¹ It was the problem of the subtitle that made me look at the transmission of the Periochae, and what alerted me to the problem was remarks of John Henderson's in a forthcoming article, of which he kindly lent me a draft. I thank John Briscoe and Leighton Reynolds for helpful comments.

² His article ‘Die handschriftliche Ueberlieferung der Periochae des Livius’, Rhein. Mus. 44 (1889), 65–103Google Scholar, gives a fuller account of the two oldest manuscripts but has almost nothing to say about others.

³ Eight of them escaped Reid, but Reid correctly knew as Burn. 202 what Jal calls Burn. 204 (i.c, cxxii). Manuscripts unknown to both Reid, and Jal, are several of Per. 33, for which see below, n. 24Google Scholar; some other partial manuscripts, for which see below, n. 43; E. Berlin Ham. 261; Eugene (Oregon) Burgess 17; Ferrara Civ. II 191; Paris B.N. Lat. 6080; Parma Palat. Parm. 2800; Rome Angel. 1434 and Vallicell. R 33; Vicenza Bertol. G 2 8 15; and four manuscripts listed by Rubio, L., Catálogo de los manuscritos clásicos latinos existentes en España (Madrid, 1984), nos. 34, 523, 543, and 691Google Scholar, namely de Osma, Burgo Arch. Capit. 19Google Scholar, Poblet Abadía 50–51–52, Salamanca Univ. 246, and Valencia Univ. 482. I have not checked the file on the Periochae at the Institut de Recherche et d'Histoire des Textes.

⁴ I agree with almost everything that Briscoe, J. says in his review, Gnomon 57 (1985), 419–24Google Scholar, about the strictly editorial part of the edition. Add that the apparatus cites the witnesses in alphabetical order, which has nothing to do with their authority or indeed with anything.

⁵ IMU 2 (1959), 159Google Scholar; pp. 87–8 of ‘Il Petrarca e gli storici latini’, in Tra latino e volgare: per Carlo Dionisotti (Padua, 1974), i. 67–145Google Scholar. Petrarch's connexion with the Periochae had been revealed by Iannelli, C., Catalogus bibliothecae Latinae ueteris et classicae manuscriptae quae in regio Neapolitano museo Borbonico adseruatur (Naples, 1827), pp. 95–7Google Scholar, and Wislocki, W., Catalogus codicum manuscriptorum bibliothecae uniuersitatis Jagellonicae Cracouiensis (Kraków, 1877–1881), p. 137Google Scholar, and discussed by de Nolhac, P., Pétrarque et l'humanisme (Paris, 1892), p. 245Google Scholar; Rossbach, ed. of Florus (Leipzig, 1896), xxii; Sabbadini, , ‘Le “Periochae Livianae” del Petrarca possedute dai Barzizza’, in F. Petrarca e la Lombardia (Milan, 1904), pp. 195–201Google Scholar; Nolhac ed. 2 (Paris, 1907), ii.36–7.

⁶ Op. cit. (n. 4), 421.

⁷ The Periochae fell between two stools in Texts and Transmission, ed. Reynolds, L. D. (Oxford, 1983)Google Scholar: no mention either under Livy, superficially the most obvious place, or under Florus, where they belong for a reason about to be given. Leighton Reynolds kindly tells me that he would have treated them under Livy if Jal's edition had not been on the point of appearing.

⁸ Similarly but more fundamentally, Kleinlogel, A., ‘Archetypus und Stemma: zur Problematik prognostisch-retrodiktiver Methoden der Textkritik’, Berichte zur Wissenschaftsgeschichte 2 (1979), 53–64CrossRefGoogle Scholar, argues that to postulate a lost witness of a certain character is to predict what may come to light, and he gives examples of predicted witnesses that have indeed come to light.

⁹ Op. cit. (n. 2), 74.

¹⁰ Jal, i.civ n. 2, says that Reid derives it from N through an intermediary, but what Reid derives from N through an intermediary is the collation.

¹¹ Bischoff, , Lorsch im Spiegel seiner Handschriften (Munich, 1974), 65–6Google Scholar.

¹² Reid's dissertation is not designed to report what the manuscripts read in any particular passage, but I happened to notice that at Per. 81.3 res is a correction in Naples Naz. IV C 32, from which it passed to two copies (p. 164); my other information comes from Jal's apparatus. True readings peculiar to N can also be found at e.g. 40.3, 49.13, 68.3, 69.3, 80.3, 112.2, 123.1, 142.5.

¹³ The two branches of the tradition agree on this much of the title, but it cannot be Florus' own, because it misrepresents the work and the use of de is late. Many manuscripts give a similar title to the Periochae, and some even attribute them to Florus, with the result that catalogues do not always make it clear which of the two works a manuscript contains. One example: ‘Florus, Abbreuiatio librorum Titi Liuii’ in the description given by de Ricci, S. and Wilson, W. J., Census of Medieval and Renaissance Manuscripts in the United States and Canada ii (New York, 1937), 1888Google Scholar, of a manuscript owned by William M. Clearwater of Tuxedo Park, New York (I do not know the present whereabouts of his collection).

¹⁴ See Olsen, B. Munk, L'étude des auteurs classiques latins aux XI^e et XII^e siècles: Catalogues des manuscrits classiques latins copiés du IX^e au XII^e siècle i (Paris, 1982), 387–8Google Scholar.

¹⁵ Bischoff, pp. 175—7 of ‘Italienische Handschriften des neunten bis elften Jahrhunderts in frühmittelalterlichen Bibliotheken ausserhalb Italiens’, in Il libro e il testo, ed. Questa, C. and Raffaelli, R. (Urbino, 1984), 169–94Google Scholar; but he does not mention B.

¹⁶ For the main arguments that lead to it see pp. 287–9, 292–4, 410–12.

¹⁷ In fact he writes [z2][z9] in place of [z16], but according to his earlier stemma (p. 412) they derive from [z210], which together with [z15] derives from [z16]. Many of his lost hyparchetypes would surely disappear if the manuscripts were viewed with a more historical eye.

¹⁸ His main arguments against descent from P can be found on pp. 166–70, 282–4, 408–9, 490–1.

¹⁹ de la Mare, A. C. in Livy, ed. Dorey, T. A. (London, 1971), pp. 183 + 191 n. 31Google Scholar, and in Garzelli, A., Miniatura fiorentina del Rinascimento 1440–1525: un primo censimento i (Indicie cataloghi toscani 18, 1985), p. 530Google Scholar.

²⁰ De la Mare, , The Handwriting of Italian Humanists i (part 1) (Oxford, 1973), 100, 103 no. 16Google Scholar.

²¹ Ibid. 127 no. 31.

²² This manuscript puts after Florus and the Periochae three letters of Caesar, namely Ad Att. 9.7c, 13a, 16, and four poems, namely Anth. Lat. 709, 392–3, 660. Ambros. S 16 sup., described by Sabbadini, , SIFC 11 (1903), 361–2Google Scholar, puts the same texts between Florus and the Periochae, and they recur in a manuscript of more diverse content, Besançon 840. As Sabbadini observes, ‘l'uso di estrarre dalle Epist. ad Att. le lettere di Cesare rimonta al Petrarca’; he refers to Nolhac, , op. cit. (n. 5) ed. 1, pp. 249–50 = ed. 2, ii. 42Google Scholar, who also points out, ed. 1 p. 245 n. 5 = ed. 2, ii. 36 n. 5, that Venice Marc. Lat. Z 368 appends the letters of Caesar to the Periochae.

²³ Boese, H., Die lateinischen Handschriften der Sammlung Hamilton zu Berlin (Wiesbaden, 1966), p. 132Google Scholar.

²⁴ Other manuscripts that have Per. 33 include Burn. 200, Modena Est. Lat. 464 (a X 1 4), and Wroclaw Rehd. 96. After Donato's death, Gasparino Barzizza from Padua tried hard to obtain his Livy, which was ‘carior mihi … oculis meis’; see Bertalot, L., Beiträge zur Forschung N.F. 2 (1929), 81Google Scholar = Studien zum italienischen und deutschen Humanismus, ed. Kristeller, P. O., ii (Rome, 1975), p. 95, no. 48Google Scholar, quoted by Billanovich, , IMU 5 (1962), 133–4Google Scholar. It is not known whether he succeeded.

²⁵ IMU 2 (1959), 159 n. 1Google Scholar. For more about the manuscript see below, n. 53.

²⁶ Billanovich, , JWCI 14 (1951), 174Google Scholar, GSLI 130 (1953), 323–4, 332–3Google Scholar; Martellotti, G., Diz. Biog. degli Italiani 1 (Rome, 1960), 611–13Google Scholar; Sottili, A., IMU 6 (1963), 189–90Google Scholar.

²⁷ See the scholars cited in n. 5 above, especially Sabbadini, who quotes the Petrarcan postille.

²⁸ Cf. n. 17.

²⁹ Pellegrin, É., La bibliothèque des Visconti et des Sforza (Paris, 1955), p. 103 A 141Google Scholar. She failed to identify the Periochae from the words that occurred in principio ultimi capituli.

³⁰ See e.g. Rizzo, Silvia, La tradizione manoscritta delta Pro Cluentio di Cicerone (Genoa, 1979), p. 33Google Scholar.

³¹ See Martellotti, G., Diz. Biogr. degli Italiani 7 (1965), 34–9Google Scholar. Rome Vallicell. R 33, a composite manuscript in which the Periochae were written at Piacenza in 1463, actually attributes them to Barzizza; see Kristeller, P. O., Iter Italicum ii (London/Leiden, 1967), p. 133Google Scholar. As Reid missed it, I know nothing about its text.

³² After the text he wrote topical notes up to 1462, of which the first concerns an earthquake that happened in regno Sicilie et maxime in ciuitate Neapolis on December 4th inditionis quinte while he was himself at Naples, the second the contingent sent from Buda on July 26th 1456 to join the papal forces against the Turks. The earthquake happened in 1456; see Baratta, M., I terremoti d'Italia (Turin, 1901), pp. 66–74 (I thank Charles Melville for this reference)Google Scholar. It would be more convenient if he wrote the manuscript in the neighbourhood of Barolo, which is in Piedmont.

³³ Sabbadini, , Le scoperte dei codici latini e greci ne' secoli XIV e XV ii (Florence, 1914), pp. 180–1Google Scholar; Hankey, Teresa, Diz. Biog. degli Italiani 5 (1963), 707–9Google Scholar.

³⁴ Hearne printed the life from Oxford New Coll. 277 in his edition of Livy (Oxford, 1708), i b3v–b4r, and Massèra, A. F. from Laur. 63.8 in Giovanni Boccaccio: opere latine minori (Bari, 1928), pp. 257–8 (cf. 369–70)Google Scholar. See also Billanovich, , GSLI 130 (1953), 328 n. 1Google Scholar.

³⁵ Casella, M. T., Tra Boccaccio e Petrarca: i volgarizzamenti di Tito Livio e di Valeria Massimo (Padua, 1982), pp. 278–86Google Scholar; oddly, the table on p. xiv gives a somewhat different date, c. 1341–2. She quotes or discusses citations on pp. 55–7, 63–4, 78, and already in IMU 6 (1963), 114–17, 124–5Google Scholar. On p. 64 she suggests a connexion between Boccaccio's manuscript and Vallicell. R 33, mentioned above (n. 31); but her three arguments are at best weak.

³⁶ Rossbach, ed. of Florus, p. xxii; Coulter, Cornelia C., IMU 3 (1960), 282–3Google Scholar; Antonia Mazza, ibid. 9 (1966), 47–8, 51.

³⁷ Rizzo, , op. cit. (n. 30), 23–47, especially 44–6Google Scholar.

³⁸ RHT 14–15 (1984–1985), 55 and n. 8Google Scholar.

³⁹ Before I became interested in the Periochae, I examined Valencia Univ. 482 (c. 1480, Florence), which contains the fourth decade of Livy, Floras, and the Periochae. I happen to have noted that it has a gap after Per. 135 and in the margin Hie deficiunt duo libri nec repperi quenquam habentem.

⁴⁰ IMU 25 (1982), 343Google Scholar.

⁴¹ The scribe of 63.9 wrote 63.8 of the third decade, copied at least in part from the editio princeps; see Rivista di Filologia 115 (1987), 154 n. 1Google Scholar. On 63.33 see Rivistadi Filologia 114 (1986), 167Google Scholar.

⁴² Rivista di Filologia 114 (1986), 167Google Scholar.

⁴³ Oxford New Coll. 277 and Vat. Ottob. Lat. 2044 (both mentioned above) include Per. 1–21, Vat. Chig. H VIII 254 Per. 11–20, Esc. g I 3 Per. 1–10, Modena Est. Lat. 385 Per. 22–30 (it has lost Per. 21); see also above, n. 24. Vat. Reg. Lat. 1847 (s. xv¹), which breaks off in the middle of Per. 3, is a French manuscript of Florus and may belong to the group about to be discussed.

⁴⁴ I have encountered the same problem in the Periochae. At 2.22, for instance, Jal's apparatus implies that P omits pedibus, and Reid agrees (pp. 111–12), but Reid also says that N omits debilis; Rossbach says nothing either in his article or in his apparatus. Incidentally, the unwanted et just before should be changed to id, not deleted.

⁴⁵ Pellegrin, É., BEC 103 (1942), 85–6, 98CrossRefGoogle Scholar; it has an erased ex libris of the Collège de Navarre, Paris. By some muddle, probably unthinking subtraction, Jal cites Z in 14 places after 2.21.9.

⁴⁶ Jal omits Leiden Voss. Lat. O 5 from his stemma but says elsewhere that it is a twin of L (i.cli n. 5); Reid takes the same view in the Periochae but derives them both from G (pp. 73–97). Reid also derives M from Harl. (pp. 63–71). In his edition of the Periochae, i.xcvii, Jal describes yet another descendant of Λ, Paris B.N. Lat. 5794 (s. xv). Three more northern-European manuscripts that need to be brought into the picture, all of s. xii, are Cambridge Corpus Christi Coll. 313 (not a descendant of e), Leiden Voss. Lnt O 70A + W. Berlin Lat. Q 306 (fragments), and Pelplin Semin. Duchownego 2574; see Olsen, Munk, op. cit. (n. 14), pp. 383–8Google Scholar.

⁴⁷ For his dating of N see op. cit. (n. 11), 96. He very kindly tells me that in his forthcoming catalogue of 9th-century manuscripts he will attribute B to ‘Loire-Gegend (?), ix. Jh., ca. 2. Viertel’.

⁴⁸ Rhein. Mus. 22 (1867), 472Google Scholar.

⁴⁹ For the text see Rossbach, p. xxviii, Malcovati, p. xxii, Jal i.cliv. I agree with Jal, i.clv, that it is not ancient. One descendant of Λ, B.L. Harl. 2620 (s. x/xi), has the last part of it, in the form nec melius … hoc Anneo quisquam componere potuit, alongside the passage on northern invaders (1.37 2 = 3.1.2, fol. 16v).

⁵⁰ Les manuscrits classiques latins de la Bibliothèque Vaticane ii (part 2) (Paris, 1982), 226–7Google Scholar. As H is the earliest witness to the preamble, its date matters more than usual; I have given the date adopted in the catalogue just mentioned, but Bischoff, , op. cit. (n. 11), 96–7Google Scholar, had adopted s. x/xi. H contains Solinus, Florus, and a fragment of Orosius, and between about 1544 and 1553 Vinetus collated for editions of Solinus, Florus, and Eutropius, a manuscript at the Dominican library in Bordeaux that had the same subscription as H in Solinus and like H included the preamble to Florus. Could H not be the beginning of Vinetus's manuscript? He had no need to mention Orosius if he did not intend to edit him.

⁵¹ On Y see Maslowski, T. and Rouse, R. H., IMU 22 (1979), 97–122Google Scholar, and on Q the opinion of F. Avril cited there, 121 n. 2; further on Q, Texts and Transmission, ed. Reynolds, L. D. (Oxford, 1983), pp. 76, 110–11, 160, 172, 402Google Scholar. Richard Rouse has kindly shown me a copy of a forthcoming article on Philip of Harcourt and his library.

⁵² Manitius, pp. 145—6 of ‘Handschriften antiker Autoren in mittelalterlichen Bibliothekskatalogen’, Zentralblatt für Bibliothekswesen Suppl. 67 (1935)Google Scholar. In Neues Archiv 7 (1882), 533–4Google Scholar and then in his Geschichte der lateinischen Literatur des Mittelalters, for instance i (Munich, 1911), p. 666Google Scholar, iii (1931), pp. 255, 389, 638 n. 1, he attempted to document use of Florus in various authors of the ninth and twelfth century, but he ignored the problem of indirect quotation; to take one example, variants show beyond doubt that the passage about the Gauls quoted from ‘Eutropius libro tertio Romanae historiae’ by Ralph, of Diss, ed. Stubbs, W. (London, 1876), i. 25Google Scholar, came not straight from Florus 1.20.1–2 = 2.4.1–2, as Stubbs and Manitius say, but through Jordanes, , Rom. 178Google Scholar and Paulus Diaconus, Hist. Rom. (an expansion of Eutropius) 3.5.1 have not seen Grunauer, E., De fontibus historiae Frechulphi episcopi Lixouiensis (1864)Google Scholar, but Jordanes was certainly one source that Freculph used.

⁵³ Nolhac, , op. cit. (n. 5), ed. 1, pp. 244–5 = ed. 2, ii. 34–5Google Scholar. The precise date of Paris Lat. 5690 between c. 1290 and c. 1330 continues to arouse controversy, mainly because it includes three decades of Livy and if late enough may reflect work of Petrarch's on the text; see Rivista di Filologia 115 (1987), 420–30Google Scholar, and Conti, A., Prospettiva 44 (1986), 76–7Google Scholar.

⁵⁴ See the older series of the Catalogue général des manuscrits des bibliothèques publiques des Départements iv (Paris, 1872), p. 359Google Scholar.

⁵⁵ Ullman, B. L., The Humanism of Coluccio Salutati (Padua, 1963), pp. 196–7Google Scholar.

⁵⁶ In the lower margin of the second page (fol. 93v) it has the biographical preamble and at the end of the text (fol. 128v) the note Lucius Anneus Florus senatui populoque Romano … xviii annum agens, reported from Laur. Fies. 181 by Bandini, , Bibliotheca Leopoldina Laurentiana iii (Florence, 1793), p. 121Google Scholar; from Reims 1327 (Jal's β) in the Catalogue général des manuscrits des bibliothèques publiques de France: Départements xxxix (Paris, 1904), p. 474Google Scholar; and from Vat. Reg. Lat. 1564 and 1847 and Vat. Lat. 1860 in Les manuscrits classiques latins de la Bibliothèque Vaticane ii 1 (1978), pp. 288, 449Google Scholar.

⁵⁷ Sabbadini, , op. cit. (n. 5), p. 197Google Scholar, mentions only the first transposition, but both are reported from the closely related manuscript Kraków 416 by Passowicz, P., p. 387 of ‘De Flori codice Cracouiensi’, Rozpr. Akad. Umiej., Wydz. Filol. ii 12 (Kraków, 1898), 376–99Google Scholar.

⁵⁸ Reviewing Jal's, edition of Florus, CR 83 (1969), 303Google Scholar, F. R. D. Goodyear remarked with his usual trenchancy that Rossbach's apparatus gave pretty well all the information necessary for constituting the text.

⁵⁹ Athenaeum 25 (1937), 294–6Google Scholar; more summarily in her edition, pp. xvi–xvii.

⁶⁰ Athenaeum 27 (1939), 152Google Scholar; AJP 62 (1941), 371–2Google Scholar.

⁶¹ Athenaeum 28 (1940), 264Google Scholar; nothing relevant in her second edition (Rome, 1972) or in ‘Floriana’, Athenaeum 61 (1973), 141–5Google Scholar. I have no t seen the edition of J. Giacone Deangeli (Turin, 1969), but from what I can gather it does not render my comments superfluous.

⁶² Gnomon 42 (1970), 275Google Scholar. In fact Jal draws six dotted lines, and she does not explain why more are needed. The contamination mentioned by Axelson, , Gnomon 17 (1941), 272Google Scholar, is not what Jal's reference implies (i.clix n. 2), namely contamination in the extant manuscripts, but contamination of B and the other family in editorial choices, for instance the extraction of repurgauit from purgauit (B) and repugnauit (N).