^{page 339 note 2} It has nowhere been better narrated than by Lehmann, P., Philologus lxxxix (1934), 349 ffGoogle Scholar. (= Erforschung des Mittelalters, ii (Stuttgart, 1959), 1 ff.).Google Scholar

^{page 339 note 3} I hope elsewhere to discuss this side of the tradition in detail, and to show that all these mutili descend from the Bernensis (B): this originally began at 1. 1. 6, and also omits 5. 14. 12 to 8. 3. 64, 8. 6. 17 to 67, 9. 3. 2 to 10. 1. 107, 11. I. 71 to 2. 33, and 12. 10. 43 to the end. These omissions have become known as the great lacunae (cf. Bruni, Leonardo, epist. 4.Google Scholar 9 (Mehus), ‘in quibus locis vetustus deerat, hoc est in sincopis illis grandioribus’). When I write of ‘mutilus parts’ and ‘non-mutilus parts’ I refer inelegantly to those parts of the Institutio that respectively are and are not available in the Bernensis, , taken as beginning at I. I. 6.Google Scholar

^{page 339 note 4} This again I hope to discuss elsewhere.

^{page 339 note 5} I am entirely happy with the view well put by Boskoff, P. S., Speculum xxvii (1952), 71 ff.CrossRefGoogle Scholar that up to 1416 only mutilus texts with or without the addition of 10. 46–507 were available ; for the addition see also myarticle in C.Q. N.S. xii (1962), 169 ff.Google Scholar For Boccaccio's mutilus see Coulter, C. C., Speculum xxxiii (1958), 490 ff.Google Scholar

^{page 339 note 6} Hence Obrecht's use of the Strasbourg MS., Gesner's of the Gothani, Spalding's of the Guelferbytanus, Halm's of the Monacensis. For editors’ use of manuscripts see the moderately accurate summary of Fierville, C., De Quintilianeis Codicibus (Bayeux, 1874), pp. 12–54.Google Scholar

^{page 339 note 7} The early mutili were not used before Halm. A was not systematically used till Zumpt.

^{page 339 note 8} These will appear as my article proceeds. The only general survey of Quintilian's manuscripts is C. Fierville's in the book mentioned above (n. 6: to be cited in future as Fierville 1874), together with part of the introduction to his edition of Book One (Paris, 1890), pp. li ff. (to be cited as Fierville 1890). Fierville, however, knew nothing of many of the manuscripts I discuss: and of those he does mention he inspected few except those in France. Moreover, his classes 3 and 4, under which most of the late manuscripts fall, lack any real ratio dividendi (see a good review by Kiderlin, M., Neue philologische Rundschau 1891, pp. 39 ff.). At the same time his work was an invaluable foundation for my own researches, which often coincide with his findings. I do not give detailed references except where points of special interest arise.Google Scholar

^{page 340 note 1} For my methods see Appendix A below.

^{page 340 note 2} For the year see Blass, H., Rh. Mus. xxx (1875), 458–61.Google Scholar For much of what I say about Poggio and other humanists I am naturally indebted to Remigio Sabbadini, whose discoveries in connexion with Quintilian are summarized in his Storia e critica di testi latini (Catania, 1914), chapter vii.Google Scholar

^{page 340 note 3} This subscript appears at the end of the Vatican MS. Urb. Lat. 327, with the comment ‘Hec verba ex originali Poggii sumpta’. I have not found it anywhere else, to my surprise: not even, despite Clark, A. C. in the preface to his Oxford text of Asconius (1907), xiiiGoogle Scholar, in the Ambrosian MS. B153 sup. What this manuscript does append to its text of the Institutio is the letter of Poggio announcing his discovery ; the different versions of this letter are discussed by Sabbadini, , Studi italiani di filologia classica xi (1903), 351 ff.Google Scholar This letter also appears in the British Museum MS. Burney 243, the Paris MSS. Lat. 7724 and nouv. acq. lat. 1757, the Vatican MS. Urb. Lat. 1151, and the Escorial MS. R. 1. 13. See also Lindsay, W. M., Berl. Phil. Woch. 1902, 1150Google Scholar: and especially stein, N. Rubin, Italia medioevale e umanistica i (1958), 398Google Scholar, for a reference in a letter of Poggio's son, Jacopo, that is clearly related to the subscript in the apograph (ibid., p. 388): ‘Quintilianus a Poggio repertus.… Transcripsit vero turn manu propria diebus liiii sede vacante, deductusque in Italiam per eum: a quo tanquam ex equo troiano omnes Quintiliani qui apud nos sunt manarunt. Liber vero ipse apud Iacoburn filium est.’ These words are not altogether clear, but in view of the fact that Jacopo knew the wording of his father's subscript we may suppose that the giber ipse’ was the actual apograph, contrasted with its multitudinous progeny. For the later fate of the apograph see Walser, E., Poggius Florentinus (Leipzig, 1914), pp. 52 f.Google Scholar and Ullman, B. L., The Origin and Development of Humanistic Script (Rome, 1960), p. 52.Google Scholar That it came into the library of the Earl of Sunderland seems to rest solely on the assurance of ‘un homme de qualité d'Écosse (M. Fletcher de Selton, Gentilhomme d'un mérite distingué)’, recorded by the Abbé Gédoyn in the preface to his translation of Quintilian (first published 1718: in the edition of 1752 vol. i. liv-lv). I have not been able to find any trace of the manuscript in the Sunderland sales catalogue.

^{page 340 note 4} For a local scribe used by Poggio at Constance see Clark, A. C., C.R. xiii (1899), 125–9.Google Scholar For literary activity at Constance at this period see Lehmann, P., Erforschung des Mittelalters (1941), pp. 256ff.Google Scholar

^{page 340 note 5} As a note beneath the colophon states.

^{page 340 note 6} Other characteristic readings of T and t in my samples include: 1 pr. 12 sed disserendum et (so t for T1; sed disserendum est, an error for A's disserendum est): 25 demonstrando (so t for demonstrari, T's error for demonstraturi): 1. 1. 3 nemo tamen, (so t for nemo): 8 humanum (hominum): 2. 13. 2 idea omitted: 10. 1. 46 et cicero (so t for ex aceono, T's error—derived from H—for ex oceano): 52 probabilis omitted:12. 10. 11 breviter (Bruti). There are manyothers: descent from T is easy to diagnose.It goes without saying that T also passes on many errors from Bg and their intermediate link, H. But these errors also appear in F, and are not characteristic of T; in the nonmutilus parts they are, for reasons that will appear, virtually universal in Renaissance Quintilians: to give one example only, at 5. 14. 13 oporteat for oportet (derived from Bg).

^{page 341 note 1} Fierville pointed out the results of this dislocation in various manuscripts, but did not know the explanation.

^{page 341 note 2} At any rate, the quire that is now marked xxiiii was formerly marked xxiii, and conversely the quire now marked xxiii was formerly marked xxiiii.

^{page 341 note 3} Now almost illegible, like the change to -ficiunt. When the order was put right, attempts seem to have been made to remove all traces of the dislocation.

^{page 341 note 4} Except that at the change over from 7. 8. 2 to 8. 2. 21 we have at persuasit instead of at suasit as apparently intended by the corrector of T. Indeed, I have not notice in any manuscript descended in this part from T, whether or not via Poggio's transcript, anything else but ac/at/al per-. The corrector's note does not look ambiguous, but .p was clearly taken as and attached to suasit.

^{page 341 note 5} At any rate, I do not argue further about what Poggio found at St. Gall. That it was T was always a natural conjecture: see, for example, Halm, K., Sitzungsb. der Kgl. Akad. d. Wass. zu Munchen, 1866, i. 499, n. 3.Google Scholar

^{page 341 note 6} I use these as examples. There are others, even in my samples, e.g.: 1. pr. 24 affection (affectatione): I. i. 2 didicisse (defecisse): 3 eripitur (reperitur): 5. 14. 5 ration (rationis): probation (probationis): 12 quam cunt inanima (1. 8 Radermacher) for quam inanima: 13 philosophet (philosophetur): 18 permittendum for GHT punitendum (A corr. has puniendum). These behave in parallel ways to the instances quoted.

^{page 342 note 1} All these three occur in the non-mutilus parts and were less subject to correction From other sources. At 1. 1. 1. the omission of Vero can have been cured by contamination From other copies of T—for the reverse process see p. 343, n. 8. Or Poggio's apograph may have been ambiguous?

^{page 342 note 2} None of which is explicable as a mechanical error resulting from the omission of a line or lines in T. But they contain more or less the same number of letters, and are perhaps lines of Poggio's apograph.

^{page 342 note 3} Peterson, W., M. Fabi Quintiliani … Liber Decimus (Oxford, 1891), lxxi.Google Scholar

^{page 342 note 4} Beltrami, A., Memorie del R. Istituto Lombards di Scienze e Lettere, xxii (1910–13), 181 and 183Google Scholar, observed that this manuscript was descended from T, but exaggerated, I think, the extent of its contamination. Less helpfully, see Sabbadini, , loc. cit. (p. 340, n. 3), PP. 349ff.Google Scholar

^{page 342 note 5} I owe my knowledge of the present whereabouts of this manuscript to Dr. A. N. L. Munby: and I am equally grateful to the Honorary Librarian of the Chester Beatty library, who supplied me with film of some folios of the manuscript free of charge.

^{page 342 note 6} See Spalding's edition of Quintilian, r. liii, and Halm, , op. cit. (p. 341, n. 5), p. 510.Google Scholar

^{page 343 note 1} This was observed by Beltrami, , loc. cit. (above, p. 342, n. 4), p. 182.Google Scholar Zumpt (in Spalding's edition, v. xix) was wrong to say that the edition was ‘non obscure ad Floren tinum librum exacta’.2 For manuscript copies of incunabula see James, M. R., The Wanderings and Homes of Manuscripts (London, 1919), p. 43Google Scholar: and especially Bahler, C. F., The Fifteenth Century Book (Philadelphia, 1960), chapter ipassim.CrossRefGoogle Scholar

^{page 343 note 3} Another Perugia MS. (no. 1365), de spite Mazzatinti, G., Inventari dei manoscritti delle biblioteche d'Italia (Forli), v (1895), 275Google Scholar, does not contain the complete Institutio, but only the Epitome by Patrizi, Francesco, bishop of Gaeta 1460–94Google Scholar, for which see the references in Sandys, J. E., History of Classical Scholarship, ii (Cambridge, 1908), 53, n. 2. There are manuscripts of the epitome in the British Museum (Add. 11671), Paris, Milan, Naples, and elsewhere.Google Scholar

^{page 343 note 4} See Sabbadini, R., Le scoperte dei codici latini e greci ne’ secoli XIV e XV (Firenze, 1905 and 1914), i. 139, 165. The same colophon occurs in MS. Laur. Med. 47. 35.Google Scholar

^{page 343 note 5} Halm, , Rhetores Latini Minores (Leipzig, 1863), pp. 22 ff., edited it from a single Vienna MS.Google Scholar

^{page 343 note 6} Into this category falls the Quintilian procured by Gasparino Barzizza, according to Sabbadini's not quite certain conjecture from his words ‘… Quintiliano, qui ex Constantia integer ad me delatus est. Quintilianus ex vetustissimo codice in Germania transcriptus totus apud nos extat; multo minus corruptus est’ (see Studi di Gasparino Barzizza [Livorno, 1886], pp. 2 ff.).Google Scholar Sab badini (ibid, pp. 6–7) wrongly assumes that Guarino's copy of Poggio's first discovery was necessarily independent of Poggio's apograph.

^{page 343 note 7} The possibility that this was written by R. Agricola is doubted by Lehmann, P., Johannes Sichardus (Munich, 1911), p. 133.Google Scholar

^{page 343 note 8} It is particularly interesting that a second hand has introduced a number of Poggio's errors into the text of this manuscript. For example, 5. 14.1 tum has been changed to tamen: 6 adici to dici: 15 n (= enim) has been dotted. However, vero at 1. 1. I and autem at 5. 14. 4 are spared.

^{page 344 note 1} And correction was not, perhaps, Poggio's forte: see Billanovich, G., Journal of the Warburg and Courtauld Institutes, xiv (1951, 178, n. 2. Billanovich rates Valla's skill high, and contrasts it with ‘Poggio's philology, hasty and ingenuous’.Google Scholar

^{page 344 note 2} I shall say more of this in a separate discussion of the mutili in general. Typical representatives are Paris MSS. Lat. 7720 (Petrarch's) and 7722, and the Vatican MS. Ottob. Lat. 1159.

^{page 344 note 3} In ‘The Beginning of Quintilian's In stitutio’, C.Q. xvii (1967), 123 ff.Google Scholar

^{page 344 note 4} There are others in my samples, though not very many, e.g. 2. 13. 7 utilem (utile): 9 factum (affectum), which is wrong, though in Radermacher's text.

^{page 345 note 1} Ep. iv. 9 (Mehus), dated to April 1417Google Scholar (see Blass, , op. cit. [P. 340, n. 2], p. 461 with n. 1). We do not need to suppose that this ‘old codex’ was anything other than a mutilus of the previous century, in a non-humanist script.Google Scholar

^{page 345 note 2} The Didot MS. For further details see Fierville, 1890, LXIII–LXX.Google Scholar

^{page 345 note 3} The work of Johannes Poullain. The various indications of date and provenance are clearly set out by Marichal, C. Samaran-R., Catalogue des manuscrits en icriture latine, etc., ii (Paris, 1962), 425.Google Scholar The bulk of the manuscript was written, in three months, in 1465. Burman gives some read ings from it in his edition of 1720, calling it the Codex Poullainii. The mutilus in this manuscript is not typical of the fourteenth- century vulgate: see my note in C.Q. xii (1962), 173.Google Scholar

^{page 346 note 1} See Ullman, B. L., op. cit. (p. 340, n. 3), pp. 79 ff., where it is shown that there are Florentine examples of the new script dated from 1405 on.Google Scholar

^{page 346 note 2} This error does, however, occur in some representatives of the fourteenth-century vulgate.

^{page 346 note 23} This manuscript is discussed, and the letter published in full, by Sabbadini, , Riv. di Fil. xxi (1892), 142–3Google Scholar (see also Storia e critica, pp. 393–4): ‘esso più di qualunque altro deve rappresentare la lezione genuina del primo codice scoperto da Poggio.’ T itself does this better.Google Scholar

^{page 346 note 4} For whom see Martines, L., The Social World of the Florentine Humanists (London, 1963), pp. 331–2Google Scholar, and especially Ullman, B. L., op. cit. (p. 340, n. 3), p. 85: ‘He was a friend and protégé of Niccoli.’ Hence, perhaps, his access to Poggio's transcript.Google Scholar

^{page 346 note 5} Indeed it has got worse. Both in Γ and in at least some of the manuscripts related to it we find a dislocation that clearly started from the dislocation in Poggio's apograph, but is even more complicated than it. The order is (at least in part): 7. 3. 3 differentibus - 7. 1. 61 patrem—2. 26 factum - 7. to. 13 denique ac terra—8. pr. 30 dicendi vis - 7. 3. 3 propriis—7 generant (for faciunt) - 7. 8. 2 inde puniri— 10. 13 mari - 8. pr. 30 fuerit—2. 21 credantur (so t for creduntur) perficiunt - 7. 3. 7 an qui se— 8. 2 contentione veniemus - 8. 2. 21 at persuasit, etc. Halm's note at 7. 10. 13 shows that something very like this occurred in the now lost Strasbourg MS. (S), which was I imagine certainly in the class of Γ. Γ also shares with S what may seem an element of sophistication, but is in fact another sign of primitiveness-a very pure version of 10. I. 46–107 descended not from T but from the separate extracts of these sections—the version which in my article cited in p. 339, n. 5 I called the S-version. This remains so pure in S, and in Γ, because there has been no time for contamination from the G-version. What I said about the fifteenth century in that article, p. 174, remains valid. I add merely, under category 2, that the S-version dominates in classes θ and λ (for which see below), but is rather purer in the former. For example, at to. I. 58 GT Phileta: Sθ Philatas: λ Philetas (rightly). Cf. 57 esse GTλ: om. Sθ. These two cases are typical, correction (due to what I call recension ε) and contamination.

^{page 347 note 1} Moreover, the mutilus signs do not appear in 12. 10. 10–15, and this passage, like to. 1. 46–107, is clearly descended from T. This is true also, in this class, of the two Vatican MSS. and that at San Daniele.

^{page 347 note 2} Beschreibendes Verzeichnis der illuminierten Handschriften in Österreich, Folge, Neue, vi. i. (Leipzig, 1930), p. 117.Google Scholar

^{page 347 note 3} They are: 45 esse (est): affectius(affectibus): 46 nos (non) quidem: 55 a (ac) iudice: 61 cogetque (cogitque): 73 bacchatur (braccatur): 75 immerito (merito): 78 in adversa (adversa). Of two other corrections in this passage attributed by Radermacher to Renaissance conjecture, § 50 veniat (veniant) already appears in the first hand of T: and at § 75 id, not id est, is the reading of P.

^{page 348 note 1} As it is in the early mutilus called N: I suppose, coincidentally. In F₉, a corrector other than Nicholas has inserted the words.

^{page 348 note 2} Nor is this mere supposition. In the Biblioteca Riccardiana in Florence there is a manuscript (521) in which a mutilus descended from (K), of the normal fourteenth-century vulgate type, has been corrected by a second hand which imports readings un mistakably those of a mutilus of the type related to the St. John's, Cambridge MS. (J). Observe that at 1. 3. 14 this manuscript omitted iniuria(e), like (K), but the corrector inserted it. It will be by a like process that we have the word in place in F₉. Other evidence of the existence of mutili not de scended from (K) in the fourteenth century will appear in my discussion of P (p. 358, n.3).

^{page 348 note 3} Cf. P. 343, n. 8.

^{page 349 note 1} For these words see above, P. 341, n. 4.

^{page 349 note 2} Poggio, , ep. i 5 (de Tonellis)Google Scholar, ‘Haec mea manu transcripsi, et quidem velociter, ut ea mitterem ad Leonardum Aretinum et Nicolaum Florentinum.’ We know that Leonardo was at work on Quintilian in 1417: see p. 347 above. And Laur. Med. 46. 9 could be the result of emending Poggio's transcript from a mutilus, though it is perhaps more natural to see it as descended from a supplemented mutilus. For Niccoli's work on Poggio's transcript of Asconius, which was found at the same time as the Quintilian, see Clark, A. C., C.R. xiii (1899), 120Google Scholar; for his work on other authors, Ullman, , op. cit. (p. 340, n. 3), chapter 111.Google Scholar

^{page 349 note 3} Readings from this manuscript have been published by Sabbadini and Beltrami. The former (Riv. di Fil. xx [1891–2], 315 f.Google Scholar and Studi Italiani di Filologia Classica, xi [1903], 350–1)Google Scholar decided from readings of this manu script and T between 1. pr. 7 and 1. I. 37 that we can absolutely exclude the view that F₃, derives directly or indirectly from T. He comes near to a correct view by saying: ‘E. se γ [his letter for this manuscript] derivasse dal cod. misto allestito dal Bruni col Poggiano e col mutilo fiorentino ?‘ only to reject it: ’Le nostre conclusioni resterebbero egualmente, perchè il testo dei mutili comincia solo col i. 1. 7.‘ So ?—Beltrami, op. cit. (p. 342, n. 4), lacked the idea of a ‘mixture’, and got into deep water with this manuscript. For he thought (p. 185) that it is more correct than the Ambrosian MS. B153 sup.—descended from T via Poggio's apograph—because it represents Poggio's second discovery (of which I shall have more to say later). The correctness is in fact provided by emendation and by conflation with a mutilus.

^{page 349 note 4} Oddly enough, this mistake also (cf. p. 348, n. 1) is shared with N: and the scribe of the Bernensis apparently started to write it before correcting the slip.

^{page 350 note 1} Though see p. 349, n. 3.Google Scholar

^{page 350 note 2} For this see p. 346, n. 5.Google Scholar

^{page 350 note 3} Perhaps not all those that I list go back to the recension, but are due to an intermediate link. When I talk of this recension in connexion with other classes, I think rather of the presence of the recension's virtues than of the presence of all the errors listed here: though many, as we shall see, do appear in class θ and in the ‘vulgate’.

^{page 351 note 1} For whose library see Sabbadini, , Riv. di Fil. xlv (1917), 197 ff. On p. 204 he calls this manuscript (not altogether accurately) ‘una copia del primo esemplare completo scoperto da Poggio‘. See p. 199 for Zomino’s absence when Poggio found Asconius (and Quintilian).Google Scholar

^{page 351 note 2} In fact, B has plici. This was expanded in some later mutili (N, J, and the Montpellier MS. for example) to multiplici: here, more correctly, to duplici.

^{page 351 note 3} See, moreover, the Vatican catalogue ( Ruysschaert, J. [Vatican 1959], p. 369) for what had to be done to Book 4, Chapter because (K)-and Ar-had omitted 4. I. 35–39: see above, p. 344.Google Scholar

^{page 352 note 1} I am fairly sure, from soundings taken elsewhere than my samples, that there is more connexion between the mutilus parts of classes λ and θ than my samples show up: for instance, at 1. 1. 24 I observe formulas for formas in F₃ and Harl. 4829 (class λ), and also in Harl 2662: but not in the corresponding mutili. This sort of thing would perhaps be due to the effect of recension ε on the mutilus part of class θ parallel to its much more thorough influence on the mutilus part of class λ: for which see pp. 350–1.

^{page 352 note 2} And Harl. 2662 has the same treatment of the dislocations in Books 7–8 as that in class λ: the treatment I attributed to recension ε (p. 349–50).

^{page 352 note 3} Op. cit. (p. 342, n. 3), lxxiii.Google Scholar

^{page 352 note 4} Meister, F., Berl. Phil. Woch. 1900, 1051–3Google Scholar, with some selected readings. Fierville 1874. 121 ff.Google Scholar also gives readings: and for this and some other Paris MSS. see the edition of F.-G. Pottier (Paris, 1812), and F. N. Klein in J. Seebode and Friedeman, F., Miscellanea maximam partem critica, i (1822), 486 ff., 650 ff.Google Scholar

^{page 353 note 1} I do not see why Fierville, 1874, p. 12Google Scholar, thought that the Jenson edition (147r ) drew on this manuscript: it seems to me to be a pure P text. I have, however, noticed some signs in the Sichard edition of 1529 (which has, of course, other more recent elements, such as Regius emendations): for example, at 5. 14. 7 insita is given as a variant. Is an early fifteenth-century manuscript of the θ type all that Sichard's manuscript from Poland boils down to? For the latter, see Appendix B.

^{page 353 note 2} Inventor der illuminierten Handschriften … der Osterreichischen Nationalbibliothek, i (Vienna 1957), 90.Google Scholar

^{page 353 note 3} Where, indeed, it had been since the eleventh century: see Reifferscheid, A., Rh. Mus. xxiii (1868), 144.Google Scholar Reifferscheid (following Halm, , loc. cit. [p. 341, n. 5])Google Scholar showed that Poggio's discovery at St. Gall was not F, despite a Florentine tradition (see Bandini, A. M., Catalogus Codicum Latinorum Bibliothecae Mediceae Laurentianae (Florence, 1775), ii. 382–4) still canvassed by Spalding (vol. i. 1): though of course it remains possible that Poggio did discover F (see n. 5 below).Google Scholar

^{page 353 note 4} There was a ‘Quintiliano di lettera antica’ in the library of Cosimo de’ Medici in 1418 (Pintor, F., La libreria di Cosimo de’ Medici [Florence, 1902], pp. 7, 54). What was that? Perhaps a mutilus?Google Scholar

^{page 353 note 5} I should like to believe Sabbadini's view that F was found along with other manuscripts at Strasbourg in 1433, which seems the right sort of date (Scoperte, , i. 117).Google Scholar Or it could be that F was found by Poggio as he came up the Rhine in 1423 (Scoperte, i. 83).Google Scholar Even this would be better than to identify F with Poggio's second discovery if we have to date that discovery as early as 1417 (as does Reifferscheid). In fact, that Poggio did find two Quintilians seems very doubtful. The evidence seems to be (a) a letter of Poggio to Niccoli (iv. 4 de Tonellis) referring to a manuscript of Plautus: liber est illis litteris antiquis corruptis, quales sunt Quintiliani.‘ The letter is dated 1429, by which time some one else could have found F. In any case ‘quales sunt Quintiliani’ at least could mean ‘like the letters in the Quintilian’—that is, the one famous Quintilian that Poggio found. Nor is Beltrami justified in assuming (loc. cit. [p. 342, n. 4], pp. 185–6)Google Scholar that the words must refer to a Quintilian in Poggio's possession or even one that he found. (b) Guarino's letter to Poggio (83 in Sabbadini's, Epistolario di Guarino Veronese = Misc. di Storia Veneta, viii [1915], 158)Google Scholar, in which Guarino says that he had received a transcript of the Quintilian, but that it was very corrupt. ‘Sentio te aliud Quintiliani exem plar nactum esse, quod apud te est: ex quo unum nornine meo conscribi facias oro.’ The letter is not dated, nor do I feel quite sure that Poggio was at Constance when it was written (though see Sabbadini, , Rio. di Fil. xx [1891–2], 310).Google Scholar But in any case Guarino does not say that Poggio found the manu script; he could have been misinformed any way; and Poggio might—whether or not Guarino meant to say this-merely have provided himself with a fresh transcript of T.

^{page 353 note 6} When Jouffroy (below, , p. 354, IL 3) bought it in Rome.Google Scholar

^{page 354 note 1} The corrections in F on the whole look fifteenth-century, but at least some of them descend to F's progeny: for example, oratio for ratio at 5. 14. 4, or solutum est at § 13 (soluta A corr., t rightly: solutum A¹GHT¹F^I. But not, for example, I. I. 3 qui perviderit, one of F's corrections that probably comes from A: F's progeny, like F¹, have quippe viderit (and T has only quippe). F's corrections could no doubt be sorted into layers on this principle if necessary.

^{page 354 note 2} Because it was previously at Landsberg in Bavaria. Its relationship to F was observed by Törnebladh, R., Quaestiones Quintilianeae (Kalmar, 1860), p. 3Google Scholar: who noted some in fluence from A in the early books-perhaps a result of the correction of F from A (see the previous note).

^{page 354 note 3} 1874, pp. 59 ff. Fierville regarded it as being close to F and T I have not observed any relation to the latter. The manuscript has Valline notes copied in by Ferrari, Michael, like one of the Escorial MSS. (see below, p. 356)Google Scholar: and was also worked on by Jouffroy, (see p. 353, n. 6)Google Scholar, for whom see Fierville, C., Le Cardinal Jouffroy et son temps (Paris, 1874), esp. pp. 224 ff.Google Scholar, and also Sabbadini, , Epistolario di Guarino Veronese = Misc. di Storia Veneta, xiv (1919), 505–6.Google Scholar

^{page 354 note 4} For this and other Spanish manuscripts see Fierville, , Arch. des missions scientifiques, ser. iii v. (1879), 85 ff.Google Scholar

^{page 355 note 1} In conversation. See now his Catalogue of the Manuscripts of Balliol College Oxford (Oxford, 1963), pp. 159–20, and also p. xxxi.Google Scholar

^{page 355 note 2} For a comment on its script see Ullman, , op. cit. (p. 340, R. 3), p. 52.Google Scholar

^{page 355 note 3} See p. 340, n. 3. The manuscript was examined by Beltrami, , op. cit. (p. 342, n. 4)Google Scholar, who saw that it is in some sense descended from T (p. 185). But it is certainly not a copy of Poggio's apograph, as the subscript has led many to expect, and as Sabbadini, , Studi Italiani di Filologia Classica, xi (1903), 350–1 also assumed.Google Scholar

^{page 355 note 4} And not 1409, despite the Catalogus Bibliothecae Latinae … in … Neapolitano Museo Borbonico … a Cataldo Iannellio (Naples, 1827), P. 46. This is made certain by the appearance of the same formula ‘per me Ioannem de Lumel’ (not Jumel, as Jannelli read it) in Vatican MSS. Lat. 1815 (dated 1439) and 434 (dated 1460).Google Scholar

^{page 355 note 5} See Gerstinger, H., Festschrift der National bibliothek in Wien (Vienna, 1926), pp. 323–4.Google ScholarHermann, H. J., op. cit. (p. 347, n. 2), vi. 4 (Leipzig, 1933), pp. 17–18.Google Scholar

^{page 356 note 1} See Lehmann, P., Nordisk Tidskrift för Bok- och Biblioteksväsen, xxiv (1937), 108. Lehmann suggested that Ferrari's corrections could be used to find out what state P was in at this period, and to distinguish between its correctors: see further below, p. 360. See also p. 354, n. 3.Google Scholar

^{page 356 note 2} For Valla's work on Quintilian before 1444, see a letter of Giovanni Aurispa, written in December 1443: Aurispa had lent his Quintilian to Valla. Valla, of course, worked on Quintilian after 1444 also, and owned more than one manuscript himself (Sabbadini, , Biografia documentata di Giovanni Aurispa [Noto, 1890], pp. 190–1Google Scholar: Storia, PP. 402–4).Google Scholar

^{page 356 note 3} For this colophon see Billanovich, , op. cit. (p. 344, n. I), p. 139 with n. 2.Google Scholar Billanovich gives a photograph of part of a folio of P in Plate 30. Sabbadini, (Storia, p. 403), following Fierville, doubted if the colophon is in Valla's hand. Even if it is not, the date must be more or less right, because Michael Ferrari read it in 1454 (see above, p. 356: Ferrari virtually quotes the words of the colophon ‘hunc codicem sibi emendavit ipse’).Google Scholar

^{page 356 note 4} P has always had great influence on the text, because of Badius's use of it and because its progeny were popular, especially the Gothanus. Attention was first powerfully directed to the manuscript itself by Becher, F., Zum 10. Buch des Quintilian, Progr. des Kg1. Gymn. zu Aurich, 1891.Google Scholar

^{page 356 note 5} In the prefaces to both volumes of his edition (Leipzig, 1907 and 1935).

^{page 356 note 6} Vol. 1. vii.

^{page 357 note 1} The corrector of T is responsible for another striking coincidence of P and Victor, when at 7. 6. 8 he changes fuisse to voluisse, quite correctly. Actually in the context this change is not so brilliant as it sounds.

^{page 357 note 2} M. Fabii Quintiliani … capita de grammatica, ed. Niedermann, M. (Neuchâtel, 1947), praef. xii.Google Scholar Niedermann also has something to say about 5. 14. 25. In asserting that P and other Renaissance manuscripts give us only conjecture over and above readings derived from the archetype I follow Niedermann's lead: the detail into which I go about P may perhaps be justified by the apparently canonical status which has been granted to Radermacher's views (see, e.g., Büchner, K. in Geschichte der Textiiberlieferung [Zurich, 1961], pp. 408–9). See F. H. Colson's edition of Book I (Cambridge, 1924), xciv f. for earlier doubts about Radermacher's thesis.Google Scholar

^{page 357 note 3} Vol. ii, praef. iv-v.

^{page 358 note 1} In the article cited in p. 339, n. 5.

^{page 358 note 2} Nor, for that matter, any striking coincidences between P and the Anecdota Ecksteinii (from Paris MS. Lat. 7530), as claimed by Radermacher, Vol. i. vii-viii. Not, certainly, at 8. 6. 43, where Radermacher's apparatus reads: ‘carth (aginem) et num(antiam) P An. Eckst. numant. Karth. A G.‘ In fact, P has ‘numantiam et Car taginem’.

^{page 358 note 3} For a parallel see above p. 348, n. 2. Such contamination would probably occur before 1416, when there was more point in emend ing one mutilus from another. For example: 2. 5. 22 voluptate quadam prava (voluptate prava) appears in (that is, the St. John's, Cambridge, MS. and the two Vossian mutili at Leyden) and in P: and also in F₁₂ and some, though not all, texts. P J also agree on 2. 13. 5 diffusa (fusa): 5. 54. 4 factum (actum): 6 solummodo (summum). Only the last also occurs in F12: the other two may therefore be coincidental. I have noticed a large number of other instances outside my samples.

^{page 358 note 4} Where I have looked elsewhere, I have been struck by the way in which changes of word order found in F₁₂ and class , have come through to P, whereas other errors have on the whole been corrected, e.g. 1.1.20 teneris protinus P F₁₂ MS. Laur. S. Cruc. 22 Sin. 5, which also agree on § 24 statim in i psis: 31 inculcare diu: and also 33 in primis ergo, where a corrector of P has unusually changed the order back: ergo as fifth word in the sentence obviously caused qualms that other variations of word order would not.

^{page 358 note 5} Though of course P constantly agrees with B in right and doubtful readings against A b: hence the feeling one gets in looking at Radermacher's apparatus that P normally agrees with B.

^{page 358 note 6} D, however, where a number of intelligent readings are to be found, has ex consensu duorum, again by conjecture. I hope to discuss the merits of mutili other than B elsewhere.

^{page 360 note 1} Op. cit. (p. 342, R. 3), lxxiv, n. I, commenting on Becher's gloom at the difficulty of distinguishing hands in P's corrections— see op. cit. (p. 356, n. 4), p. 4. For Lehmann, see p. 356, n. i: the Escorial MS. would ofcourse be particularly useful because Michael Ferrari says in so many words that he is using Valla's manuscript at a given date; even here, of course, positive conclusions would be valuable, negative ones might mislead.Google Scholar

^{page 361 note 1} See Billanovich, , loc. cit. (p. 344, n. 1), p. 178 with n. 2.Google Scholar

^{page 361 note 2} In his second volume, praef. iv, Professor Mynors tells me that Q was written before 1469, the date of a note on the front paste-down.

^{page 361 note 3} y a un grand nombre de leçons qui ne sont que de la seconde main dans le Vallensis, sur lequel cependant il est évident qu'il n'a pas été copié.’ So Fierville, 1890, cxxi, bafflingly.Google Scholar

^{page 361 note 4} Op. cit. (p. 342, 11. 3), lxxiv.Google Scholar

^{page 361 note 5} To give but one instance: at 9. 3. 2 in the D'Orville MS. 13 ‘hic deficit codex vetustissimus’.

^{page 362 note 1} As it would seem: Nicolas's name appears amid decorations in the margin off. I^r. The name Alfonsus appears after the colophon.

^{page 362 note 2} Sold in Perugia in 1473, and once the property ‘Hieronymi Car^lịs Racanetensis,’ that is, Hieronymus de Ruvere, Bishop of Recanati from 1476, who was made a Cardinal in 1478 and died in 1507 (Eubel, C., Hierarchia Catholica, ii [1901], pp. 19 and 242).Google Scholar

^{page 362 note 3} de Marinis, T., La Biblioteca Napoletana dei re d'Aragona (Milan, 1947), i. 55 ff. and ii. 142. De Marinis dates this MS. 1473.Google Scholar

^{page 362 note 4} For whom see Campana, A. in Studi e ricerche sulla storia della stampa del Quattrocento (Milan, 1942), 1 ff.Google Scholar His Tavola X shows the last folio of the Leyden MS. On pp. 24 f.Campana adds other manuscripts that have the Valline scholia, and identifies the hand of Vaticanus Lat. 1767, which has no signa ture. I now add the Venice MS., which has the gloomy colophon ‘… expliciunt infeliciter per me Bartholomeum de Colunnis de Chio M.CCCC.LXI III Idus Decembris. Matellicae Agri Piceni.’ For Bartolomeo at Matelica at this period see Campana, p. 14.

^{page 362 note 5} 1890, cxvii-cxviii.

^{page 362 note 6} Gesner's praef. § 20 mentions another Gotha MS., and he gives some readings from it in his apparatus and addenda. At 6. 2. 6 its ‘praecipitat animus’ is quoted by Halm from the Munich MS. (It is also the reading of the corrector of F.) And at 9. 4. 85 its breviter for patitur also appears in M.

^{page 363 note 1} In his first volume, praef. x. See also his second volume, praef. iv, for remarks on the consensus of P and V against AB. Such agreement will normally derive from the common ancestors of P and V, especially, in many parts of the book, T. Radermacher, by the way, drastically misinterpreted the relation of the corrections in T to P. He says (i, praef. vi: cf. x) that ‘eum, qui Turicensem correxit, aliquo huius propaginis‘ (that is, relations of P) libro usum esse’. We know from a despairing rhyme (on 1T. 232^v–233^r) that the corrector of T was not using a manuscript to help him in his corrections; and the corrections are inherited by P (and many other fifteenth-century manuscripts).

^{page 363 note 2} I originally read this as ‘Iteo’, perhaps not without excuse, as the Vatican catalogue gives the same spelling in connexion with another manuscript written by the same man, Vaticanus Lat.1532. ‘Itro’, however, appears, again according to the catalogue, in two others of his manuscripts, Vaticanus Lat. 1539 and 1756; presumably rightly, because in our manuscript at f. 128^v we have the note Johannes de Idria id est Itro scripsit’. Idria is now in north-western Yugoslavia.

^{page 363 note 3} For example, V does not read ‘tropus’for tropis at 8. 6. 41. And the fact that ‘semel Gabbae nomen servavit’ (see Radermacher, praef. x: the reference is 6. 3. 80) is rather outweighed by the five other places where it has the spelling Galba, like the other manuscripts

^{page 364 note 1} So too in the proem to Book 6, as is clear from the apparatus given by Hahn, in the article cited above (p. 341, n. 5), pp. 517 f.Google Scholar

^{page 364 note 2} See Gerstinger, , op. cit. (p. 355, n. 5), PP. 320–1.Google Scholar

^{page 364 note 3} Op. cit. (p. 347, n. 2), vi. 4 (Leipzig, 1933), PP. 13–14.Google Scholar

^{page 364 note 4} But it is not therefore to be supposed that this manuscript is at all intimately connected with Poggio's discovery of both authors at the same time at St. Gall, as implied by Clark, A. C. in his Oxford Classical Text of Asconius (1907), praef. xiii, n. 1.Google Scholar

^{page 364 note 5} In the article cited at p. 344, n. 3.Google Scholar

^{page 365 note 1} Op. cit. (p. 342, 3), lxxv.Google Scholar

^{page 365 note 2} In the text, e.g. at 5. 14. 6, where D'Orv. has rectissimum, the impossible correction made by the second hand of A for tres summum. As a variant, e.g. at 2. 13. 16, where D'Orv. has et si rarum, A's corrector's weird change for et stratum. The latter appears also in Vat. Lat. 1761, as one would expect; the other nowhere else. This is symptomatic of the small influence of A.

^{page 365 note 3} Another Aragon book: de Marinis, , op. cit. (p. 362, n. 3), ii. 43 with Tavola 214.Google Scholar So too is Parisinus Lat. 7729 (ibid. 142).

^{page 365 note 4} Like the Paris MS. Lat. 2231, of Gregory the Great (Samaran-Marichal, , op. cit. [p. 345, n. 3], ii. i)Google Scholar, as observed by Fierville, 1890, CXV, n. 1.Google Scholar

^{page 365 note 5} This bears a helpful note at the front explaining that Anthonius Estournel obtained a manuscript in 1463 from Johannes Ricoul, Canon of Liège, which Ricoul had himself written in Italy. Estournel had this manuscript copied, leaving gaps for the Greek words, which Ricoul later wrote in for him. ‘Et quia plures sunt non expositae in textu, obtinui postmodo ab eodem ut ipse omnes huiusmodi graecas dictiones … mihi interpretaretur, earumque interpretationem in margine conscripsi.’ I am not sure that these operations are very obvious in the manuscript itself.

^{page 365 note 6} In a lecture at the Institute of Classical Studies in London in 1964, Professor B. Bischoff suggested attractively that it was A that was discovered by Bartolomeo Capra, Archbishop of Milan, apparently in the north of Italy in 1423 (Sabbadini, , Scoperte, i. 101). For use of A see above, n. 2.Google Scholar

^{page 365 note 7} For the weird activities of Gasparino Barzizza, see Sabbadini, , Studi di Gasparino Barzizza (Livorno, 1886), p. 2.Google Scholar

^{page 366 note 1} The Medieval Tradition of Seneca's Letters’ (Oxford, 1965), 68.Google Scholar

^{page 365 note 2} J.A.W. cxcii (1922), 231.Google Scholar