Scholarly Reading - Oxford Scholarship Jump to ContentJump to Main Navigation
Aulus GelliusAn Antonine Scholar and his Achievement$

Leofranc Holford-Strevens

Print publication date: 2003

Print ISBN-13: 9780199263196

Published to Oxford Scholarship Online: February 2010

DOI: 10.1093/acprof:oso/9780199263196.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE ( (c) Copyright Oxford University Press, 2015. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see Subscriber: null; date: 01 July 2016

Scholarly Reading

Scholarly Reading

(p.157) 9 Scholarly Reading
Aulus Gellius

Leofranc Holford-Strevens

Oxford University Press

Abstract and Keywords

Gellius is acquainted with Greek scholarship, particularly on Homer, but pays it far less attention than he does Latin. In particular, he makes considerable use of Marcus Terentius Varro and Nigidius Figulus, who both flourished in the mid-first century BC; he is rather less impressed with later grammarians, more than once rejecting the opinions of Verrius Flaccus, with the marked exception of Marcus Valerius Probus, whose written works he knew and whose opinions were reported to him by his elders; Gellius never disagrees with him (as he sometimes does even with Varro), though he may deny the authenticity of a report. He also makes use of Pliny the Elder (not without sometimes unfair disparagement) and Suetonius.

Keywords:   Homer, Nigidius Figulus, Pliny the Elder, scholarship, Suetonius, Varro, Verrius Flaccus

Gellius read few learned writers in Greek except philosophers. He is acquainted with Homeric scholarship,1 but the one problem discussed at length, the cπáρτa of Il. 2. 135, comes from Varro (17. 3. 4); Apion, who wrote a major work on Homer’s vocabulary, provides only marvels. No first-hand knowledge need be supposed of Aristarchus and Crates (2. 25. 4, 14. 6. 3),2 Alexander Aetolus (15. 20. 8), or Apollodorus (17. 4. 5–6). The miscellanies of Sotion (1. 8) and Pamphila (15. 17, 15. 23) may contribute more than we can tell; others too of those listed at pr. 6–9 may have been of service, despite the disdain of §11. But grammar, in its broad ancient sense,3 Gellius pursued in Latin.4

It is no surprise that Gellius should read his authors with the aid of commentaries; named scholars apart, there are references to commentaries on Ennius (18. 5. 12), Lucilius (2. 24. 5, perhaps underlying 11. 7. 9, and also 15. 21, cf. 1313 Marx), Vergil (2. 6. 1, 6. 20. 1), and historical accounts of the gens Porcia (13. 20. 17). (p.158) Nor need they all have been written by authors whom he cites; but to avoid unwarranted speculation, I confine myself to the latter.

The poet Accius’ views on authentic Plautine comedies, like others cited in 3. 3, are known to Gellius through Varro (§9); so, presumably, his assertion that Hesiod was older than Homer (3. 11. 2), and perhaps his reasons (§5),5 but the contemptuous comment on his feeble arguments (§4, cf. the lemma) may be Gellius’ own. Of the other authorities in 3. 3. 1, ‘non indicibus Aelii nec Sedigiti nec Claudii nec Aurelii nec Accii nec Manilii’, from Varro’s treatise on Plautus’ plays, L. Aelius Stilo Praeconinus is described as the most learned Roman of his day (1. 18. 1, 10. 21. 2). He was Varro’s teacher (16. 8. 2, cf. Cic. Brut. 205); all knowledge of him in the Nights is likely to come from Varro6, except for the treatise on propositional logic that Gellius turned up in the Bibliotheca Pacis but found unhelpful (16. 8. 2–3).7 Volcacius Sedigitus, a noted poet (Plin. NH 11. 244), delivers a versified ranking-list of comic dramatists in 15. 24. At 13. 23. 19 Stilo’s son-in-law Ser. Claudius or Clodius derives Nerio from ne and ira; the notice, in postscript form, does not (despite Hosius i, p. xlvii) come from Varro—who, had he mentioned this conjecture, would have refuted it as he refutes Stilo’s etymologies at 1. 18. 2—but from Claudius himself ‘in commentario quodam [cf. 6. 20. 1] Seruii Claudii scriptum inuenimus’. Gellius quotes it for what it is worth (‘cuicuimodi est’), he makes no further reference to Claudius, and does not know that he practised the very mode of Plautine criticism advocated in 3.8. Aurelius Opillus9 is cited in another postscript (1. 25. 17–18) for an etymology that Gellius mentions, as we might in a footnote, (p.159) to forestall the charge of ignorance. There is no other mention of Manilius.

All these writers belong to Varro’s boyhood or his youth: Stilo followed the unbending Numidicus into exile in 100 BC , Aurelius the upright Rutilius Rufus eight years later. The aged Accius conversed with the youthful Cicero; Manilius may be the proconsul of Gallia Transalpina, L. Manlius, defeated by the Sertorian Hir-tuleius;10 Claudius left his library to his brother (or cousin?) L. Papirius Paetus, who gave it to Cicero in 60 BC .11 To this period belongs Hypsicrates, cited by Cloatius Verus for an absurd etymology (p. 183).

Far more attention is devoted to the varied learning of Varro and Nigidius, the greatest scholars of their age (4. 16. 1, 19. 14. 1); Varro was even greater than Nigidius (4.9. 1).12 Neither is infallible: Varro falls into the same error as his teacher (1. 18. 5), gives a definition of a truce more amusing than sound (1. 25. 3),13 and includes some trivial items in his disquisition on the number seven (3. 10. 16); Nigidius’ etymologies of auarus and autumo are unconvincing (10. 5. 3, 15. 3. 5), his view of -osus as always pejorative is wrong (4.9. 12).14

Varro is cited for every field of knowledge in which Gellius takes an interest, and from a wide range of his writings: he tells of Sal-lust in love and Naevius in war, he inveighs against gluttony and prescribes for a banquet, he defines terms of geometry and logic and comments on the caesura of hexameter and trimeter.15 He is (p.160) quoted nearly eighty times, in over seventy chapters, and is the likely source for much else;16 full examination must await a comprehensive modern edition of his fragments.17 Furthermore, he is an authority on Latin usage not only as a grammarian but also as a stylist, which conflicts with the judgements of Quintilian (10. 1. 95), of Gellius’ own admirer Augustine (CD 6. 2), and of modern readers disgusted by the diction of De lingua Latina. The remnants of the satires show that Varro could write more skilfully when he tried; in any case Gellius admired the pre-classical tradition to which he belonged, anacolutha and all.18 A play on words is described as witty (‘lepide … composita sunt’, 1. 17. 5);19 the use of mutare ‘to change’ as an intransitive is praised at 18. 12. 8 (‘in-quit elegantissime’). Varro also gave it the sense ‘to di·er’, not only illustrated in the next sentence but adopted at 2. 23. 7 and 13. 3. 1;20 it is otherwise found only at Catullus 22. 11. At 1. 4. 1 and 20. 1. 34 Gellius uses ‘ad amussim’, found at Varro, RR 2. 1. 26, Sat. fr. 555 Cυbe = B.–A., in preference to the Plautine ‘examus-sim’; it has been plausibly restored in Apuleius (Met. 2. 2. 8)—who also read Varro (Apol. 42. 6)—and is otherwise confined to Gellius’ own imitators Ammianus and Macrobius. It may indeed have been used by other late Republican writers; but Gellius was always likely (p.161) to imitate an author he admired.21 No such sentiment is found in Fronto.22

6. 10 does he write ‘Nigidius Figulus’ instead of ‘P. Nigidius’.

Nor does Fronto take any notice of Nigidius Figulus, mentioned by Apuleius (Apol. 42. 7, from Varro) for his proficiency in divination. Although Gellius occasionally cites his treatises on augury and entrails, and quotes one passage each from those on winds and animals,23 he is mostly concerned with his grammatical writings, which unlike Varro’s are exclusively linguistic with no concern for literature. They were (says Gellius) far less familiar than Varro’s, being more abstruse (19. 14. 2–3): examples follow on Latin phonology, which the reader is left to work out for himself. In 17. 7, on the temporal import of the future perfect, Gellius finds it necessary to explain the argument. Elsewhere Nigidius discusses the accentual contrast in his day between Valéri genitive and Váleri vocative (13. 26. 1),24 and the articulation of personal pronouns (cf. SVF ii. 895) as evidence for the natural origin of language (10. 4. 4). He provides etymologies, and those who seek the origin and basic sense of quin are referred to him at 17. 13. 11; he was clearly a more philosophical linguist than Varro. Although his skill in saying the same thing three times over is observed at 11. 11. 4,25 he is not a model stylist,26 though his technical vocabulary is of interest.27

Nepos, a major source of the chronological chapter, occasionally (p.162) contributes anecdotes. In 15.28 Gellius notes that neither his scholarship nor his friendship with Cicero saved him from misstating the latter’s age when he defended Sex. Roscius, and suggests, without disapprobation, that he was improving on the truth for his friend’s sake (§5).28 Despite his date, Nepos is never cited for his language or his literary merit as at Fronto, Ver. 2. <1>. 17.

The chief Triumviral and Augustan scholars to appear are Tul-lius Tiro, Gavius Bassus, Verrius Flaccus, and Julius Hyginus. Bas-sus’ etymology of persona is praised at 5. 7. 1 (‘lepide mi hercules et scite’), but when that of parcus finds no favour at 3. 19.3 Favorinus has harsh words for ‘your friend Gavius Bassus’.29 Gellius disagrees quite sharply with Tiro (6. 3, cf. 13. 9) and Hyginus (7. 6. 1, cf. 5. 18. 2), and describes a Verrian etymology as ‘too forced and absurd’ (16. 14.4). Indeed, of the six chapters in which Verrius is cited, his opinion is rejected in three; nor is he ever accorded the respectful qualifications that Gellius bestows on other writers, even when he di·ers.30 Although Verrius’ dictionary, like any other, might well be used without acknowledgement, Gellius does not follow the common ancient practice of citing his source only in dissent; we should be slow to posit use of Verrius unless Festus or Paul exhibits the same doctrine and no common or intermediate source is plausible.31 Seemingly Gellius held him in no high regard.32

(p.163) Of post-Augustan writers, Cornutus is cited in only two chapters, and only to be refuted: that he is ‘not in general unlearned’ (9. 10. 5, cf. 2. 6. 1) does not save him in specific instances. An ‘ancient’ grammarian whose name is irretrievably corrupted (Elydis, gen. sing.) records the emperor Claudius’ inclusion of the Aventine within the pomerium (13. 14. 7).33 The illustrious and infamous Q. Remmius Palaemon is not mentioned.

The greatest grammarian of the first century ad was Probus, whose writings Gellius assiduously collected and whose opinions were reported to him by former associates such as Annianus.34 He is described as a man eminent in his age for learning (4. 7. 1), and skilful in the study of ancient authors (9. 9. 12).35 On no occasion does Gellius disagree with him, though Favorinus is made to disbelieve a report as out of character (3. 1. 6), and Gellius himself denies that a certain absurd opinion was found in his works (15. 30. 5; cf. 12. 14. 3–4 for false ascription to Nigidius).

In 13. 21 Probus is said to have stated that the choice between the accusatives plural urbis and urbes should be made not by following pedantic rules but by consulting one’s ear, as Vergil had done.36 At Georg. 1. 25 he had found, in a manuscript corrected (he said) by the poet’s own hand, ‘urbisne inuisere’, where urbes would sound crass; but at Aen. 3. 106 ‘centum urbes habitant magnas’, urbis would be thin and bloodless. Whereas the questioner’s inability to (p.164) appreciate the di·erence roused Probus’ disdain, Gellius not only understands the principle but applies it to other instances.37 Contempt for school grammar, use of allegedly autograph manuscripts, and considerations of sound are all found elsewhere in Gellius; scorn for ‘finitiones illas praerancidas’ and ‘fetutinas grammaticas’ sits easier with Probus and Gellius, in their di·erent circumstances, than with those who taught such stale and stinking rules for their living; and although euphony was not invented by Probus, such discussions are not found in Fronto.

Probus the textual critic appears to have had a taste for conjecture, in particular for moralistic emendation, reading ‘infusum’ for ‘infusus’ at Aen. 8. 406 to remove an obscenity that others considered to have been decorously avoided;38 at NA 1. 15. 18 Gellius reports that late in life Probus had read ‘loquentiae’ for ‘eloquen-tiae’ in the statement at Sall. Cat. 5. 4 that Catiline had eloquence enough but little wisdom (‘satis eloquentiae, sapientiae parum’), because the former was inconceivable without the latter.39 Gellius is too much in thrall to philosophical piety, as well as Probus’ authority, (p.165) to detect the flaw in the argument; but whatever a Cicero might prescribe for the ideal orator, the clear-eyed Sallust could have flirted with no such fatuity40 However, Gellius does not venture to play such games himself: his own textual suggestions are always based on at least the report of a manuscript reading.

Probus’ conception of grammar extended to writing a monograph on Caesar’s use of code (17. 9. 5); but for the broad range of scholarship, Varro’s heir was Pliny, whose remarks on his own and others’ titles Gellius ironically echoes. The implication that he can do better is supported by the ensuing contrast between his predecessors’ indiscriminate accumulation and his own rigorous selection (pr. 11–12). Gellius artfully ascribes the fault especially to the Greeks, but the Roman reader will have associated unselective bulk first and foremost with the Naturalis Historia. Pliny had complained (NH pr. 24) that though the Greek books had marvellous titles, there was nothing in them; Gellius adopts and adapts this same complaint, saying in e·ect that there is both nothing in the other miscellanies and far too much. In §12 he claims to have spent every stolen moment of leisure in reading; Pliny (pr. 18) had written his book in his spare time. This was a commonplace of Roman literary life; Pliny’s ‘subsiciuis(que) temporibus’, which Gellius later resumes and refreshes with a coinage of his own (‘subsiciua et subsecundaria tempora’, pr. 23), can be traced back to Cicero (Leg. 1. 9). But the imitation reinforces Gellius’ contrast between the industrious but uncritical compiler—what better summation of the man, we are meant to think, than Heracleitus’ dictum?—and himself, the judicious and discriminating collector whose work was worth reading.41

It is in this spirit that Gellius uses Pliny’s information. He cites him for facts (3. 16. 22–4, 17. 15. 6), but a·ects disgust at his talk of wonders (10. 12. 6, cf. 9. 4. 12), denying that Democritus wrote the account of the chameleon from which they are taken. One is left to conclude that Pliny not only filled his pages with rubbish, but did not realize it was rubbish. In fact, Pliny, though uncritical enough to accept the text as authentic, was not stupid enough to (p.166) suppose it true; but even now authors are often debited with opinions they rebut.42 Whereas Pliny had enjoyed a good laugh at the lies of the Greeks (NH 28. 112 ‘non sine magna uoluptate nostra cognitis proditisque mendaciis Graecae uanitatis’, cf. ‘deridiculae uanitatis’ NA 10. 12. 4), Gellius has to assure us—with the high moral tone of one who would touch pitch and not be defiled—that he finds this matter utterly boring (§1 ‘ex quibus pauca haec inuiti meminimus quia pertaesum est’) but reproduce it as a public service, to demonstrate the harm such things do to those who love knowledge (§4).

At 9. 4. 13 he observes that Pliny supports one marvel with his eyewitness—as Apion had the tale of Androclus and the Lion (5. 14. 4; cf. 6. 8. 4–5),43 describing him as a man of great authority for his intellect and rank; at 10. 12. 4 he implicitly includes him among ‘the cleverest minds, especially those intent on learning’; at 9. 16. 1 he states that Pliny was considered the greatest scholar of his age, but promptly takes him to task for missing a counter-argument in a controuer si a. He never denies his erudition, but insinuates that the great polymath would have done better to learn less and think more.44

Suetonius is named only twice,45 but will also be the source for Augustus’ ‘Hasten slowly’ (Aug. 25. 4, NA 10. 11. 5)—though not for its reportedly being in his letters46—and the measures taken against philosophers and rhetors (Gram. 25. 1, NA 15. 11. 1–2). Other ascriptions made by Ruske rest on Rei·erscheid’s over-ambitious collection of the fragments47 and the source-critic’s calculus, (p.167) in which two and two make fifteen. Indeed, Gellius may have used him oftener than we can tell;48 yet Suetonius (whom Fronto cites at Amic. 1. 13) was perhaps too well known for frequent copying to be prudent.

His contemporary, Caesellius Vindex,49 was much criticized by later authors: both the Hadrianic grammarian Terentius Scaurus and Gellius’ teacher Apollinaris had joined in the sport, which was not always conducted fairly (6. 2. 2). Against Caesellius’ statement that verbal adjectives in -bundus were simply equivalent to present participles (11. 15. 2), Scaurus maintained that they denoted simulation of the action,50 which Gellius saw to be absurd (‘What was meant by acting or imitating one at play, I would rather appear not to understand than suggest that Scaurus himself did not understand’, §4). Instead, he should have blamed Caesellius for not examining whether the suffix conveyed a nuance of its own: remembering a passage of Sisenna in whichpopulabundus, ‘laying waste’, is used with a direct object, Gellius consults Apollinaris, who derives -bundus from abundantia. Gellius applauds (‘εύεπιβόωc hercle’); by contrast, in 2. 16, despite repeating his teacher’s criticism of Caesellius, he is not convinced by his alternative theory.

Yet despite their eagerness to do down Caesellius, many scholars (p.168) had failed to notice that he had taken cor in a line of Ennius’ for a masculine through not connecting it with what followed. In NA 6. 2 Gellius sets out Caesellius’ discussion and refutes it;51 he throws in an absurd defence that does not merit a reply, thus bolting the door against all foreseeable objections.52 He need not depend on Scaurus or Apollinaris either here or at 3. 16. 11, where Caesellius wrongly takes Livius Andronicus’ ‘Morta’ for the name of an individual Fate and not generically ‘quasi Moeram’;53 Gellius could observe that the original had Moipa, perhaps even that Homer never names his Moipai. This error earns Caesellius the contemptuous qualification ‘homo minime malus’;54 at 20. 2. 2 he is ‘homo ingenuae ueritatis’ for admitting that he does not know what instrument a siticen played.55 At 18. 11. 1 he is ‘by no means unlearned’; but prompt rebuke follows for wantonly and ignorantly (‘petulanter insciteque’) censuring the verbal coinages of Furius Antias. Only once is he vindicated outright: at 9. 14. 6 he states that Cicero, at Sest. 28, used the genitive singular dies, which Gellius confirmed from ancient manuscripts. For all Caesellius’ faults, he had a pair of eyes.

Scaurus does not appear in any other chapter; his contemporary Velius Longus is cited only at 18. 9. 4. A schoolmaster and a scholar are arguing over a word in Cato: the former maintains that it should be insequenda,56 and that Ennius had written inseque in the imperative. (p.169) The latter answers that Velius should be believed when he says that Ennius had written insece; that was why the ancients had called narratives insectiones (our only evidence for this word); and Varro, reading sectius at Plaut. Men. 1047, had interpreted it as ‘worthy to be narrated’.57 It is formally uncertain whether the references to insectiones and sectius belong to the quotation or the narrative,58 but probable that Velius is the source.59

Gellius himself settles the argument to his own satisfaction in §5 by citing insece from a ‘really old’ manuscript of Livius Andronicus he had seen at Patrae, no doubt taken there by one of Augustus’ colonists into whom the text had been beaten as it had into Horace. He adds that sectius proves nothing (§6)60, ° that insece, though smoother and lighter (‘lenius leuiusque’, cf. Cat. 84. 8) than inseque (§7), was derived from the same root, that of sequor, just as secta was (§8), and that experts in Greek took Homer’s ἓυυεπτε and ἓcπετε for its kindred, as also ἓπη ‘words, verses’, derived from ἓπεcθaι ‘to follow’ and ει̑πεîv ‘to say’ (§§9–11).61 We need neither deny that Gellius personally inspected the manuscript of Livius and devised the arguments of §§6–8 for himself, nor make Velius the source of §§9–11 (indeed his insece conflicts with §9 inseque); much less invoke Verrius Flaccus on the not strength but weakness of PF 99. 10 ‘inseque apud Ennium dic. insexit dixerit’.62

Since Gellius’ pinpricks against Pliny are not always justified, (p.170) we may suspect unfairness or sleight of hand in his dealings with other authors; but we are not entitled to suppose it in any given case merely because it is requisite for our speculations, only to deduce it from solid evidence for what stood in his source. This is not a presumption of innocence (scholarship is not a lawcourt), but a precaution against wild theorizing. He is capable of independence even towards Varro and his teachers; if any grammarian holds him in subjection it is Probus;63

Gellius’ hopes of passing on his erudition were amply fulfilled in an age perhaps later than he had envisaged. Nonius Marcellus draws heavily on him even while calling him an ‘old writer of obscure authority’;64 and never acknowledging him by name;65 the debt of the great Donatus may be measured by the Gellian material in Servius and in Deutero-Servius66, though it remains strange that no echoes have yet been found in his most famous pupil Jerome. Macrobius owes him as much for content as Ammianus for language, though neither names him. One C. Aurelius Romulus, given at least a partial copy, expressed his appreciation in verses found at the end of bk. 9 in our manuscripts (AL 904).

Cecropias noctes, doctorum exempla uirorum,

donat habere mihi nobilis Eustochius;

uiuat et aeternum laetus bona tempora ducat,

qui sic dilecto tanta docenda dedit.

Th’ Athenian Nightes, the Schollers Paragon,67

Gentle Eustochius maketh bee mine Own.

May hee for euer florishe in good Heart,

That gaue his Freind soe manie thynges t’impart.

(p.171) It is evident from docenda that the work is to be used as a resource for teaching others, not merely for perfecting Romulus’ own erudition.68


(1) . NA 2. 6. 11 ∼Σ Il. 4. 223–4, Eust. 465. 24 (but cf. Hermog. Id. 1. 12, p. 306. 8–10 Rabe); NA 3. 16. 17∼Eust. 1682. 27–8 (unless really from Favorinus); NA 6. 20. 5∼ ΣT (‘Towneley’) Il. 22. 152, Σ Od. 11. 596, Eust. 1702. 19–22 (cf. Holford-Strevens, ‘Nola’, 393 n. 1); NA 13. 21. 25∼ΣT Il. 16. 583, 17. 755; NA 13. 25. 21∼ΣT Il. 2. 8; NA 18. 9. 9∼ΣT Il. 2. 484, Eust. 1381. 28 (the equation of inseque and ἓννεπε probably came from a commentary on Ennius). The moralistic comments at 1. 15. 3, 11 on ἒρκoc ?ΔÓΝΤΩΝ (cf. ΣT Il. 4. 350) and ἐκoλῴa (cf. Eust. 205. 7–8) will be due to Favorinus, the interpretation of aἴθoπa oἶvov at 17. 8. 10 (cf. Eust. 135. 34–6, 1449. 20–2, 1854. 3–4) to Taurus. Cf. also Ch. 8 n. 25.

(2) On whose debate see Alessandra Buonajuto, A8&R2 41 (1996), 1–6.

(3) That is, the study of literary texts and all things necessary to their understanding: e.g. Dion. Thrax, §1 with 2, Quint. 1. 4. 2, S.E. Math. 1. 91–3.

(4) Knowledge of Greek technical terms, e.g. ἀπaρέμφτoν ‘infinitive’ 1. 7. 6, came with the study of Greek; they were the more necessary since Latin usage was not yet fixed (note ‘per infinitum modum’ 15. 13. 9 of the impersonal indicative ‘est ueritum’, in contrast to Quintilian’s use for the infinitive, cf. Italian infinito; modus infinitiuus in our sense not till Char. 216. 3 Barwick). Reference to Attic usage, real or supposed (pp. 231–2), reflects the fashion of the day but not a serious study of the Attic authors.

(5) ‘inuolgatum esset’, from a verb never found before Gellius, scarcely after him, but used at five other places in the Nights, indicates the wording is his own; ‘itidem’ is common in Early Latin, but also characteristic of Gellius, and ‘procul dubio’ is unlikely for Accius (Ch. 13 n. 38).

(6) Explicitly in 1. 18, 10. 21; perhaps through the additional filter of Verrius 2. 21. 8 and of Sinnius Capito 5. 21. 6. At 12. 4. 5 the citation is confessedly indirect, presumably through Varro (Kretzschmer 52–3), through whom comes too Stilo’s judgement (Quint. 10. 1. 99) that if the Muses spoke Latin they would speak like Plautus.

(7) For his comment that Stilo had written it ‘rather to remind himself than to teach others’ (16. 8. 3) cf. 17. 7. 5 on Nigidius.

(8) Cic. Fam. 9. 16. 4. Ritschl 92 n. 1 allowed, and Mercklin 643–4 asserted, that Gellius had access to Claudius’ index; the mode of reference, as an item in a list, suggests otherwise, nor is there reason to suppose that Claudius wrote after Varro (cf. Kretzschmer 43 n. 1).

(9) The correct form of the cognomen, despite ‘Opilius’ in Gellius’ MSS: see Suet.Gram. 6. 4, Funaioli, GRF 86. Gellius does not remark that stribiligo, said at 5. 20. 1 to be an old word for soloecismus, was Opillus’ term (fr. 17).

(10) See Schanz–Hosius i4. 605–6, cf. E. Rawson 273 n. 43.

(11) Cic. Att. 1. 20. 7, cf. Starr, ‘Used Book Trade’, 154–5.

(12) Cf. Serv on Aen. 10. 175; perhaps from a common source (the Suetonian life posited by Della Casa, ch. 1?), but Gellius knew both well enough to form his own judgement, even if it confirmed another’s. It proves nothing that only here and at 7.

(13) Stephanus to Vulcob 16 states that Gellius expresses ‘tribus elegantissimis modis’ the notion se non è υero è ben troυato; he does not cite the instances, of which there are five (1. 25. 3, 12. 14. 7, 15. 3. 5, 18. 1. 12, 20. 1. 35; cf. also 1. 23. 8, 3. 19. 4, 7. 8. 5, 15. 9. 7). Cf. Plat. Euthyd. 305 E , 489 B , Arist. Cael. 290b14, 295b16, Pol. 1291a11; Satyrus, Vit. Eur. fr. 39 xviii 17–20, Lucr. 2. 644–5, and other passages cited by Lightfoot on Luc. Syr. D. 16.

(14) See Otto Schonwerth and Carl Weyman, ALL v. 212–13 (and p. 203 on Nigidius’ bibosus, cf. NA 3. 12).

(15) Sallust 17. 18 (cf. p. 253), Naevius 17. 21. 45. Gluttony 4. 19, banquets 13. 11 (cf. 6. 16, 13. 31). Geometry 1. 20 (also 16. 18, see §6), logic 16. 6, caesura 18. 15(cf. 16. 18. 5, Ch. 16 n. 106; but the Ennian term for hexameters, ‘longis uersibus’ (Op. inc. 20 Skutsch) may be Gellius’ own touch.

(16) Most obviously at 18. 14, which will come from Varro, De disciplinis; in general see Kretzschmer 44–54, Ruske 21–33.

(17) Such an edition—replacing that in the Bipontina of 1788, drawn not without error from Ausonius Popma (Leiden, 1601) and the Berewout Opera omnia (Dordrecht, 1619)—has been begun by Marcello Salvadore with a ‘Supplementum’ (Hildesheim, 1999), covering fragments not available in modern collections.

(18) Varro was not an archaizer, using old expressions for e·ect, but a stick-in-the-mud, writing as Romans had written in his youth; cf. Lebek 331 n. 96. In his defence, and emphasizing his variety of styles, H. D. Jocelyn, RFIC 108 (1980), 121–2.

(19) ‘Vitium uxoris aut tollendum aut ferendum est’ (Rolfe: ‘A wife’s faults must be either put down or put up with’). The literal meanings of the verbs overlap, the figurative are in contrast.

(20) In Early Latin mutare is intransitive only in the sense ‘to change one’s mind’ (Plaut. Rud. 865, cf. Cic. Fam. 16.1. 1, and ‘non demutabo’ Plaut. Pseud. 555, 566, cf. Stich. 725); in the general sense of changing it is frequent in Livy, present in the Senecas and Tacitus, and a mannerism of Tertullian’s;cf.μετxaλλάττεν. At Lucr. 1. 787 ‘mutare’ means ‘to change places’, cf. mutitare at NA 2. 24. 2; 18. 2. 11 in connection with the Megalensia and Verrius Flaccus’ mutitationes at Fast. Praen. for 4April. At 2. 23. 7 ‘quantumquemutare’, like Man. 4. 56, substitutes the intransitive for the passive of Verg. Aen. 2. 274–5.

(21) On other echoes of Varronian usage see Ch. 3 n. 22, p. 117, Hache passim, Knapp 159 (on paupertinus), Foster 42 (ruminor). Aper mocks those who prefer Sisenna and Varro to Aufidius Bassus and Servilius Nonianus (Tac. Dial. 23. 2).

(22) At Nep. am. 2. 3 ‘fata a fando appellata aiunt’ Fronto is repeating what everyone knew, not specifically citing Varro, LL 6. 52.

(23) At 6. 9. 5 the paradosis is deligitur CPR; V’s diligitur, ably championed by Timpanaro, Contributi, 120–3, will be error correcting error.

(24) See Mariner, ‘Paradoja’; Leumann 425; and on the whole chapter Bernardi Perini, ‘Sistema’; Belardi–Cipriani, and Cavazza’s comm.

(25) The Annaei, who had the misfortune to be born in an unfavoured age, reap harsher words from Fronto (Orat. 4–6).

(26) 3. 12 his form bibosus, ‘boozy’, is ‘noua et prope absurda’ (cap.), irregularly derived and supported only by Laberius (cf. Ch. 10 n. 34); despite the lemma, his o·ence was to recognize rather than to use it.

(27) He calls planets errones (3. 10. 2 from Varro, 14. 1. 11), barbarous language rustic (13. 6: Greeks corrected foreigners, Romans polished the country cousin), accents uoculationes (13. 26. 3, cf. 13. 6. 1), and the genitive the casus interrogandi (13. 26. 3–4; cf. 20. 6. 8). On this the bibliography is long and depressing (see Della Casa 83–5, Bernardi Perini, ‘Sistema’, 2 n. 4=Mincio 122 n. 4); against suggestions of misunderstanding see Cavazza, comm. 246–52. Least bad of explanations is that alone among the cases properly so called the genitive may by itself constitute the predicate in a question about ownership or other relations (J. C. Scaliger, Probl. C , pp. 31–2). For another Nigidianism see Ch. 2 n. 35.

(28) Cf. 1. 3. Mercklin’s rule, and the representation of Asconius as correcting only Fenestella (whom in the surviving commentaries he cites five times, thrice in disagreement), suggest that Nepos (whom Asconius is not known to mention) was used directly.

(29) ‘Superstitiose … et nimis moleste atque odiose confabricatus commolitusque magis est originem uocabuli Gauius iste Bassus quam enarrauit’. With 5. 7. 2 (fr. 8 Fun.) cf. Boeth. Eutych. 3 (Chadwick, Boethius, 193), GLK viii. 202. 2–4, 248. 33–249. 2 and n.

(30) NA 5. 18. 1–2 (but Verrius has his doubts); 16. 14. 4, 17. 6, where Gellius’ view is widely accepted (e.g. Kaser, Priv. i. 330 n. 7, Sblendorio Cugusi on Cato fr. 113=158 Malc.; otherwise Luigi Labruna, Index, 17 (1989), 167–84). Courtesy in dissent: 2. 6. 1, 6. 3. 8, 9. 10. 5, 18. 11. 1; cf. [Long.] 4. 6; Quint. 9. 1. 18; Macr. Sat. 5. 19. 2–3. Similar compliments are bestowed on anonymi at 13. 29. 2, 14. 6. 1, 18. 5. 2.

(31) For alleged parallels see Mercklin 701 n. 1, ‘Capita’, 4–11, review of Kretz- schmer, 723–4. The most convincing is 10. 3. 10∼PF 28. 19–21; the connection between 2. 21. 8–10 and Fest. 454. 36–456. 11 may be use of Varro, between 4. 6. 8 and Fest. 242. 11–19 of Capito (but see Strzelecki’s edn. of Capito, p. xxiv n. 6). (Is Festus the source of Aldhelm, Ep. 4 at MGH Auct. Ant. Xv. 483. 7?) Although Ser. Sulpicius will be cited at 4. 3. 1–2 through Neratius, §3, tacked on with little relevance, conceivably comes from Verrius, cf. Fest. 248. 2–6; but Festus may have used Gellius or some intermediary Leopoldo Gamberale, RFIC 90 (1970), 194–8 derives ‘profestum et profanum’ (pr. 20) from Verrius (cf. Fest. 298. 31), but the collocation came readily to the ear (cf. 2. 17. 5). For divergence from Verrius cf. Ch. 11 n. 142, Ch. 16 n. 87.

(32) Kretzschmer 69 (yet even he strains to derive 5. 18. 6–9 from Verrius, pp. 70–1, along with notices in Servius and Isidore), cf. ‘Zu A. Gellius’, 368. But ‘si qui sunt’ (17. 6. 4) is not contemptuous; cf. 12. 3. 4.

(33) Cf. the Lex de imperio Vespasiani, ll. 14–16 (where Cola di Rienzo, cit. Crawford i. 551, took pomerium in its medieval sense ‘giardino’).

(34) I do not assert against Jocelyn I 465 that Gellius’ accounts of the oral tradition, even after sifting (3. 1. 6, 15. 30. 4), accurately represent Probus’ opinions, nor against id. I 471, III 472–4 that Donatus’ and Servius’ explanations of his readings are authentic; but, finding in those accounts and explanations a coherent quasi- personality, I treat it as the Probus in whom Gellius believed and by whom he was influenced. The quest of the historical Probus I leave to others, and with it the question by whom that other Probus was created.

(35) Against Zetzel’s suggestion (pp. 41–54) that Probus touched no author older than Terence see NA 4. 7, Don. on Ter. Phorm. 372 (though Plaut. Truc. 265 is missed).

(36) Cf. J. S. Th. Hanssen, SO 22 (1942), 80–106; N. I. Herescu, AC 22 (1953), 89–95. But each instance was decided by the poet as it arose, not always with the like care or by the same considerations; rational analysis is not always possible (Quint. 9. 4. 110).

(37) ‘Nos autem aliud quoque postea’ §10 recalls 3. 3. 7–8, where Gellius interrupts his summary of Varro (for self-interpolation cf. Ch. 6 n. 75, Ch. 11 n. 4, p. 254) and 1. 7. 18, where Gellius reports finding other accusatives for ablatives than those adduced by his friend (see 17. 2. 11, cf. Caec. 170 Ribbeck at 2. 23. 21, Enn. fr. 34 Courtney at 19. 8. 6); he then passes from the legitimacy of the construction to its rhythmical motivation in Cicero, and that of the verb-form explicauit, citing the same speech by title twice (§§16–20) without adverting to the fact. Various notes have been tacked together; the friend, having linked the accusatives of §§16–17 with the infinitives of §§1–15 as usages wrongly impugned, is discharged when Gellius changes theme. Yet the source-critics assign the whole chapter to Probus!

(38) The words coniugis infusus gremio, superficially ‘spread out on his wife’s lap’, hint at a second sense, ‘poured into his wife’s womb’ (D.S. notes the ‘enphasin coitus’; for ἒμφacιc cf. Ch. 7 n. 33), of which the prudish Probus disapproved (cf. Don. on Ter. Phorm. 1005) but not the saintly Augustine (De nupt. et coni. 2. 8 19, CSEL 42. 271. 18–19; CD 14. 26), and which is reinforced by the verbal patterning (infusus between coniugis and gremio, the entire phrase centred within the splayed placidum … soporem). Others objected that Probus’ infusum was meaningless and gave soporem two epithets; Ti. Claudius Donatus had no difficulty with marital intercourse: ‘per amplexus cetera quoque quoque consecuta intellegenda sunt … infusus est ergo quem amoris liquefecerat calor. See too Annianus at NA 9. 10. 1 (p. 149) with Hosius’ parallels. For profound reflections, see Montaigne, Essais 3. 5.

(39) Jocelyn, observing Suetonius’ silence (I 466), denied that Probus’ readings could be proved conjectures (III 472); but what proof could we now have? Tim-panaro, Stor. fil. virg. 87 allowed that ‘infusum’ was one, but ibid. 122–6 took ‘loquentiae’ for an error of transcription, as if ‘sic a Sallustio relictum’ could point only to MS evidence. It is not known to Plin. Ep. 5. 20. 5 (who credits another with the coinage), Fronto, Ep. M. Ant. 3. 1. 1, Orat.1, Sidon. Ep. 9. 9. 2, Grillius 12. 87–8 Jakobi, Prisc. GLK iii. 85. 4.

(40) Note the ‘Machiavellian’ use of uirtus at Cat. 53. 6, cf. Syme, Sallust, 242, Jocelyn on Enn. Trag. fr. LXXI. Probus adds that Sallust was a word-coiner (which, be it said, does not prove him the source of NA 4. 15).

(41) On the relation between the prefaces see Schröder 57-60; Stevenson 129–31. In turn, some 250 years later, Macrobius would appropriate Gellius’ language for a preface promising order and completeness: Macr. Sat. pr. 2–3 ∼Gell. NA pr. 2–3, 12.

(42) Gellius himself so su·ers at D.S. on Georg. 1. 260∼NA 10. 11. 1, Rabelais, Gargantua, ch. 3 (i. 40 Lefranc) ∼NA 3. 16. 16; cf. Jac. Gronovius on 2. 6. 5 (edn. 135 n. 8). See too Ch. 6 n. 110. In G. K. Chesterton’s story ‘The Invisible Man’, John Turnbull Angus’s uncle ‘had disinherited him for Socialism, because of a lecture which he had delivered against that economic theory’ (The Innocence of Father Brown, ed. Gardner, 102).

(43) Pluris est oculatus testis unus quam auriti decem, and Gellius does not like to give the lie direct. With Pliny’s tale of a sex-change cf. Phlegon, Mir. 9, dated to 116; Apion is telling a well-known folktale (Thiele, edn. of ‘Romulus’, pp. xxiv f.; Stith Thompson B381; yet cf. Sen. Ben. 2. 19).

(44) It makes no di·erence that Gellius owes Pliny more than he admits (e.g. 17. 15 comes from NH 25. 48–52, 61; cf. Mercklin 670, but see Ch. 13 n. 27). How much more is uncertain: Beck’s hypotheses are unconvincing; Jorge Avilés, Helmantica, 29 (1978), 91–8 advances us no further.

(45) NA 9. 7. 3 (see Ch. 16 n. 117), 15. 4. 4.

(46) Cited directly at 15. 7. 3; cf. 10. 24. 2, Gascou 501–2. For the adage and its fortune see Ch. 10 n. 20.

(47) Ruske 44; but see Wallace-Hadrill 41–2.

(48) Rei·erscheid’s argument (edn. 424–5, cf. Wallace-Hadrill 58) for making him the source of 15. 28—that since Jerome, whose Chronica took Roman dates from Suetonius’ De uiris illustribus, despite the misdating of Pro Quinctio to 83 BC (and of Pro Sex. Roscio to Ol. 175. 2 = 79 BC; cf. Helm 28–9), concurs as to Cicero’s age (Ol. 174. 2 ‘XXVI anno aetatis suae Cicero Quintium defendit’) with Asconius, whom Suetonius/Donatus cites in the Vita Vergilii, against Nepos and Fenestella, who appear in Suetonius’ life of Terence, therefore Gellius too used Suetonius, not Nepos (but see n. 28) and Asconius—is at any rate better than Della Casa’s (p. 35 n. 52) for deriving 19. 14. 3 from Suetonius because in uulgus exire (cf. NA 12. 12. 3, 14. 3. 3; Fronto, Amic. 2. 2) and obscuritas (Ciceronian; cf. NA 19. 10. 8) are found in that author. To be sure, Asconius is cited without specific reference; but if the place was his preface to Sex. Rosc., it hardly needed stating. However, Gellius’ sources may be Nepos and Suetonius. NA 13. 2. 1 may suggest Suetonius rather than Varro; the story is noted by Jerome, Chron. Ol. 160. 2. Suetonius may underlie the ‘plerique alii’ of 3. 3. 14, cf. Jerome, Chron. Ol. 145. 1 (whence, not from Gellius, Boccaccio, Geneal. deor. gent. 14. 4, ii. 690. 9–14 Romano).

(49) Whose other fragments confirm both his interest in the ancient authors and (if GLK vii. 206. 26–7 is authentic) his capacity for misunderstanding them. M. T. Vitale, SRIL 1 (1977) 221–58 is more than fair to him; she discusses the Gellian references on pp. 224–40. Cf. A. M. Tempesti, ibid. 179–84.

(50) So too the pack of Latin grammarians; Hosius ad loc. omits Serv on Aen. 4. 646, against whom Beroaldus, Ann. Seru., sig. [b5]v–[b6]r cites ‘moribundo similis’ (Val. Max. 5. 7. ext. 1; Q. Curt. 9. 5. 10); see too Valla, De linguae Latinae elegantia 1. 9 (Opera i. 13), Pianezzola 18–21.

(51) Ann. 371–3 Skutsch: ‘Hannibal audaci cum pectore de me hortatur | ne bel- lum faciam, quem credidit esse meum cor | suasorem summum et studiosum robore belli’: Hannibal, whom Antiochus had thought the keenest advocate of war, counsels against it. Caesellius, ignoring 373, and taking ‘cum’ for a conjunction, understood a question ‘quem credidit esse meum cor?’, ‘what did he believe my [heart =] intelligence to be?’, i.e. ‘what kind of fool does he take me for?’ Nonius copies the wrong interpretation at 195. 17–20 (i. 287).

(52) Cf. 1. 25. 18, 9. 14. 26, 17. 13. 7–9.

(53) Fr. 23 MBB (from Od. 2. 99–100 and parallels or 10. 175, not 3. 237–8, where the subjunctive is generic). Whether Caesellius had some right on his side is disputed (Latte 53 n. 1; G. Radke, KlP iii. 1431 and literature there cited). ‘… men set up the Deity of Morta’, Sir Thomas Browne, Hydriotaphia, ch. 4 (p. 115 Martin).

(54) i.e. ‘uir simplex’ (Proust). See Cic. In Caec. 45, Rep. 3. 26, . 3. 39; for the notion of naïυeté (already implicit at Plaut. Mil. 356) cf. ἒκaκoc, εύήθηc, bonus (e.g. NA 7. 16. 3), German harmlos, treuherzig, and English ‘innocent’. Cf. Liv 40. 14. 5 ‘minime malus ac suspicax’ (ironical); Querolus 64 (‘à’ ce brave homme’ Jacquemard-Le Saos); Cavazza, ‘Appunti’.

(55) Cf. Cic. Orat. 230‘ uirum simplicem, qui nos nihil celet’.

(56) Since the beginning of this chapter is known only from the corrupt (and now vanished) β, we do not know the context of the word, missing from what remains of the citation: but it should mean ‘to be narrated’.

(57) The direct tradition exhibits setius, used as at Capt. 417, which makes better sense. (β’s minus, not supported by the paraphrase, must be apost-Gellian corruption for mihi.)

(58) ‘Velio Longo … fidem esse habendam qui … scripserit non inseque apud Ennium legendum sed insece, ideoque a ueteribus quas narrationes dicimus insectiones esse appellatas, Varronem quoque uersum hunc … sic enarrasse’: are ‘esse appellatas’ and ‘enarrasse’ parallel to ‘esse habendam’ or ‘legendum’?

(59) If Gellius took the whole of §9 from Velius for the starting point of his story, he would have been less conscious of the ambiguity, which could have been avoided by quod clauses (Ch. 3 n. 37). Cf. Kretzschmer, De A. Gellii fontibus, 93.

(60) Nihil in alteras partes [Hertz: β plus] argumenti habet’; for alter after a negative =‘either’ cf. Liv. 25. 28. 4, 28. 18. 4, 40. 20. 4. See Holford-Strevens, ‘Aduer-saria’, 112.

(61) inseque is congeneric with ἒvvεπε (< *enseque), pl. ἒcπετε (< *ensquete), related to sequi/ἒπεcθaι, ‘say’, and ‘see’; but εἰπεîv (Fειπεîv) and ἒπoc are related to uox. Gellius’ ‘sequo [cf. Prisc. GLK ii. 396. 21; ἒπω Σ Il. 2. 484] … et sectio’ suggests a link with secare, cf. Non. 404. 7–9 (ii. 649), Serv on Aen. 10. 107, Isid. Orig. 19. 19. 8. J(avier) Velaza, RPh3 72 (1998), 259–68 unwarrantably calls insece ‘una invención arcaizante de Gelio’; see too Gerald M. Browne, Mnem.* 53 (2000), 711–12.

(62) Mercklin, ‘Citiermethode’, 660–1.

(63) Yet since in 6. 7 the intermediation of Annianus permits Gellius to disagree with Probus yet not say so, the enthralment was less than total; see too Ch. 11 n. 78. By whatever routes, besides direct reading (15. 30. 5), Probus’ real or supposed opinions reached Gellius (Paul Wessner, RE2 iva. 741–3 opted for Caesellius and Apollinaris) they made a deep impression on him. But others will have attempted to apply the master’s methods for themselves, as Gellius does in 13. 21; not everything of a Probian spirit need come from Probus.

(64) apud ueterem auctoritatis obscurae’ (493. 3–4, iii. 791), citing 1. 17. 2 (cf. Lindsay’s index, edn. iii. 943–9); but at 121. 21 (i. 175) ‘honeste ueteres dixerunt’ is followed by a quotation credibly supposed to come from 8. 3.

(65) For details see Mercerus, edn.2, comm. 89–90 et passim; Hertz, A. G. und N. M.’; and the analyses in Lindsay’s monograph (cf. ibid. 104 n. f).

(66) For the latter see Van Hoye, rightly excluding a common source.

(67) I assume that the exempla are not those left by the learned men whom Gellius cites but the Attic Nights themselves as a model for emulation by the learned men who shall read them.

(68) Even were Bücheler’s fruenda palaeographically plausible, docenda, in a line with two other cs and three ds, is protected by Romulus’ taste for sound-play (‘noctes doctorum’, ‘aeternum laetus’), which I confess to not replicating in my translation. (Its Elizabethanizing style is my riposte to a reviewer who regretted the lack of whimsy in my first edition; but it also proved the most congenial for rendering the epigram.)