² Habet … nimis multa, quae interpolatorem et correctoris ingenium sapiant (Ernesti), degener et maiorum auctoritatem parum curans (Walther), recentissimus et mire interpolatus (Ruperti), recentis aeui multumque interpolatus (Ritter). See also the notes of Jac. Gronovius on Ann. ii. 18 and 15. 53. Lipsius had known some readings of the codex Agricolae and been inclined to think they were conjectures.

² Mendell, C. W. and Ives, S. A., ‘Ryck's Manuscript of Tacitus’, A.J.Ph. lxxii (1951), 337–45.Google Scholar

⁴ I use arabic numerals for books of the Annals, roman for the Histories.

⁵ A.J.Ph. lxxv (1954), 250–70Google Scholar and Tacitus, , The Man and his Work (1957), pp. 325–44.Google Scholar

⁶ Philologus civ (1960), 92–115.Google Scholar

⁷ C.Q. N.s. xiv (1964), 109–19.Google Scholar

⁸ Also perhaps than the first of Puteolanus's editions. See Martin, , loc. cit., p. 111. Many good corrections in L are shared with Puteolanus. We cannot establish con- clusively to whom they belong.Google Scholar

¹ I use this term rather freely to describe anyone who deliberately altered the text. The justification for its use will appear in what follows. I do not mean to imply that only one fifteenth-century scholar was involved. Alterations from at least two sources are present in L.

² Here and in various other places I have tried to give enough of the text of M to enable the reader to grasp the sense, but, to save space, I have cited only that part of the text of L which differs from M. It may generally be assumed that in these passages L otherwise agrees with M.

¹ A few examples (14. 45. 1 minitante L: minante M and iv. 44. 3 exercitata L: exercita M) are exceptions to the general pattern. minitante at 14. 45. 1 is plausible. But minitor is much more frequent in Tacitus than minor and minitante is therefore lectio facilior.

¹ I shall say something later about the recentiores other than L. At iii. 66. 4 and iv. 53. 4, as in scores of other places, most modern editions give no indication of the provenance of the correction.

² Other examples of assimilation are to be found at 12. 33 cateruaque maiorum pro munimentis constiterant M: cateruaque maiore L: cateruaeque armatorum Lipsius, Freinsheim; 13. 57. 2 prosperum … exitiosius M: exitiosum L; i. 10. 3 occulta fati et ostentis ac responsis destinatum M: occultis fatis L.

¹ It is almost incredible that anyone could take this nonsense seriously, yet Mendell (A.J.Ph. lxxv [1954], 260) says ‘the logic of the situation calls for a guard to watch Silius and Messalina. Either reading is possible’.Google Scholar

² Mendell (ibid. 266) reminds us that Nero was consul at the time and says L's reading is possible.

¹ Mendell shows better judgement on this matter than Koestermann. He rightly notes of L that ‘it is the exception when a proper name at all unusual is rendered correctly’.

¹ I find little, apart from what Koestermann has collected, which bears in L's favour, but I have noted the following: 12. 36. 1 Cartimanduae Puteolanus: Cartimcmdae L: Cartimandus M (the correction could have come from iii. 45. 1); iii. 6. 2 alamque, cui Sebosianae nomen L (Nipperdey): Sebonianae M; ii. 24. 2 Castorum L: castrorum M.

² Admirers of L may want to make something like comesatione scurrarum to set against Lipsius's correction, but (a) the bad spelling comesatione may speak against Tacitean provenance, (b) comissatio does not give as good a sense as conuersatio in the context, and (c) it would not be so closely supported by Suet, . Claud. 5Google Scholarcontubernio sordidissimorum hominum.

¹ It gives an appearance of sense: nec uim medicaminis statim intellectam socordia mentis a Claudio; bili an uinolentia simul soluta alms subuenisse uidebatur. But (a) simul is ambiguous and in an awkward position, if it is to connect the two clauses, (b) the sense of bili an uinolentia is questionable and speculation by Tacitus on this sort of detail unlikely, (c) a Claudio is oddly placed. The text remains in doubt. To apply socordia to ‘the dullness of those watching’ (Furneaux) is unconvincing, as far at least as Agrippina is concerned. J. Jackson's explanation of M's reading from dittography and a superscribed Claudii

Claudii

(thus: socordiane an ui an uinolentia) has much to commend it.

² M's tendency to omit a syllable within a word is well known. Here are a few examples: n. 12. 3 alteram for adulterum; 11. 16. 1 Italus for Italicus; 11. 25. 1 dictor for dictator; 12. 36. 2 spectalum for spectaculum.

¹ The corrections of M¹ are sometimes superior to those of L, as at n. 8. 2 praeparaberat M: properauerat M¹: praeparauerat L: parauerat Halm. The choice is between properauerat and parauerat. For the former cf. II. 37. 1 and 13. 17. 2, for the latter 13. 1. 1. and 4. 54. 2. There is no comparable support for L.

² It is likely that many more readings than we know at present are shared with other manuscripts. See Martin, , loc. cit., pp. 111 ff.Google Scholar Martin rightly complains of the inadequacy of available information about the recen- tiores. No edition gives anything like a full account of their readings. I have gleaned what information I can from the editions of Orelli, Ruperti, and Walther.

¹ There is a simple example at n. 16. I

ic

Italus M: Italus L. The correct form is preserved at II. 17. 1. Having found it there, the scribe went back and added the missing syllable above the line in the earlier passage.

² Some of L's readings are suspect because of the difficulty of the corruption to the reading of M, for instance 12. 46. 3 praesidium ammis M in fine uersus: praesidium tradituros L: omissuros Freinsheim: amissuros C. F. W. Miiller.

¹ More detailed study might, I believe, reveal various patterns and perhaps tell us something about fifteenth-century views on Latin word-order. I have not explored the question, for differences in word-order are not likely to be decisive as to the authority of L or otherwise. Similarly I have not concerned myself with differences in ortho- graphy.

² One of the worst features of Koestermann's edition is that he haphazardly adopts the word-order now of M, now of L.

¹ For a more balanced discussion cf. Koestermann, in Philologus civ (1960), 111.Google Scholar

² Koestermann argues that, if this were so, ‘müsste man erwarten, daβ auch der Mime mit in das Verderben hineingezogen wurde’. It does not seem to me psychologically incredible that a Messalina should kill a rival and die rival's accomplices, but keep her lover.

¹ I draw this information from Walther and reproduce it in the form in which he gives it.

² We shall never know the last words of Agrippina. Of our three authorities Tacitus is the most likely to have altered the tradition to his own taste.