¹ See, e.g.Butcher, S. H., Harvard Lectures on the Originality of Greece (London, 1911), pp.257–258Google Scholar;Atkins, J. W. H., Literary Criticism in Antiquity: A Sketch of its Development (Cambridge, 1934), ii.335–336Google Scholar. The standard reference for a study of ancient comparative criticism is still Focke, F., ‘Synkrisis’, Hermes 58 (1923), 327–68, at pp.339–348Google Scholar.I have not seen Hense, O., Die Synkrisis in der antiken Literatur (Diss., Freiburg, 1893).Google Scholar

² For definitions of the exercise: Quint. Inst. 2.4.21; Theon, Prog. 9, Spengel, Rhet., ii. 112; Apth. Prog. 10, ibid., p. 42. See Reichel, G., Quaestiones Progymnasmaticae (Diss., Leipzig, 1909), pp.95–97Google Scholar;Focke, , Hermes 58 (1923), 331–339.Google Scholar

³ See further Walbank, F. W., A Historical Commentary on Polybius, vol. II (Oxford, 1967), pp.410–111.Google Scholar

⁴ Cf. the similar comparison in Book 1 of Cicero's de Oratore and Quint. Decl.Min. 268. On the quarrel between rhetoric and philosophy, see Kennedy, G. A., The Art of Persuasion in Greece (Princeton, 1963). pp.321–325Google Scholar. For similar comparisons between history and poetry in the Hellenistic age: Praxiphanes fr. 18 Wehrli IX, on which see Brink, K. O., ‘Callimachus and Aristotle’, CQ 40 (1946), 11–26CrossRefGoogle Scholar, at p. 24. See further Ullman, B. L., ‘History and Tragedy’, TAPhA 73 (1942), 25–53Google Scholar. Cicero's friend Roscius composed a book ‘quo eloquentiam cum historia compararet’ (Macr. Sat. 3.14.12), and a similar comparison seems to have been the subject of Cicero's own Hortensius.

⁵ Walbank, Cf. (n. 3), vol. I (Oxford, 1957), p.40Google Scholar. On comparisons in Polybius see further Lorenz, K., Untersuchungen zum Geschichtswerk des Polybios(Stuttgart, 1931), pp.15,81 notes73–74.Google Scholar

⁶ Vita Arati pp. 143, 150 Maas (= pp. 5, 17 Martin) with Maas, E., Aratea (Berlin, 1892), pp.385–386.Google Scholar

⁷ Cf. Nep.Han. 13.4: fr. 58 Marshall = fr. 9 Funaioli (cf. Veil. 2.41.1). See Leo, F., Die griechisch-römische Biographie nach ihrer litterarischen Form (Leipzig, 1901), pp.149–151Google Scholar;Focke, Hermes 58 (1923), 351Google Scholar –366;Geiger, J., Cornelius Nepos and Ancient Political Biography (Stuttgart, 1985), pp.118–119.Google Scholar

⁸ Cf. Porph. ap. Euseb.Praep.Evang. 10.3.23, 467D = FGrHist 70 T 17Google Scholar, and see Walbank, F. W., ‘Alcaeus of Messene, Philip V and Rome’, CQ 37 (1943), 1–13, at p.3.CrossRefGoogle Scholar

⁹ The agon of Frogs has been taken to parody (a) ‘ syncritic’ critical treatises by Gorgias or members of his circle (Pohlenz, M., ‘Die Anfange der griechischen Poetik’, NGG 1920, 142–178 =Kleine Schriften [Hildesheim, 1965], ii.436–472)Google Scholar; (b) treatises on tragedy similar to the Certamen of Homer and Hesiod; and (c) Sophocles′ περ xορο (Harriott, R., Poetry and Criticism before Plato [London, 1969], p.131)Google Scholar. There is, however, no evidence of Gorgias′ having been engaged in any kind of comparative criticism. The Certamen depicts an agon similar to that represented in the Frogs (and to the dramatic competitions of the festivals themselves for that matter) and is also set in an amoebaean form, yet it does not go beyond reporting of the competition and its result and contains no critical discussion (cf. the certamen of erotikoi logoi in Plat. Phdr. 230E–257C). It therefore cannot be the object of Aristophanes′ parody on the methods of contemporary literary criticism. See also Ugolini, G., ‘L'evoluzione della critica letteraria d'aristofane II’, SIFC 3 (1923), 259–291Google Scholar, at pp. 283–4;dayman, D. L., ‘The Origins of Greek Literary Criticism and the Aitia Prologue’, WS ² 11 (1977), 27–34Google Scholar, at p. 29;Dover, K., Aristophanes′ Frogs (Oxford, 1993), p.32.Google Scholar

¹⁰ See Brink, CQ 40 (1946), 11–26CrossRefGoogle Scholar;Podlecki, A. J., ‘The Peripatetics as Literary Critics’, Phoenix 23 (1969), 114–137.CrossRefGoogle Scholar

¹¹ See Fraser, P. M., Ptolemaic Alexandria (Oxford, 1972), i.547–550Google Scholar;Russell, D. A., Criticism in Antiquity (Berkeley & Los Angeles, 1981), pp.37, 173–176.Google Scholar

¹² Cf. Cic. de Orat. 2.94; Oral. 42; D.H. Dem. 32, i.200f. U.R.; Quint. Inst. 10.1.33; Plin. Ep. 5.8.10. The same observation on the styles of the two orators is attributed to Philip of Macedon in the Life of Demosthenes preserved in the pseudo-Plut.Lives of the Ten Orators (Mor. 845D) and in Photius (Bibl. 265, 493b24–30 =Vita Demosth. I 75–80, pp. 283–4 Westermann). A comparison with Demosthenes might also be implied in a remark of Cleochares nearcontemporary Hieronymus of Rhodes on Isocrates′ disregard for delivery (Phld. Rh. 4 col. xvi, 13ff., i.198 Sudhaus; D.H. hoc. 13, i.73 U.-R.). See Drerup, E., Demosthenes im Urteile des Altertums (Wüirzburg, 1923), pp.94–97.Google Scholar

¹³ Cf. the two similar studies (of uncertain date) by Platonius, and see Kaibel, CGF, pp. 3–6;Pfeiffer, R., History of Classical Scholarship I: From the Beginnings to the End of the Hellenistic Age (Oxford, 1968), p.160.Google Scholar

¹⁴ Like the elder Seneca (Con. 10.pr. 13), Quintilian tends to distinguish only theprinceps (who is not necessarily the auctor as well) from the rest of the ordo; e.g. Inst. 10.1.61, 73, 76, 81, 85 etc. On the history of ‘canons’ see now Scotti, M., ‘I “canoni” degli autori greci’, Esperienze letterariel (1982), 74–91.Google Scholar

¹⁵ A classification of Herodotus, Thucydides and Xenophon according to the three rhetorical is preserved in the (4th century A.D. ?) Marcellinean Life of Thucydides (sect. 39–40, lines 204–11, p. 194 Westermann), but to assume a Hellenistic origin for this classification would involve reopening the vexed question of the origins and date of the ‘tripartite division of styles’, which is beyond the scope of this study.

¹⁶ For the association of Fronto's discussion with the same Varronian source Gellius uses, see Warren, M., ‘On a Literary Judgment of Fronto’, TAPhA 25 (1894), xlii–xlv. H. Dahlmann's attribution of the fragment to Varro's περί χαρακτήρων is plausible, though as fragments 40 and 99 Funaioli show, similar discussions could also have been included in a Menippean Satire or the de sermone latino; see his Zu Varros Literaturforschung besonders in de poetis Entretiens de la Fondation Hardt 9 (1962), 3–20, at pp. 6–8.Google Scholar

¹⁷ ‘The Origin and Meaning of the Ancient Characters of Style’, AJPh 26 (1905), 249–90, at pp.270–271. For a different classification of the three philosophers, see Cic. de Orat. 2.157–61.Google Scholar

¹⁸ Varro seems to be using here Aristotle's classification of the τραγῳδίς μέρη (Po. 6, 1449b31–50a14). For the origins of Varro's method, see Dahlmann Varros Schrift'de Poematis′ unddie hellenistisch-römische Poetik (Wiesbaden, 1953), pp.30–32, 60–72 (=AAMainz 1953, 116–18, 146–58).Google Scholar

¹⁹ On which see Grilli, A., Tulli, M.Ciceronis Hortensius (Varese & Milan, 1962), pp.63–66. For similar classifications of prose writers in Cicero cf. de Orat. 3.28; Hort. fr. 25 Müller (= fr. 15 Grilli) and infra.Google Scholar

²⁰ For the state of the question, see Brink, C. O., Horace on Poetry III (Cambridge, 1982), pp.83–91Google Scholar. Horace's objection to the ‘contrast and label’ method of the critics, especially prominent in lines 56 and 59, is present throughout the section Ep. 2.1.50–59, where it is combined with his censure of two other critical fallacies: blind preference of the ancient over the new, and the tendency to couple Roman poets with Greek ones (see Brink, ibid., 57). In lines 50–54 Ennius is labelled ‘sapiens et fortis’, coupled with Homer, and contrasted with Naevius, whose sole merit is being older (cf. Ennius‘ quality in Volcacius Sedigitus’ canon of comic poets [fr. 1 Morel]: ‘decimum addo causa antiquitatis Ennium’); ‘doctus’ Pacuvius is contrasted with ‘altus’ Accius; and the coupling of Afranius and Plautus with Greek models also serves to contrast the features of ancient comedy present in Plautus’ Palliatae with the Nea-type Togatae of Afranius. If this interpretation is correct, Varro or his fellow critics join a long tradition of comparisons between ancient and new comedy: Arist. E.N. 4.8.6, 1128a22ff. (cf. Tract. Coislin. 7); Cic. Off. 1.104 and perhaps Hort. fr. 89 Müller; Plut. Mor. 711F–12D; 853A–4D; Marcus Aurelius, Med. 11.6; see further Janko, R., Aristotle on Comedy (London, 1984), pp.203–208Google Scholar;Schenkeveld, D. M., Gnomon 58 (1986), 216Google Scholar. Porphyrio's note to line 56 seems to imply that the differentiation between Pacuvius and Accius has also been adapted from a standard critical practice of contrasting Sophocles′ style with those of Aeschylus and Euripides.

Horace himself is not entirely free of such pigeonholing of writers; see C. 2.13.29ff. (cf. Ov. Her. 15.29–30) with Penna, A. La, ‘Sunt qui Sappho malint: note sulla σύγκρισις di SafFo e Alceo nell'antichità’, Maia 24 (1972), 208–215Google Scholar. And, despite his criticism, the method remained a favoured constituent of Roman critical discourse throughout the first century A.D. (e.g. Veil. 2.9.3; Quint. Inst. 10.1.52–3, 73, 87, 105–6, 108, 118; Tac. Dial. 25) and into the second, when it is represented in enumerations of the characteristic qualities of authors in Fronto (e.g. Fer.Als. 3.1, p. 227 v.d. Hout ²) and Apuleius (Apol. 95), as well as in a list of the peculiar faults of some prominent Latin erotic poets that Gellius puts in the mouth of a company of Greek συμπ⋯ται (19.9.7).

²¹ See Susemihl, F., Geschichte der griechischen Litteratur in der Alexandrinerzeit (Leipzig, 1891–1892), i.507–508.Google Scholar

²² See Roberts, W. Rhys, ‘Caecilius of Calacte’, AJPh 18 (1897), 302–312, at pp. 310–11, though in view of the evidence collected above, I find the central role he ascribes to Caecilius in the history of comparative criticism somewhat exaggerated. Apart from these two treatises, ‘Longinus’ testifies to a comparison of Caecilius between the styles of Lysias and Plato, probably in his own irepl περί (‘Long.’ 32.8; cf. fr. 95 Ofenloch).Google Scholar

²³ E.g. ‘Long.’ 9.11–14, 12.3, 33.4–5, 34; Petr. 55; Quint. Inst. 10.1.73, 87, 101, 105; Tac. Dial. 18, 26; Plin. 1.20.4; ‘Lucian’, Dem.Enc. 4–7; see also Demetr. Eloc. 80, 112–13. If the author of the Trepl περί ‘Eρμηνείας is of the Hellenistic age, as many scholars nowadays believe, these instances further testify to the abundance of comparative remarks in the literary inquiry of that era.

²⁴ In spite of Phrynichus′ censure of the non-Attic use of synkrisis in the title of this work (on which see Lobeck, C. A. in his edition [Leipzig, 1820], p. 278), the term, as shown above, is in accordance with earlier terminology of literary critics, and is similarly employed by Plutarch in Mor. 331C. In view of the abundant evidence for the employment of both the term synkrisis and the procedure it represents in many fields of Hellenistic scholarship including literary criticism, I see no reason to associate the application of this critical technique to literature with the progymnasma of synkrisis, as has sometimes been suggested. As the descriptions and exemplary models of the exercise offered by rhetoricians reveal, it was always focused on moral evaluations, either of specific persons or of abstract ways of life (see n. 2 supra), and even when the persons compared happened to be orators, as in Libanius′synkrisis of Demosthenes and Aeschines (Prog., compar. 3, Foerster viii.342; cf. Nicol. Exemp. Prog., compar. 2, Walz, Rhet.Gr. i.358), it is the orators′ moral conduct and political career which are compared, rather than stylistic or aesthetic values of their speeches (for such comparisons see also Gel. 15.28.6–7; Philostr. V.S. 1.18).Google Scholar

²⁵ Cf. Quint. Inst. 12.1.22; Suet.Claud. 41.3; and for the title of Gallus′ work, Plin. Ep. 7.4.3–6.Google Scholar

²⁶ Though Dionysius′ earlier writings (such as Imit. and Lys.) abound in general comparative criticism, they contain no discussion of specific passages. The method is first employed in the de Isocrate, and only in the later works (esp. Isae., Demosth. and Thucyd.) does he appear to fully realize the potential of the technique and make ample use of it. See Bonner, S. F., The Literary Treatises of Dionysius of Halicarnassus: A Study in the Development of Critical Method (Cambridge, 1939), pp.54–55, 62–68, 93, whose reconstruction of the chronological order of Dionysius′ works I follow here.Google Scholar

²⁷ The advantage of comparisons of passages sharing a common or similar theme is also acknowledged in Plat.Phdr. 236D; ‘Lucian’, Dem.Enc. 5; Hermog. Id. 397–8 Rabe.Google Scholar

²⁸ Cf. ‘Long.’ 15.7 and 32.5 (possibly borrowed from Caecilius; see fr. 95 Ofenloch) where the passages compared are not cited but referred to. See also D.Chr. Or. 2.3–13; ‘Lucian’, Dem.Enc. 5–6; Hermog. Id. 243, 353–354, 397–398 RabeGoogle Scholar. A possible explanation for the relative rarity of comparisons of passages in rhetorical treatises is the competing (and often more handy) method of illustration by contrasting examples cited from literature with the rhetor's own recasting of them in different words, metre or word order. Found as early as Plat. Rep. 3, 392E–394B, this is Demetrius′ favourite method of illustration, and is also practised by Cicero (e.g. Oral. 214–15), Dionysius (e.g. Comp. 4, ii.15f. U.-R.), ‘Longinus’ (40.3, 43.4), and Hennogenes (e.g. Id. 230, 250, 301, 326 Rabe), though vehemently rejected by Philodemus of Gadara; see Greenberg, N. A., ‘Metathesis as an Instrument in the Criticism of Poetry’, TAPhA 89 (1958), 262–270. For the Hellenistic controversy between rhetoricians who illustrate their propositions by citing from well known authors and those in favour of inventing their own examples, see ad Her. 4.1–10 and H. Caplan's note in the Loebed. (London & Cambridge, MA, 1954), pp. 242–3 note c. Gellius attributes a similar method of illustrating the euphonic qualities of literary passages to Probus (Gel. 13.21 passim; cf. 6.20.2, 6).Google Scholar

²⁹ It is tempting to speculate on the possibility that Hellenistic inquiries such as the περί τῶν παρ’ Eύριπίδη και Σοϕοκλεί of Heracleides of Pontus (fr. 180 Wehrli VII) were similarly formed. For the form of Dio's synkrisis, see Luzzatto, Maria T., Tragedia greca e cultura ellenistica: VOr. Llldi Dione di Prusa (Bologna, 1983), pp.29–37. Her emphasis on the deviation of Dio's ‘triadic’ comparison from the regular rhetorical practice of comparing only two authors at a time seems to ignore Gellius’ ‘locorum conlatio’ of three orators in 10.3, on which see infra.Google Scholar

³⁰ See note 9 supra. The comparison of ‘best passages’ rather than of passages sharing a common theme is typical of certamina, which are more concerned with the result of the competition than with the process and criteria of evaluation; cf.Certamen 321.

³¹ E.g. Tac. Dial. 18.2–4; Plin. Ep. 1.20.4; Fro. Eloq. 1.2, 2.12, pp. 134.3–4, 140.10–11 v.d.Hout ²; Apul. Apol. 95.

³² Gellius studied with two rhetors: T. Castricius (esp. 11.13, 13.22.1) who reveals some familiarity with C. Gracchus (11.13) and Antonius Iulianus (1.4.8, 9.15, 18.5.1, 19.9.2). He also attended some kind of extracurricular conversations with didactic intent held by Fronto (19.8.1, 16).Google Scholar

³³ See Stemplinger, E., Das Plagiat in der griechischen Literatur (Leipzig & Berlin, 1912), pp.121–158Google Scholar; W. Kroll, RE Suppl. VII, pp. 1113–16, s.v. ‘Rhetoric’ 35;D'Alton, J. F., Roman Literary Theory and Criticism (London, 1931), pp.426–432Google Scholar;Bompaire, J., Lucien écrivain: imitation et création (Paris, 1958), pp.59–121Google Scholar;Reiff, A., Interpretatio, Imitatio, Aemulatio: Begriff und Vorstellung literarischer Abhängigkeit bei den Römern (diss., Koln, 1959)Google Scholar;Russell, D. A., ‘De Imitatione’, in West, D. &Woodman, T. (eds.), Creative Imitation and Latin Literature (Cambridge, 1979), pp.1–16.Google Scholar

³⁴ Probably not from the eighth-century B.C. poem by Aristeas of Proconnesus; see D. A. Russell's commentary (Oxford, 1964), p. 103. Aratus′ imitation of the passage is also mentioned in Schol. (Bt) 15.628 and pseudo-Plut. Life and Work of Homer 160, but, as often in ancient literary exegesis, the commentators are content with the bare mention of the borrowing, without attempting any judgement of the passages; for this type of observation on literary dependence: Stemplinger (n. 33), pp. 6–12, 27–31.Google Scholar

³⁵ For these comparisons of‘ Longinus’, and especially for his judgement of the Aeschylean and Euripidean lines in 15.5–6, see Russell, in Creative Imitation (n. 33), pp. 13–14. The elusive Demetrius, already working under some kind of theoretical distinction between discreditable μετάθεσις (cf. the similar use of μετατίθηνι by Agatharchides, ap. Phot. Bibl. 250.21, 446b28), and permissible μιμησις, in which the borrowed material is tinged with the imitator's own personal style and altered to suit a new context (Eloc. 112–14), illustrates the latter by bringing together Homer's use of the epithet περίρρυτος to describe the vastness of Crete (Od. 19.172–173), and the adoption of the same epithet by Thucydides, to emphasize the unity of Sicily (Thuc.)

³⁶ The passage is in fact taken from Ps. Demosth. in Ep. Phil. 13, a work not noted for its conciseness. But although this may dull the point of Seneca's criticism, for the purpose of this study his critical observations should be investigated according to the critic's own assumptions, without regard to their accuracy. The same also applies to the discussion in Con. 7.1.27, in which Varro Atacinus is probably not at all the model of Vergil in these lines (though the original lines of Apollonius might have been), and to the assumption that Vergil's depiction of Etna (Aen. 3.570–77) is modelled on Pindar Pyth. 1.21–6 rather than on Lucretius 6.639–702 or on Vergil's ownG. 2.308–9 (Gel. 17.10).

³⁷ See also Sat. 5.17.5–6, 6.3.4 for cases in which Vergil is deemed superior to his models (Apollonius and Ennius), and 5.3.2, 5.15, 5.17.7–14 (almost a verbatim citation of Gel. 17.10) for cases in which he is judged inferior.

³⁸ On which see Gamberale, L., La traduzione in Gellio (Rome, 1969), p.111 n.98.Google Scholar

³⁹ For doubts as to the potential of imitations to equal their models, see Sen. Con. l.pr.6; Quint. Inst. 10.2.9–11; and for an attribution of the a priori inferiority of Latin imitations of Greek originals to the egestas verborum of the Romans′ language, Plin. Ep. 4.18.1

⁴⁰ Ar. Ra. 797, 799: For κανὡν in this sense, see e.g. Cic. Fam. 16.17.1; D.H. Lys. 2, i.9 U.R., 18, i.29 U.R.; Pomp. 3, ii.239 U.R.; Phot. Bibl. 61, 20b.25 and, for similar employment regarding the plastic arts, Plin. N.H. 34.55. See further H. Oppel, ‘κανὡν’ zur Bedeutungsgeschichte des Wortes und seiner lateinischen Entsprechungen (regula-norma)′, Philologus Suppl. 30.4 (1937), 43–50; Pfeiffer (n. 13), i.207.

⁴¹ Gamberale (n. 38), esp. pp. 168–72 but also 43, 62–3, 94–5 and n. 69, 108 n. 93; see also Marache, R., La critique litteraire de langue latine et le développement du goût archaïsant au He siècle de notre ère(Rennes, 1952), pp.305–308Google Scholar; Reiff(n. 33), pp. 101–2;Holford-Strevens, L.Aulus Gellius (London, 1988), pp.149–150Google Scholar

⁴² For the terms see Richter, H. E., Übersetzen und Ubersetzungen in der romischen Literatur (Diss., Erlangen, 1938); Reiff(n. 33), esp. pp.104–411Google Scholar. The terms imitatio and aemulatio, as well as the often despised interpretari, appear to have lost much of their distinct significance in the course of the first century; see Reiff, ibid.; Russell, ad ‘ Long.‘ 13.2; Gamberale (n. 38), pp. 97–99,119–20. As a critical term, Gellius usually employs aemulatio in the sense of ‘ a successful imitation or translation’ (‘cum earn tragoediam verteret, non sane incommode aemulatus est’, 11.4.3), without the further notion of rivalry (but see 20.17.8 on a translation exercise). His employment of interpretari is confined to translations focused on rendering the meaning of the original, at times, as often in Cicero (e.g. N.D. 2.47; Fin. 2.5, 3.35), in reference to the Latin equivalents of single words (5.18.7, 11.16.3, 13.9.4, 17.3.cap. 1, 19.2.2); on which see further Reiff (n. 33), pp. 46–50.

⁴³ For the syntactical structure cf. Cic. An. 11.15.3: ‘Prorsus nihil abest quin sim miserrimus’.

⁴⁴ Gel. 9.9 is one of the chapters of the N.A. for which it is particularly difficult to adopt J. Kretzschmer's principle ‘singula capita, quoad licebit, ad unum auctorem revocanda sunt’ (De A. Gelliifontibus, I. De auctoribus A. Gelliigrammaticis [Diss. Greifswald; Posen, 1860], pp. 6–11), and because of the different critical methods and value concepts employed in sections 12–17 of this chapter, I prefer to follow Holford-Strevens (n. 41), pp. 53–54 in regarding Probus′ criticism as a postscript appended to the first part of the chapter because of the similar topic.Google Scholar

⁴⁵ Gellius′ text lacking πέτρας in v. 23, he seems to have taken κρονούς from the next sentence as the direct object of ϕέρει, hence his blunder in assuming it was rendered by Vergil's ‘globos flammarum’. See Holford-Strevens (n. 41), p. 172 n. 50.

⁴⁶ For transferre in the sense of translate′, see Gamberale (n. 38), pp. 110–112, notes 98, 100. For the technical term άκύρως, see Quint. Inst. 8.2.3; cf. Cic. Fam. 16.17.1; Gel. 1.22.12 (again on the language of Vergil), 19.7.3; and for later employment Moore, J. L.,‘Servius on the Tropes and Figures of Vergil I’, AJPh 12 (1891), 157–192, at pp. 176–177.Google Scholar

⁴⁷ On interpretari see n. 42 above

⁴⁸ This, as noted already by Dionysius Thrax (p. 83 Uhlig), is often marked by the use of evaluative adjectives in the comparative.

⁴⁹ For the distinction between low types of humour, traditionally associated with mime or the scurra, and the more respectable ones, based on facetiae, see e.g. Cic. Off. 1.104; De Oral. 2.239, 244; Hor. S. 1.10.6; Quint. Inst. 6.3.8, 82; Euanth. 3.5; cf. Arist. E.N. 4.8.3, 1128a4ff.; Tract.Coislin. 12, 18; and see further D'Alton (n. 33), pp. 358–363; Janko (n. 20), pp. 216–218.

⁵⁰ For ancient acclamation of Menander's realism, especially in the portrayal of characters, see Ar.Byz. apud Syrian. In Hermog. ii.23.8–11 Rabe; Man. 5.474–475; D.Chr. Or. 18.7; Quint. Inst. 10.1.69; Hermog. Id. 323.22ff. Rabe.

⁵¹ Especially in Dionysius of Halicarnassus, but also among the Neo-Atticists of the Second Sophistic, e.g. Lucian, Rh.Pr. 9. The expectation that imitations should reproduce the most characteristic virtue of their original might be related to a typically Roman aspect of the idea of imitatio, which regards rivalling a Greek model in the very field of its excellence the utmost achievement of aemulatio; see, for example, Seneca's above-mentioned comparison of Sallust's brevitas with that of ‘Thucydides’ (Con. 9.1.13), and further in Reiff (n. 33), pp. 77–82.

⁵² In spite of the rejection of verbatim translation of the speeches of Demosthenes and Aeschines in Opt. Gen. 14, the aim that the author sets for himself in this treatise is also to reproduce all the (essential) virtues of his originals (23).

⁵³ Beall, S. M., Civilis eruditio: Style and Contents in the ‘ Attic Nights’ ofAulus Gellius (Diss., Berkeley, 1988), pp.151–208.Google Scholar

⁵⁴ E.g. Cic.de Oral. 2.90–1; Hor. Ep. 1.19.15–16; D.H.Vet.Orat., praef. 4, i.6.23–4 U.R.; Sen. Ep. 114.17.

⁵⁵ Such a unilateral examination of correspondence would also explain Gellius‘ apparent blindness to the fact that some of the alterations he censures are necessitated by the different function an ’ imitation‘ comes to serve in its new context, as often noted by modern critics. See, for example, the observations on Probus’ criticism of Aen. 1.498ff. in the commentaries of R. G. Austin (Oxford, 1971), p. 167 and R. D. Williams (London and New York, 1972), p. 199, and in Marache (n. 41), p. 307.Google Scholar

⁵⁶ For the ability of comparisons to modify judgements cf. D.H.Pomp. 1, ii.223 U.R.

⁵⁷ Holford-Strevens (n. 41), p. 89, and pp. 53–4 for the possible attribution of ch. 13.27 to Probus. It is perhaps significant that Favorinus had some familiarity with the teachings of Probus (Gel. 3.1.6). H. Nettleship assumed that at least some of Gellius′ discussions of translation had been taken from ‘ a manual in which the whole question of translation was discussed’, which he believes to be Octavius Avitus′ ὁνοι⋯τηες (‘The Noctes Atticae of Aulus Gellius’, AJPh 4 [1883], 391–415, at p. 404), but see Gamberale (n. 38), pp. 55–9.Google Scholar

⁵⁸ Sidonius Apollinaris testifies to the employment of this practice in the instruction of Latin Palliata (Ep. 4.12.1–2) and Juvenal ridicules the use of such learned comparisons by the undeservedly educated (i.e. women) to parade erudition at dinner parties (6.436–8); cf. Macrobius 6.2.4, 15. For similar comparisons in the Latin scholia, see Georgii, H., Die antike Äneiskritik aus den Scholien undanderen Quellen hergestellt (Stuttgart, 1891), p.562Google Scholar;Rabbow, P., ‘De Donati commento in Terentium specimen observationum primum’, NJPhP 155 (1897), 305–342, at pp.314–320Google Scholar;Nettleship, H., ‘On Some of the Early Criticisms of Virgil's Poetry’, in Connington, J. and Nettleship, H. (eds.), Vergil, Operasup ⁵ (London, 1898). pp.xxix–1.Google Scholar

⁵⁹ For which see Stemplinger (n. 33), pp. 33–80.

⁶⁰ For Servius′ replies to this type of criticism, see Georgii (n. 58), pp. 49–50, 102 and the index, p. 562; idem, ‘Die antike Vergilkritik in den Bukolika u. Georgika’, Philologus Suppl. 9 (1907), 209–328, at pp. 313, 315.Google Scholar

⁶¹ Cf. Porphyrio ad Hor. S. 1.5.1, 87–88, 2.1.30; Ep. 1.1.45; C. 1.15; Don. An. 204, 483, 592, 801; Eu. 46; Hec. 58; Serv. Eel. 5.32; Aen. (SD) 1.198, 4.1, 10.361, 11.483, (SD) 12.176; and for notes on divergence from the original: Don. An. 977; Ph. 339; Ad. 351; cf. Schol. (A) Il. 2.670, 3.371, 9.70, 22.351. See further Fraenkel, E., ‘Zur römischen Komödie’, MH 25 (1968), 231–242, at pp. 239–242.Google Scholar

⁶² E.g. Don. Eun. 689: ‘colore mustellino: erravit Terentius non intellegens Menandricum illud [fr. 163 Koerte]. ait autem stellionem, quod animal lacertae non dissimile est maculoso corio; namque ad id genus fades exprimitur eunuchorum corporis, quia plerique lentiginosi sunt. hie ergo erravit ideo, quia γαλῆ mustella dicitur, γαλεὡτης stellio’. Cf. Schol. (A) II. 4.439–40.

⁶³ E.g. Don. Hec. 440: ‘imperite Terentium de Myconio crispum dixisse aiunt, cum Apollodorus calvum dixerit (fr. 12 CAF iii.284), quod proprium Myconiis est′. Cf. Serv. Aen. 3.623.

⁶⁴ E.g. Don. Phor. 9 1:‘ Apollodorus tonsorem ipsum nuntium facit, qui dicat se nuper puellae comam ob luctum abstulisse [fr. 16 C4.F iii.285], quod scilicet mutasse Terentium, ne externis moribus spectatorem Romanum offenderet’.

⁶⁵ E.g. Serv. Eel. 2.25:‘nuper me in litore vidi: negatur hoc per rerum naturam posse fieri; sed Theocritum secutus est, qui hoc dicit de Cyclope (Id. 6.35)’. Cf. Serv. Aen. 5.517, 12.725.

⁶⁶ For an example in which Vergil is said to have improved on his Latin original, see Serv. Aen. 9.503: ‘at tuba terribilem sonitum: hemistichium Ennii. nam sequentia iste mutavit. ille enim ad exprimendum tubae sonum ait taratantara dixit [v. 140 Vahlen3 = 451 Skutsch]. et multa huius modi Vergilius, cum aspera invenerit, mutat. bene tamen hie electis verbis imitatur sonum tubarum’. Cf. Don. An. 301, 891; Eu. 539 (but see Fraenkel, MH 25 [1968], 239–242); Hec. 825; Ad. 81, 938; Serv. Eel. 2.23, 2.51, 3.8; Aen. 2.7, (Serv.D.) 9.267; 9.801. For cases in which the imitation is found inferior: Serv.D. Aen. 1.92:‘reprehenditur sane hoc loco Vergilius quod improprie hos versus Homeri transtulerit [Od. 5.297–8]... nam solvuntur frigore membra longe aliud est quam γύτο γονατα, et duplices tendens ad sidera palmas talia voce refert molle, cum illud magis altum et heroicae personae . praeterea quis interdiu manus ad sidera tollit, aut quis ad caelum manum tendens non aliud precatur potius quam dicit o terque quaterque beatil et ille intra se, ne exaudiant socii et timidiores despondeant animo; hie vero vociferatur.’ (cf. Don. Eu. 1001; Serv. Aen. 8.670, 12.266). In this relatively detailed comparison, censure of deviation from the original (in the first argument) is joined to criticism of Vergil's depiction on the basis of common criteria applied to both model and imitation, by which his hero's behaviour is found unmanly, unconvincing, and not befitting a commander.

⁶⁷ Though in his preface to the Georgia Servius does make a distinction between Vergil's closer adherence to his model in the Eclogues and the imitations‘ longo intervallo’ of the Aeneid (cf. Cic. Fin. 1.7).

⁶⁸ See Serv. Aen. 5.517 for Urbanus′ further contribution to the discussion of this passage in Vergilian exegesis.

⁶⁹ Gellius is familiar with at least some of Velius Longus‘ works (Gel. 18.9.4), though he does not mention his commentary on Vergil. It is impossible to decide whether he could have come across the commentary of Aemilius Asper, as this grammarian is dated between Cornutus (whom he mentions) and Iulius Romanus (by whom he is mentioned). Since all we know of Iulius Romanus‘ time is that he, in turn, is mentioned by Charisius in the middle of the fourth century, Asper's can be any time between the later part of the first century and the end of the third; but see Geymonat, M., Enciclopedia virgiliana I (Rome, 1984), pp.373–374, s.v. ‘Aspro’, who attempts a more precise dating.Google Scholar

⁷⁰ Esp. An. 15–21;Eu. 7–13;Ad. 6–1 4; on which see Traina, A., Vortit Barbare: Le traduzioni poetiche da Livio Andronico a Cicerone (Rome, 1970), pp.61–65Google Scholar;Posani, Maria R.,‘Osservazioni su alcuni passi dei prologhi terenziani’, SIFC 37 (1965), 85–113Google Scholar;Garton, C., Personal Aspects of the Roman Theatre (Toronto, 1972), pp.40–72, 89–90.Google Scholar

⁷¹ Theon, Prog, praef., Spengel, Rhet., ii.62–64;Quint. Inst. 10.5.2–11; Hermog. Meth. 24, Spengel, Rhet., ii.445–446. See also: Cic. de Orat. 1.154–5; Sen. Con. 9.1.13;Suet. Gramm. 4, 25; ftia.Ep. 7.9.1–3; D.Chr. Or. 18.12; Fro. Eloq. 5.4, p. 152.1 v.d.Hout ²; Gel. 17.20.cap.; Lucian, Rh.Pr. 9; August. Conf. 1.17; Hieron.Chron. praef. See further Reichel (n. 2), pp. 18–19, 112, 128–30; Stemplinger (n. 33), pp. 118–121, 212–215.Google Scholar

⁷² Quint. Inst. 1.9.2; D.Chr. Or. 18.19. See Beudel, P., Qua ratione Graeci liberos docuerint, papyris, ostracis, tabulis in Aegypto inventis illustratur (Münster, 1911), pp.51–56Google Scholar;Richter (n. 42), pp.71–76Google Scholar;Marrou, H.I., Histoire de l'éducation dans l'antiquite ⁶ (Paris, 1965), pp.387Google Scholar, 596. For the adaptation of similar methods to exegetical practices:Rutherford, W. G., A Chapter in the History of Annotation, being Scholia Aristophanka Vol. III(London, 1905), pp.336–346Google Scholar. And for the relation between this exercise and Latin literary translations:Jocelyn, H. D., The Tragedies of Emius (Cambridge, 1967), pp.25–26.Google Scholar

⁷³ For Greek translations and paraphrases of Latin authors on papyri: Cavenaile, CPL nos. 1–7, 13, 21, 22 (=Pack ² 2940, 46, 50, 48, 51, 22, 23; cf. nos. 2519, 3026e); for Latin translations: Cavenaile, CPL 38–40 (= Pack ² 3010, 52, 172); and see Radermacher, L., ‘Aus dem zweiten Bande der Amherst Papyri’, RhM 57 (1902), 137–151Google Scholar, at pp. 142–145;Moore, C. H., ‘Latin Exercises from a Greek Schoolroom’, CPh 19 (1924), 317–328Google Scholar;Roberts, C. H., P.Ryl. iii.478; for a possible translation exercise by a Roman: Roberts, ‘A Fable Recovered’, JRS 47 (1957), 124–125.Google Scholar

⁷⁴ Quint. Inst. 10.5.5: ‘Neque ego paraphrasin esse interpretationem tantum volo, sed circa eosdem sensus certamen atque aemulationem’; cf. D.Chr. Or. 18.19:

⁷⁵ This holds true even if Gellius′ friend is a renowned author such as Apuleius, as has sometimes been argued; see Dahlmann, H., ‘Ein Gedicht des Apuleius? (Gellius 19, 11)’AAMainz 1979, no. 8.Google Scholar

⁷⁶ These distinctions of Quintilian and Gellius are more akin to Jerome's differentiation between ‘verbum e verbo’ and ‘sensum de sensu’ (Ep. 57.5.2; cf. Schol. Pers. 1.4: ‘verba potius quam sensum secutus’) than to that between ‘verbum pro verbo’ and ‘genus omne verborum vimque’ in Opt. Gen. 14, in which verbatim translation is opposed not to a more general semantic correspondence but rather to equivalence in stylistic qualities and in the effect on the audience. For the variety of Roman notions of translation, see now Seele, A., Römische Übersetzer Nöte, Freiheiten, Absichten: Verfahren des literarischen Übersetzens in der griechischrömischen Antike (Darmstadt, 1995)Google Scholar. Close verbatim equivalence was, of course, always required of translations of legal and business documents and of sacred texts (see Brock, S., ‘Aspects of Translation Technique in Antiquity’, GRBS 20 [1979], 69–87, at pp.69–79)Google Scholar, and there is evidence that faithful translations were also needed by some people for technical manuals (e.g. Vitr 5.4, and see further Richter [n. 42], pp. 77–89). It is possible that the increasing prominence of the requirement of faithfulness from translations resulted from a gradual change in the function of literary translation due to a decline in the knowledge of Greek among the Romans from the second century on; see Russell, D. A., ‘Greek and Latin in Antonine Literature’, in idem (ed.), Antonine Literature (Oxford, 1990), pp.4–9.Google Scholar