¹ Based on an edition published at Göttingen between 1807 and 1809, to which I do not have access.

² I cite from the second edition, except when it is necessary to place Madvig's ideas in the context of conjectures published between 1860 and 1877.

³ I have tried to cite the original place of publications of all conjectures made after the appearance of Crévier's edition; for earlier proposals I rely on Drakenborch.

⁴ See Wex, C., Emendationum livianarum promulsis (Aschersleben, 1832), pp. 2–3Google Scholar. Wex has been followed by e.g. W-M, Conway and Walters, and Foster.

⁵ This remains true even if one adopts Wex's punctuation.

⁶ Further passages where it is claimed that a verb of motion has been omitted in ellipse are discussed below in the note on 41.3.4–6. It is difficult to formulate general rules as to when this phenomenon is permissible, and when it should be eliminated by conjecture; but it may be noted that most of the instances assembled at K-S ii.552 come from familiar rather than elevated prose, and thus cannot serve as parallels for alleged instances in Livy; in the words of K-S ‘Die Verben des Gehens und Kommens werden besonders in der familiären Rede nicht selten weggelassen’. One potentially relevant passage which they do cite is Tac. ann. 4.57.1 inter quae diu meditato prolatoque saepius consilio tandem Caesar in Campaniam, specie dedicandi templa apud Capuam Ioui, apud Nolam Augusto, sed certus procul urbe degere, but here Otto suggested Campaniam 〈concessit〉, and I am in full agreement with what Martin, R. H. and Woodman, A. J. write ad loc. in their commentary (Cambridge, 1989, p. 223Google Scholar): ‘the omission of a verb of motion seems out of place in elevated historical narrative, and it would be strange if T. avoided explicit mention of one of the cardinal events of the reign. We cannot know what verb has dropped out, but concessit is more likely than most’.

⁷ Sigonius stated that he found this reading in a MS.

⁸ J. F. Gronovius suggested qua fama [in] Sabinos and qua fama tum Sabinos, but neither proposal expedites the difficulties of the passage.

⁹ Heumann's conjecture (listed by Drakenborch) is recorded by Ogilvie in his apparatus, but not discussed in his commentary.

¹⁰ Gronovius had earlier considered but rejected a more violent transposition: qua fama in Sabinos quoue praesidio unus per tot gentes dissonas sermone moribusque peruenisset? aut quo linguae commercio quemquam ad cupiditatem discendi exciuisset?

¹¹ On the other hand one may argue that here the destination in question is easily deduced from the previous sentence, and that a passage like 5.47.1–2 is comparable:…interim arx Romae Capitoliumque in ingenti periculo fuit. namque Galli, seu uestigio notato humano qua nuntius a Veiis peruenerat seu…(where the previous sentence allows one to deduce that the Capitol was the destination of the messenger from Veii).

¹² Thus, rightly, e.g. Ruperti.

¹³ perlata could also be placed after in Sabinos, but the corruption would then be less easy to explain.

¹⁴ I had thought perlata my own improvement on Ruperti, but then found that Ruperti, though actually proposing lata, glossed qua fama in Sabinos with ‘sc. perlata, quae in Sabinos penetrasset’.

¹⁵ The merits of accisae were strongly argued by Rhenanus (see Drakenborch's note for the views of sixteenth century editors on this passage), and thereafter it was generally accepted by editors until Conway and Walters; they deemed it uix necessario, and returned to the reading of the MSS. Hey rightly includes our passage in his entry for accido in TLL 1.299.26–7.

¹⁶ See his 1888 Berlin Weidmann edition of books 21–5. There he deletes in in the text, but comments in the apparatus ‘in 〈statiuis〉 circa conicias’.

¹⁷ The collocation in statiuis is found also at 6.14.1 and 22.23.9. Another possibility is in 〈castris〉.

¹⁸ P actually reads ratus, but 〈i〉ratus (found e.g. as a correction in M [Flor. Laur. lxiii. 20] and L [Paris Lat. 5690]) is certain.

¹⁹ quin was first introduced into the text as a replacement for cum by the humanists: the lost MS. known as Θ read pugnaeque 〈proposuisset〉 ‘quin immo…’ (this reading is not reported by editors, but Professor Reeve has kindly checked Florence Laur. Conv. Soppr. 263, and Venice Marc. Lat. Z 364, two of the MSS from whose readings Θ may be reconstructed, and the same reading is found as a correction—in Petrarch's hand—in A (London B.L. Harl. 2493). On the relationship of A² and Θ (whose readings show that many of the conjectures which Petrarch wrote in A are probably not his own), see Reeve, M. D., RFIC 115 (1987), 424–30.Google Scholar

²⁰ See TLL 7.1.479.40–63.

²¹ Madvig², p. 285 n. 1 argued that we have to deal with a lacuna of more than one word, since Flaminius could never have begun a speech to troops who had not been present in the consilium with immo…; but this applies over-rigid logic.

²² But M. Müller punctated with a colon after hostis, and this implies a somewhat better understanding of Livy's pattern of thought.

²³ Note also ne mediocrem rem actam arbitremini at Phil. iv. 1.

²⁴ For further instances of the idiom, see e.g. Lebreton, J., Études sur la langue et la grammaire de Cicésron (Paris, 1901), pp. 302Google Scholar–3 and K.-S ii.233–4.

²⁵ Including W-M, Luchs, i. pp. 5–6, M. Müller, and Foster.

²⁶ Our passage was noted by Housman, A. E., M. Manila Astronomicon Liber Primus² (Cambridge, 1937), p. lviiiGoogle Scholar, in his list of corruptions of this kind.

²⁷ To be taken as referring to the preceding parti, and not as agreeing with litem. M. H. Crawford, ap. Laffi, U. in Cébeillac-Gervasoni, M. (ed.), Les ‘bourgeoisies’ municipales italiennes aux ii^e et i^er siècles av. J.-C. (Paris and Naples, 1983), p. 62 n. 13Google Scholar, suggested that the preceding eas and ei make it likely that earn should be taken with litem; but (i) secundum litem iudices dare postulates a sense of lis unknown to any Latin author, (ii) the parallels for litem dare cited in the next note show that it is regularly followed by secundum and an accusative. Though I cannot assent to Professor Crawford's interpretation, I am grateful to him for drawing my attention to this passage.

²⁸ iudicem or iudices dare is extremely common; see e.g. Cic. Verr. II. 2. 30 si… praetor …det quem uelit iudicem, 38 aduersarii postulant ut in eam rem iudices ex lege Rupilia dentur, 39, 40, Tull 41, and the passages cited at TLL vii.2.598.72–5. litem dare is rather more rare, but see Cic. Rose. com. 3 nunc tuas (sc. tabulas), C. Fanni Chaerea, solius flagitamus et quo minus secundum eas lis detur non recusemus, Val. Max. ii.8.2 itaque, Lutati,… secundum te litem do, Luc. 8.333–4 secundum/Emathiam lis tanta datur, and Gell. v. 10. 10 nam si contra te lis data erit, merces mihi ex sententia debebitur, quia ego uicero; sin uero secundum te iudicatum erit, merces mihi ex pacto debebitur, quia tu uiceris (all cited at TLL vii.2.1497, 4–9).

²⁹ Furthermore, as Professor Reeve has pointed out to me, it is somewhat surprising to find the appositional iudices with litem dare, when Livy has not employed patronos or aduocatos with eas causas suscipere and ei semper parti adesse.

³⁰ Alternatively, one may place a comma after omnium; but this would make a more difficult readjustment for the reader at belli auida.

³¹ My examples all come from K.-S ii.479–80, where a fuller discussion of the idiom may be found; a similar kind of ellipse is discussed by Fordyce on Catullus 45.5.

³² Luchs, i. p. 11.

³³ Note also e.g. 9.45.7, 23.7.2, 24.1.13, 25.16.7, and 23.4.

³⁴ In its support one might adduce 31.31.7 urbem agrosque suaque omnia cum libertate legibusque Reginis reddidimus, but Livy doubtless thought the repetition sua…suis would have been unnecessary or inappropriate.

³⁵ Though Jal returns to what L. seems most likely to have written, there is need for a thorough discussion of this crux in which the full evidence is cited.

³⁶ That the corrector of P was not the original scribe and thus has no independent authority is shown by the appearance of P's corruption in its descendants, several of which are similarly ‘corrected’.

³⁷ For interpolations in the first decade, see below on 41.14.1.

³⁸ See his Teubner edition of 1888 and Jahrbücher für classische Philologie Suppl. 16 (1888), 343–4.Google Scholar

³⁹ This note provides the reasons for a conjecture which I communicated to D r Briscoe before the publication of his edition, and which he records in his apparatus.

⁴⁰ Dr Briscoe drew my attention to Hertz's conjecture when I communicated my proposal to him.

⁴¹ This is Briscoe's conjecture for uere; other proposals are listed in his apparatus.

⁴² In general on such ellipses, see n. 6 above.

⁴³ Drakenborch's 〈data sunt〉 is diplomatically much less probable. Dr Briscoe, who tells me that he is much less willing than I am to emend away instances of ellipse, has asked me to adduce a parallel for the assonance mandata data. Though I cannot yet meet the challenge, I do not find the assonance difficult, and observe that its effect is diminished by the second a of mandata being long, but the first a of data being short.

⁴⁴ Briscoe actually cited Plin. epist. 1.15.9, but this is a misprint.

⁴⁵ ad loc. in Madvig, J. N. and Ussing, J. L., edition of Livy pars IV.1 (Copenhagen, 1864).Google Scholar

⁴⁶ See Weissenborn, W., Lectionum livianarum particulae (Eisenach, 2 vols), i.17–18.Google Scholar

⁴⁷ Sigonius' nuntius 〈opprimendum uenit〉 is impossible; the variation by which later editors just supplied uenit has less to commend it than the conjecture of Weissenborn.

⁴⁸ Madvig¹, p. 495 = ², p. 604. Hertz later suggested militem, but the collective singular seems less appropriate.

⁴⁹ See e.g. 36.3.13, 44.37.5, and many further examples at Packard iii. p. 65.

⁵⁰ These two last objections to Madvig's conjecture were put to me by Dr Briscoe.

⁵¹ Briscoe, however, is probably correct to suspect that this last is a misprint in W-M and not a new proposal by H. J. Müller.

⁵² Gitlbauer, M., De codice liviano vetustissimo vindobonensi (Vienna, 1876), pp. 97–101.Google Scholar

⁵³ Harant, A., Emendationes et adnotationes ad Titum Livium (Paris, 1880), p. 223Google Scholar. Harant seems to have been unaware of the proposal of Gitlbauer.

⁵⁴ von Hartel, W., ZöG 17 (1866), 1.Google Scholar

⁵⁵ This was adduced by Hartel (but with a faulty reference). It has to be admitted that further parallels from books 41–5 are not easily produced; but at 41.8.2 sed ea propter belli magnitudinem prouincia consularis facta; Gracchus earn sortitur, Histriam Claudius Drakenborch must have been correct to delete Gracchus… Claudius (cf. 9.1 and 9.8), and Briscoe's excision of the whole passage is attractive; and at 44.19.6. sacrificio rite petfecto, which is interpolated before primi, looks like a marginal summary which reproduces a phrase found in §4.

⁵⁶ Most are listed on pp. vii–viii of the 1919 Oxford Classical Text of C. F. Walters.

⁵⁷ Note also that the π-group of MSS interpolate numerus caesorum apud uolsinios at 10.37. 2, the γ-group quemadmodum Satricum capiatur a Camillo at 6.8.9 (for other small interpolations in γ, see Ogilvie, R. M., CQ 1 [1957], 69Google Scholar). See in general Drakenborch's note on 7.40.4.

Further parallels may be adduced from the text of Velleius Paterculus. Watt, W. S., Vellei Paterculi Historiarum Libri Duo (Leipzig, 1988), p. 9Google Scholar, provides a useful list of passages where interpolation is probable or possible; and of these 1.8.6 hic centum homines electos appellatosque patres instat habuit consilii publici; hanc originem nomen patriciorum habet. [raptus uirginum Sabinarum] is closest to the corruption postulated in our passage (though whatever Velleius said about the rape of the Sabine women has been lost in a lacuna after Sabinarum). See also above on 29.22.10.

⁵⁸ Jal does not even bother to record the reading of V.

⁵⁹ In Latin as a whole this coupling is extremely common, and Cic. har. resp. 4, Phil. 13.2, 14.8, Pomp. Trog. ap. lust. 38.6.6, Quint, inst. 2.16.2, Flor. 2.15.1, Arn. nat. 1.65, and lust. 40.1.1 are cited in TLL.

⁶⁰ On the other two occasions (2.17.2, 39.51.4) it is coupled with odium; a further instance would be given by Alschefski's attractive supplement at 26.13.9 odii 〈inexpiabilis〉 exsecrabilisque.

⁶¹ All are listed by Briscoe.

⁶² See Weissenborn, op. cit. (n. 46), ii. p. 9.

⁶³ See Madvig¹, p. 510 = ²p. 620.

⁶⁴ See Seyffert, M., NJPhP 7 (1861), 835.Google Scholar

⁶⁵ See Vahlen, J., ZöG 12 (1861), 250–1Google Scholar = Gesammelte philologische Schriften (Leipzig and Berlin, 1911–1923), i. pp. 592–3.Google Scholar

⁶⁶ See Müller, M., NJPhP 15 (1869), 349Google Scholar. Goldbacher, A., Kritische Beiträge zum XLI., XLII. u. XLIII. Buche des T. Livius (Vienna, 1919), pp. 40–1Google Scholar proposed the same conjecture, apparently unaware that he had been anticipated by Müller.

⁶⁷ See Madvig², p. 620 n. 1.

⁶⁸ Reported by Müller, H. J., JPhV 9 (1883), 340.Google Scholar

⁶⁹ See Roobol, J., Exercitationes criticae in T. Livi Libros XLI–XLV (Diss. Utrecht, 1916), p. 17.Google Scholar

⁷⁰ See Damsté, P. H., Mnem. 50 (1922), 46.Google Scholar

⁷¹ A possible objection to this argument (as Dr Briscoe has pointed out to me) is that the Greeks may not have thought so rationally; but I do not think that it is strong enough to protect Weissenborn's conjecture.

⁷² As parallels for obiecta Müller cited several passages, including 6.1.12, 22.34.6, 42.6, and 34.9.4, but none much helps his case. Schlesinger translates the proposal of Müller and Goldbacher (which he misreports) ‘or because they did not wish all things to become completely subject to the Romans’: but obiecta esse cannot mean ‘to become completely subject’. Jal also follows Müller and Goldbacher and translates ‘ou parcequ'ils ne voulaient pas que 〈tout〉 fût à la merci des Romains’; but this is equally improbable.

⁷³ He does use the imperfect of uelle after nec, but that is rather different; imperfect forms of nolle are found six times in his work.

⁷⁴ Not, however, those of Grynaeus and Seyffert; but we have seen that there are other reasons for impugning these.

⁷⁵ See Briscoe's apparatus and note.

⁷⁶ F. Luterbacher, ap. Müller, H. J., JPhV 8 (1882), 298 proposed 〈e〉 dixit for dixit, and he has been followed by e.g. Heraeus and Jal, but not by Briscoe. This conjecture may be correct: edicere is found in all the passages cited above, except §12 and 36.8.2, and also at 35.3.2 Minucuis consul Arretium die quam edixerat ad conueniendum militibus uenit; and further support comes from passages like 40.26.6 …militibus… diem edicerent quo Pisas conuenirent. Nevertheless, §12 and 42.48.4 P. Licinio consuli…mandatum ut exercitui diem primam quamque diceret ad conueniendum support the paradosis; and, rather than introduce edicere in all three passages, it is probably easier to accept that Livy occasionally used dicere in this context. (Apart from the passages discussed in this note ad conueniendum is found in his work elsewhere only at 33.41.2, which is rather different.)Google Scholar