Chapter I

§1. cognitioni, Harl. 4995: Burn. 243 (and so Gothanus, Spald.). Cogitationi G and most codd., probably mistaking a contraction in the ancient text.

§2. sciet G. The reading scierit (Harl. 4995 and many codd.) is probably due to H, which gives sciuit (so FT).

quae quoque sint modo dicenda. So GHFTL, and Halm. The alternative reading is quo quaeque s. m. d., S and all my 15th cent. MSS: Spalding and Meister, with the approval of Becher. See note ad loc. In the parallel passages i. 8. 1 Halm adopts Spalding’s reading (ut sciat) quo quidque flexu ... dicendum for quid quoque ABMS, and i. 6. 16 (notatum) quo quidque modo caderet for quid quoque BMS, and so Meister: Fierville returns to the reading of the MSS. In support of quo quaeque other exx. might be cited: v. 10. 17 quo quaeque modo res vitari vel appeti soleat, and vi. 4. 22 quo quaeque ordine probatio sit proferenda. But the parallel instances in the Tenth Book quoted in the notes (1 §8: 7 §§5 and 6) seem to guarantee the correctness of the reading of the oldest MSS.: though it is better to take quoque as the ablative of quisque than (as Halm) as the relative with que.

tamen: GHFT Harl. 4950: tanquam Harl. 2662, 11671, 4995, 4829, L S Bodl. Ball. Burn. 243 Dorv. In Burn. 244 tanquam is corrected to tamen. Paratam explains in procinctu: so that tanquam is not so necessary as velut in xii. 9. 21.

§3. ante omnia est: so all codd., and Halm. Hirt (Jahresb. des philol. Vereins zu Berlin viii. p. 69 sq. 1882: ix. p. 312 sq. 1883) conjectured ante omnia necessarium est, and this is approved by Kiderlin (Blätter f. d. bayer. Gymn. 1887, p. 454): cp. necessarium just above, and necessaria in §1. Schöll (Rh. Mus. 34, p. 84) first challenged the MS. reading, and suggested that the original may have been ante omnia stat atque, corrupted into ante omniast [at] atque: for which use of sto, see Bonn. Lex. s.v. ii. γ. As an alternative suggestion he put forward ante omnia necesse est, and this was adopted by Meister. Becher (Phil. Rundsch. iii. 14. 428) proposed ante omnia sciet, though more recently he has signified his adherence to the tradition of the MSS. Maehly suggested ante omnia opus esse. Perhaps the true reading may be ante omnia prodest.

The question depends to some extent on the treatment of the following passage. GH agree in giving proximam deinde inimitationem novissimam scribendi quoque diligentia. This Halm converted into proximum deinde imitatio est, novissimum ... diligentia,—where the est is certainly superfluous (cp. i. 3. 1), while it may be doubted (comparing ii. 13. 1 and iii. 6. 81—Kiderlin l.c.) whether proxima deinde imitatio, novissima &c. would not be a sufficient change: Kiderlin compares ‘proxima huic narratio,’ ii. 13. 1, and ‘novissima qualitas superest,’ and objects to the citation of ‘proximum imitatio,’ in 1. 3, in support of the neuter, on the ground that there ‘signum ingenii’ is to be supplied.

Kiderlin’s proposed modification of Gemoll’s conjecture (l.c. p. 454 note, cp. Rhein. Mus. 46 p. 10 note) proximum deinde multa lectio is adopted by Krüger (3rd ed.), who thinks that the sequence of thought makes the special mention of legere (alongside of dicere and scribere) a necessity: multa corresponds to diligentia in what follows: cp. multa lectione §10. But legere has already been touched on in §2, and moreover is included under imitatio (sc. exemplorum ex lectione et auditione repetitorum).

§4. iam opere. So Harl. 4995 and Regius: all other codd. iam opere iam. Becher reports iam opere also from the Vallensis.

qua ratione. For qua in oratione, the reading of all MSS., Hirt conjectured qua exercitatione. Schöll proposed to reject in oratione as a gloss: but qua by itself (sc. via) is only used by Quint. with verbs of motion: see on 7 §11.

In his latest paper (Rheinisches Museum, 46, pp. 10-13, 1891), Kiderlin subjects the whole of §4 to a searching and destructive analysis. He translates: ‘doch nicht darüber, wie der Redner heranzubilden ist, sprechen wir in diesem Abschnitte (denn dies ist genügend oder wenigstens so gut, als wir konnten, besprochen worden) sondern darüber, durch welche Art von Uebung der Athlet, welcher alle Bewegungen von seinem Lehrer bereits genau erlernt hat, für die Kämpfe vorzubereiten ist.’ He doubts whether such passages as §33 and 7 §1 can be cited to justify the abrupt transition from orator to athlete, on the ground of the formal antithesis in which the two stand to each other,—‘orator’ coming in at the end of one clause, and ‘athleta’ standing at the head of another, in front of ‘quo genere exercitationis.’ And yet it is just the ‘orator’ who is to be understood in the ‘athleta.’ As to the sentence introduced by ‘Igitur eum,’ if by ‘athleta qui omnes iam perdidicerit a praeceptore numeros’ we are to understand one who has mastered the whole theory of rhetoric, then it adds nothing to what has been said already, and is therefore altogether superfluous.

Kiderlin proposes to read: sed ut (so L and S,—also Harl. 2662, 4995) athleta, qui omnes iam perdidicerit a praeceptore numeros, multo (nonnullo?) varioque (numuro quae G,—also H: num muro quae T: numeroque F L; nimirum quo S) genere exercitationis ad certamina praeparandus erit (sit, the codd.) ita (so S,—also Harl. 2662, 4995 and Bodl.) eum, qui ... perceperit, instruamus, qua in praeparatione (qua in oratione, the codd.) quod didicerit facere quam optime, quam facillime possit. Ut may easily, he contends, have fallen out before at: and the running of three words into one (numeros multo vario—numero) is paralleled by such a case as §23, where it will be found that Kiderlin sees ut duo tresque in utrisque. For ‘multo varioque’ he compares viii. 5. 28 multis ac variis: x. 5. 3 multas ac varias: xi. 3. 163 varia et multiplex: xii. 1. 7 totae tam variis; and, for ‘varioque,’ vii. 3. 16 latiore varioque, and xii. 10. 36 sublimes variique. ‘Vario genere’ actually occurs i. 10. 7, and multo may easily have been written in the singular, like nonnullus vi. 3. 11 (hoc nonnullam observationem habet) and elsewhere. The motive for changing que, quae, into quo and erit (est?) into sit may have been the analogy of the foregoing quomodo sit. As for ut (sicut) ita (sic), it is so favourite a form with Quintilian that he uses it seven times in the first nineteen paragraphs of this chapter. Qua in oratione, the reading of all MSS., may have resulted from qua in praeparatione more probably than from qua ratione, which appears first in the ed. Col. 1527, and is not so appropriate to the context as qua in praeparatione (cp. praeparandus above, and parandae below). Quintilian is detailing in this Book on what preparation (cp. praeparant §35, comparant §67, praeparetur 6 §6, praeparantur 7 §19) the orator may best and most easily carry out in practice what he has learnt theoretically. For the preposition (in praeparatione) cp. viii. pr. 22: ut in hac diligentia deterior etiam fiat oratio.

The text of Quintilian, especially of this part of the Tenth Book, is admittedly very defective, and invites emendation: there is a great deal to be said for the theory that in many places several words must have dropped out. Kiderlin’s attempts to remedy existing defects are always marked by the greatest ingenuity: they are all well worth recording as evidences of critical ability and insight, even though it may be that not all of them will be received into the ultimate text. Here there seems no reason why Quintilian, who was notoriously a loose writer, should not have said in the concluding sentence of the paragraph what he had already said, in the form of a metaphor, in the clause immediately preceding. Indeed the word igitur seems to suggest that after indulging in his favourite metaphor (sed athleta, &c.) he wishes to resume, as it were, and is now going on to say what he means in more ordinary language. It may not be artistic: but it is Quintilian. If he had had some of his modern critics at his side when preparing a second edition of the Institutio some of his angularities might have been smoothed away.

§5. Non ergo. Meister and ‘edd. vett.’: I find this reading in Harl. 4995, and Burn. 243. So Vall. Halm. has Num ergo, and so most codd. (including HFT Bodl. and Ball.).

§6. ex his. Qy. ex iis? so §128: cp. Introd. p. xlix.

§7. quo idem, Meister and ‘edd. vett.’: quod idem Halm, supported by Becher and Hirt, perhaps rightly. Nearly all my MSS. agree with GLS in quod: quo occurs in Harl. 4995 only.

§8. quod quoque GH Halm, Meister: quid quoque (as 7 §5) occurs in L S, also in Bodl., Ball. For quid Zumpt cites also Par. 1 and 2: i.e. 7723 and 7724 (Fierville). Aptissimum (strangely mangled in most codd.—e.g. locis ita petissimum G) is given rightly in Dorv.

§9. omnibus enim fere verbis. This reading, ascribed by Meister to Badius, and by Halm to ed. Colon. (1527), I have found in Harl. 4995 (A.D. 1470): ferebis vel G H: fere rebus vel L S Harl. 2662, 4950, 4829. From the Vallensis Becher reports fere verbis vel.

intueri, ed. Col. 1527. In Harl. 11671 I find interim intueri: Harl. 2662 L S Ball., Dorv., Bodl., interim tueri.

quae nitidiore in parte occurs first in ed. Col. 1527: Vall.² Harl. 4995 Goth. Voss. ii. shows quae cultiore in p.: GH quaetidiorem in p.: LS Harl. 2662 Guelf. Bodl. quae utiliore in p.

§10. cum omnem, &c. cum omnem misermonem a. pr. accipiamus GH: cum omnem enim, most codd. Osann, followed by Gemoll and Krüger (3rd ed.), suggested omnem enim sermonem a. pr. accipimus.

§11. alia vero, Frotscher: aliave GH: aliaque Harl. 4995. This last Becher now prefers (alia que Vall.: alia quae Regius), comparing ix. 3. 89 and ix. 4. 87.

τροπικῶς quasi tamen, Spalding, Zumpt, Meister and Krüger (3rd ed.): tropicos quare tam GH, quare tamen, later MSS. Halm obelized quare tamen: Mayor only quare. Becher recommends tamen by itself. Gensler (Anal. p. 25) reads tamen quasi, and is followed by Hild, who takes quasi with feruntur in the sense of referuntur (μεταφορά): Zumpt took it with eundem intellectum. Gemoll approves of the exclusion of quare, which he thinks must have arisen from a gloss figurate (either marginal or interlinear) on τροπικῶς. Kiderlin adopts this and thinks the quare tam of GHL a mutilation of the gloss figurate: gurate and quare tā are not far apart.

§12. figurarum G (per compendium): figuranus H. Kiderlin suggests mutuatione figurarum, sc. ostendimus: after which Quintilian continues ‘sed etiam ex proximo mutuari licet.’ Cp. Cic. de Or. iii. 156 translationes quasi mutuationes sunt. Kiderlin adds (Rhein. Mus. 46, p. 14 note) that in iii. 4. 14 all MSS. wrongly give mutantes for mutuantes, and in i. 4. 7 A¹ has mutamur for mutuamur.

§15. hoc sunt exempla potentiora. Hoc is a conj. of Regius (also Vall.²), all the MSS. giving haec (hec). Hoc appears in the Basle ed. of 1555 and in that of Leyden 1665. It is challenged by Schöll (Rhein. Mus. 44, p. 85), who says quia stands too far away from hoc to allow of such a construction, and thinks the context has been misunderstood. According to him haec exempla (those derived from lectio and auditio) are set over against those which one gets in theoretical books and lectures: they are more telling, because they act directly on the mind, and are not served up as dry theory in the form of extracts (‘quia quae doctor praecepit orator ostendit’). He therefore understands ‘ipsis (exemplis) quae traduntur artibus,’ but admits that ‘etiam’ is thus otiose, and would therefore read quam ipsis quae traduntur artibus.

Schöll is supported by Hirt (Jahresb. des philol. Vereins zu Berlin, 1882, p. 70), who thus gives the sense of the passage: ‘Der Wortschatz wird durch Lektüre und vieles Hören erworben. Aber nicht nur seinetwegen soll man lesen und hören; man soll es auch noch aus einem anderen Grunde. In allem nämlich, was wir lehren, sind diese Beispiele, d.h. diejenigen, welche uns die Lektüre und der Vortrag bieten, wichtiger selbst als die Beispiele welche die Handbücher und Vorlesungen darbieten, weil, was der Lehrer nur als Forderung aufstellt, bei dem Redner That geworden ist und sich durch den Erfolg bewährt hat.’

Iwan Müller (Bursian’s Jahresb. vii. 1879, 2, p. 168) objects that if Quintilian had wished to convey this meaning he would have said, not haec exempla, but hinc ducta (petita) or quae hinc ducuntur (petuntur) exempla; and he rightly desiderates also quam quae (in) ipsis traduntur artibus. Meister also opposes Schöll (Philol. xlii. p. 149): the order quam ipsis quae traduntur artibus is in fact impossible.

On the whole it seems much better to keep hoc, and to understand: ‘in all instruction, example is better than precept: the doctor relies only on precept, the orator on example.’

Gertz conjectures nam omnium quaecunque docemus hinc (cp. v. 10. 5: xii. 2. 31) sunt exempla, potentiora (i.e. quae potentiora sunt) etiam ipsis quae traduntur artibus. But with hinc, as Kiderlin observes, some other verb than sunt would be expected: v. 10. 15 is an uncertain conjecture, the MSS. giving nihil, and in xii. 2. 31 hinc belongs to bibat and sumptam. Kiderlin himself at first proposed haec praestant exempla, potentiora: this he now withdraws, however, (Rhein. Mus. 46, p. 15) in favour of haec suggerunt exempla, potentiora, &c. By haec he understands legere and audire, and gives the sequence of thought as follows:—‘Aber wenn auch auf diese Weise eine Fülle von Ausdrücken erworben wird, so ist das doch nicht der einzige Zweck des Lesens und Hörens. Denn von allem was wir lehren (nicht nur von den Ausdrücken) liefert dieses (das Lesen und Hören) Beispiele, welche noch wirksamer sind als die vorgetragenen Theorieen selbst (wenn der Lernende so weit gefördert ist, dass er die Beispiele ohne Beihilfe verstehen und sie bereits aus eigener Kraft befolgen kann), weil der Redner das zeigt, was der Lehrer nur vorgeschrieben hat.’ For suggerere Kiderlin compares i. 10. 7 artibus, quae ... vim occultam suggerunt, and v. 7. 8 ea res suggeret materiam interrogationi: cp. also §13 quorum nobis ubertatem ac divitias dabit lectio, and ii. 2. 8 licet satis exemplorum ad imitandum ex lectione suppeditet.

§16. imagine et ambitu rerum: so Harl. 2662 L S Ball. Burn. 243 and Bodl.: followed by Spalding, Frotscher, Herbst, and Bonnell. GH give imagine ambitu rerum. Halm (after Bursian) bracketed ambitu: but it is more probable that imagine is a gloss on ambitu than vice versa (so Hirt and Kiderlin), and Meister accordingly (followed by Krüger 3rd ed.) reads [imagine] ambitu rerum. It seems just as likely, however, that et has fallen out. Hertz suggested imagine ambituve rerum: Maehly thinks that ambitu was originally tantum.

nec fortune modo. Gertz proposed nec forma modo: pro Mil. §1 movet nos forma ipsa et species veri iudicii.

§17. accommodata ut: ed. Col. 1527, and so Meister and Krüger (3rd ed.): commodata ut Halm (after Bursian): commoda ut Spald., Frotsch., Herbst, and Bonnell. GHS give commoda aut: L and all my MSS commoda ut (except Burn. 243 which shows comendat ut).

et, ut semel dicam. Kiderlin would delete et, rendering ‘Stimme, Aktion, Vortrag ist, um es kurz zu sagen, alles in gleicher Weise belehrend.’

§18. placent—laudantur—placent: so Halm and most edd., following S, with which all my MSS. agree. The emphasis gained by the opposition of placent and non placent makes this reading probable. But GH give laudetur: and so Meister and Krüger (3rd ed.) prefer to follow Regius in reading placeant—laudentur—placent.

§19. e contrario. This reading, which Meister adopts from ‘edd. vett.,’ occurs in Harl. 2662, 4995, 4829, 11671, Burn. 243, 244, Bodl. and Dorv. Becher reports it also from the Vallensis. Halm wrote contrarium.

actionis impetu, Spald. and Krüger (3rd ed.): actionis impetus GH and all MSS. (except Vall., in which the s in impetus has been deleted): ut actionis impetus Halm and Meister.

tractemus GHL: tractamus all my MSS.: retractemus Spald., Halm, Meister. Becher (Phil. Rundsch. iii. 14. 429) supports tractemus, arguing that the phrase is a sort of hendiadys = repetendo tractemus (cp. Frotscher, and Bonn. Proleg. to Lex. p. xxxviii), or that the re of repetamus is to be supplied in thought with tractemus: cp. Cic. de Div. 1 §1 ‘praesensionem et scientiam rerum futurarum.’ Tractamus in 5 §8 also supports this reading.

iteratione, Harl. 4995 and Vall.²: most MSS. altercatione (as G) or alteratione (as Harl. 2662).

§22. illud vero. The MSS. vary between illa (GH) and illud (Harl. 4995 Vall.²). Kiderlin suggests illa ... utilissima.

§23. Quin etiam si ... tamen: so all MSS. Meister and Krüger (3rd ed.) accept Eussner’s proposal to exclude quin. Becher on the other hand objects (Bursian’s Jahresb. 1887. xv. 2, p. 9). From some points of view the deletion would be an improvement: it would bring out better the chiastic arrangement, utilissimum ... utrimque habitas legere actiones and easdem causas ... utile erit scire. But (1) such careless repetition (quin etiam—quin etiam) is not unusual in Quint.: and (2) si when followed by tamen often = etiamsi: Cic. pro Leg. Man. §50: pro Deiot. §25: Sall. Bell. Iug. 85, 48 &c., so that it is not necessary to connect etiam with it like etiamsi ... tamen xi. 3. 48. The sentence (as recommending the reading of the ‘minus pares actiones’) forms an exception to the rule otherwise consistently followed, ‘non nisi optimus quisque legendus,’ &c.

Again Spalding, Bonnell, and Hild put the comma before, not after aliquae, which they take with requirentur (‘yet in some cases’). But this does not square with ‘quoties continget utrimque habitas legere actiones,’—words which are distinctly against any idea of selecting from the ‘minus pares.’

causas ut quisque egerit utile erit scire, Halm and Meister following ed. Ald., and ed. Colon. 1527: causas utile erit scire Vall.: all other codd. causas utrisque erit scire. Meister thinks non inutile would be more in accordance with Quintilian’s usage. Gemoll suggests causas ut plures egerint intererit scire, Kaibel ut quisque egerit e re erit scire. Perhaps (with Becher) causas ut quisque egerit intererit scire.

Kiderlin’s treatment of the passage merits a separate notice. He accepts the first quin etiam, as the reading of the MSS., and also as quite appropriate to the context (‘in cases even where the combatants are not equally matched—as were Demosthenes and Aeschines’). But he doubts whether Quintilian could have written two sentences running, each beginning with quin etiam, and relies greatly on the undoubted fact that in the second all the MSS. have quis etiam,—quin being an emendation by Regius. The MS. reading is quis etiam easdem causas utrisque erit scire: this Kiderlin would at once convert into ‘quis etiam illud utile neget (or, negat esse utile) easdem causas ut quisque egerit, scire’?—comparing xii. 10. 48 ceterum hoc quod vulgo sententias vocamus ... quis utile neget? But ut quisque does not quite satisfy him. In the sequel reference is made to cases in which two and even three orators have handled the same theme: Kiderlin therefore proposes ut duo tresque for the MS. utrisque. The passage would then run: ‘quis etiam illud utile neget (negat esse utile?) easdem causas ut duo tresque (tresve?) egerint, scire?’ The position of easdem causas is due to a desire for emphasis: and for the isolated position of scire cp. v. 7. 2 quo minus et amicus pro amico et inimicus contra inimicum possit verum, si integra sit ei fides, dicere.

§28. poeticam ostentationi comparatam. This is Schöll’s conj. for the MSS. genus ostent. comparatum, which is however defended by Becher in Bursian’s Jahresb. (1887), p. 40: he contends that the feminine participles below (adligata, depulsa) refer to poesis, present in the mind of the writer, and that the text of the MSS. is simply a case of constr. κατὰ σύνεσιν: cp. ix. 2. 79: ib. 3 §3, and such passages as Cic. Or. §68 ego autem etiamsi quorundam grandis et ornata vox est poetarum, tamen in ea (sc. poesi), &c. This would support also the traditional reading nescio an ulla §65 below, where see note. Becher explains the MS. reading as = genus (sc. poeticum or hoc genus) ostent. comp. (esse)—Halm prints genus * * * ostent., and supposes that poeseos has fallen out.—For genus cp. §68: de Or. ii. §55, where genus hoc = history.

Schöll’s argument (Rhein. Mus. 34, p. 86) is that Quintilian cannot have passed from genus to adligata: Halm’s genus poeseos is not probable, in the light of Quintilian’s avoidance of the word poesis (cp. xii. 11. 26, where it occurs once, and there only in A in rasura—GM giving poetas, which was probably at first the reading also of A: there Halm and Meister now read poetica). The text may have been altered by interpolation from viii. 3. 11: namque illud genus (sc. demonstrativum) ostentationi compositum solam petit audientium voluptatem,—from which passage genus may have been written in where the Greek ποιητικήν had fallen out, giving rise to comparatum. Meister, who adopts poeticam, thinks it probable that the Greek word started the corruption. Other suggestions are praeter id quod, genus ost. comp., sol. petit vol. (Hild),—a transposition which does no good, especially as it leaves no subject to ‘iuvari’: figurarum sed esse hoc eloquentiae genus ost. comp. et ... iuvari (Binde); fig., ingenuam ost. comparatam artem (Gemoll); Kiderlin (Hermes 23, p. 164) thinks we ought to assume a lacuna, and would read poeticam (or poesin?) ut illud demonstrativum genus, ostentationi comparatam: cp. ii. 10. 11: v. 10. 43: iii. 7. 28: viii. 3. 11.

§30. neque ego: Spald., Frotscher, Herbst, Halm, Meister. Neque ergo all MSS. Bonnell and Frieze retain the reading of the MSS., the latter explaining ergo ‘viz. because I have given this caution to the orator about too close imitation of the poetic manner.’

§31. quodam uberi: Spald. for quodam moveri of GH and all MSS. except Harl. 4995, Vail.² and Burn. 243, which give quodam molli. Kiderlin suggests quodammodo uberi, thinking that uberi became ueri, while the letters mo (in moveri) point to modo: cp. ix. 1. 7 where A has quomo for quomodo, and xi. 3. 97 where b has homo for hoc modo. In the margin of Bodl. and Dorv. (both which have moveri) I find quodammodo vero.

est enim, H, which (like G) has est also after solutum. Halm adopts Osann’s conjecture etenim: Kiderlin suggests ea enim or ista enim, which may be right. Becher defends the double est (GH), comparing ix. 3. 7 quod minus mirum est, quia in natura verborum est, and i. 3. 14 (reading servile est et ... iniuria est).

poetis, H, following b: poesi Spald. ‘recte ut videtur,’ Halm.

§33. adde quod, Regius followed by Meister and Krüger (3rd ed.). audeo quia GH; audio quia L S Bodl. Ball. Harl. 2662, &c. Halm adopted Geel’s conj. ideoque: and the Bonn. Meister ed. reads adeo. Becher proposes quid? quod: Kiderlin id eo magis (fortius) dicere audeo. The last conj. revives what I find is the reading of some old edd. (e.g. ed. Col. 1527 and Riccius 1570) quod dicere fortius audeo quia, except that from id eo the eye might pass more easily to audeo.

opus, accepted from Spalding (who conjectured it independently) by Halm and Meister, already appears in ed. Col. 1527 and in that of Riccius 1570.

§34. rerum exemplorumque. Kiderlin suspects a lacuna after rerum and suggests ex cognitione rerum enim venit copia exemplorum. His argument is that while ‘ex cognitione rerum’ might serve as a sort of explanation of ‘ex historiis,’ ‘exemplorumque’ must also be accounted for, and that after ‘locum’ we expect to hear what advantage is derived from historical literature, not from what that advantage arises. The omission by a copyist of enim venit copia explains how exemplorum comes to be joined with rerum: cp. xii. 4. 1 in primis vero abundare debet orator exemplorum copia cum veterum tum etiam novorum, and esp. ii. 4. 20 et multa inde cognitio rerum venit exemplisque, quae sunt in omni genere potentissima, iam tum instruit, cum res poscet, usurum. For ne omnia (Badius and Vall.²) the codd. give nec omnia, which Becher prefers.

§35. vitio factum est oratorum. G gives est orum with al. oratorum written in above by the hand which Halm calls b. H (with FTLS Bodl.) gives est alia oratorum,—one of many strong indications that it was copied from G: for alia some MSS. give alias. Halm (ii. p. 369) thinks that orum in G may have stood for rhetorum.

quae sunt istis. GHLS and Vall. all give sint. But iniusta, inhonesta, inutilia are as definite as their contraries.

Stoici supplied by Meister, whom Krüger follows. Kiderlin would place it after maxime, just as Socratici stands after optime. Perhaps Stoici and Socratici are both glosses. Quint. may simply be saying that philosophical reading improves the matter of oratory (de iustis, &c.) and also the form (by altercationes and interrogationes). Stoici looks appropriate to de rebus divinis (see note): and argumentantur acriter is quite in place as referring to the Stoic logic, renowned for its acuteness (Zeller, Epic. & Stoics, p. 118): but on the other hand interrogationibus would be as apt in regard to them as to the Socratics. Cp. de Or. i. §43 Stoici vero nostri disputationum suarum atque interrogationum laqueis te inretitum tenerent.

On the alternative explanation of the passage mentioned in the note, altercationibus and interrogationibus are taken as datives (as often in Quint. after praeparo), referring to two well-understood parts of the duty of a counsel in an action-at-law. As regards the altercatio indeed, previous writers on rhetoric had not stated any special rules for its conduct, probably (as Quint., in his treatment of the subject, suggests vi. 4. 1) because it was sufficiently covered by precepts of a more general kind. In a court-of-law, the altercatio was a discussion carried on between opposing advocates in the way of short answers or retorts: it followed (when resorted to) the examination of the witnesses, which was in Roman usage preceded by the main speeches for the prosecution and defence, embracing all the facts of the case (Cic. in Verr. i. 1 §55). Cp. Cic. Brut. §159 iam in altercando (Crassus) invenit parem neminem.—See Poiret, L’éloquence judiciaire à Rome pp. 212-216.

§37. qui sint legendi. Halm, Meister: GHL and all MSS. qui sint. Legendi appears in ed. Col. 1527, and I have found it also inserted by a later hand above the line in the Bodleian codex. It may have fallen out because of legendo above, and Spalding is probably right in regarding it as indispensable. There seems however no reason for eliminating the asyndeton by reading et quae (with Meister) or quaeque (Halm). Kiderlin (Hermes, 23, 1888 p, 160) suggests that the original may have run qui sint qui prosint: cp. 2 §14 tum in ipsis quos elegerimus quid sit ad quod nos efficiendum comparemus: xii. 2. 4 quid sit quod memoriam faciat. This suits the context, cum tantum utilitatis in legendo iudicemus, and §40 paucos enim ... utilitatis aliquid. Cp. ii. 5. 20 nec prodesse tantum sed etiam amari potest (Cicero).

§38. [quibuscum vivebat] is bracketed by Krüger (3rd ed.), as it had already been by Frotscher and Herbst. This reading first appears in the Aldine edition: the only MS. in which I have been able to find any trace of it is Burn. 243, where quibuscum convivebat is inserted as a correction. Some have refused to recognise it as a gloss, in spite of the uncertainty of the MSS., and have sought to interpret it ‘with whom he lived in close, familiar intercourse’ (opp. to quos viderim §§98, 118): cp. Cic. de Off. i. §143 quibuscum vivimus, ib. §46. But in Brut. §231 Cicero distinctly says in hoc sermone nostro statui neminem eorum qui viverent nominare, whence Jeep was led to conj. qui quidem viverent: Hortensius, for example, was ‘aetatis suae,’ but had died four years before the date of the Brutus. So Geel conjectured qui tum vivebant (a reading which however I find in the ed. Col. 1527 and Riccius 1570): Törnebladh qui quidem tum vivebant, Wrobel qui tunc vigebant (cp. §122), Zambaldi ut quisque tum vivebat, and Kiderlin qui quidem nondum e vita excesserant; see Rhein. Mus. 46, p. 23. Andresen proposed to read qui quidem sescenti erant.

G (and practically H) gives quidqui convivebit. FT part company with H, the former reading quod quid convivabit, the latter quidque contuuebit (man. sec. quod quisque contuebat). Many MSS. (e.g. Bodl. Ball. Harl. 2662, 4995 LS) have quid quisque convivebat (convivabit L). The Carcassonensis gives quid quod convivabit.

persequamur [et philosophos]. Persequamur is a conj. of Regius adopted by Meister: all MSS. give et Graecos omnes et philosophos (philosophis HFT). In Harl. 4995 (which is dated A.D. 1470) I have however found et philosophos exequar: and so (Becher) a later hand in Vall. The reading of the ed. Col. 1527 is Graecos omnes et philosophos et poetas persequi velim.

Schmidt, followed by Halm, rejected et philosophos as a gloss, as both here and in the next sentence Quint. is evidently speaking of orators only. Certainly, if it stood, we should expect the poets and historians to come in also. Accordingly Claussen (Quaest. Quint. p. 335) suspected a lacuna consisting both of the finite verb and the poets and historians: Krüger (3rd ed.) adopts his conjecture and reads si et illos et qui postea fuerunt et Graecos omnes persequamur et poetas et historicos et philosophos? He cps. 1 §25 nam si, quantum de quaque re dici potest, persequamur, finis operis non reperietur: v. 10. 91: viii. 5. 25. So Andresen (Rhein. Mus. 30, p. 520), except that he omits ‘persequamur,’ and proposes to read above de Romanis tantum et oratoribus for et in sense of ‘and that’: cp. §§51, 94. Gertz suggests et Graecos omnes persequi velis nec oratores tantum, sed etiam poetas et historicos et philosophos. Kiderlin (Berl. Jahr. xiv. 1888, p. 62 sq.) prefers persequamur because of iudicemus and adiungamus above. If the verb could be dispensed with, he would propose ‘et praeter hos oratores etiam omnes poetas et historicos et philosophos,’—arguing that et praeter hos and philosophos may have run together in the eye of the copyist and so caused the lacuna. For et philosophos Jeep suggested explico novos.

§39. fuit igitur, all codd.: fuerit, Regius. That the difficulty of the passage was felt by the early editors is obvious from this emendation, and also from the fact that in §40 the traditional reading has been non est tamen (for non est): sed non est, Spalding: at non est Osann.

Taking §§37-45 as they stand the sequence of thought seems to be this: ‘If I am asked to recommend individual writers I shall have to take refuge in some such utterance as that of Livy. His dictum was “read Demosthenes and Cicero first, and let others follow in the order of their resemblance to Demosthenes and Cicero.” Mine is that there is some good to be got out of almost every author,—except of course the utterly worthless. But (sed non quidquid, &c. §42) the particular object I have in view itself supplies a limitation for what would otherwise be an endless task (infiniti operis §37). My business is the formation of style. In regard to this matter there is a difference of opinion—a cleavage between the old school and the new (see esp. §43). This opens up the whole question of the various genera dicendi, a detailed examination of which I must postpone: for the present I shall take the various departments of literature (genera lectionum §45) and mention in connection therewith certain representative writers who may serve as models for the students of style ((iis) qui confirmare facultatem dicendi volent).’

This seems satisfactory enough, especially in the case of so loose a writer as Quintilian. §§39 and 40 are parallel, instead of being antithetical: §39 says ‘Livy’s prescription was the safest,’ while §40 gives a general utterance on the part of Quintilian. In each deliverance brevitas is meant to be the distinguishing characteristic of individual representatives of poetry, history, oratory, and philosophy.

In his Beiträge zur Heilung der Ueberlieferung in Quintilians Institutio Oratoria (Cassel, 1889), Dr. Heinrich Peters makes some very drastic proposals in regard to the sections under discussion. He fails to see any satisfactory connection between the purport of §§40-42 and that of §§37-39. And he thinks the statement of a summa iudicii in §40 is inconsistent with the special treatment of individual authors which begins at §46. On these and other grounds he proposes to transfer §§40-42 (down to accommodatum) to §44 and read: interim non est dissimulanda nostri quoque iudicii summa. Summa iudicii then furnishes the antithesis to disseram diligentius: nostri quoque iudicii receives additional point from the reference to conflicting views which immediately precede it: an explanation is gained of the emphasis laid in §§40-41 on the distinction between the veteres and the novi,—the later sections §§43-44 explain the preceding (§§40-42): and the transition from Livy’s dictum in §39 to verum antequam de singulis in §42 is natural and easy. Then Dr. Peters would propose to continue: quid sumat (for summatim, see below) et a qua lectione petere possit qui confirmare facultatem dicendi volet attingam. This gives a very satisfactory and even a necessary sequel, he thinks, to non quidquid ... accommodatum. Sections 40-42 are then addressed, not to the student of rhetoric, but to the disputants who quarrel over the comparative merits of the veteres and the novi: Quintilian says ‘something may be learned from everybody.’ Then he continues ‘for the formation of style a selection is necessary, and that I now proceed to make under the two heads of what the student is to appropriate and to whom he is to go for it.’

quae est apud Livium, &c. Schöll unnecessarily conjectured qua praecipit Livius (cp. ii. 5. 20) or qua apud Livium in ep. ad fil. praescribitur,—doubting if brevitas could have an acc. and infin. depending on it. But see note. G gives quae apud Livium epistula, in being inserted by the second hand, which H as usual follows.

§42. ad faciendam φράσιν. This is the reading now proposed by Kiderlin (in Hermes, vol. xxiii. p. 161), though φράσιν appeared as early as the edition of Riccius (1570). The following are the MSS. readings ad farisin G: ad faciendam etiam ad farisin H (affaresim S. Harl. 2662 Bodl. Ball. apharesim Harl. 4295) ad faciendam affarisin L. Meister adopts the vulgate, ad faciendam etiam phrasin: Halm reads ad phrasin.

The parallel passage in §87 clearly makes for faciendam. The probability is that ‘phrasin’ was originally written in Greek, as at viii. 1 §1: cp. ἕξις in §1: §59: 5 §1, where the MSS. vary between ex his, lexis, exitum, &c.: τροπικῶς §11. Cp. on §87. Two Paris MSS. (acc. to Zumpt) show ἀφέρεσιν. Etiam Kiderlin rejects: perhaps however the true reading may be protinus et ad faciendam φράσιν.

de singulis loquar, G man. 2 H L and Vall. Halm omits loquar, with G.

§44. tenuia atque quae. In a very interesting note (Programm des königlichen Gymnasiums zu Aurich, 1891, p. 8) Becher establishes the correctness of this reading, instead of the traditional tenuia et quae. The Vallensis has tenuia atque que (i.e. atque quae): for what may appear a cacophony, Becher compares i. 3. 8 atque ea quoque quae, Cic. de Leg. Agr. ii. 33. 90 atque qui. ‘That V (Vall.) has preserved the true reading is confirmed by the other codices: not only S, which gives tenia atque que, but also GL [and H], tenui atque, which is nothing else than tenui AtQUE, i.e. tenuia atque quae.’ In the Rh. Mus. xi. (‘zur Kritik der ciceronischen Briefe’ pp. 512-13) Buecheler says, ‘One of the commonest sources of corruption in the Florentine codex is that when two “consonant syllables” follow each other, one is omitted. The reason of this phenomenon is probably the fact that in the archetype of which this MS. is an indirect copy the sounds which were to be repeated were distinguished by letters of a larger size.’ Becher finds the same phenomenon in the manuscripts of Quintilian, and gives the following examples, selected at random from many others: §45 aliquos G(H)LSV, i.e. aliQUOS = aliquos quos: §54 reddit G(H)V, i.e. redDIt = reddidit (so cod. Almen.): §79 auditoris S (audituris G, also H), i.e. auditorIs = auditoriis (as Vall. M: also Ball. Dorv. Burn. 244 Harl. 4829, 4995): ibid. comparat GMS (and all my codd.) i.e. compARat = compararat: §84 probandoque G (and H) = probandoQUE: §89 etiam sit G (see Crit. Note ad loc.) = etiam SIt. Especially significant is ix. 4. 41 o fortunatam me consule Romam AGM, i.e. o fortuNATAM me consule Romam.—Becher finds a further ground for atque, as connecting ‘quae minimum ab usu cotidiano recedunt’ more closely than et, in the fact that already in Cicero tenuis is used of a person of the commoner sort, ‘unus de multis,’ de Leg. iii. 10. 24.

lenis ... generis. For lenis Krüger (3rd ed.) reads levis, adopting a conj. of Meyer (Halm ii. p. 369) for which cp. §52 (levitas verborum) and v. 12. 18 (levia ac nitida): supported by Becher Phil. Runds. iii. 14. 430. In this sense levis (λεῖος) is opp. to asper: cp. de Orat. iii. §171 struere verba sic ut neve asper eorum concursus neve hiulcus sit, sed quodam modo coagmentatus et levis: cp. §172: Orat. §20: Quint. ii. 5. 9 levis et quadrata compositio: de Orat. iii. §201 levitas coniunctionis: Brut. §96: de Opt. Gen. Or. §2: Quint. viii. 3. 6.

interim. H. Peters would prefer nunc (if the text stands as it is), comparing v. 11. 5; 14. 33: ix. 4. 19.

summatim quid et a qua. Kiderlin approves of Meister’s retention of the vulgate: petere must have an object. So Krüger, 3rd ed. The original reading in G is sumat et a qua, corrected to sumat quia et a qua, which occurs in HFTL. Bodl. Ball, and my other MSS. agree with S in reading summa for sumat. Even if the text stands (without his proposed inversion) H. Peters would prefer quid sumat et a qua, as nearer the MSS.

§45. paucos enim qui sunt eminentissimi. Meister and Krüger 3rd ed. have paucos (sunt enim em.) =‘nur wenige’: cp. hos (sc. tantum) §91. Halm reads paucos enim (sunt autem em.) GH give paucos enim sunt em. L and the British Museum MSS. all read paucos sunt enim. The text is that of ed. Col. 1527 adopted by Zambaldi, and approved by Kiderlin: cp. §101 qui sunt dulciores: ix 4. 37 quae sunt asperiores. Osann proposed paucos enim, sunt enim.

his simillimi, Halm, supported by Becher, who compares §39: his similes Meister and Krüger (3rd ed.). G has hi similibus, corrected by the same hand to simillimis: H gives his simillimis: all the other MSS. his simillimi.

plures is the common reading, and occurs in Harl. 4995, and also Vall. (Becher). GHFT give plurimis: LS and the later MSS. generally plurimos. Kiderlin proposes pluris iis as being nearer plurimis. The pronoun, he argues, is not superfluous, because Quintilian is distinguishing between ‘qui confirmare fac. dic. volent’ (i.e. those who have finished their rhetorical studies and want practice) and the ‘studiosi’ (young men busy with theory). The latter will read more authors than those for whom this book is intended, its aim being (§4) to instruct the young orator (after the stage of theory) how best and most readily to use what he has acquired.—For aliquos quos see on tenuia atque quae §44 above.

qui a me nominabuntur, ed. Col. 1527; GH have quia nom.: Vall. LS qui nom. Hertz rejects a me, and he may be right.

§46. omnium fluminum. GHL Bodl. annium: S Harl. 2662, 4950, Ball. amnium vim. Halm, following Osann, read omnium amnium: but though omnium is necessary (cp. πάντες ποταμοί Il. 21. 196), Quintilian would surely have avoided such a cacophony as omnium amnium. Wölfflin conjectured omnium fluminum (Rhein. Mus. 42, Pt. 1, 1887, p. 144), and this is now accepted by Meister (vol. ii. p. 362 and Pref. to Book x, p. xiii). Wölfflin supposes that the archetype had omnium fontiumque, fluminum having fallen out: omnium was then corrected into amnium. Amnis however is rare, and fluminum not only secures an apt alliteration, but is constantly found: cp. §78 puro fonti quam magno flumini propior: viii. 3. 76 magnorum fluminum navigabiles fontes: Lucr. iv. 1024: v. 261, 945 (‘fluvii fontesque’): Ovid Met. i. 334.

§47. ac consiliorum L: hac con. G: et con. Prat. Put. atque con. 7231, 7696.

§48. operis sui ingressu: operis si ingressus GH: operis sui Bodl.: operis Prat. Put. S Harl. 2662, 4995, 4950, Dorv. Ball. Badius conj. ingressu, and Halm added in, which is however unnecessary: cp. iv. 1. 34 operum suorum principiis: iv. pr. 4 initiis operum suorum. Becher keeps ingressus, but makes it a genitive dependent on versibus.

Two Oxford MSS (Bodl. and Dorvilianus) give nam for non, and in the former case the nam looks very like viam. It is possible that viam may be the true reading: cp. ii. 10. 1 quarum (materiarum) antequam viam ingredior ... pauca dicenda sunt,—though there the phrase refers to entering on the regular treatment of a subject. Age vero is not always found with questions, Hand Turs. i. p. 211. Without non, the reading may possibly be age vero viam utriusque operis ingressus, in paucissimis, &c. The si after operis may have arisen from operi s ingressus. The MSS. are unanimous for ingressus, and the awkwardness of operis sui ingressu in pauc. vers. makes it very probable that something is wrong. Utrumque opus ingressus would have been more natural: viam utriusque operis ingressus is not far off it. Perhaps however it would be preferable to keep the question and read nonne viam ut. op. ingressus.

nam benevolum. nam et ben, Put. 7231, 7696: so too the Carcassonensis.

§49. ceteraque genera. GHL and the Brit. Mus. MSS. give ceteraque quae: so too Bodl. and Ball. Genera was conjectured by Caesar (Philol. xiii. p. 757). Schöll (in Krüger 3rd ed.) proposes ceteraeque viae ... multae: Kiderlin ceteraque, quae probandi ac refutandi sunt, nonne sunt ita multa ut ... petant? For quae ... sunt he compares §106 omnia denique quae sunt inventionis.

§50. ut magni sit. G Burn. 243: Ball.: Bodl.: sint H: ut magni sit viri Prat. Put. 7231, 7696, S, Harl. 2662, 4995, 4950, 4829, Dorv., Burn. 244 (sint L): ut magnum sit, Gensler: ut magni sit spiritus, Kiderlin (cp. i. 9. 6).

§51. et in omni: et om. Prat. and Put.

clarissima LS and most codd.: durissima GHT Prat. Put. 7231, 7696, Dorv.

§52. utiles circa praecepta, &c. Kraffert proposed utilis circa praecepta sententiasque levitas verborum ... With praecepta may there not have been a genitive in the original text: utilis circa praecepta sapientiae (pr. §19: i. 4. 4: xii. 1. 28), or perhaps utiles circa morum praecepta sententiae (xii. ii. 9)?

§53. secundum Prat. Put. 7231, 7696, Vall. LS Harl. 2662, 4995 Dorv. Ball.: om. GHFT Bodl. Halm, following Hertz, gives parem (cp. §127 pares ac saltem proximo): aequalem would be as probable, and is given by some MSS. in §55. Schöll now thinks secundum an old interpolation, and conjectures quam sit aliud atque aliud proximum esse, cp. i. 7. 2: ix. 4. 90.

§54. poetarum iudices Prat. Put. 7231, 7696, LS Ball. iudicium G, iuditium H. Halm suspected it to be a gloss introduced from the margin (cp. laus Ciceronis §109) and Mayor removed it from the text.

reddidit cod. Almen.: reddit GHFT Vall. Harl. 4995 Bodl. Burn. 243. Edidit is given in Prat. Put. 7231, 7696 Harl. 2662, 4950, 4829 Dorv. and Ball., besides L and S.

sufficit MSS.: Halm would prefer suffecit (cp. §123). For parem many MSS. give equalem, which must have been a gloss: S has equalem credidit parem, and so Prat. (Fierville Introd. p. lxxix) Harl. 2662 (A.D. 1434) and 11671 (A.D. 1467).

§56. Macer atque Vergilius. Unger suggested Valgius for Vergilius. This is however unnecessary, though it has been proposed to insert the comma after Vergilius instead of after idem below.

§59. adsequimur GHS Prat. Put. 7231, 7696 Bodl. Ball. Dorv. and British Mus. MSS. (except 4950 which gives C and L’s assequatur and 4829 which has assecuntur). Halm reads adsequamur, and is followed by Meister. Krüger (3rd ed.) proposes ut adsequamur.

§60. quibusdam quod quoquam minor est. GH give quibus for quibusdam: Prat. Put. S and all my MSS. have quibusdam quod quidem minor est: (minoris Bodl. Burn. 243): quod quodam 7696. Wölfflin (Rhein. Mus. xlii. Pt. 2, p. 310) proposes quod idem amarior est: amarus (§117) indicates the excess of acerbitas (§96) which might be alleged against Archilochus for his lampoons on Lycambes. Cp. iamborum amaritudinem Tac. Dial. 10. But quoquam (Madv. 494 b) does not necessarily imply that there is any one superior to the great Archilochus, though, outside the range of iambographi, Homer is always present (§65) to the writer’s mind. Quoquam is not to be restricted to the narrow circle of iambic writers, otherwise materiae would have no point. Quintilian means that Archilochus must be ranked immediately after Homer, if indeed the disadvantage of his subject-matter forbids us to place him alongside of Homer. That he had a schoolmaster’s liking for an ‘order of merit’ is shown by §§53, 62, 85, 86.

§61. spiritu, magnificentia, Put. 7696 S Harl. 2662, 4995, 11671, Dorv.: spiritus H (sps.) Prat. 7231 Harl. 4950 Burn. 243 Bodl. Ball., and so Halm and Meister. The strongest argument for the abl. is that the nouns go together in pairs,—spiritu magnificentia, sententiis figuris, copia ... flumine. So Claussen (Quaest. Quint. p. 334), who compares Dion. Hal. ἀρχ. κρ. 2. 5, p. 420 R ζηλωτὸς δὲ καὶ Πίνδαρος ὀνομάτων καὶ νοημάτων εἵνεκα, καὶ μεγαλοπρεπείας καὶ τόνου, καὶ περιουσίας .... καὶ σχηματισμῶν.

§62. Stesichorum Badius: iste sichorus GH: Stesichorus Bodl. 7696: Stesicorus Harl. 4995: other MSS. Terpsichorus or Terpsicorus.

§63. magnificus et diligens et plerumque oratori similis: GH magnificus et dicendi et plerumque orationis similis; so Burn. 243 and Bodl. (orationi); most other MSS. et diligens plurimusque (plurimum or plurimumque) Homero similis: plurimumque oratio, Prat. Put.: plerumque orationis 7231, 7696. Halm gives dicendi vi, which, after in eloquendo, would be strange. Wölfflin proposes elegans et (for dicendi et, diligens et): cp. §§78, 83, 87, 93, 114, and Dion. Hal. l.c. Ἀλκαίου δὲ σκόπει τὸ μεγαλοφυὲς καὶ βραχὺ καὶ ἡδὺ μετὰ δεινότητος ... καὶ πρὸ πάντων τὸ τῶν πολιτικῶν πραγμάτων ἦθος. Halm’s dicendi vi rested on μετὰ δεινότητος, but we need not suppose that Quintilian translated word for word from Dionysiusin eloquendo, diligens seems quite appropriate: i. §3 cum sit in eloquendo positum oratoris officium.

Sed et lusit, Prat. Put. Voss. 1 and 3: sed et eius sit GH: sed in lusus MS Ball. Dorv.: sed editus sit Bodl.

§64. eius operis: ei GH: eius M Bodl. Burn. 243: eiusdem Prat. Put. 7231, 7696 S, Harl. 2662, 4995, 4950, 4829, Burn. 244, Dorv., Ball. In Prat. and Put. the order is in hac parte omnibus eum eiusdem operis.

§65. est et in. The MSS. give etsi est: Wölfflin conjectured est et, and Halm, (following some old edd.) inserted in, comparing §§64 and 68. So too Meister. Etsi may have crept into the text to anticipate tamen (ii. 5. 19): or the true reading may be est et etsi in. Schöll suggests (Krüger, 3rd ed. p. 92) that the passage ought to run as follows:—ant. com. cum sincera illa sermonis Attici gratia prope sola retinet vim (dum G, tum vulg.) fac. libertatis, et si est in insect. vitiis praecip., plur. tamen, &c.

nescio an ulla. This is the reading of Prat. Put. 7231, 7696, M, S, Harl. 2662, 4995, 4950, 11671, Dorv. Ball., and if it can be sustained, the sense it gives is quite satisfactory. We must suppose that poesis (probably the only fem. noun that would suit) was present in the writer’s mind: see on poeticam §28 above.

But in Quint. poesis occurs only once (cp. on §28),—at xii. 11. 26, where it is not used of a special branch of poetry, as here; and even there a doubt has been expressed about the reading. Kiderlin therefore urges (Hermes 23, p. 163) that it is incredible that Quintilian would have left his readers to supply for themselves a word which he uses only once, if at all: ullum genus would surely have occurred to him, as both genus and opus are constantly used to denote departments of literature. Again the text gives post not praeter Homerum. Founding on the reading an illa (GHFT Burn. 243 Bodl.) Kiderlin therefore suggests an illa poeta ullo post &c.: ‘und ich weiss nicht, ob nicht jene mehr als irgend ein Dichter (nach Homer jedoch, &c.).’ The copyist would easily wander from poet. to post, and it is not unusual to compare old comedy &c. with the poets and not their works (cp. similior oratoribus: historia proxima poetis est §31: at non historia cesserit Graecis §101); especially as here post Homerum follows at once. For ullo cp. §60 quod quoquam minor est. An alternative emendation would be poesi ulla.

The aut ... aut immediately below is very much against this conjecture, which however Krüger (3rd ed.) has received into the text: we should expect rather nescio an illa quisquam, or nullus poeta, or keeping illa as nominative nescio an illa poeta ullo. Quintilian’s use of nescio an (like that of post-Augustan writers generally) is vague: it is usually an expression of doubt, the an meaning either ‘whether,’ or ‘whether not’ indifferently. Cp. ix. 4. 1: vi. 3. 6: viii. 6. 22: xii. 10. 2: i. 7. 24. (Mayor cites also Plin. Ep. i. 14. 9: iii. 1. 1: iv. 2. 1: v. 3. 7: vi. 21. 3: vii. 10. 3: 19. 4: viii. 16. 3: ix. 2. 5; and adds ‘In all these instances nescio an (dubito an) is ‘I doubt whether’; in Cicero the meaning is always ‘I rather think.’’) Andresen proposed nescio an ulla poeseos pars. The passage closely resembles §28, and must be emended on the same lines.

§66. tragoedias. Thurot (Revue de Phil. 1880, iv. 1, p. 24) conjectured tragoediam: cp. §67 hoc opus. He is followed by Dosson, against all MS. authority. Becher points out that we must supply with hoc opus in §67 the words ‘tragoedias in lucem proferendi,’ so that opus and tragoedias square well enough with each other.

§68. quod ipsum reprehendunt, Meister, Krüger (3rd ed.) and Becher. This reading also occurs in the Codex Dorvilianus. Other readings are quod ipsum quod GHT Burn. 243, Bodl.: quo ipsum MS Harl. 2662, 4995, 4950, Ball. Halm conjectured quem ipsum quoque, and was followed by Mayor and Hild. But as no fault has been found with Euripides in the foregoing, quoque seems out of place.

Founding on the reading of GHT, &c., also on that of F (which gives quod ipsum qui) Kiderlin (Hermes 23, p. 165) proposes to read quod ipsum quidam, comparing §98, where for quem senes quem (GT) Spalding rightly conjectured quem senes quidem, and 7, §21, where Bn, Bg give quod for quosdam. He then goes on, in an interesting paper, to reconstruct the whole passage, which is open to suspicion, especially in respect that sublimior stands as predicate with gravitas and cothurnus, as well as with sonus. The admirers of Sophocles consider his elevation of tone more appropriate than the strain of Euripides. Sublimior is therefore perhaps not the predicate of the sentence, however suitable it may be as the attribute of sonus. The predicate may have dropped out, and sublimior may have been transferred from its real place to supply it. It is striking that GFTM (also H and Bodl.) all give sublimior erit. Kiderlin imagines that a copyist who missed the predicate wrote in the margin ‘sublimior erit ponendum post esse’: and then another inserted sublimior erit after esse in the text. For the predicate, magis accommodatus might stand: in copying, the eye may have wandered from magis accommodatus to magis accedit: for magis accomm. cp. ii. 5. 18 and x. 1. 79. Kiderlin therefore boldly proposes to make the parenthesis run, ‘quod ipsum quidam reprehendunt quibus gravitas et cothurnus et sublimior sonus Sophocli videtur esse magis accommodatus’: ‘was gerade manche tadeln, welchen das Würdevolle, der Kothurnus, und der erhabenere Ton des Sophokles angemessener zu sein scheint.’

et dicendo ac respondendo 7231, 7696: dicendo ac respondo GH: in dicendo et in respondendo Prat. Put. S (et respondendo M).

praecipuus. Hunc admiratus maxime est. This is Meister’s reading, except that for eum I give (with Prat. Put. 7231, 7696 Harl. 2662 and 4995) hunc, which is commoner in Quint. at the beginning of a sentence (§§46, 78, 91, 112). The following are the readings of the MSS.: GH praecipuus et admiratus miratus: M Bodl. Harl. 4950, 4829, Burn. 244, C, Burn. 243 Ball. Dorv. praecipuus et admirandus: S praecipuum. Nunc admiratus et: Prat. Put. Harl. 2262 and 11671 praecipuus hunc admiratus et maxime est ut saepe test. et sec. quamvis: Harl. 4995, hunc admiratus max. ut s. test. et eum secutus quamquam. Halm gives praecipuus est. Admiratus maxime est: Kiderlin insists on the est after praecipuus, to correspond with accedit, though it seems better to take all that comes after accedit as an explanation of the statement magis accedit oratorio generi: he also retains the et of most MSS. and reads praecipuus est. hunc et admiratus (Blätter f. d. bayer. Gymn. 24, p. 84). Wölfflin (partly followed by Krüger 3rd ed.) proposed a more radical change (Rhein. Mus. 1887, 2 H. p. 313) praecipuus. Hunc imitatus, quoting in support of the conjunction imitatus ... secutus §122, eos iuvenum imitatur et sequitur industria: 5 §19, deligat quem sequatur, quem imitetur: Ovid, Fasti v. 157, ne non imitata maritum esset et ex omni parte secuta virum. But Kiderlin (l.c.) aptly remarks that if Quintilian had written imitatus, he would not have said ut saepe testatur but ut ex multis locis patet (apparet, videmus): while vii. 4. 17 (on which Wölfflin relies) is not really to the point. Moreover Quintilian, would never have separated such synonyms as imitatus and secutus by ut saepe testatur.

Charisi nomini addicuntur, Frotscher: Charis in homine adductura GH: Charisii nomine eduntur Prat. Put. 7231, 7696 Harl. 2662 Dorv.

§70. aut illa iudicia Prat. Put. 7231, 7696 Harl. 4995. GH Harl. 4950 give aut illa mala iudicia: Bodl. Burn. 243 aut alia mala iud. S Harl. 2662 Dorv. and Ball. aut alia iudicia. The edd., following Gesner, have generally given (with Harl. 4950) aut illa mala iudicia (so Halm and Meister), and have taken mala as predicate, though the order of the words makes that impossible. Becher approves of Andresen’s deletion of mala. Krüger (3rd ed.) prints mala [illa] iudicia, thinking that illa arose by dittography, and that then the order was changed in the codd. to illa mala iudicia. Kiderlin (in Hermes 23) gives as an alternative to deleting mala the conjecture illa simulata iudicia (‘jene erdichteten nachgemachten Gerichtsverhandlungen’; cp. xi. 1. 56: cum etiam hoc genus simulari litium soleat). A similar mutilation occurs, e.g., xi. 1. 20, where b gives secum M secus instead of consecutum.

§71. filiorum militum, most codd.: filiorum maritorum militum Prat. Put. 7231, 7696 S.

§72. si cum venia leguntur. The reading of the MSS. is upheld by Iwan Müller, Meister, and Kiderlin. Spalding suggested cum verecundia: Schöll cum iudicio: Becher cum ingenio. Becher points out (Bursians Jahresb. 1887) that the expression is meant to cover decerpere as well as legere, and decerpere indicates careful and intelligent reading (cp. §69, diligenter lectus): cum ingenio = ‘mit Verstand’: cp. Cic. ad Fam. xiii. 10. 2 quod versabatur in hoc studio nostro .. et cum ingenio .. nec sine industria: Ulp. Dig. 1. 16. 9 patientem esse proconsulem oportet, sed cum ingenio, ne contemptibilis videatur. Finally, Krüger (3rd ed.) proposes cum acumine or cum vigilantia (cp. v. 7. 10).—Prat. Put. 7231, 7696 S Harl. 2662 all give Osann’s conjecture legantur.

prave GH Harl. 4995, 4950 Burn. 243 Bodl.: pravis Regius, Halm, Meister, Becher draws attention to the parallelism between the clauses: ut prave praelatus est sui temporis iudiciis, ita merito creditur (= meruit credi) secundus consensu omnium.

§76. nec quod desit ... nec quod redundet: H Burn. 243 and Bodl. give quod .. quod: Prat. Put. MS Harl. 2662, 4995, 4950, Burn. 244, Dorv. C, and Ball, quid .. quid. The latter reading is supported by Becher (Phil. Rund. iii. 434). For quod cp. xii. 10. 46: (xii. 1. 20 where for quod adhuc BM give quid adhuc): on the other hand, in vi. 3. 5 the MSS. are in favour of quid, though Halm reads quod (followed by Meister). For quid cp. Cic. pro Quint. §41, neque praeterea quid possis dicere invenio.

§77. grandiori similis. So all MSS.: Halm and Meister. Several conjectural emendations have been put forward. Comparing 2 §16 (fiunt pro grandibus tumidi), Becher suggests grandi oratori,—an easy change, if the copyist used contractions, but without point: above in §74, ‘oratori magis similis’ is appropriate enough in speaking of historians, but ‘oratori’ would be inappropriate here. This is accepted, however, by Hirt (Berl. Jahr. ix., 1883, p. 312; cp. P. Hirt, Subst. des Adjectivums, p. 12). Schöll proposes to read gladiatori similis, in view of the close connection with what follows, strictus ... carnis ... lacertorum: but plenior and magis fusus are a bad introduction to gladiatori, and if Aeschines had plus carnis and minus lacertorum, he cannot really have resembled a gladiator. This reading is, however, adopted by Krüger (3rd ed.). Finally, Kiderlin (Hermes 23, p. 166 sq.) has conjectured et grandi (or grandiori) organo similis, and applies the figure throughout: ‘voller und breiter lässt Aeschines den Ton hervorströmen, einem grossen Musikinstrumente gleich’: ‘einer Orgel gleich,’—he is grandisonus. The translation appears to limit unnecessarily the meaning of plenus and fusus: though the former is used of tone i. 11. 6 (cp. xi. 3. 15 of the voice: ib. §§42, 62: and §55 of the breath): while fusus is used of the voice xi. 3. 64. For such a use of grandis cp. §58 (cenae): §88 (robora): xi. 2. 12 (convivium): 3. 15 (vox): 68 (speculum): and for organum, i. 10. 25: ix. 4. 10: xi. 3. 20 (where there is a comparison between the throat and a musical instrument): probably also i. 2. 30. There is an antithesis in the two parts of the sentence between fulness and breadth, on the one hand, and real strength on the other; and for the transition to the second figure Kiderlin compares §33.

§78. nihil enim est inane: perhaps ‘nihil enim est in eo inane’ (Becher), or nihil enim inest.

§79. honesti studiosus. Becher’s proposal to alter the punctuation of this passage is discussed in the note ad loc.—For auditoriis and compararat, see on tenuia atque quae §44, above.

§80. quem tamen. Kiderlin, in Hermes (23, p. 168), raises a difficulty here. Tamen shows that the clause cannot go with the main statement (fateor), and its position forbids us to take it with the quamquam is primum clause: it can only go with quod ultimus est, &c., ‘though Demosthenes is ultimus fere, &c., yet Cicero, &c.’ To prevent so awkward a joining of the clauses, Kiderlin proposes to read eumque tamen: pointing out that the quae of the MSS. (GH) may have arisen out of que, and that Quintilian may have written eumque; cp. vi. 2. 13, where Halm makes utque out of quae (G), and xi. 2. 32, where Meister reads estque. The meaning will then be: Demetrius is worthy of record as being about the last, &c., and yet Cicero gives him the first place in the medium genus.—It seems better, however, to give tamen a general reference: ‘yet, in spite of all that can be said on the other side’ (e.g., inclinasse eloquentiam dicitur). Cp. §99 quae tamen sunt in hoc genere elegantissima.

§81. prosam (prorsam) orationem et all MSS.; Halm, Meister, Krüger (3rd ed.) omit et. I find that Becher supports the view stated in the note ad loc.: he would however write prorsam, which the best MSS. give also in Plin. v. 31, 112 D.

quodam Delphici videatur oraculo dei instinctus: so Frotscher, followed by Krüger (3rd ed.). On the other hand Claussen (Quaest. Quint., p. 356) and Wölfflin (followed now by Meister, pref. to ed. of Book x., p. 13) propose to delete Delphici, of which Becher also approves. But the MS. evidence cannot be disregarded. The following are the various readings: GH quaedam Delphico videatur oraculo de instrictus, and so FT, the former giving also (by a later hand) de instinctus, the latter dei instructus. Bodl. gives quodam delphico videatur oraculo dei instructus. The most frequent reading is that of Prat. Put. 7231, 7696 Harl. 2662, 4995, 4829, 11671, Ball. and most edd., quodam delphico videatur oraculo instinctus: S agrees, but is reported to have delphico after oraculo: Harl. 4950 and Burn. 244 have the same reading, with institutus corr. to instinctus: Burn. 243 gives instructus. Delphico was originally deleted by Caesar: Phil xiii, p. 758. Halm read tamquam Delphico videatur oraculo instinctus: but Quintilian would take no trouble to avoid the repetition of quidam (cp. divina quadam, above).—For the arrangement of words, Krüger (3rd ed.) compares §41 qui ne minima quidem alicuius certe fiducia partis memoriam posteritatis speraverit.

§82. quandam persuadendi deam. Nettleship (Journ. of Philol., xxix, p. 22) conjectures Suadam [persuadendi deam], comparing Brutus, §59, quoted ad loc. Persuadendi deam would thus become a gloss on Suadam: but the expression in the text is quite in Quintilian’s style.

§83. eloquendi suavitate: eloquendi usus (or usu) suav. GH and all codd. except Harl. 4950, and Dorv., both of which give simply eloq. suav. Halm admitted into his text Geel’s conj. for usus, ‘eloquendi vi ac suavitate,’ and this has met with some acceptance (Iwan Müller and Becher). But the parallel from Dion. Hal., Ἀρχ. κρ. 4 is hardly conclusive: τῆς τε περὶ ἑρμηνείαν δεινότητος ... καὶ τοῦ ἡδέος. Hirt properly remarks that the agreement between the two is not so great as to allow of correcting the one by the other. Kiderlin conjectures eloquendi vi, suavitate, perspicuitate.

tam est loquendi. See note ad loc. for Kiderlin’s conj. tam manifestus est. Though Meister’s tam est eloquendi is probably a misprint, it is found in some MSS.—Harl. 4950: Burn. 244.

§84. sane non affectaverunt. Bodl. and Vall. (veru subpunctuated in the latter: affectant Prat. Put. 7231 MS Ball. Dorv. Harl. 2662, 4995, 4829, 11671: sene non adfectitacuerunt GH Burn. 243: adfectarunt 7696: adfectitant Harl. 4950, and so Burn. 244 (corrected from affectant).

§85. haud dubie proximus. Halm inserted ei after dubie, though it is not found in any MS.: Regius had suggested illi. Kiderlin (Hermes 23, p. 170) points out that if propiores alii in §88 is allowed to stand without a dative, ei is not necessary here. He suggests, however, illi before alii in §88: both passages must be dealt with in the same way.—For haud (Vall.), GHS have aut: M haut. Cp. on 3 §26.

§86. ut illi ... cesserimus: cum illi GHFT Harl. 4995 Burn. 243: ut illi Prat. Put. 7231, 7696: and so S Harl. 4950 (with caelesti atque divinae): ut ille M Harl. 2662. Kiderlin (Hermes, p. 170) proposes to go back to the reading of the older MSS. cum illi, and instead of cesserimus to read cesserit, so as to make Vergil the subject throughout. Cum cannot, he contends, be a copyist’s error, motived by ita; and it is probable, therefore, that at first cesserit a was inadvertently written for cesserit; then (in G or some older MS.) cesserimus ita was made out of that, to correspond with vincimur below: and then in the later MSS. cum was changed to ut, because of ita. For the transition, with this reading, from cesserit to the plural (vincimur, pensamus), he compares §107, where, after speaking of Demosthenes and Cicero, Quintilian passes to vincimus.

§87. sequentur MS Halm and Meister: sequenter G seq̅nt’ H: sequuntur Prat. Put. 7231, 7696.

φράσιν id est. These words are omitted in the Pratensis, which is Étienne de Rouen’s abridgement of the Beccensis, now lost. This is an additional proof that φράσιν was originally written in Greek: cp. on §42.

§88. propiores H Prat. Put. Vall. Harl. 2662, 4495, 11671, Burn. 243. Bodl., Halm: propriores GMS 7231, 7696, Harl. 4950, C, Burn. 244, Dorv., Meister. In Cicero and Quintilian magis proprii would be more usual for the latter.

§89. etiam si sit. This conjecture of Spalding’s (for etiam sit GH Bodl. &c.: etiam si M Harl. 4950 Dorv.: etiam sic Prat. Put. S Harl. 2662) I have found in the Balliol codex. 7231 and 7696 give etiam si est. Cp. note on tenuia atque quae §44, above.

ut est dictum. These words were bracketed as a gloss by Halm, and are now omitted altogether by Krüger (3rd ed.): see however note ad loc. Döderlein proposed to place them after poeta melior, Fleckeisen after etiam si.

Serranum is Lange’s conjecture for ferrenum GHM: farrenum 7231, 7696 Harl. 2662, 11671: Pharrenum Prat. Put. Some MSS. (e.g. Vall. Harl. 4995, Burn. 243 and 244) give sed eum, but it is obvious that the criticism of Severus stopped with the word locum.

§90. senectute maturuit ed. Col. 1527 and so 7231, 7696 (Fierville): senectutem maturbit GH: senectute maturum Prat. Put. MS Harl. 2662, 4995, 4950, Burn. 244, Dorv. and Ball.: senectus maturavit Bodl., Burn. 243.

et, ut dicam. Halm’s sed instead of et has been rejected by later critics. Cp. Claussen (Quaest. Quint., p. 357 note): sed ‘sententiam efficit ab hac operis parte alienam. Nam cum oratori futuro exempla quaerantur oratoria virtus in quovis scriptore laudi vertitur (§§46, 63, 65, 67, 74, &c.). Itaque propter huius censurae consilium Quintilianus Lucani elocutionem oratoriam laudat, sed ingenium poeticum una reprehendit.’

§91. propius H Prat. Put. Burn. 243, Harl. 2662 and other codd.: Bodl. Ball. Harl. 4950 proprius. Reisig conjectured propitius, which also is apt; but in spite of industrius, necessarius, cited in its support (cp. iv. 2. 27: vii. 1. 12), it is too uncertain a form to be received into the text. Iwan Müller thinks it would have to be magis propitiae. Halm gives promptius: Wölfflin pronius: while Schöll now suggests propitiae potius (cp. iv. pr. §5: 2 §27: vii. 1. 12).

§92. feres G Prat. Put. 7231, 7696 S Harl. 2662, 4829, Dorv., Ball., Halm.: feras H, Harl. 4950, Burn. 243, Bodl. C and M, Meister and Krüger (3rd ed.). Harl. 4995 has fere: from Vall. Becher reports feras, ‘probably at first feres.’

elegea GH 7696, and so A² BN Put. S at i. 8. 6.

§94. abunde salis G Prat. Put. M and all my MSS. except H, Burn. 243, Bodl. which have abundantia salis.

multum est tersior. The variety of MS. readings seems to point to an et wrongly inserted after multum, perhaps from a confusion with ‘multum et ver gloriae’ below. GH give multum et est tersior: M Harl. 4950, Bodl. Ball. C Dorv. Burn. 243 and also Harl. 4829 multum etiam est t.: Prat. Put. 7231, 7696 S Harl. 2662, 11671 multum est tersior: while Harl. 4995 (and Vall.) has multo et est tersior. Osann proposed multo eo est tersior: Wölfflin multo est tersior: Halm and Meister print multum eo est tersior. For multum, cp. multum ante xii. 6. 1: and see Introd. p. li.

non labor GH Burn. 243 Bodl. and Meister: nisi labor 7231, 7696 S Harl. 2662, 4995, 4950, 4829, 11671, Burn. 244, Dorv. Ball. C, and Halm. Prat. and Put. have mihi labor.

hodieque et qui: H, Prat., Put., 7231, 7696, Harl. 2662, 4829, Bodl. Dorv.: hodie et qui Burn. 243: hodie quoque et qui Vall. Harl. 4995, 4950: hodie quod et qui S.—Becher is of opinion that the text will not bear the explanation given in the note, and would read hodie quoque et qui: ‘es giebt auch heute noch berühmte Satirendichter, die einst &c.’ Et qui he takes with clari, not with hodie quoque, the et being omitted in translation: clari (hodie quoque) qui (olim) nominabuntur.

§95. etiam prius. Founding on the classification given in Diomedes (see note ad loc.), according to which the satura of Pacuvius and Ennius preceded and was distinct from that of Lucilius, Horace, and Persius, Claussen (Quaest. Quint., p. 337) thinks that the true reading here may be Alterum illud et iam prius Ennio temptatum saturae genus, &c. For the satura of Ennius, cp. ix. 2. 36. Iwan Müller points out that Ennius is not mentioned below (§97), beside Attius and Pacuvius, probably because neither in tragedy nor in satire did Quintilian consider him to have produced anything helpful for the formation of an oratorical style. Other unnecessary conjectures are etiam posterius, Gesner: etiam proprium, Spald.: etiam amplius, L. Müller: etiam verius, Riese: alterum illud Lucilio prius sat. genus, Krüger (3rd ed.).

sola: solum Prat. and Put.

collaturus quam eloquentiae. These words, omitted in GHS Bodl. Burn. 243, occur in all my other codd.

§96. sed aliis quibusdam interpositus: sc. carminibus, Christ. In H the reading is quibusdam interpositus: so 7231, 7696 Bodl. and Burn. 243: but M Harl. 4950, 4829 Burn. 244 Dorv. and Ball, give a quibusdam interpositus: S cuiusdam: Prat. and Put. opus interpositus. Osann conjectures sed quibusdam, and so Hild. In the margin of Harl. 4995 is the variant aliquibus interpositis.

In Hermes, vol. 23, p. 172, Kiderlin makes a fresh conjecture. Recognising that something must have fallen out before quibusdam, but dissatisfied with Osann’s sed and Christ’s sed aliis, he proposes to read ut proprium opus, quibusdam aliis tamen carminibus (or versibus) a quibusdam interpositus. The eye of a copyist may easily, Kiderlin thinks, have wandered from the first to the second quibusdam: cp. v. 10. 64, ut quaedam a quibusdam utique non sunt, &c., and for quibusdam aliis xi. 3. 66, et quibusdam aliis corporis signis.

intervenit, which is a conjecture of Osann, I have found in Harl. 2662, 11671 Prat. Put. 7231, 7696.

lyricorum. Kiderlin thinks there may be something wrong in the text here. The last sentence (sed eum longe, &c.) shows clearly that Quintilian had a high opinion of the lyrists of his day: if Bassus was legi dignus, they were even more so. Would he then have said ‘of the Roman lyrists Horace is almost the only one worth reading’? Perhaps we should read lyricorum priorum: after -ricorum, priorum might easily fall out, and it gives a good antithesis to viventium. Bassus (quem nuper vidimus) forms the transition: and the next paragraph begins Tragoediae scriptores veterum, &c.

§97. clarissimi. This reading is stated by Halm to be ‘incerta auctoritate,’ and is referred by Meister to the Aldine edition. It occurs in Prat. 7231, 7696 Harl. 2662 (A.D. 1434) Vall. 4995, 4829, 11671, Dorv. and Ball.: Put. gives clarissime: G has gravissima: HFTS gravissimus, and so also Harl. 4950, Burn. 243, Bodl. and C. Halm prints grandissimi: Ribbeck (Röm. Trag. p. 337, 3) inclines to accept the sing. grandissimus, M, of Pacuvius alone.

Kiderlin (in Hermes 23, p. 173) rejects all the above readings. Gravissimus and gravissima are obviously due, he says, to gravitate following: but the word before gravitate must have begun with the same letter, and so clarissimi cannot stand, especially as it is inappropriate to the context. For ceterum shows that the sentence before it must have contained some slight censure: some defect, or quality excluding others equally good, must have been mentioned. He therefore conjectures grandes nimis, in preference to grandissimi, which in tragedy would hardly be a fault. Attius and Pacuvius, Quintilian says, are ‘zu grossartig, sie kümmern sich zu wenig um Zierlichkeit (Eleganz) und die letzte Feile (d.h. Sauberkeit im Kleinen); doch daran ist mehr ihre Zeit schuld als sie selbst.’ He evidently thinks more of the ‘Thyestes’ of Varius and Ovid’s Medea: cp. Tac. Dial. 12. With this judgment Kiderlin compares §§66, 67 tragoedias primus in lucem Aeschylus protulit, sublimis et gravis et grandiloquus saepe usque ad vitium, sed rudis in plerisque et incompositus ... sed longe clarius inlustraverunt hoc opus Sophocles atque Euripides, and is of opinion that the parallelism cannot be mistaken. For the position of nimis he compares ix. 4. 28 longae sunt nimis: v. 9. 14 longe nimium: xii. 11. 9 magna nimium.

§98. quem senes quidem parum tragicum. So Spalding, Bonnell, Halm, Meister, and Krüger. Quidem occurs in no MS.: GH have quem, M Vall., Harl. 4995, Burn. 244, Ball, omit it: Bodl. Burn. 243 and Dorv. show the corruption Pindarum. Becher would exclude quidem, regarding quem in G as an instance of the tendency of copyists inadvertently to repeat, after a particular word that by which it has been immediately preceded, e.g. §68 quod ipsum quod (G): ix. 4. 57 ut cum ut (G): iv. 1. 7 ipsis litigatoribus ipsis (b): iv. 2. 5 aut ante aut (bT): x. i. 4 iam opere iam (G).—But here the authority of the Pratensis and its cognates may be invoked. In the archetype from which they are derived something must have stood before parum, as Prat. Put. 7696, 7231 all give quem senes non parum tragicum: so Harl., 2662 (A.D. 1434), and 11671. Above in §96, G Prat. Put. 7231, 7696 have si quidem for si quem.

§100. linguae suae. So Köhler (v. Meister pref. to Book x. p. 13): suae supplies an antithesis to ‘sermo ipse Romanus’: GH give linguae quae: so Harl. 4950: S Burn. 243, Bodl. linguae: while Harl. 2662, 4995, 4829, 11671, Dorv. and Ball. omit it altogether: M has ligweque.

§101. Titum: GH Prat. Put. M. 7231, 7696.

commendavit: Halm and Meister give commodavit, which is approved also by Hirt. Halm compares §69 where Menander is said to be ‘omnibus rebus personis adfectibus accommodatus.’ But this would require the meaning ‘appropriately treated,’ and there is no instance in Quintilian of the verb used absolutely in this sense. Nor is there any example to support Hild’s interpretation praestitit, which would be moreover extremely weak. The recurrence of the word so soon after accommodata tells against Halm’s reading, though Quintilian is negligent on this head.—On the other hand, in vi. 3. 14 the reading ‘ad hanc consuetudinem commodata’ is rightly accepted against ‘commendata’ most edd.

§102. immortalem GS Meister: illam immortalem Prat. Put. M Halm: immortalem illam Vall.

velocitatem. So all MSS, except S, Burn. 243, and Bodl., which have civilitatem. Kiderlin (in Hermes 23, p. 174) thinks that we might have expected ideoque immortalem gloriam quam velocitate Sallustius consecutus est: ‘und darum hat er die velocitas durch (von der velocitas) verschiedene Vorzüge erreicht.’ Consequi cannot mean ‘to supply the place of’: and immortalis is inappropriate as an attribute of velocitas: besides, Quintilian has not spoken of Sallust’s velocitas, even indirectly. Schlenger conjectured claritatem: Andresen auctoritatem (‘klassisches Ansehen,’ cp. iv. 2. 125: xii. 11. 3): Kiderlin now proposes divinitatem, which in Cicero = Vortrefflichkeit, Meisterschaft: cp. xi. 2. 7. Judged by the previous sentences the expression is not too strong. For immortalem divinitatem cp. §86 illi ... caelesti atque immortali: and for consecutus est iii. 7. 9 quod immortalitatem virtute sint consecuti.

clarus vi ingenii. This is a conjecture of Kiderlin’s, which I find has been adopted also by Krüger (3rd ed.). GHFT give clarius ingenii: Prat. Put. clari ingenii vir: 7231, 7696 clari vir ingenii: MS Harl. 4995, 4950, 4829, Burn. 243 and 244, Dorv. C and Ball, clarus ingenio; Harl. 2662 and 11671 clarus (?) or claret vir ingenii. Spalding had already pointed out that clarus is not found with ingenium, except where ingenium is used of a person: e.g. §119 erant clara et nuper ingenia: he therefore wrote elati vir ingenii (following Goth. elatus ingenio and Bodl. elatus ingeniis). Kiderlin compares §70 sententiis clarissimus, and for vis ingenii i. pr. 12: ii. 5. 23: x. 1. 44: xii. 10. 10. The reading clarus vi ingenii points the contrast to what follows in ‘sed minus pressus,’ &c.: it was his style that did not altogether suit the dignity of history.

§103. genere ipso, probabilis in omnibus, sed in quibusdam. Till Kiderlin made this happy conjecture (see Hermes 23, p. 175) genere had always been joined with probabilis, and the text was twisted in various directions. GHS, Burn. 243, Bodl. give in omnibus quibusdam: M Harl. 2662, 4995, 4950, Burn. 244, Dorv. in omnibus sed in quibusdam, and so apparently Prat. Put. 7231, 7696. Out of omnibus Halm gives on Roth’s suggestion, operibus: afterwards he decided for partibus, and this (though omnibus to partibus is not an easy transition) is adopted by Meister. Kiderlin’s punctuation makes everything easy: ‘Anerkennung verdienen seine Leistungen alle, manche stehen hinter seiner Kraft zurück.’ Even these last, Quint. means, are probabiles (cp. viii. 3. 42 probabile Cicero id genus dicit quod non plus minusve est quam decet); but they do not show the great powers that distinguish his other writings. It is uncertain whether Quintilian wrote in quibusdam or sed in quibusdam (M). The easiest explanation of the omission in the other MSS. is to suppose that he wrote in omnibus in quibusdam: perhaps the copyist of M saw that omnibus and quibusdam were antithetical, and inserted sed. Kiderlin notes Quintilian’s liking for chiasmus, without any conjunction: cp. §106 in illo, in hoc (where in hoc is wanting in M).

suis ipse viribus: ed. Col. 1527 (Halm), and so (Fierville) 7231, 7696. In Harl. 2662 and 11671 (A.D. 1434 and 1467) suis already appears, corrected from vis GH. The Juntine ed. (1515) has suis viribus minor: so Prat. and Put.

§104. et exornat. Vall. and (apparently) Prat. Put. 7231, 7696, and most edd.: et ornat M Halm, Meister, Krüger: exornat GHS. Becher remarks that et exornat might easily pass into exornat.

nominabitur: Weber and Osann proposed nominabatur (which appears in Harl. 2662, but corrected to -itur). Krüger at first accepted this in support of his theory that the whole passage refers to Cremutius, who ‘in former days (olim), while his works were under a ban, was only named (i.e. was a mere name, but now is known and appreciated).’ The parallel passage (§94) is sufficient to dispose of any such interpretation: sunt clari hodieque et qui olim nominabuntur.

Cremuti. Nipperdey, Philol. vi, p. 193, Halm, and Meister: remuti H Prat. Put. 7231, 7696 remremuti G, rem utili Burn. 243: remitti S. Bodl.: nec imitatores uti Harl. 2662, 4995, 4950, 4829, 11671. A review of the various explanations of the whole passage (Superest—quae manent) will be found in Holub’s Programm ‘Warum hielt sich Tacitus von 89-96 n. Chr. nicht in Rom auf?’—Weidenau, 1883: but his conjecture remoti (i.e. relegati) for remuti is not to be thought of.

dividendi: first in the Aldine edition: all MSS. have videndi, except M (indicendi) and Prat. Put. Harl. 4995 (vivendi). Cp. i. 10. 49, where the case is the same.

§105. In the Aurich Programm, Becher gives a more recent statement of his views: ‘wie zu cum causale, so tritt praesertim auch zu cum concessivum, in diesem Falle wiedenzugeben mit, “was um so auffallender ist, als.” Der Sinn ist also: “Ich weiss sehr wohl, welchen Sturm des Unwillens ich gegen mich errege, und dies (dieser Sturm) ist um so auffallender, als ich jetzt gar nicht die Absicht hege, meine (in Potentialis gesprochene) Behauptung (fortiter opposuerim) wahr zu machen, resp. comparando durchzuführen. Ich lasse ja dem Demosthenes seinen Ruhm—in primis legendum vel ediscendum potius.”’

§106. praeparandi. For Kiderlin’s conj. praeparandi, narrandi, probandi see ad loc.

[omnia] denique, GH, Burn. 243, Bodl. omit omnia (which is in all my other MSS.), and Meister now approves (following Spalding, Osann, and Wölfflin), on the ground that Demosthenes and Cicero were not alike in everything that belongs to inventio. Halm thinks that omnia is to be found in racioni of the older MSS.: but Kiderlin points out that this error may have arisen from the carelessness of a copyist who, after thrice writing the termination i, gave it also to the fourth word.

illi—huic Prat. M, S Vall. Harl. 4995, 2662 Bodl. &c.: illic—hic GH Put. 7231, 7696, Halm.

§107. vincimus, H, G², and most MSS.: (cp. §86): vicimus G.

§109. ubertate Harl. 4995. This is also the reading of codd. Vall. and Goth.: all the other MSS. give ubertas.

totas virtutes Bn Bg N Prat. Ioan. 7231, 7696: totas vires M b.

§112. ab hominibus Halm and Meister: ab omnibus Bn Bg HFT Ioan. Prat. 7231, Sal. and most codd.: hominibus S Harl. 4995 Bodl.

§115. urbanitas. Kiderlin proposes to read et praecipua in accusando asperitas et multa urbanitas: cp. §117: §64: 2 §25: ii. 5. 8.

Ciceroni, for Ciceronem of the MSS. In the Rev. de Phil. (Janv.-Mars, 1887) Bonnet quotes from the Montpellier MS. a note of the sixteenth century deleting the name as a gloss (on inveni). Certainly all codd. give Ciceronem, not Ciceroni. Bonnet thinks that the insertion does not accord with Quintilian’s habitual deference towards Cicero: ‘Quintilien se trouvant dans le cas de contredire Cicéron ne le nomme pas.’—Becher reports Ciceroni, a correction in the Vallensis.

castigata, B (i.e. Bn and Bg) Ioan. Prat. 7231, 7696 Harl. 2662, 4995, 11671: custodita H M b F T Alm. Harl. 4950, 4829, Burn. 243, 244, Bodl. Dorv. and Ball. For gravis (bH M Vall. and seemingly Prat.) B Sal. 7231, 7696 and Ioan. give brevis.

si quid adiecturus sibi non si quid detracturus fuit, Vall. Harl. 4995. For the repetition, see on haud deerit 3 §26. Halm and Meister print si quid adiecturus fuit—(sc. virtutibus suis, cp. §§116, 120)—the reading of B (i.e. Bn and Bg), which is also that of Ioan. Prat. N 7231 Harl. 2662, 11671: while M Harl. 4950, 4829, Burn. 244 have si quid adiecturus fuit, non si quid detracturus. The reading of H is si quid adiecturus sibi non si quid detracturus [Sulpicius insignus] fuit ut servius sulpicius insignem &c.: so also T, Burn. 243, Bodl. The brackets in H are by a later hand, indicating a gloss which arose from a mistake made by the copyist of H. In Bg the passage stands:—

sibi non si quid detracturus

si quid adiecturus: fuit et servius sulpicius

The words added above the line are by the hand known as b.

In copying H wrote: si quid adiecturus sibi non si quid detracturus (then omitting fuit continues) et Serv. Sulp. (then goes back and resumes) fuit et servius &c. This is the origin of the confusion which exists in all the MSS. of this family.

§117. et fervor. This is Bursian’s conjecture, adopted by Halm and Krüger (3rd ed.), and now approved by Becher. BM have et sermo, which is also the reading of N Prat. Sal. 7231, 7696 Ioan. Harl. 2662, 4950 and Ball.: Hb et summo: Harl. 4829, 11671, Burn. 244 et smo: while Bodl., Dorv., and Burn. 243 give the correction in T eius summa, out of which the second hand in the Vallensis (Laurentius Valla) made et vis summa, a reading which occurs also in Harl. 4995. Meister reads et sermo purus; while Kiderlin proposes et simplex sermo (cp. iv. 1. 54: viii. 3. 87: ix. 3. 3: 4. 17: viii. pr. 23: x. 2. 16).

ut amari sales. Francius conjectured ut amantur sales, but this loses the antithesis between amari and amaritudo ipsa. Kiderlin’s ut amantur amari sales (viii. 3. 87: vi. 1. 48) is an improvement; but if ridicula is taken in a good sense it seems impossible that after censuring Cassius for giving way unduly to stomachus, Quintilian should go on to say, ‘moreover, though bitter wit gives pleasure, bitterness by itself is often laughable.’ Is it possible that we ought to read ut amari sales risum movent ita amaritudo ipsa ridicula est? Such an antithesis might have been written ‘per compendium,’ and the words risum movent may then have dropped out. See the note ad loc.: and cp. especially vi. 1. 48 fecit enim risum sed ridiculus fuit, and οὐ γέλωτα κινεῖ μᾶλλον ἢ καταγελᾶται, quoted in the note on 1 §107.—Krüger (3rd ed.) adopts frequentior for frequenter, which gives a good sense, except that freq. amar ipsa is awkward.

§121. lene Halm and Meister: leve B Prat. N 7231 M 7696 C. Here again Becher prefers leve, comparing Cic. de Orat. iii. §171, quoted on §44 above: levitasque verborum 1. 52: and levia ... ac nitida, v. 12. 18.

§123. scripserint. So Bn Bg H Ioan. Prat. 7231, 7696 Vall. Harl. 4995, 2662, 11671, Bodl., Dorv., Spalding, and Bonnell. Becher compares among other passages 2 §14 (concupierint), and points out that Quintilian is not thinking of individual writers on philosophy, but of the class, as opposed to the class of orators, historians, &c.—Halm, Meister, and Krüger have supersunt (Put. M, Ball. Burn. 243 Harl. 4950).

§124. Plautus, Prat. N, 7231 Ioan. Harl. 2662, 4829, 11671: plantus M Harl. 4950: Plantatus Sal.: plaustus Hb: Plancus edd. vett. and Harl. 4995.

Catius. The name is rightly given in Harl. 4995.

§126. iis quibus illi. Iis is the conjecture of Regius, followed by Halm, Meister, and Krüger. Becher would retain in quibus illi,—the reading of BN Prat. Ioan. Vall. M Harl. 4995, 2662, 4950, 11671, Burn. 244 Dorv. Ball. The difficulty of construing probably led to the omission of in in bH Bodl. Burn. 243, 7231, 7696, Spalding and Bonnell.

ab illo B Ioan. 7231, 7696 Sal. Harl. 2662, 4950, 4829: ab eo bHM Burn. 243.

§127. foret enim optandum: fore enim aliquid optandum bHFT. Spalding conjectured alioqui optandum, which Kiderlin approves.

ac saltem all MSS.: Meister has aut saltem, probably relying on a wrong account of the Bambergensis: see Halm vol. ii, p. 369.

illi viro B: illi virus bHM: illi virtutibus Halm: illi viro eos (or viro plurimos) Kiderlin.

§128. multa rerum cognitio: so all codd. except Ioannensis and Harl. 4995, which have multarum rerum cognitio. b omits cognitio and is followed by HFT.

§130. si obliqua contempsisset, si parum recta non concupisset. I adopt the reading recently proposed for this vexed passage by Ed. Wölfflin in Hermes, vol. xxv (1890), pp. 326-7, though it is right to note that he was partly (as will be seen below) anticipated by Kiderlin. Obliqua seems thoroughly appropriate in reference to Seneca’s unnatural, stilted, affected style,—‘jene unnatürliche, durch unmässigen Gebrauch von Tropen und Figuren auf Schrauben gestellte Ausdrucksweise, welche statt der Klarheit ein Schillern zur Folge hat.’ Wölfflin compares ix. 2. 78 rectum genus adprobari nisi maximis viribus non potest: haec diverticula et anfractus suffugia sunt infirmitatis, ut qui cursu parum valent flexu eludunt, cum haec quae adfectatur ratio sententiarum non procul a ratione iocandi abhorreat. Adiuvat etiam, quod auditor gaudet intellegere et favet ingenio suo et alio dicente se laudat. Itaque non solum si persona obstaret rectae orationi (quo in genere saepius modo quam figuris opus est) decurrebant ad schemata ... ut si pater ... iacularetur in uxorem obliquis sententiis. This passage supplies (what is indeed suggested by obliqua itself) the antithesis parum recta: cp. ii. 13. 10 si quis ut parum rectum improbet opus.

In the Jahrbücher f. Philologie (vol. 135, 1887: p. 828) Kiderlin had previously dealt with the passage on similar lines. The traditional reading si aliqua contempsisset (b) he considers too indefinite, though not impossible: in point of authority, though preferable to the si nil aequalium cont. of the later MSS., it cannot rank so high as the reading of Bn and Bg, which give simile quam without any attempt at emendation. This Kiderlin thinks must be nearest the original: he therefore rejects such conjectures as Jeep’s si antiqua non, on the ground that it is improbable that simile quam arose out of antiqua. He introduces his own conjecture by referring to ix. 2. 66 and 78 (see above), and to the contrast between schemata and rectum genus, recta oratio; the former are called lumina or lumina orationis (xii. 10. 62). Cp. viii. 5. 34. He would read: nam si mille ille schemata (or illas figuras) similiaque lumina contempsisset, si parum rectum genus (or sermonem) non concupisset, &c. Similiaque occurs ix. 4. 43: mille (for sescenti) is used v. 14. 32: for contempsisset cp. ix. 4. 113. Si mille illa and similiaque may easily have run together, when schemata (or figuras) would fall out: quam in the older MSS. may represent que lumina, which again reappears in the qualium of the later codd. (si nil aequalium). As an alternative for parum rectum genns (or sermonem) Kiderlin suggests Wölfflin’s reading parum recta: and compares ix. 2: ii. 5. 11: v. 13. 2: ix. 1. 3; 3. 3: x. 1. 44; 89: ii. 13. 10.

Of the MSS. Prat. 7231 Sal. 7696 N Ioan. Harl. 2662 and 11671 agree with Bn and Bg in giving simile quam: b has si aliqua: HFT, Burn 243, Bodl. aliqua: M Harl. 4995, 4950, 4829, Burn. 244, Dorv. C si nil aequalium. Among previous conjectures are si multa aequalium, Törnebladh: si ille quaedam, Halm (where ille is surely superfluous): si antiqua non, Jeep. Meister accepts the reading si aliqua non: Becher thinks that si nil aequalium may be right.

It is generally admitted that a word must have fallen out after parum: the codd. all give si parum non concupisset. Jeep proposed si pravum (= corruptum: cp. ii. 5. 10) non conc.: on which Halm, comparing omnia sua, remarks, ‘debebat saltem prava.’ But prava seems too strong a word for Quintilian to have used in a criticism where he is so studiously mixing praise and censure. Halm suggested si parum sana, and is followed by Meister: cp. Fronto’s ‘febriculosa’ of Seneca, p. 155 n. Sarpe proposed si prava or parva or plura: Buttmann si parum concupiscenda (or convenientia): Herzog si parvum: Madvig si partim or partem (i.e. paulo plus quam aliqua, and in opp. to omnia sua, below): Hoffmann si opiparum: Seyffert si garum: Kraffert si non parum excussisset (cp. §101, §126: v. 7. 6; 7. 37; 13. 19: xii. 8. 13, &c.): Gustaffson si parva (cp. i. 6. 20 frivolae in parvis iactantiae): Andresen si similem ei quem contempsit se esse (sc. concupisset; cp. Tac. Ann. xiii. 56: xii. 64: Hist. i. 8: Livy xlv. 20. 9) si parem non concupisset (i.e. si Ciceronianum genus dicendi imitari quam diverso genere gloriam eius aemulari maluisset): or, nam si similem ei quem contempsit se esse, non parem concupisset: Krüger (3rd ed.) si parum arguta: Hertz (who argues that the word which has fallen out must, with parum, correspond to corrupta above) si parum pura.

utrimque Meister and Becher, following old edd., Spalding, and Bonnell: utrumque B N 7231, 7696: virumque M: utcumque Halm, ‘in every way,’ ‘one way or another,’—proposed by Gesner at 6 §7.