² T&T(see n. 1 above), pp. 121–2. To the literature there cited should be added Beeson, C. H., ‘The “lost” manuscript of Cicero's De Amicitia’,CP 21(1926),120–132.Google Scholar

³ B. Munk Olsen, Vfctude des auteurs classiques latins aux XF et XII’ siecles (Paris, 1987), vol. 3.2, pp. 20 and 30 (cf. vol. 1, pp. 83,258 and 318).Google Scholar

⁴ See n. 1. The rediscovery was also subsequently reported byHolford-Strevens, L.in CQ 42 (1992),577–8.CrossRefGoogle Scholar

⁵ Combès was unable to trace P itself, but made use of Laurand's photographs (Combes, introd. p. lviii, n. 4).Google Scholar

⁶ For bibliography seeT&T, p. 118, n. 9.

⁷ Chatelain, E.,Paleographie des classiques latins I(Paris 1884) 11, plate 42.Google Scholar

⁸ Laurand, L.,Musee Beige 30(1926),33–42. On the Cato Maior text in this MS, see my edition (Cambridge 1988), p. 41 (the MS is there referred to as M).Google Scholar

⁹ Laurand (n. 8), pp. 131–7.Google Scholar

¹⁰ For bibliography, see T&T, p. 119, n. 15.

¹¹ Cf. T&T, p. 122, n. 4.

¹² Finch, C. E., TAPA 95(1964),66.Google Scholar

¹³ Gerstinger, H.,Festschrift fur Rudolf Egger, III (Klagenfurt, 1952), pp.388–91. I rely on Gerstinger's collation.Google Scholar

¹⁴ Cf. T&T, pp. 122–3.

¹⁵ Mommsen, T., RhM 18(1863),594–601.Google Scholar

¹⁶ K. Simbeck, Teubner edition (1917).Google Scholar

¹⁷ Wunder, E.,Variae lectiones librorum aliquot M. Tulli Ciceronis ex codice Erfurtensi enotatae(Leipzig,1827).Google Scholar

¹⁸ A. D. Michnay, Neue Jahrbiicher fur klassische Philologie, Suppl. 2.1 (1883), 138ff.Google Scholar

¹⁹ As noted by Combes in his edition, introd. p. lxii, a number of the variants of GHVSB (to which we may now add Z) are transpositions of word order: 12 reductus ad vesperum: ad vesperum reductusGHVSBZ; 13 haec nuper. nuper haecGHVSBZ; 22 locis pluribus: pluribus locisGHVSB; 32 amicitias conglutinaret: conglutinaret amicitiasGHVSB; 35 odia etiam: etiam odiaGHVSB; 58 vera amicitia: amicitia veraGHVSB. Generally speaking, where a choice can be made on grounds of sense or rhythm, the order in GHVSB is inferior, hence in cases of doubt we may assume that this family is less trustworthy in matters of word order. Other errors of GHVSB: 14 Manilius: manliusGVSBZ (malliusH); 22 qui (potest esse): quaeG'HVSB; 43 futura: futura sitGHVSB; 51 secuta: consecutaHVSB, consecutataG; 82 pro altero: ab alteroG'HVSIB; 100 exardescit: ardescitGHVSB. Additionally, in sections 71–81, where V's text is missing, we have 74 quorum: eorum quoqueGHSB; 78 horum vitiorum: vitiorum horumGHSB; 80 sibi quisque: quisque sibiGHSB. For agreements in error of GHVSB and LQ, see below, n. 27.

²⁰ The variants of HVSB are too numerous to list in full; again many of them are alterations of word order. In the first ten sections of the text alone we have 1 multa ab eo: ab eo multaHVSB (and, perhaps coincidentally, F); prudenter disputata: disputata prudenterHVSB; et commode, etom. HV'SB; 5 nemo fere: fere nemoHVSB; 8 id: hocHVSB; ut ulla: ut iliaHV'SB; 9 hi quidem nee catoni comparanturHVSB (see below); 10 nihii. nihilenimHVSB.

²¹ Laurand, articles cited above, nn. 8 and 9; Buè6 edition (1928).

²² P. Venini, ed. Paraviana (Turin, 1959); P. Fedeli, ed. Mondadori (Milan, 1971); id., RhM 115 (1972), 156–9.Google Scholar

²³ An exception is 99 aperte enim adulantem: aperte enim et adulantemPMF'L1: apertum enim et adulantemQ. This passage is discussed by Fedeli (article cited in previous note), p. 173, who argues that the intrusive etstood in the archetype, and proposes to restore sense by inserting a word before it, e.g. assentantem.This is possible, but etcould just as well have been an error in the archetype. There are also one or two perhaps fortuitous agreements of LQ with M: 53 amicosom. MLQ; 91 et monere: etom. MLQ

²⁴ Mommsen (n. 15).

²⁵ One could surmise that this error derived ultimately from an uncial examplar, in which p and i could easily be confused.

²⁶ T&T 122; Dr B. C. Barker-Benfield confirms that nothing is known of O's history which is incompatible with this identification.

²⁷ Note also: 39 memoriae: memoriaLQGV'SB (but V2 and H correct this); 41 potuimusPO'FH2: posuimusLQGVSB together with O2 and R: possumusH1; 42 in magna aliqua rePO'F: in magnam aliquant remLQGHV'SB and O2 (all MSS have the mistakenly added p(ublicam),deleted by the correctors of O and V); 47 reapse: re ipsaLQGHVS (post com: reom. S1) and B; 71 si: sicLQGSB (V unavailable in 71–81) and R1; 81 se diligantLQGHSB and R; 86 esse nullam: esse nullam sentiuntLQGHVSB and tfRF; 99 vidisse: vicisseLQGHVSB and RF.

²⁸ Shared errors in LQG or LQG2: 3 induxi: introduxiLQG2; 54 parare: parereL'QG; multorum: malorumLQG2; 81 diligit: diligatLQG; 97 contione: contentioneLQG..

²⁹ Error in LQ and the consensus of HVSB, while G preserves truth: 23 iunctionemPAOKG1: coniunctionemLQG2HVS (-eB); in this passage, iunctionemis shown to be preferable by the clausula; see also below on R. 41 secuti: secuti suntLQVSB; 54 viderePO'KMG1: videriLQO2G2HVSB; 54 fuerunf.fuerantLQHVSB2; 61 caput: de capiteLQG2HVSB (but caputG1); 63 consecuti sint: consecuti suntLQV'SB; 70 maximusPOMG: permaximusLQHVSB..

³⁰ Errors in Q and the consensus of GHVSB but not in L: 32 dissentimusQVSBZ; 40 amicitiaeQHVSB; 52 eruntom. QHVSB (replaced in VSB immediately before audiendi;H has audiendi sunt);55 paranturQGVSB; 88 cum evitareQGH^B; in sublevandoQG2V'SB..

³¹ Agreement of L in error with the xfamily, while Q preserves the correct reading: 38 accedentPOKFL; 48 in sapientem: insipientemPML1; 74 ingeniis et: etom. POKRL; 75 quaedam: quadamPOKFL.

³² Finch (n. 12).

³³ See T&T,pp. 118, 122. However, I was probably wrong to imply in that context that O and K depend on a common exemplar at one remove from the parent of the x family: there is nothing to prove that they are not direct copies of x.

³⁴ Cf. B. C. Barker-Benfield, T&T, p. 226

³⁵ It should be remembered that A exists for sections 1–29 only, and M for sections 44–end; thus there is no passage where all five of these MSS are extant. But the pattern is clear enough.

³⁶ T&T, p. 122.

³⁷ Cf. my 1990 edition, p. 169

³⁸ Fedeli (n. 22), pp. 166–7.

³⁹ Ibid., p. 172.

⁴⁰ This was noticed already by Philippson, R., reviewing Laurand in BPW49 (1929),969ff., and reinforced by Venini (ed. Paraviana [1959], p. xxix) and Fedeli ([n. 22], pp. 156–9). The variants of LQ have been listed by Fedeli, and there is no reason to reproduce them here.Google Scholar

⁴¹ Laurand, REL 6(1928),336–7.Google Scholar

⁴² For the errors of HVSB see above, n. 20. H does not share all the errors of VSB: see, for example, 4 maximeom. VSB; 6 M. om. VBS2 (modoom. HS1); iamom. VSB; 7 solitus esses: solitus esVSB; 10 decessu: discessuVS'B; 11 fuerit: fuitV'SB; 16 si quemadmodum: si quidem quemadmodumVS'B; etc. On the other hand, there are also passages in which H has a more advanced corruption than is found in VSB; e.g. 5 ad senem senex: ad senem te ego senexH: ad senem ego senexV'SB, egodel. V2; te ipse: te ipsumQGVSB: tu te ipsumH. In two passages an error is shared by GVSB but not by H: 82 ipsumom. GVSB; 89 ferri: fieriGVSB; cf. also 71 si: sicLQGSB but siH (V absent). One must assume that the writer of H made these easy corrections of his own accord. Cf. also Fedeli's edition (1971), pp. 22–3..

⁴³ See my edition of the Cato Maior, pp. 40–1.

⁴⁴ Knecht, D.(L'Antiquite Classique 61 [1992],386, reviewing my edition) defends the reading of the majority of MSS. I admit that it is not an open-and-shut case, but I still feel that the text reads more smoothly without et.Google Scholar

⁴⁵ See my edition of the Cato Maior, p. 43

⁴⁶ Harley 4927, Vatican Reg. lat. 1439 and 1574. On Harley 4927, a manuscript which later belonged to Petrarch, see E. Pellegrin, and G. Billanovich,Scriptorium 8(1954),115–17. The reading is firmly entrenched in older editions such as those of Madvig, Baiter-Halm, and Seyffert-Muller, and is rightly supported by Fedeli (n. 22), pp. 161–2, who attributes it to recentiores but does not specify particular manuscripts. All older MSS read iudicatum.CrossRefGoogle Scholar

⁴⁷ See my edition of the Cato Maior, pp. 37–8.

⁴⁸ Cf. n. 20 above

⁴⁹ See my edition of the Cato Maior, p. 39.

⁵⁰ This interpolation recurs in E (cf. n. 17 above), demonstrating its relationship to O; have different interpolations, ut dicis or sicut dicis; cf. Fedeli (n. 22), p. 160.

⁵¹ Fedeli, Hommages M Renard 1(Collection Latomus 101, 1969), pp.339–49.Google Scholar

⁵² Catalogue générale des départements III (1861), 548.

⁵³ Munk Olsen (n. 3), 3.2, pp. 42–3.

⁵⁴ This MS was another of Laurand, 's discoveries (CPh 21[1926],149). Combes attributes to it the correct reading neutris in section 53, otherwise found only in S, but neutri Sel. according to Laurand; neutris Laur. 45.2 according to Laurand. I suspect confusion of sigla.Google Scholar

⁵⁵ Fedeli (n. 22), p. 162.

⁵⁶ In my 1990 edition, before I had consulted this MS, Knecht (n. 44) objected that de die in diem does not occur in Cicero; I now try diem e(x) die, cf. Att. 7.26.3, which could even more easily be corrupted to deinde.

⁵⁷ See n. 55.

⁵⁸ On most of these points see the Appendix to my 1990 edition, pp. 167–72.

⁵⁹ This is the same policy as I followed, in dealing with a somewhat similar tradition, in my edition of the Cato Maior de senectute (Cambridge Classical Texts and Commentaries, 28 [1988]). Fedeli, P.(Gnomon 62[1990],689–92) criticized this policy; but I should make it clear that I deliberately confined myself (after an exhaustive search through the library catalogues available to me) to a thorough examination of the earliest manuscript sources for the text, and did not simply rely on those manuscripts that had been used by previous editors.Google Scholar

⁶⁰ See R. Freundlich, ‘Aliquid causae est...’, in S. Perlman and B. Shimron (edd.), AQPON: sive Commentationes... Benzioni Katz... dedicatae (Tel-Aviv, 1967), pp. 7–20; I am most grateful to Professor John Glucker and Mr Ivor Ludlam for providing me with an English translation of this article published in Hebrew.