = schraffirnadel, C.F. Hermann caelatura, which does not seem to be a Ciceronian word. Halm's aeque introduces a construction with ceteris omnibus which is not only not Ciceronian, but not Latin at all. I read atque, taking ceteris omnibus to be the abl. neut. "all the other implements." Formerly I conj. ascra, or atque in, which last leading would make omnibus = om. statuis. Alexandros: Lysippus alone was privileged to make statues of Alexander, as Apelles alone was allowed to paint the conqueror, cf. Ad Fam. V. 12, 7.

§86. Anulo: cf. 54. Aliqui: n. on 61. Gallinarium: cf. 57. Adhibes artem: cf. 20 adhibita arte. Pictor ... tibicen: so in 20. Simul inflavit: note simul for simul atque, cf. T.D. IV. 12. Nostri quidem: i.e. Romani. Admodum: i.e. adm. pauci cf. De Leg. III. 32 pauci enim atque admodum pauci. Praeclara: evidently a fem. adj. agreeing with natura. Dav. and Ern. made the adj. neuter, and understanding sunt interpreted "these arguments I am going to urge are grand, viz. quanto art. etc."

§87. Scilicet: Germ. "natürlich." Fabricata sit: cf. 30, 119, 121 and N.D. I. 19. Ne modo: for modo ne, a noticeable use. Physicis: probably neut. Contra sensus: he wrote both for and against συνηθεια; cf. R. and P. 360 and 368. Carneadem: Plut. Sto. Rep. 1036 B relates that Carneades in reading the arguments of Chrysippus against the senses, quoted the address of Andromache to Hector: δαιμονιε φθισει σε το σον μενος. From Diog. IV. 62 we learn that he thus parodied the line qu. in n. on 75, ει μη γαρ ην Χρυσιππος ουκ αν ην εγω.

§88. Diligentissime: in 48—53. Dicebas: in 52 imbecillius adsentiuntur. Siccorum: cf. Cic. Contra Rullum I. 1 consilia siccorum. Madere is common with the meaning "to be drunk," as in Plaut. Mostellaria I. 4, 6. Non diceret: Orelli was induced by Goer. to omit the verb, with one MS., cf. 15 and I. 13. The omission of a verb in the subjunctive is, Madv. says on D.F. I. 9, impossible; for other ellipses of the verb see M.D.F. V. 63. Alcmaeo autem: i.e. Ennius' own Alcmaeon; cf. 52. Somnia reri: the best MSS. have somniare. Goer. reads somnia, supplying non fuisse vera. I have already remarked on his extraordinary power of supplying. Halm conj. somnia reprobare, forgetting that the verb reprobare belongs to third century Latinity, also sua visa putare, which Bait. adopts. Thinking this too large a departure from the MSS., I read reri, which verb occurred in I. 26, 39. Possibly putare, a little farther on, has got misplaced. Non id agitur: these difficulties supply Sextus with one of his τροποι, i.e. ‛ο περι τας περιστασεις; cf. P.H. I. 100, also for the treatment of dreams, ib. I. 104. Si modo, etc.: "if only he dreamed it," i.e. "merely because he dreamed it." Aeque ac vigilanti: = aeque ac si vigilaret. Dav. missing the sense, and pointing out that when awake Ennius did not assent to his sensations at all, conj. vigilantis. Two participles used in very different ways not unfrequently occur together, see Madv. Em. Liv. p. 442. Ita credit: MSS. have illa, which Dav. altered. Halm would prefer credidit. Itera dum, etc.: from the Iliona of Pacuvius; a favourite quotation with Cic.; see Ad Att. XIV. 14, and T.D. II. 44.

§89. Quisquam: for the use of this pronoun in interrogative sentences cf. Virg. Aen. I. 48 with the Notes of Wagner and Conington. Tam certa putat: so Sextus A.M. VII. 61 points out that Protagoras must in accordance with his doctrine παντων μετρον ανθρωπος hold that the μεμηνως is the κριτηριον των εν μανιαι φαινομενων. Video, video te: evidently from a tragedy whose subject was Αιας μαινομενος, see Ribbeck Trag. Lat. rel. p. 205. Cic. in De Or. III. 162 thus continues the quotation, "oculis postremum lumen radiatum rape." So in Soph. Aiax 100 the hero, after killing, as he thinks, the Atridae, keeps Odysseus alive awhile in order to torture him. Hercules: cf. Eur. Herc. Fur. 921—1015. The mad visions of this hero, like those of Orestes, are often referred to for a similar purpose by Sext., e.g. A.M. VII. 405 ‛ο γουν ‛Ερακλης μανεις και λαβων φαντασιαν απο των ιδιων παιδων ‛ως Ευρυσθεος, την ακολουθον πραξιν ταυτηι τη φαντασιαι συνηψεν. ακολουθον δε ην το τους του εχθρου παιδας ανελειν, ‛οπερ και εποιησεν. Cf. also A.M. VII. 249. Moveretur: imperf. for plup. as in 90. Alcmaeo tuus: cf. 52. Incitato furore: Dav. reads incitatus. Halm qu. from Wesenberg Observ. Crit. ad Or. p. Sestio p. 51 this explanation, "cum furor eius initio remissior paulatim incitatior et vehementior factus esset," he also refers to Wopkens Lect. Tull. p. 55 ed. Hand. Incedunt etc.: the MSS. have incede, which Lamb. corrected. The subject of the verb is evidently Furiae. Adsunt: is only given once by MSS., while Ribbeck repeats it thrice, on Halm's suggestion I have written it twice. Caerulea ... angui: anguis fem is not uncommon in the old poetry. MSS. here have igni. Crinitus: ακερσεκομης, "never shorn," as Milton translates it. Luna innixus: the separate mention in the next line of Diana, usually identified with the moon, has led edd. to emend this line. Some old edd. have lunat, while Lamb. reads genu for luna, cf. Ov. Am. I. 1, 25 (qu. by Goer.) lunavitque genu sinuosum fortiter arcum. Wakefield on Lucr. III. 1013 puts a stop at auratum, and goes on with Luna innixans. Taber strangely explains luna as = arcu ipso lunato, Dav. says we ought not to expect the passage to make sense, as it is the utterance of a maniac. For my part, I do not see why the poet should not regard luna and Diana as distinct.

§90. Illa falsa: sc. visa, which governs the two genitives. Goer. perversely insists on taking somniantium recordatione ipsorum closely together. Non enim id quaeritur: cf. 80 n. Sext. very often uses very similar language, as in P.H. I. 22, qu. in n. on 40. Tum cum movebantur: so Halm for MSS. tum commovebantur, the em. is supported by 88.

§§91—98. Summary: Dialectic cannot lead to stable knowledge, its processes are not applicable to a large number of philosophical questions (91). You value the art, but remember that it gave rise to fallacies like the sorites, which you say is faulty (92). If it is so, refute it. The plan of Chrysippus to refrain from answering, will avail you nothing (93). If you refrain because you cannot answer, your knowledge fails you, if you can answer and yet refrain, you are unfair (94). The art you admire really undoes itself, as Penelope did her web, witness the Mentiens, (95). You assent to arguments which are identical in form with the Mentiens, and yet refuse to assent to it Why so? (96) You demand that these sophisms should be made exceptions to the rules of Dialectic. You must go to a tribune for that exception. I just remind you that Epicurus would not allow the very first postulate of your Dialectic (97). In my opinion, and I learned Dialectic from Antiochus, the Mentiens and the arguments identical with it in form must stand or fall together (98).

§91. Inventam esse: cf. 26, 27. In geometriane: with this inquiry into the special function of Dialectic cf. the inquiry about Rhetoric in Plato Gorg. 453 D, 454 C. Sol quantus sit: this of course is a problem for φυσικη, not for διαλεκτικη. Quod sit summum bonum: not διαλεκτικη but ηθικη must decide this. Quae coniunctio: etc. so Sext. often opposes συμπλοκη or συνημμενον to διεζευγμενον, cf. esp P.H. II. 201, and Zeller 109 sq. with footnotes. An instance of a coniunctio (hypothetical judgment) is "si lucet, lucet" below, of a disiunctio (disjunctive judgment) "aut vivet cras Hermarchus aut non vivet". Ambigue dictum: αμφιβολον, on which see P.H. II. 256, Diog VII. 62. Quid sequatur: το ακολουθον, cf. I. 19 n. Quid repugnet: cf. I. 19, n. De se ipsa: the ipsa, according to Cic.'s usage, is nom. and not abl. Petrus Valentia (p. 301, ed Orelli) justly remarks that an art is not to be condemned as useless merely because it is unable to solve every problem presented to it. He quotes Plato's remarks (in Rep. II.) that the Expert is the man who knows exactly what his art can do and what it cannot. Very similar arguments to this of Cic. occur in Sext., cf. esp. P.H. II. 175 and the words εαυτου εσται εκκαλυπτικον. For the mode in which Carneades dealt with Dialectic cf. Zeller 510, 511. The true ground of attack is that Logic always assumes the truth of phenomena, and cannot prove it. This was clearly seen by Aristotle alone of the ancients; see Grote's essay on the Origin of Knowledge, now reprinted in Vol II. of his Aristotle.

§92. Nata sit: cf. 28, 59. Loquendi: the Stoic λογικη, it must be remembered, included ‛ρητορικη. Concludendi: του συμπεραινειν or συλλογιζεσθαι. Locum: τοπον in the philosophical sense. Vitiosum: 49, n. Num nostra culpa est: cf. 32. Finium: absolute limits; the fallacy of the sorites and other such sophisms lies entirely in the treatment of purely relative terms as though they were absolute. Quatenus: the same ellipse occurs in Orator 73. In acervo tritici: this is the false sorites, which may be briefly described thus: A asks B whether one grain makes a heap, B answers "No." A goes on asking whether two, three, four, etc. grains make a heap. B cannot always reply "No." When he begins to answer "Yes," there will be a difference of one grain between heap and no heap. One grain therefore does make a heap. The true sorites or chain inference is still treated in books on logic, cf. Thomson's Laws of Thought, pp 201—203, ed 8. Minutatim: cf. Heindorf's note on κατα σμικρον in Sophistes 217 D. Interrogati: cf. 104. In 94 we have interroganti, which some edd. read here. Dives pauper, etc.: it will be easily seen that the process of questioning above described can be applied to any relative term such as these are. For the omission of any connecting particle between the members of each pair, cf. 29, 125, T.D. I. 64, V. 73, 114, Zumpt Gram. 782. Quanto addito aut dempto: after this there is a strange ellipse of some such words as id efficiatur, quod interrogatur. [Non] habemus: I bracket non in deference to Halm, Madv. however (Opusc. I. 508) treats it as a superabundance of negation arising from a sort of anacoluthon, comparing In Vatin. 3, Ad Fam. XII. 24. The scribes insert and omit negatives very recklessly, so that the point may remain doubtful.

§93. Frangite: in later Gk. generally απολυειν. Erunt ... cavetis: this form of the conditional sentence is illustrated in Madv. D.F. III. 70, Em. Liv. p. 422, Gram. 340, obs. 1. Goer. qu. Terence Heaut. V. 1, 59 quot incommoda tibi in hac re capies nisi caves, cf. also 127, 140 of this book. The present is of course required by the instantaneous nature of the action. Chrysippo: he spent so much time in trying to solve the sophism that it is called peculiarly his by Persius VI. 80. inventus, Chrysippe, tui finitor acervi. The titles of numerous distinct works of his on the Sorites and Mentiens are given by Diog. Tria pauca sint: cf. the instances in Sext. A.M. VII. 418 τα πεντηκοντα ολιγα εστιν, τα μυρια ολιγα εστιν, also Diog. VII. 82 ‛ησυχαζειν the advice is quoted in Sext. P.H. II. 253 (δειν ‛ιστασθαι και επεχειν), A.M. VII. 416 (‛ο σοφος στησεται και ‛ησυχασει). The same terms seem to have been used by the Cynics, see Sext. P.H. II. 244, III. 66. Stertas: imitated by Aug. Contra Ac. III. 25 ter terna novem esse ... vel genere humano stertente verum sit, also ib. III. 22. Proficit: Dav. proficis, but Madv. rightly understands το ‛ησυχαζειν (Em. 184), cf. N.D. II. 58. Ultimum ... respondere: "to put in as your answer" cf. the use of defendere with an accus. "to put in as a plea". Kayser suggests paucorum quid sit.

§94. Ut agitator: see the amusing letter to Atticus XIII. 21, in which Cic. discusses different translations for the word επεχειν, and quotes a line of Lucilius sustineat currum ut bonu' saepe agitator equosque, adding semperque Carneades προβολην pugilis et retentionem aurigae similem facit εποχη. Aug. Contra Ac. trans. εποχη by refrenatio cf. also Lael. 63. Superbus es: I have thus corrected the MSS. responde superbe; Halm writes facis superbe, Orelli superbis, which verb is hardly found in prose. The phrase superbe resistere in Aug. Contra Ac. III. 14 may be a reminiscence. Illustribus: Bait. with some probability adds in, comparing in decimo below, and 107, cf. however Munro on Lucr. I. 420. Irretiat: parallel expressions occur in T.D. V. 76, De Or. I. 43, De Fato 7. Facere non sinis: Sext. P.H. II. 253 points the moral in the same way. Augentis nec minuentis: so Halm for MSS. augendi nec minuendi, which Bait. retains. I cannot believe the phrase primum augendi to be Latin.

§95. Tollit ... superiora: cf. Hortensius fragm. 19 (Orelli) sed ad extremum pollicetur prolaturum qui se ipse comest quod efficit dialecticorum ratio. Vestra an nostra: Bait. after Christ needlessly writes nostra an vestra. αξιωμα: "a judgment expressed in language"; cf. Zeller 107, who gives the Stoic refinements on this subject. Effatum: Halm gives the spelling ecfatum. It is probable that this spelling was antique in Cic.'s time and only used in connection with religious and legal formulae as in De Div. I. 81, De Leg. II. 20, see Corss. Ausspr. I. 155 For the word cf. Sen. Ep. 117 enuntiativum quiddam de corpore quod alii effatum vocant, alii enuntiatum, alii edictum, in T.D. I. 14 pronuntiatum is found, in De Fato 26 pronuntiatio, in Gellius XVI. 8 (from Varro) prologium. Aut verum esse aut falsum: the constant Stoic definition of αξιωμα, see Diog. VII. 65 and other passages in Zeller 107. Mentiris an verum dicis: the an was added by Schutz on a comparison of Gellius XVIII. 10 cum mentior et mentiri me dico, mentior an verum dico? The sophism is given in a more formally complete shape in De Div. II. 11 where the following words are added, dicis autem te mentiri verumque dicis, mentiris igitur. The fallacy is thus hit by Petrus Valentia (p. 301, ed Orelli), quis unquam dixit "ego mentior" quum hoc ipsum pronuntiatum falsum vellet declarare? Inexplicabilia: απορα in the Greek writers. Odiosius: this adj. has not the strong meaning of the Eng. "hateful," but simply means "tiresome," "annoying." Non comprehensa: as in 99, the opposite of comprehendibilia III. 1, 41. The past partic. in Cic. often has the same meaning as an adj. in -bilis. Faber points out that in the Timaeus Cic. translates αλυτος by indissolutus and indissolubilis indifferently. Imperceptus, which one would expect, is found in Ovid.

§96. Si dicis: etc. the words in italics are needed, and were given by Manut. with the exception of nunc which was added by Dav. The idea of Orelli, that Cic. clipped these trite sophisms as he does verses from the comic writers is untenable. In docendo: docere is not to expound but to prove, cf. n. on 121. Primum ... modum: the word modus is technical in this sense cf. Top. 57. The προτος λογος αναποδεικτος of the Stoic logic ran thus ει ‛ημερα εστι, φως εστιν ... αλλα μην ‛ημερα εστιν φως αρα εστιν (Sext. P.H. II. 157, and other passages qu. Zeller 114). This bears a semblance of inference and is not so utterly tautological as Cic.'s translation, which merges φως and ‛ημερα into one word, or that of Zeller (114, note). These arguments are called μονολημματοι (involving only one premise) in Sext. P.H. I. 152, 159, II. 167. Si dicis te mentiri, etc.: it is absurd to assume, as this sophism does, that when a man truly states that he has told a lie, he establishes against himself not merely that he has told a lie, but also that he is telling a lie at the moment when he makes the true statement. The root of the sophism lies in the confusion of past and present time in the one infinitive mentiri. Eiusdem generis: the phrase te mentiri had been substituted for nunc lucere. Chrysippea: n. on 93. Conclusioni: on facere with the dat. see n. on 27. Cederet: some edd. crederet, but the word is a trans. of Gk. εικειν; n. on 66. Conexi: = συνημμενον, cf. Zeller 109. This was the proper term for the hypothetical judgment. Superius: the συνημμενον consists of two parts, the hypothetical part and the affirmative—called in Greek ‛ηγουμενον and ληγον; if one is admitted the other follows of course.

§97. Excipiantur: the legal formula of the Romans generally directed the iudex to condemn the defendant if certain facts were proved, unless certain other facts were proved; the latter portion went by the name of exceptio. See Dict. Ant. Tribunum ... adeant: a retort upon Lucullus; cf. 13. The MSS. have videant or adeant; Halm conj. adhibeant, comparing 86 and Pro Rabirio 20. Contemnit: the usual trans. "to despise" for contemnere is too strong; it means, like ολιγωρειν, merely to neglect or pass by. Effabimur; cf. effatum above. Hermarchus: not Hermachus, as most edd.; see M.D.F. II. 96. Diiunctum: διεζευγμενον, for which see Zeller 112. Necessarium: the reason why Epicurus refused to admit this is given in De Fato 21 Epicurus veretur ne si hoc concesserit, concedendum sit fato fieri quaecumque fiant. The context of that passage should be carefully read, along with N.D. I. 69, 70. Aug. Contra Ac. III. 29 lays great stress on the necessary truth of disjunctive propositions. Catus: so Lamb. for MSS. cautus. Tardum: De Div. II. 103 Epicurum quem hebetem et rudem dicere solent Stoici; cf. also ib. II. 116, and the frequent use of βραδυς in Sext., e.g. A.M. VII. 325. Cum hoc igitur: the word igitur, as usual, picks up the broken thread of the sentence. Id est: n. on I. 8. Evertit: for the Epicurean view of Dialectic see R. and P. 343. Zeller 399 sq., M.D.F. I. 22. E contrariis diiunctio: = διεζευγμενον εξ εναντιων.

§98. Sequor: as in 95, 96, where the Dialectici refused to allow the consequences of their own principles, according to Cic. Ludere: this reminds one of the famous controversy between Corax and Tisias, for which see Cope in the old Journal of Philology. No. 7. Iudicem ... non iudicem: this construction, which in Greek would be marked by μεν and δε, has been a great crux of edd.; Dav. here wished to insert cum before iudicem, but is conclusively refuted by Madv. Em. 31. The same construction occurs in 103. Esse conexum: with great probability Christ supposes the infinitive to be an addition of the copyists.

§§98—105. Summary. In order to overthrow at once the case of Antiochus, I proceed to explain, after Clitomachus, the whole of Carneades' system (98). Carneades laid down two divisions of visa, one into those capable of being perceived and those not so capable, the other into probable and improbable. Arguments aimed at the senses concern the first division only; the sapiens will follow probability, as in many instances the Stoic sapiens confessedly does (99, 100). Our sapiens is not made of stone; many things seem to him true; yet he always feels that there is a possibility of their being false. The Stoics themselves admit that the senses are often deceived. Put this admission together with the tenet of Epicurus, and perception becomes impossible (101). It is strange that our Probables do not seem sufficient to you. Hear the account given by Clitomachus (102). He condemns those who say that sensation is swept away by the Academy; nothing is swept away but its necessary certainty (103). There are two modes of withholding assent; withholding it absolutely and withholding it merely so far as to deny the certainty of phenomena. The latter mode leaves all that is required for ordinary life (104).

98. Tortuosum: similar expressions are in T.D. II. 42, III. 22, D.F. IV. 7. Ut Poenus: "as might be expected from a Carthaginian;" cf. D.F. IV. 56, tuus ille Poenulus, homo acutus. A different meaning is given by the ut in passages like De Div. II. 30 Democritus non inscite nugatur, ut physicus, quo genere nihil arrogantius; "for a physical philosopher."

§99. Genera: here = classifications of, modes of dividing visa. This way of taking the passage will defend Cic. against the strong censure of Madv. (Pref. to D.F. p. lxiii.) who holds him convicted of ignorance, for representing Carneades as dividing visa into those which can be perceived and those which cannot. Is it possible that any one should read the Academica up to this point, and still believe that Cic. is capable of supposing, even for a moment, that Carneades in any way upheld καταληψις? Dicantur: i.e. ab Academicis. Si probabile: the si is not in MSS. Halm and also Bait. follow Christ in reading est, probabile nihil esse. Commemorabas: in 53, 58. Eversio: cf. D.F. III. 50 (the same words), Plat. Gorg. 481 C ‛ημων ‛ο βιος ανατετραμμενος αν ειη, Sext. A.M. VIII. 157 συγχεομεν τον βιον. Et sensibus: no second et corresponds to this; sic below replaces it. See Madv. D.F. p. 790, ed. 2. Quicquam tale etc.: cf. 40, 41. Nihil ab eo differens: n. on 54. Non comprehensa: n. on 96.

§100. Si iam: "if, for example;" so iam is often used in Lucretius. Probo ... bono: it would have seemed more natural to transpose these epithets. Facilior ... ut probet: the usual construction is with ad and the gerund; cf. De Div. II. 107, Brut. 180. Anaxagoras: he made no ‛ομοιομερειαι of snow, but only of water, which, when pure and deep, is dark in colour. Concreta: so Manut. for MSS. congregata. In 121 the MSS. give concreta without variation, as in N.D. II. 101, De Div. I. 130, T.D. I. 66, 71.

§101. Impeditum: cf. 33, n. Movebitur: cf. moveri in 24. Non enim est: Cic. in the vast majority of cases writes est enim, the two words falling under one accent like sed enim, et enim (cf. Corss. Ausspr. II. 851); Beier on De Off. I. p. 157 (qu. by Halm) wishes therefore to read est enim, but the MSS. both of the Lucullus and of Nonius agree in the other form, which Madv. allows to stand in D.F. I. 43, and many other places (see his note). Cf. fragm. 22 of the Acad. Post. E robore: so Nonius, but the MSS. of Cic. give here ebore. Dolatus: an evident imitation of Hom. Od. T 163 ου γαρ απο δριος εσσι παλαιφατου ουδ' απο πετρης. Neque tamen habere: i.e. se putat. For the sudden change from oratio recta to obliqua cf. 40 with n. Percipiendi notam: = χαρακτηρα της συγκταθεσεως in Sext. P.H. I. 191. For the use of the gerund cf. n. on 26, with Madv. Gram. 418, Munro on Lucr. I. 313; for propriam 34. Exsistere. cf. 36. Qui neget: see 79. Caput: a legal term. Conclusio loquitur: cf. historiae loquantur (5), consuetudo loquitur (D.F. II. 48), hominis institutio si loqueretur (ib. IV. 41), vites si loqui possint (ib. V. 39), patria loquitur (In Cat. I. 18, 27); the last use Cic. condemns himself in Orat. 85. Inquit: "quotha," indefinitely, as in 109, 115; cf. also dicit in 79.

§102. Reprehensio est ... satis esse vobis: Bait. follows Madv. in placing a comma after est, and a full stop at probabilia. Tamen ought in that case to follow dicimus, and it is noteworthy that in his communication to Halm (printed on p. 854 of Bait., and Hahn's ed. of the philosophical works, 1861) Madv. omits the word tamen altogether, nor does Bait. in adopting the suggestion notice the omission. Ista diceret: "stated the opinions you asked for." Poetam: this both Halm and Bait. treat as a gloss.

§103. For this section cf. Lucullus' speech, passim, and Sext. P.H. I. 227 sq. Academia ... quibus: a number of exx. of this change from sing. to plural are given by Madv. on D.F. V. 16. Nullum: on the favourite Ciceronian use of nullus for non see 47, 141, and Madv. Gram. 455, obs. 5. Illud sit disputatum: for the construction cf. 98; autem is omitted with the same constr. in D.F. V. 79, 80. Nusquam alibi: cf. 50.

§104. Exposuisset adiungit: Madv. on D.F. III. 67 notices a certain looseness in the use of tenses, which Cic. displays in narrating the opinions of philosophers, but no ex. so strong as this is produced. Ut aut approbet quid aut improbet: this Halm rejects. I have noticed among recent editors of Cic. a strong tendency to reject explanatory clauses introduced by ut. Halm brackets a similar clause in 20, and is followed in both instances by Bait. Kayser, who is perhaps the most extensive bracketer of modern times, rejects very many clauses of the kind in the Oratorical works. In our passage, the difficulty vanishes when we reflect that approbare and improbare may mean either to render an absolute approval or disapproval, or to render an approval or disapproval merely based on probability. For example, in 29 the words have the first meaning, in 66 the second. The same is the case with nego and aio. I trace the whole difficulty of the passage to the absence of terms to express distinctly the difference between the two kinds of assent. The general sense will be as follows. "There are two kinds of εποχη, one which prevents a man from expressing any assent or disagreement (in either of the two senses above noticed), another which does not prevent him from giving an answer to questions, provided his answer be not taken to imply absolute approval or absolute disapproval; the result of which will be that he will neither absolutely deny nor absolutely affirm anything, but will merely give a qualified 'yes' or 'no,' dependent on probability." My defence of the clause impugned is substantially the same as that of Hermann in the Philologus (vol. VII.), which I had not read when this note was first written. Alterum placere ... alterum tenere: "the one is his formal dogma, the other is his actual practice." For the force of this see my note on non probans in 148, which passage is very similar to this. Neget ... aiat: cf. 97. Nec ut placeat: this, the MSS. reading, gives exactly the wrong sense, for Clitomachus did allow such visa to stand as were sufficient to serve as a basis for action. Hermann's neu cui labours under the same defect. Various emendations are nam cum (Lamb., accepted by Zeller 522), hic ut (Manut.), et cum (Dav. followed by Bait.), sed cum (Halm). The most probable of these seems to me that of Manut. I should prefer sic ut, taking ut in the sense of "although." Respondere: "to put in as an answer," as in 93 and often. Approbari: sc. putavit. Such changes of construction are common in Cic., and I cannot follow Halm in altering the reading to approbavit.

§105. Lucem eripimus: cf. 30.

§§105—111. Summary. You must see, Lucullus, by this time, that your defence of dogmatism is overthrown (105). You asked how memory was possible on my principles. Why, did not Siron remember the dogmas of Epicurus? If nothing can be remembered which is not absolutely true, then these will be true (106). Probability is quite sufficient basis for the arts. One strong point of yours is that nature compels us to assent. But Panaetius doubted even some of the Stoic dogmas, and you yourself refuse assent to the sorites, why then should not the Academic doubt about other things? (107) Your other strong point is that without assent action is impossible (108). But surely many actions of the dogmatist proceed upon mere probability. Nor do you gain by the use of the hackneyed argument of Antiochus (109). Where probability is, there the Academic has all the knowledge he wants (110). The argument of Antiochus that the Academics first admit that there are true and false visa and then contradict themselves by denying that there is any difference between true and false, is absurd. We do not deny that the difference exists; we do deny that human faculties are capable of perceiving the difference (111).

105. Inducto ... prob.: so Aug. Cont Ac. II. 12 Soluto, libero: cf. n. on 8. Implicato: = impedito cf. 101. Iacere: cf. 79. Isdem oculis: an answer to the question nihil cernis? in 102. Purpureum: cf. fragm. 7 of the Acad. Post. Modo caeruleum ... sole: Nonius (cf. fragm. 23) quotes tum caeruleum tum lavum (the MSS. in our passage have flavum) videtur, quodque nunc a sole. C.F. Hermann would place mane ravum after quodque and take quod as a proper relative pronoun, not as = "because." This transposition certainly gives increased clearness. Hermann further wishes to remove a, quoting exx. of collucere without the prep., which are not at all parallel, i.e. Verr. I. 58, IV. 71. Vibrat: with the ανηριθμον γελασμα of Aeschylus. Dissimileque: Halm, followed by Bait., om. que. Proximo et: MSS. have ei, rightly altered by Lamb., cf. e.g. De Fato 44. Non possis ... defendere: a similar line is taken in 81.

§106. Memoria: cf. 22. Polyaenus: named D.F. I. 20, Diog. X. 18, as one of the chief friends of Epicurus. Falsum quod est: Greek and Latin do not distinguish accurately between the true and the existent, the false and the non existent, hence the present difficulty; in Plato the confusion is frequent, notably in the Sophistes and Theaetetus. Si igitur: "if then recollection is recollection only of things perceived and known." The dogmatist theory of μνημη and νοησις is dealt with in exactly the same way by Sext. P.H. II. 5, 10 and elsewhere, cf. also Plat Theaet. 191 sq. Siron: thus Madv. on D.F. II. 119 writes the name, not Sciron, as Halm. Fateare: the em. of Dav. for facile, facere, facias of MSS. Christ defends facere, thinking that the constr. is varied from the subj. to the inf. after oportet, as after necesse est in 39. For facere followed by an inf. cf. M.D.F. IV. 8. Nulla: for non, cf. 47, 103.

§107. Fiet artibus: n. on 27 for the constr., for the matter see 22. Lumina: "strong points." Bentl. boldly read columina, while Dav. proposed vimina or vincula. That an em. is not needed may be seen from D.F. II. 70. negat Epicurus (hoc enim vestrum lumen est) N.D. I. 79, and 43 of this book. Responsa: added by Ernesti. Faber supplies haruspicia, Orelli after Ern. haruspicinam, but, as Halm says, some noun in the plur. is needed. Quod is non potest: this is the MSS. reading, but most edd. read si is, to cure a wrong punctuation, by which a colon is placed at perspicuum est above, and a full stop at sustineat. Halm restored the passage. Habuerint: the subj. seems due to the attraction exercised by sustineat. Bait. after Kayser has habuerunt. Positum: "when laid down" or "assumed."

§108. Alterum est quod: this is substituted for deinde, which ought to correspond to primum above. Actio ullius rei: n. on actio rerum in 62, cf. also 148. Adsensu comprobet: almost the same phrase often occurs in Livy, Sueton., etc. see Forc. Sit etiam: the etiam is a little strange and was thought spurious by Ernesti. It seems to have the force of Eng. "indeed", "in what indeed assent consists." Sensus ipsos adsensus: so in I. 41 sensus is defined to be id quod est sensu comprehensum, i.e. καταληψις, cf. also Stobaeus I. 41, 25 αισθητικη γαρ φαντασια συγκαταθεσις εστι. Appetitio: for all this cf. 30. Et dicta ... multa: Manut. ejected these words as a gloss, after multa the MSS. curiously add vide superiora. Lubricos sustinere: cf. 68 and 94. Ita scribenti ... exanclatum: for the om. of esse cf. 77, 113 with notes. Herculi: for this form of the gen. cf. Madv. on D.F. I. 14, who doubts whether Cic. ever wrote -is in the gen. of the Greek names in -es. When we consider how difficult it was for copyists not to change the rarer form into the commoner, also that even Priscian (see M.D.F. V. 12) made gross blunders about them, the supposition of Madv. becomes almost irresistible. Temeritatem: προπετειαν, εικαιοτητα.

§109. In navigando: cf. 100. In conserendo: Guretus interprets "εν τω φυτυεσθαι τον αγρον," and is followed by most commentators, though it seems at least possible that manum is to be understood. For the suppressed accus. agrum cf. n. on tollendum in 148. Sequere: the fut. not the pres. ind., cf. 61. Pressius: cf. 28. Reprehensum: sc. narrasti. Id ipsum: = nihil posse comprehendi. Saltem: so in 29. Pingue: cf. Pro Archia 10. Sibi ipsum: note that Cic. does not generally make ipse agree in case with the reflexive, but writes se ipse, etc. Convenienter: "consistently". Esse possit: Bait. posset on the suggestion of Halm, but Cic. states the doctrine as a living one, not throwing it back to Antiochus time and to this particular speech of Ant. Ut hoc ipsum: the ut follows on illo modo urguendum above. Decretum quod: Halm followed by Bait. gives quo, referring to altero quo neget in 111, which however does not justify the reading. The best MSS. have qui. Et sine decretis: Lamb. gave nec for et, but Dav. correctly explains, "multa decreta habent Academici, non tamen percepta sed tantum probabilia."

§110. Ut illa: i.e. the decreta implied in the last sentence. Some MSS. have ille, while Dav. without necessity gives alia. Sic hoc ipsum: Sext. then is wrong is saying (P.H. I. 226) that the Academics διαβεβαιουνται τα πραγματα ειναι ακαταληπτα, i.e. state the doctrine dogmatically, while the sceptics do not. Cognitionis notam: like nota percipiendi, veri et falsi, etc. which we have already had. Ne confundere omnia: a mocking repetition of Lucullus phrase, cf. 58. Incerta reddere: cf. 54. Stellarum numerus: another echo of Lucullus; see 32. Quem ad modum ... item: see Madv. on D.F. III. 48, who quotes an exact parallel from Topica 46, and sicut ... item from N.D. I. 3, noting at the same time that in such exx. neither ita nor idem, which MSS. sometimes give for item, is correct.

§111. Dicere ... perturbatum: for om. of esse cf. 108, etc. Antiochus: this Bait. brackets. Unum ... alterum: cf. 44. Esse quaedam in visis: it was not the esse but the videri, not the actual existence of a difference, but the possibility of that difference being infallibly perceived by human sense, that the Academic denied. Cernimus: i.e. the probably true and false. Probandi species: a phenomenal appearance which belongs to, or properly leads to qualified approval.

§§112—115. Summary. If I had to deal with a Peripatetic, whose definitions are not so exacting, my course would be easier; I should not much oppose him even if he maintained that the wise man sometimes opines (112). The definitions of the real Old Academy are more reasonable than those of Antiochus. How, holding the opinions he does, can he profess to belong to the Old Academy? (113) I cannot tolerate your assumption that it is possible to keep an elaborate dogmatic system like yours free from mistakes (114). You wish me to join your school. What am I to do then with my dear friend Diodotus, who thinks so poorly of Antiochus? Let us consider however what system not I, but the sapiens is to adopt (115).

§112. Campis ... exsultare ... oratio: expressions like this are common in Cic., e.g. D.F. I. 54, De Off. I. 61, Orat. 26; cf. also Aug. Cont. Ac. III. 5 ne in quaestionis campis tua eqitaret oratio. Cum Peripatetico: nothing that Cic. states here is at discord with what is known of the tenets of the later Peripatetics; cf. esp. Sext. A.M. VII. 216—226. All that Cic. says is that he could accept the Peripatetic formula, putting upon it his own meaning of course. Doubtless a Peripatetic would have wondered how a sceptic could accept his formulae; but the spectacle of men of the most irreconcilable opinions clinging on to the same formulae is common enough to prevent us from being surprised at Cicero's acceptance. I have already suggested (n. on 18) that we have here a trace of Philo's teaching, as distinct from that of Carneades. I see absolutely no reason for the very severe remarks of Madvig on D.F. V. 76, a passage which very closely resembles ours. Dumeta: same use in N.D. I. 68, Aug. Cont. Ac. II. 6; the spinae of the Stoics are often mentioned, e.g. D.F. IV. 6. E vero ... a falso: note the change of prep. Adhiberet: the MSS. are confused here, and go Halm reads adderet, and Bait. follows, while Kayser proposes adhaereret, which is indeed nearer the MSS.; cf. however I. 39 adhiberet. Accessionem: for this cf. 18 and 77. Simpliciter: the opposite of subtiliter; cf. simpliciter—subtilitas in I. 6. Ne Carneade quidem: cf. 59, 67, 78, 148.