¹ The Graecizing sense ‘ineffable’(ἄϕατος) in the negative theology of ‘Apul.’ Plat. 1.5(190) need not detain us.

² A perversity typical of Mosellanus, who finds the same sense at Hor. AP 130 and detects in our passage a paradox like Jerome, Ep. 14.1 (CSEL 54.45.16–17) ‘affatim diues est, qui cum Christo pauper est’.

³ Carrio: ‘at vt; inscitum sit quod non sciatur, quemadmodum hie capi existimo, plane nouum est. et erunt qui earn uocem in vulgato significatu volent exaudire. quibus non repugno.’

⁴ ‘Specimen emendationum in A. Gellii Lib. I, II, III & IIII’, in Krause's, J. G.Nova litteraria anni M DCC XXI (Leipzig, 1721), 138–43 at 140Google Scholar; at Amoenitates philologicae (Amsterdam, 1729), iii.224Google Scholar, he cites the parallel in answer to D. W. Triller, who in a letter of 12 June 1722 (Copenhagen, Det Kongelige Bibliotek, MS. E don. var. 6,2°, p. 5), gave indictum the unattested sense λγου οὐκ ἄξιον, and to Lackmann, A. H., who in a lecture at Eutin in 1722 asserted without argument: ‘indictum si per non dictum explicatur, sensui certe omnia optime conveniunt’ (Primitiae Utinenses (Lübeck, 1725), 55).Google Scholar

⁵ Cretic + molossus is a favourite Gellian clausula, cf. Holford-Strevens, L. A., Aulus Gellius (London, 1988), 44Google Scholar; for the adnominatio ‘sequi…adsequi’ cf. ibid. 20 n. 5.

⁶ Note too ‘inculcauit’ §12 ∼ ‘inculcantes’ Off. 1.111. If despite Holford-Strevens, Aulus Gellius, 68 n. 45 the five πθη of 19.12.3 derive from Off. 1.69 (note ‘carere et uacare’ ∼ Cic. ‘uacandum’), then perhaps anger had already been interpolated into Cicero's text (where it produces bad balance and worse doctrine) from 1.89.

⁷ Still so, for C omits the end of §4 by saut du même au même from ‘necessum est’ to ‘uisum est’.

⁸ LCM 15 (1990), 144Google Scholar, commenting on Holford-Strevens, Aulus Gellius, 37 n. 26.

⁹ Printed by Ascensius (whose own editions all refuse the correction) and sold by Jean Petit (who had also published Aegidius Maserius' edn., cit. Marshall on 2.2.7 dum inspicimus as ed. in Bellouisu facta 1508); reissued by Barth. Trot (Lyon, 1512), and thrice more in Paris. Inscriptionis also appears in Florence, Laur. 54.11, 25, 26, 31; London, BL Burney 175; Venice, Marc. lat. XIII.58 (3917); Lucca, Archivio arcivescovile, Bibl. Feliniana 478 (ipsiusque inscriptionis); Vat. Barb. lat. 123, 145, Ottob. lat. 2062 (ipsis inscriptionis). Florence, BNC II.i.72, Munich, Clm 5359, and Vatican, Chigi H VII.237 substitute it for the second scriptionis instead of the first.

¹⁰ Chicago, Newberry Lib. f90; London, BL Burney 176; Oxford, Bodl. Lib. E. D. Clarke 20; Rome, Bibl. Casanatense 1085 (‘ipsum ac’); Vat. Barb. lat. 145, Ottob. lat. 2019 (corr. from ipsos), Vat. lat. 1537. Cf. LCM 14 (1989), 151Google Scholar. Hosius and Marshall write ‘ipsum ς’; the siglum is better reserved for the text inherited by the recentiores, not readings in this or that MS.

¹¹ ‘For they say that in the one (ἔννεπε the ν has been doubled, in the other (ἔσπετε) the σ has undergone metathesis [i.e. *σπετε > ἒσπετε].’

¹² Hertz also reports that for tralatum Z reads translatum.

¹³ Preserved in the Herzog-August-Bibliothek, Wolfenbüttel (363 Gud. Lat. 8°).

¹⁴ Both books, held ‘in bibliotheca publ. Hamburgensi’ (Hertz, ed. mai. ii, p. lxxix) – now the Staats- und Universitätsbibliothek Hamburg Carl von Ossietzky – were destroyed by Allied bombs (ex rel. Dr Eva Horváth of the Handschriftenabteilung, letter of 12 Aug. 1992).

¹⁵ Cf. Tornaesius' note: ‘s. vet. editae. Tor. [= F] paulo aliter.’

¹⁶ So Cesena, Bibl. Malatestiana S XVI 4; Chicago, Newberry Lib. f90; Florence, Laur. 54.30; London, BL Burney 175, 176; Oxford, Bodl. Lib. E. D. Clarke 20 m.^lpc (LCM 16 [1991], 48 errs); Rome, Bibl. Casanatense 1085; Vat. Barb. lat. 123 (s. erased and followed by g.s.), Chigi H VIII.262 (.S. with superior .g.), Vat. lat. 1537 (g with superior s), Vat. lat. 3453 (s subpuncted, in marging g). Others read is: London, BL Harley 2495, 2768; Lucca, Bibl. Feliniana 478; Vat. lat. 1535; Venice, Marc. lat. XIII.58 (3917).

¹⁷ Cf. enim for .n., London, BL Harley 4859; Vat. lat. 1532; also in Connellus, along with alteros for altero .s. Georgius de Caballis reads scilicet .s. translatum.

¹⁸ For Afanasij and his translation see Holford-Strevens, Aulus Gellius, 246–7, where at n. 30 for ‘Dnepropetrovsk’ read ‘Dnipropetrovs'k (Ukraine)’.

¹⁹ See Weissenborn–Müller ad loc., cf. Kühner–Stegmann i.754.

²⁰ The paraphrase at Macr. Sat. 7.13.8, ‘et ideo uisum ueteribus ut ille digitus anulo tamquam corona circumdaretur’, is far too free to help.

²¹ Including respectively the comparative inscitiores (19.10.14) and the superlative inscitissime (10.16.5).

²² At 18.5.7 non does not negate inscite.

²³ Hermes, 48 (1913), 157–9Google Scholar, cit. Iancu, Fischer in Nopţile atice, trans. David, Popescu, (Bucharest, 1965)Google Scholar, ad loc. (who, as if confusing piano-keys with intervals, makes C–E five semitones and F–B seven), and Maria Laura, Astarita, La cultura nelle ‘Noctes Atticae’ (Catania, 1993), 165 n. 27Google Scholar; Bernardi-Perini also adopts this explanation, without giving a source. These scholars have overlooked Maas's ‘Berichtigung’ in the same volume, p. 636 (appended to the article in his Kleine Schriften, Wolfgang, Buchwald (ed.), [Munich, 1973], 593–5).Google Scholar

²⁴ As Dr Bonnie J. Blackburn (pers. comm.) points out, instead of dividing the string by ratios and sounding the longer part, Maas took for model one of the three instruments he mentions, namely the guitar, on whose fingerboard the semitones are separated by frets. The note sounded at the fifth fret, which divides the whole string in 1/4 and 3/4, yields the perfect fourth; the next seven frets, which occupy the same space on the fingerboard and in Maas's diagram as the first five, cover the adjoining fifth, reaching to the mid-point of the string, which yields the octave.

²⁵ Hosius, edn. i, p. lv; but Astarita 165 takes 18.14 for Gellius'own work, intended to clarify the next chapter. If ‘de numeris Latine scripserunt’ (§2) denotes book 5 of de disciplinis (NA 10.1.6) rather than Atticus (Cens. 2.2) or the nine books de principiis numerorum (see Jerome's catalogue), Latin translations already existed when Varro wrote, let alone Gellius (Holford-Strevens, Aulus Gellius, 169 n. 26, where add Aug. Mus. 2.10.18 – ignored by Guillaumin, J.-Y., RPh³ 53 (1989), 105–9Google Scholar – and the generic term sesquatus, Ter. Maur. 1577; Aug. Mus. 1.10.17 ‘ueteri iam nomine’).

²⁶ So ‘Alexander’, confirmed by Bassett, Samuel E., CPh 11 (1916), 458–60Google Scholar; Holford-Strevens, Aulus Gellius, 117 n. 13 should be corrected.

²⁷ The former is exemplified by 3:2, 15:10, 30:20, the latter by 4:3, 12:9, 40:30; the simple terms apart, only 12:9 has musical relevance, forming part of the continued ratio 12:9:8:6.

²⁸ For the mathematical conflict see Ptol. Harm. 1.10, cf. Porph. Comm. 2.10, Boeth, Mus. 3.3. Ancient attempts at reconciling the two (on which see West, M. L., Ancient Greek Music (Oxford, 1992), 239Google Scholar) do not avail us here.

²⁹ JbClPh 8 (1862), 335–7Google Scholar; ibid. 13 (1867), 132–3; Études de littérature et de rythmique grecques (Paris, 1902), 142–4.Google Scholar

³⁰ But see Manfred, Simon, Das Verhältnis spätlateinicher Enzyklopädien der artes liberates zu Varros disciplinarum libri novem (Diss. Jena, 1963), 83–4Google Scholar, on fr. 27 (GRF 220 = 116 GS).

³¹ Cf. 6.10.27: ‘…in uersu occultiore consideratione numerorum ea quae inaequalia membra iunguntur uim aequalitatis habere inueniantur’. Augustine does not remark that the hexameter embodies the ‘Pythagorean’ 3,4,5 triangle.

³² The implicit exclusion of hephthemimeral caesura is due to summary exposition; at Mus. 5.5.9 Augustine admits it, yet at 3.2.3 the Master asserts that however far the Pupil proceeds in the Aeneid he will find word-end ‘in quinto semipede, id est in duobus pedibus et semisse’ (Aen. 1.9 refutes).

³³ Giraldus, Cambrensis, Speculum duorum, ed. Yves, Lefèvre and Huygens, R. B. C. (Cardiff, 1974), 202.212–14.Google Scholar

³⁴ Dictio in natale Laurenti Mediolanensis episcopi 1.5 (MGH AA vii. 1.25–7; CSEL vi.424.16–17).

³⁵ See Goddu, A. A. and Rouse, R. H., ‘Gerald of Wales and the Florilegium Angelicum’, Speculum, 52 (1977), 488–521 at 513–14CrossRefGoogle Scholar; cf. , R. H. and Rouse, M. A., ‘The Florilegium Angelicum: Its Origin, Content, and Influence’, in Medieval Learning and Literature: Essays Presented to Richard William Hunt (Oxford, 1976), 66–114Google Scholar = eid., Authentic Witnesses: Approaches to Medieval Texts and Manuscripts (Notre Dame, 1991), 101–52Google Scholar. With the sentiment contrast NA 2.23.22.

³⁶ Which also cites Gellius on t he topic (11.18.18): ‘Fures, inquit [Cato, or. fr. 172 SbC = ORF ⁴ 224 Malc.], priuatorum furtorum in neruo atque in compedibus aetatem agunt, fures publici in auro atque in purpura’.

³⁷ E.g. Miroslav, Flodr, Incunabula classicorum (Amsterdam, 1973), 178–9Google Scholar; numbers as in Hain, L., Repertorium bibliographicum, 4 vols. (Milan, 1948).Google Scholar

³⁸ Found by Lamberto, Donati, Archivio storico per la Dalmazia, 2 (1927–1928), 4/22, pp. 181–98Google Scholar to exist in two very different editions: Brescia, Biblioteca Queriniana D.H.12 = A, ibid. Lechi 152= B, ‘quella definitiva’ (p. 195). To B belong Bodleian Library Auct. O.5.2 and British Library 167.e.14; ‘cicerrnis’ (sig. [a5]^r24) has been corrected to ‘ciceronis’ in the former, ‘futur’ (sig. AA1^V28) – if not Donati's error – to ‘futuR’ in both.

³⁹ MS. Lat. 59: s. xv; Florentine; scribe of Genoa, Durazzo B.III.18 (Caesar), Vat. Urb. lat. 51 (Jerome, Epp.); Greek by John Scutariota; presented, apparently in 1465, by Robert Fleming, dean of Lincoln Cathedral, who may have bought it from Vespasiano da Bisticci. See Weiss, R., Bodleian Quarterly Record, 8 (1935–1937), 343–59, esp. 349, 356 n. 61Google Scholar; id. Humanism in England during the Fifteenth Century, 3rd edn. (Oxford, 1967), 97–105, esp. 104Google Scholar; de la Mare, A. C., Lincoln College Record, 1962–1963, 16Google Scholar; ead. in Garzelli, A. (ed.), Miniatura fiorentina del Rinascimento (Florence, 1985), i. 550Google Scholar; Alexander, J. J. G. and Temple, E., Illuminated Manuscripts in Oxford College Libraries (Oxford, 1985), 99, no. 953.Google Scholar

⁴⁰ The letter, dated 18 Aug. 1670 (apparently Old Style; not 1 Sept., as Hertz, ed. mai. ii, pp. exxv f. n. ***), was copied by Johannes Fredericus at the end of the collation in Leiden, Bibliotheek der Rijksuniversiteit, MS Gron. 57, fasc. VI; it begins: ‘Ecce tibi, Pater carissime, omnia, in quibus editionem, ut scripsi ante, Venetam anni 1491, a MS. Gellio collegii Lincolniensis discrepare adverti.’ The earlier letter is lost.

⁴¹ H7517–18 have no book-numbers, H7519–20 number the extant books 1–19 in text and table, H7521–2 in the running heads as well; none shows any awareness that a book is missing.

⁴² fo. 72^r ‘Sunt nonnulli qui existimant octauum librum noctium atticarum non reperiri: eumque qui nonus est uicem octaui tenuisse: quod si ita est uiderint doctiores. Asseruntque se subsequentia excerpsisse ex uetusto exemplari: quae quidem magis placuit sehorsum imprimere quam ordinem pristinum librorum infringere’, followed by headings to bk. 8; cf. H7524, fo. 48^r ‘A. Gelii Noctium Atticarum Liber Octauus hac aetate non reperitur. Sed ex uetustissimo exemplari haec capitula exterpta [sic] sunt’, reproduced misprint and all in H7526, fo. 48^r. Whereas H7517–22 and H7525–7 are the work of well–known houses, the printers of H7524 produced only one other book, HI2206/12210 = Ovid, , Heroides (Venice, 14 Dec. 1493)CrossRefGoogle Scholar; the great Ottaviano Scotto, established 1479, disdained to learn from fly-by-nights (e.g. 14.6.4 sit an H7517–20, H7524, sittun H7521–2, sit tum H7525).

⁴³ F reads doctrinis omnigenus, with the left-hand minim of the u erased to leave an i.

⁴⁴ Although the readings noted above for H7526 all recur in H7527, it is harder to see how Pincius' ‘M.ccccc.’ could be so misinterpreted.

⁴⁵ Shelfmark K.9.23; inscribed in 16th-C. hand ‘Liber collegij Lincoln ex dono Gulielmi Ruswell’; bound with a defective copy of H14569 = Opera agricolationum, Beroaldus, Ph. (ed.), (Reggio Emilia, 18 Sept 1496)Google Scholar; inside front cover a Library bookplate of 1703. On the donor see Joseph, Foster, Alumni Oxonienses, 1500–1714, 4 vols. (Oxford, 1891–1892), iii. 1281Google Scholar; Green, V. H. H., The Commonwealth of Lincoln College, 1427–1977 (Oxford, 1977), 606Google Scholar. All cited readings of H7526 were confirmed by me for this copy; those of the other incunables were taken from the copies in the Bodleian and British Libraries.

⁴⁶ The compilers of the Gesamtkatalog der Wiegendrucke, ix (Stuttgart and Berlin, 1991), col. 342Google Scholar, observe that H7523 is ‘Nicht nachweisbar’ and exclude it from their numeration: H7522 = GW 10598, H7524 = GW 10599.