¹ Ruysschaert's book was published by the ‘Bibliothèque de l'Université’, Louvain, 1949. Prof. A. Momigliano reviewed it in JRS XXXIX (1949), 190Google Scholar, the present writer in CR LXIV (1950), 120Google Scholar. In the following pages reference is made to the chapters and lines of C. D. Fisher's Oxford text, for the Annals and the Histories, and, for the Opera Minora, to the chapters and sections of Furneaux's Oxford text. The sections introduced in the recent editions of the Annals do not always square. It would be convenient if Fisher's meritorious editions could be brought up to date, and, at the same time, be divided into sections.

² Information on this score in Wilamowitz's and Sandys's histories of classical scholarship is not very enlightening and occasionally incorrect. According to Wilamowitz (Geschichte der Philologie 26), Lipsius was the first editor of Tacitus to use the Medicean manuscripts—which he was not. Sandys is more correct about the Medicei, but in the one specimen of Lipsius's textual criticism that he offers, in his History of Classical Scholarship 11, 303, he confuses conjecture and transmitted text: for both points, see below, pp. 33 ff. and 51. Lipsius's interest in the criticism of the text of Tacitus continued while his other interests changed. His Tacitus went through eight or more editions, and many changes, from 1574 to the posthumous folio of 1607. (The actual number is not easily available as the two fullest lists do not quite agree: see Ruysschaert XI f., and H. Goelzer in his large edition of the Histories 1920, 1, p. xviii, n. 3, to which Prof. G. B. A. Fletcher draws my attention.) Yet, according to Sandys, l.c., only two editions appeared in his lifetime.

³ The list, in fact, contains either too much or too little. If Ruysschaert had restricted his collections to Lipsius's own emendations the list would have shrunk considerably, and a clearer picture of the humanist's ars emendandi would have emerged. If, on the other hand, it was his purpose to show the progress made by Lipsius as against earlier editions or, indirectly, as against the Medicei, he should have recorded the emendations due to Puteolanus, Beroaldus, and Rhenanus which appeared by the hundred in Lipsius's text as ‘concealed emendations’. The same applies to the items taken over from Pichena for Lipsius's posthumous edition of 1607. As it is, Ruysschaert's figure of 1,064 ‘corrections lipsiennes’ is likely to cause confusion. There are after all in the Annals not anything like that number of ‘corrections lipsiennes’.

⁴ This has been recognized since Orelli, Baiter, and Grat, Andresen. F., Mélanges d'Archéol. et d'Hist. de l'École Franç. de Rome XLII (1925Google Scholar), considered one of the Vaticani (1958) an independent witness. I do not feel convinced by his arguments nor by the review of his article by H. Goelzer, Bull. de l'Assoc. Guill. Budé 1925, no. 8, p. 24. Mendell, C. W., Yale Class. Stud. VI (1939Google Scholar), has gone to the trouble of classifying the apographa.

⁵ Ferrettus may have turned the pages of the Mediceus, but the two readings which he reports in his Annotatiunculae of 1541 (cited by Ruysschaert 125, n. 2) are both incorrect: 1, 13, 14 apud tȩ, not caput, and 20, 11 intus, not invictus.

⁶ See, for example, above p. 32, n. 2.

⁷ Lipsius died in 1606.

⁸ There are, for instance, no more than nine textual changes in the first book; in only five of these cases readings of Pichena and the Mediceus are restored, in one case a manuscript's reading is ascribed to Mercerus, who had restored it by conjecture, and there are three new emendations.

⁹ In his Notae of 1575 (Introd. to Book XI, cited by Ruysschaert p. 32, n. 2) Lipsius expressed himself very clearly on Med. no. 1: ‘Unicum exemplar manuscriptum Europa habet, reconditum in Bibliotheca Medicaea, quod accurate et cum fide, ut opinio mea fert, Philippus Beroaldus exprimi curavit. Ait et Ferretus vidisse. Quorum fide nitar. Nam mihi inspiciundi eius occasio non fuit, et, ut vere dicam, post alios ne cupiditas quidem.’

¹⁰ For the evidence, see Ruysschaert l.c. 138–143. Lipsius's attitude is shown by two revealing quotations (ib. 139)—the first from a letter written in 1600: ‘Vidi Curtiana (i.e. Curtius Pichena's Notae of 1600) ad Taciturn, et bona insunt: sed plura, hercules, a tarn vetusto exemplari exspectabam. Illud mihi delectationi, et paene dicam gloriae, vel centenis locis comprobari ab eo coniecturas nostras, quas solo ingenio duce, et timide saepe, ponebamus’; and again in the introduction to the edition of 1607 where he remarks, ‘hoc in paucis sed bonis notis e Gallis Josias Mercerus fecit, hoc Curtius Pichena ab Italia, sed et e Britannia contulit Savillus’. He then goes on: ‘Pichena tamen super omnes, adiutus a Florentino bonae notae codice, qui in Medicaea bibliotheca asservatur, et qui centenis circiter locis coniecturas nostras, quod gaudeam, confirmavit.’

¹¹ l.c. 26 ff. In addition, Lipsius occasionally referred to other evidence known to him from printed editions or through the good services of his friends: see Ruysschaert 23 ff., 30 ff., 123 ff.

¹² l.c. 113 ff.

¹³ Libri, Lib. vet., mss., vet., sincerus, optimus, ille, appear without a hint at the identity of the codices. At times he specifies Vat. or Farn.

¹⁴ Ernesti's criticism is not without interest in this matter. Before the remark quoted in the text he says: ‘vulgatus turn textus erat Rhenani. itaque cum dudum deprehendissem lectionem, quam ipse (i.e. Lipsius) vulgatam vocat, non semper consentire cum lectione editionis primae Rhenani, cuius exemplum in manibus erat, putabam secundae Rhenani lectiones intelligi. sed ea comparanda didici in nonnullis quidem locis ita esse ut suspicatus eram, sed plures, quas ille vulgatas lectiones vocet, nee ibi nee in ulla alia superiorum, quas haberem, reperiri.’

¹⁵ P. Cornelii Taciti equitis ro. ab excessu Augusti Annalium libri sedecim, ex castigatione Aemylii Ferretti, Beati Rhenani, Alciati, ac Beroaldi … Lugduni apud Seb. Gryphium, 1542.

¹⁶ See Ruysschaert, l.c. 115 ff., on variants erroneously attributed to one of the manuscripts; 117, on the distinction between transmitted text and conjecture.

¹⁷ The position as regards the last (posthumous) edition of 1607 is slightly different. From the last edition but one I note the following references in the first book: 1, 8, Vertranius; 4, 15, Muretus (though his emendation is not in the text); 32, 17, Rhenanus; 74, 1, Mercerus (not in the text). The reference to Ferrettus at 13, 14, concerns a reading of the Mediceus. Conjectures of his own are abandoned at 4, 19 and 7, 4 with reference to Muretus, at 8, 21, to Cujacius, and at 70, 21, to an anonymous still unidentified. The well-known reference to Muretus at 5, 8, may also be mentioned here.

¹⁸ For the evidence, see Ruysschaert's list, pp. 172 ff.

¹⁹ See above, p. 34.

²⁰ Ruysschaert 21, n. 3, cites these emendations from Rhenanus's text: 1, 8, 28, inprospere repetitae; 35, 3, universi; 56, 9, metuebantur; 79, 17, concederetur: 11, 36, 13, honorem; 56, 17, Servaeus; 60, 16, Lycium: IV, 8, 22, confirmaret; 66, 9, conexus (correct in OCT): VI, 10, 3, Fufii: XII, 43, 2, prorutae.

²¹ Compare, for example, his remarks on Muretus at 1, 4, 15, ‘M. Antonio … Mureto, cuius scripta Venus inhabitat pariter cum Musis’ (he did not, however, put the conjecture in the text); or ib. 74, 1, ‘sidus exoriens suae Galliae Jos. Mercerus.’

²² See, for example, above, n. 17.

²³ The ordinary reader would certainly be misled by such borrowings from printed sources. Even erudite scholars, like Ernesti, tried in vain to establish the nature of Lipsius's vulgate: see above, p. 34. Much painstaking effort was needed in order to identify Beroaldus's and Rhenanus's contributions— and from Lipsius's text and notes it would certainly be impossible to give his due to either of his predecessors or, indeed, to Pichena. If several centuries had to pass until some of Rhenanus's property was discovered in Lipsius's keeping (cf. above, n. 20), it is hard to understand the meaning of these words of Ruysschaert's: ‘mais les emprunts de cette sorte étaient trop aisément identifiables pour des lecteurs tant soit peu avertis pour qu'on puisse taxer Lipse de plagiat à ce propos’ (p. 152).

²⁴ See Ruysschaert 144 ff., and below, p. 50.

²⁵ See below, p. 51.

²⁶ Ruysschaert has discovered (l.c. 149 f.) that Lipsius went to the length of erasing the name of Chifflet in the margins of his working copy so as to cover his traces.

²⁷ JRS XXXVIII (1948), 122Google Scholar—areview of H. Fuchs's edition of 1946 (Edit. Helvet., Ser. Lat. IV, 1).

²⁸ Witness Löfstedt's Spätlateinische Studien 1908, Tertullians Apologeticum textkritisch untersucht 1915, Arnobiana 1917, and the magnum opus of the Swedish School, Philologischer Kommentar zur Peregrinatio Aetheriae, 1911.

²⁹ Persson, P., Krit.-exeg. Bemerkungen zu den Kleinen Schriften des Tacitus, Uppsala, 1927Google Scholar; Eriksson, N., Studien zu den Ann. des Tac., Lund 1934Google Scholar; and Sörbom, G., Variatio sermonis Tacitei, Uppsala 1935Google Scholar. A similarly conservative attitude to the manuscript was shown by Lenchantin de Gubernatis in his edition of Annals I–VI (Rome 1940). Koestermann's, E. retractatio of Andresen's text (Teubner 1936Google Scholar) may also be compared.

³⁰ 26, 2, mandata Clementi centurioni quae perferret (praeferret); 31, 10, una et vicesimanis, misspelt as three words by Lipsius (undevicesimanis); 32, 5, prostratos verberibus mulcant (multant); 37, 10, legiones nihil cunctatas, misspelt contatas by Lipsius (contatus); 49, 6, et quidam bonorum caesi. postquam, intellecto in quos saeviretur, pessimi quoque arma rapuerant (… caesi. postquam intellectum, etc.). Lipsius retained, however, the wrong punctuation after caesi; Ruysschaert's entry, p. 174, is incorrect. In addition, two cases may be mentioned when, after Pichena's publication, Lipsius accepted correct readings of the Mediceus. 39, 1, legati ab senatu regressi (regressum) … Germanicum adeunt; at 37, 3, largitio differebatur in hiberna cuiusque. non abcessere quintani, etc., the full stop after cuiusque, and cuiusque for cuiusquam, were restored by Mercer, and accepted by Lipsius in his last edition without a mention of the fact that cuiusque had been found in the manuscript, and published by Pichena.

³¹ 49, 16, ‘tramittit duodecim milia e legionibus, sex et viginti socios cohortis (e legionibus sex, et).’

³² At 1, 3, ‘neque decemviralis potestas ultra biennium,’ Lipsius eschewed Vertranius's triennium, at 8, 8, Muretus's insertion of the antiquarian's account of Augustus's legacy (Suet., Aug. 101, 2Google Scholar) into the more general statement of the historian, and at 14, 7, Vertranius's ‘aeraque adoptionis’ for the correct text ‘aramque adoptionis’.

³³ Muretus's conjecture became respectable in the nineteenth century. It was accepted by most editors, and is still found in Andresen, Furneaux, Fisher, and Goelzer. Exulem was kept in most of the earlier editions, and is back in the texts of Koestermann, Lenchantin, and Fuchs. The accusative was defended by Thomas, P., Mnemos. 49 (1921), 43Google Scholar, who in support quoted Suet., Tib. 12, 2, ‘tune non privatum modo, sed obnoxium et trepidum egit’ (Tiberius), Pliny, , Ep. I, 17Google Scholar, 1, ‘qui defunctorum quoque amicos agant,’ and a passage from Sidonius Apollinaris. See also N. Eriksson's wordy defence, Stud. zu den Ann. 1934, 109 ff.

³⁴ Ago, with the accusative of a personal noun indicating the part taken by the agent, according to Gerber and Greef, Lexicon Taciteum p. 60, B, is found five times in Tacitus, : Hist. I, 30, 4Google Scholar; 11, 83, 2; IV, 2, 3; Ann. 1, 4, 15, and XVI, 28, 11. I cannot find the idea of pretence in any of these passages—for instance, what sense would there be in Domitian pretending to be the son of the Emperor (Hist. IV, 2, 3) when he was precisely that, but ‘nondum ad curas intentus‘ ‘fulfilled his part’ only ‘stupris et adulteriis’ ? The Thes. L.L. offers interesting material s.v. ago, col. 1399. Compare also Vell. Pat. 11, 124, 2, ‘ut potius aequalem civem quam eminentem liceret agere principem.’

³⁵ Chap. 3, 24, ‘sed quo pluribus munimentis (monumentis) insisteret’ sc. Tiberi domus: the emendation is also claimed by Muretus, , Var. Lec. XI, 1Google Scholar; 5, 8, gnarum (C. Nauum, corr. G. nauum M) id Caesari: also claimed by Muretus, see below, p. 51; 10, 20, Iullos (Iulios): misspelt Iulos by Lipsius, cf. Mommsen's, article Hermes 24 (1889), 155Google Scholar, as cited in the notes on the passage; 13, 26, ‘donec Haterius Augustam oraret eiusque (et usque vulgo etusque M) … precibus protegeretur’; 22, 13, hi (ii); 25, 1, ‘postquam vallum introiit (introii),’ despite Draeger, Pfitzner, and Gerber and Greef's Lexicon p. 1149, B; on the present after postquam restricted to certain verba sentiendi, see Nipperdey and Andresen ad l., Hofmann Syntax 734; 26, 10, ‘numquamne ad se nisi (nisi ad se) filios familiarum venturos’; 31, 4, ‘daretque se legionibus vi sua cuncta tracturis (tracturus) ’; thus Lipsius, 1585 (Ruysschaert p. 173), before Freinsheim; 57, 5, ‘quanto quis audacia promptus, tanto magis fidus rebusque motis (rebus commotis) potior habetur’; 58, 22, ‘sedem vetere (uetera) in provincia pollicetur’: vetere M² and Lipsius; 76, 10 ‘in vulgus formidolosum’: in added by M² and Lipsius; 76, 16, quamquam id quoque (quod) dictum est’; 77, 2, ‘occisis non modo e plebe sed (et) militibus et centurione’: sed M² and Lipsius. Here I also mention 28, 14, hi where Lipsius indicated at least the right way by suggesting ii for in.

³⁶ Lindsay, Contrac. in early Lat. Minuscule 48, and Notae Lat.§412, mentions three usual symbols for con, namely c, ɔ, or 7, all of them well known to the reader of manuscripts. Any one of these, especially the third, would easily be mistaken for q, the symbol for que.

³⁷ Ruysschaert conveniently lists the changes found in the various editions: he also shows at p. 86 that ‘sa connaissance de style a donc progressé d'une réédition à l'autre’.

³⁸ Wolfflin's argument in favour of motu, for metu, 40, 1, has carried no conviction; 54, 6, ‘ludos Augustalis tune primum coepta (for coeptos) turbavit discordia’ perhaps still deserves some attention in view of 77, 1, ‘theatr i licentia, proximo priore anno coepta’: see below, p. 41.

³⁹ See Walther's note on the passage, and R. Syme, Roman Revol. 433, n. 4, JRS XXXVIII (1948), 129Google Scholar, XXXIX (1949), 7. Syme, (JRS XXXVIII, 130Google Scholar) also doubts the relevance of Dio 55, 33, 2, to Tac., Ann. I, 38, 4Google Scholar, M'.Ennius (Nipperdey) for mennius (M), i.e. M. Ennius; and (JRS XXXIX, 12 and 14) considers the evidence for the names Hispo 1, 74, and Falanius 1, 73.

⁴⁰ N. Eriksson l.c. 115, following in the main Walther's note in his edition of 1831, which was also approved by Ruperti (1834), Pfitzner (1869), and recently by Syme, , JRS XXXVIII, 129Google Scholar.

⁴¹ Since both traits—Rufus the hard worker and Rufus the disciplinarian—are anyhow expressed in the sentence I can make but little of Walther's contention that the latter alteration is an inepta tautologia.

⁴² For the skipping of letters and syllables in the Mediceus, see Rostagno's preface to the Leyden Facsimile p. XIII, also Ruperti's edition 1, 227.

⁴³ Lex. Tac. 663, B. There are seven passages all told, and they all are about military matters. In Gerber and Greef's first quotation attentus is a misprint for intentus.

⁴⁴ Eriksson l.c. 116 cites Sen., De Clem. 11, 5, 3 (and Ausonius), for the genitive with attentus, and rightly says that words like ferox provide parallels in Tacitus. Cf. Draeger, Synt. d. Tac., 3rd ed., §71, Furneaux Tac., Ann., Introd. chap. V, § 33.

⁴⁵ According to the Lexicon Taciteum, misceo is used by Tacitus with the accusative (one or several), or accusative and ablative. Also, adulatio is used in singular or plural.

⁴⁶ The conjecture has been discussed here because it goes under Lipsius's name, but Ruysschaert (p. 172) has pointed out that Lipsius himself attributes it to Divaeus: ‘sans doute une allusion à une conversation de l'auteur avec Pierre van Dieve (1535–1585), l'historien louvaniste. Cf. Biogr. Nat. Belg.’

⁴⁷ JRS XXXVIII (1948), 128Google Scholar.

⁴⁸ Omitting the common instances for the suprascript stroke wrongly put or missed, I mention 1, 8, 4, augustu or augusta instead of Augustum, wrongly altered to augustȩ (= Augustae) by the first hand of M: see Rostagno, in the preface to the Leiden Facsimile, p. XV; Fisher, however, ascribes the alteration to M². Equally, exacta is put for exactum at 11, 85,7. At 11, 15, 5, pars onusta vulneribus tergũ the opposite mistake is found; the stroke is wrongly put on what ought to be the letter a, but was taken to be u. Because of the structure of the following clause, I still believe Muretus's conjecture terga to be right, and deplore Lenchantin's printing of the manuscript's reading. At VI, 16, 1, redactu may be either redactum or redacta, probably the former.

⁴⁹ For the Augustalia, see Wissowa, Religion (2nd ed.) 453, 457, and the other works cited there, and P-W 11, 2361. Wissowa refers to the very instructive parallel of the Ludi Apollinares, as reported by Livy 25, 12, 9–12; 26, 23, 3; 27, 11, 6; 27, 23, 5–7.

⁵⁰ I skip Dio 56, 29, 1 (A.D. 13, the year before Augustus's death): the incident reported there happened either at the Augustalia in October or at the Games in honour of his birthday in September.

⁵¹ See above, p. 39, n. 38. The conjecture is based on the parallel between 54, 6, ‘ludos Augustalis tune primum coeptos turbavit discordia ex certamine histrionum,’ and 77, 1, ‘theatri licentia, proximo priore anno coepta.’

⁵² Compare Orosius's paraphrase VI, 10, 3, ‘cum … instrument a ruralia non haberent, gladiis concidendo terram et sagulis exportando … vallum … et fossam … perfecerunt.’ Horace, (Epode V, 31Google Scholar) describes a similar action for a different purpose, ‘ligonibus duris humum exhauriebat’.

⁵³ Both Caesar's and Tacitus's descriptions are graphic enough to enable a reader to recognize the same operation when it is described in Vegetius's, Epitoma rei militaris III, 8Google Scholar, ‘primum in unius noctis transitum et itineris occupationem leniorem (castra muniuntur) cum sublati caespites ordinantur et aggerem faciunt, supra quern valli, hoc est sudes vel tribuli lignei, per ordinem digeruntur. Caespes autem circumciditur ferramentis qui herbarum radicibus continet terram,’ etc.

⁵¹ Examples for egero in this sense are cited in the Thesaurus p. 242, col. 2, 41 ff.

⁵⁵ See Forcellini and De Vit; also Georges, Lexicon (8th ed.), s.v.

⁵⁶ Cic., Ad Att. III, 15, 5Google Scholar, pergo praeterita IV, 11, 1, perge reliqua, and (probably) De Leg. 11, 69, perge cetera, are cited in corroboration. The ellipse of a verb like dico is, of course, very frequent in Cicero.

⁵⁷ E. Löfstedt has discussed many cases of this kind, mainly, but not exclusively, found in later Latin, : Syntactica 1 (2nd ed. 1942)Google Scholar, chaps, XIV and XV. Students of grammar differ as regards the early history of transitive and intransitive verbs. Rego and its compounds (pergo among them) happen to figure in this controversy in which the present writer has felt more convinced by the arguments set out by Wackernagel, J., Vorles. über Synt. 11 (2nd ed.), 179Google Scholar, than by J. B. Hofmann's discussion of the same problems, Syntax, 27 and 378.

⁵⁸ This idea was applied to the present passage by Boetticher, Lex. Tac. (1830) 19. Compare also Ruperti and Furneaux ad l.

⁵⁹ So Cic., Pro Quinct. 76 and Livy XXII, 38, 13. The parallels are adduced in Draeger and Heraeus's note on the passage.

⁶⁰ There, however, the end of a line intervenes between non and prudentem, and the reason for the mistake may be different from that in the first book.

⁶¹ Jean Chifflet's objection ‘defectus enim non fit repente sed paulatim’ (Lips., Epist. ed. Burmann 1, 727) is refuted by such passages as Cic., Rep. I, 23Google Scholar, ‘quod serena nocte subito candens et plena luna defecisset’ (cited by Furneaux), and ib. 25, ‘cum obscurato sole tenebrae factae essent repente.’

⁶² The alternating order of words, thus brought about, resembles the arrangement of adjectives and nouns at 10, 5, ‘simulatam Pompeianarum gratiam partium,’ which is not quite so praeposterus as H. Fuchs maintains in his note on the passage; cf. Fletcher, G. B. A., CR LIX (1945), 67Google Scholar.

⁶³ Lipsius on the vulgate at 1, 28, 2: ‘ “luna clariore paene coelo uisa.” Quid hoc clariore paene caelo ? Cassa palearum, si examinas. A Beroaldo ea lectio est, quam sperne: et substitue “luna claro repente caelo”. Facit sententia, et prisca scriptura “clamore pena caelo”. Claro autem caelo is Lunae languor: et ideo sequitur, “postquam ortae nubes offecere visui.” ’

⁶⁴ See above, p. 37.

⁶⁵ VI, 26, 5, cited by Gronovius.

⁶⁶ 1, 17, 2, cited by Andresen, who also compared Agr. 25, § 3, magno paratu, maiore fama.

⁶⁷ The adverb qualifies the verb only indirectly, like a predicative adjective, at Ann. IV, 47, 6, quidam audentius … visebantur which means ‘they were seen to be daring’. So also XV, 45, 5, prospere aut in metu sacraverat. These are two out of three parallels cited by Walther in support of the MS. reading gravius, in our passage of the Annals. But gravius, if it were correct, would here be an attribute, not a predicative. In Walther's third passage, Germ. 5, § 4, ‘simplicius et antiquius … utuntur,’ I can find nothing but the usual employment of the adverb. This use of the adverb has proved a trap for eminent textual critics; cf. Madvig on Cic., De Fin. IV, 63Google Scholar, and Kühner and Gerth on the corresponding Greek idiom, Griech. Gram. § 497, 4.

⁶⁸ Tacitus sometimes combines adverbial expressions of the predicative and attributive classes, as at Hist. 11, 98, 2, ‘palam epistulis … Vitellium, occultis nuntiis Vespasianum fovens’—where palam can go with epistulis as much as with fovens. Despite such passages it is expedient to distinguish the two usages. For unless this is observed the restricted character of the attributive group is easily forgotten. This Lenchantin perhaps failed to remember when he sought to justify the reading gravius in our passage of the Annals by referring to Sörbom, Variatio serm. Tac. 96 f. Sörbom indeed gives many examples of the variatio of adjective and adverb, but more than that is needed to justify the manuscript's reading. Cf. Hofmann, Syntax 461 and 467, with bibliography.

⁶⁹ This point is made pace Dr. E. A. Lowe who has advanced the view that the Mediceus no. 11 was directly copied from a codex in capital writing; see The unique Manuscript of Tac. Hist. 271 (Montecassino 1929). It is hard on this hypothesis to account for the mistakes in that codex, due to the misreading of minuscule lettering, such as s/f.

⁷⁰ Chap. 13, 20, ‘flexit paulatim, non ut fateretur suscipi a se imperium, sed ut negare et rogari desineret.’ Gerber and Greef's references, Lex. Tac. 961, B, 4, and 962, A, 8, suggest that the two passages from the Annals, and Dial. 31, § 1, non ut … nec ut … sed ut, are the only examples with ut of this kind in Tacitus. There would have been nothing inherently unlikely in the construction non ut, sed + subjunctive, but Tacitus seems not to have employed it.

⁷¹ Tacitus's own practice is stated in note no. 70. Another case of what is usually, and I believe rightly, regarded as an ut missed out in the Mediceus is declared by Syme to be due to Tacitus's pen—an act of violence which, he says, now finds approval: l.c. 123. The passage is 1, 9, 16, ‘non aliud discordantis patriae remedium fuisse quam ab uno regeretur’—the ut added by Ferrettus, and approved by Lipsius. Eriksson l.c. 92 defended the manuscript's reading; so did, unnoticed by him, Baehrens, W. A., Beiträge zur lat. Syntax 375 (Philol., Supp. XII, 1912)Google Scholar. Both authorities seem to overlook that non alius quam ut is not identical with non potius (or similar comparatives) quam ut — subjunctive. In the latter expression ut was never more than gratuitous, even in indirect statements, and more often than not spurned. This has been well brought out by Stegmann in Kühner's, Lat. Gram. vol. 11 (2nd ed.), 300–2Google Scholar, and by Hofmann, Syntax 731 f. After comparatives colloquial, or late Latin, usage sometimes has quam to mean quam cum, quam si, or the like. On the other hand, non alius quam followed by a subjunctive without ut has not been noticed anywhere before Florus (1, 13, 10)—a passage which itself may or may not be sound; cf. Thes. L.L. 1, 1634, 2 ff. The negative point is borne out by Tacitus's own usage. Ut is freely omitted after potius quam (Lex. Tac. 1155, A) though not after other comparatives (l.c. 1247, A). After non alius quam or the like no case of a missing ut is recorded apart from our passage, but there are eight instances of ut (66, B and 1246, B) and some others of quam ne, quam si, quam quia, and quam quod.

⁷² Dio 57, 2, 6, οὐχ ὡς καὶ τὸ τρίτον ἕξοντός σου, ἀλλ᾿ ὡς ἀδύνατον ὄν τὴν ἀρχὴν διαιρεθῆναι, τοῦτό σοι προέτεινα.

⁷³ This number comprises both et si quis alius and sive (seu) quis alius.

⁷⁴ For et quis alius and the like, see Lex. Tac. 67, A.

⁷⁵ A similar problem is found also in the isolated ablative veteri (for vetere) at 60, 3; cf. Sörbom, Variatio serm. Tac. 27, note.

⁷⁶ Neue, and Wagener, , Formenlehre III (3rd ed.), 413Google Scholar ff.

⁷⁷ Landgraf, Festgruss an die 41. Versammlung, etc. (Wilhelmsgymnasium, Munich). I have not seen this article.

⁷⁸ Bell. Alex. 28, 3, and the present passage in Tacitus.

⁷⁹ Livy IX, 37, 7, considerant in all MSS.; XXVIII, 12, 15, considerunt as a variant. Conway and Johnson have this note on the latter passage: ‘cf. 9, 37, 7 adn. (ubi contra codd. -sed- legimus) et Neue-Wagener III, p. 414 sqq. (in his duobus tantum locis -sid- in codd. nostris inuenitur).’

⁸⁰ The evidence is presented by Gerber and Greef, Lex. Tac., consido, insido, and praesideo. Cf. also Fuchs, H., Tac. Ann. 1 (ed. Helv., , 1946), 199Google Scholar, note on chap. 30 considerant.

⁸¹ Tac., Ann. 1, 76, 10Google Scholar, praesidit: M, for praesedit; VI, 47, 12, praesidiis se: M, for praesedisse; XII, 56, 17, praesidere for praesedere in Med. no. 11. It may also be remembered, as Fuchs notes, that supersedeo in the second Mediceus is misspelt supersideo, at XV, 63, 20.

⁸² Lenchantin's statement on p. XXXV in the introd. of his edition of Books I–VI, perhaps errs on the side of brevity.

⁸³ Insidere: Ann. III, 61, 8; insiderant XVI, 27, 2.

⁸⁴ 45, 7, ‘arma classem socios demittere Rheno parat’; 63, 11, ‘legiones classe, ut advexerat, reportat’; 70, 21, ‘quo Caesar classe contenderat.’

⁸⁵ Thus immediately before our passage, 60, 7, ‘ipse inpositas navibus quattuor legiones per lacus vexit: simulque pedes eques classis ‘… convenere.’

⁸⁶ Lenchantin quotes Germ. 2, § 1, ‘classibus advehebantur qui mutare sedes quaerebant’ picked up at the end of the sentence by ‘raris ab orbe nostro navibus aditur’. So also Germ. 44, § 2, ‘Suionum ‘hellip; civitates ‘… praeter viros armaque classibus valent. forma navium eo differt quod, etc.’

⁸⁷ Lenchantin's second passage comes from A. XIV, 11, 13, where he says the plural classes is used de una classe Misenensi. But in the same clause the plural cohortes is used of one cohort, and Furneaux rightly explains both cohortĩs and classĩs as rhetorical exaggerations for the one praetorian cohort in attendance, and for the fleet of Misenum. No one would say that therefore the plural of cohors means one cohort in Tacitus. Lenchantin did not refer to the other passage sometimes claimed to have the same extended meaning—thus Lex. Tac. 178 and Thes. L.L. III, 1284, both s.v. classis. The passage occurs at A. 11, 75, 3: ‘Agrippina … ascendit classem cum cineribus Germanici et liberis.’ But Agrippina naturally sailed in style: she has got a fleet (cf. III, 1, 10 and 14), and nevertheless disembarks navi, not classe, III, 1, 17. There is no harm in Englishing classis by ‘boat’ at 11, 75, 3, but classis does not mean ‘boat’. Compare 11, 79, 2, where the several ships of her fleet are said to carry her, ‘obviis navibus quae Agrippinam vehebant.’ It may be added that the relevant portion in the Thesaurus needs shortening.

⁸⁸ Cf. Walther's notes ad l., and on 28, 3, ac suis—there, however, the Med. has asuis for suis. The a may repeat the first letter of the preceding word accepit, or else belongs to that class of unexplained initial letters on which W. Heraeus entered a timely caveat in Lindsay's Palaeog. Lat. IV, 1925, 14. The MS. reading is strangely defended by Lenchantin.

⁸⁹ Hist. IV, 54, 1, and below, n. 92.

⁹⁰ Beiträge, etc., 1907, 37; Spätlat. Studien, 1908, 27; Pereg. Aeth., 1911, 61; also, Syntactica 11, 1933, 219; Vermischte Studien, 1936, 56.

⁹¹ The various redundant uses of connective particles have, in fact, become a favourite topic in recent grammatical lore; the references are too numerous to be given here. Hofmann has provided useful surveys, Syntax §§ 227–237, and Thes. L.L., s.v. et; especially 906, 32, on redundant et … -que. For the use of connective particles in Tacitus, see Gerber and Greef, Lex. Tac., N. Eriksson, Stud, zu d. Annalen, etc., 72, and G. Sörbom, Variatio serm. Tac. 50.

⁹² E. Tidner, De particulis copulativis ap. Scrip. Hist. Aug. (1922) 120, n. 2. The passages are these: H. 1, 80, 9, ‘fremit miles et tribunos et’ (et deleted by the same hand) ‘centurionesque proditionis arguit’; 11, 2, ‘inter spem et metumque’ (the reading comes from Laur. 68, 4, only; Giarratano does not even trouble to mention it in his full apparatus); IV, 53, 18, ‘argenti et aurique stipes et metallorum primitiae’; IV, 54, 1, ‘per Gallias et Germaniasque’; Ann. 1, 65, 16, as cited in the text.

⁹³ Apart from inscriptions, there are, so far as I can see, only two examples of redundant et … -que alleged to occur in prose before Tacitus: Bell. Afr. 33, 1, and Bell. Hisp. 42, 4, in both of which the text is doubtful. But even their being accepted by editors would be of little importance for the text of Tacitus, as this is not the kind of literature he considered a model of style. In verse, the first examples cited are Ciris 79 (v.l.) and the Carmina Epigraphica.

⁹⁴ There seems to be only one exception in Tacitus, namely hodieque meaning ‘also nowadays’ (Germ. 3, § 3). But this is normal Silver usage, cf. Hofmann, Syntax 657.

⁹⁵ Lex. Tac. 1278, B: Ann. 11, 3, 8, segue regnumque; at XVI, 16, 2, double -que connects two clauses ‘meque ipsum satias cepisset aliorumque taedium expectarem’: cf. Hofmann 656.

⁹⁶ Lex. Tac. 395, B and 1278, B: ‘seque et arma, seque et coniugem or libertum, seque et delatores, seque et domum, seque et cohortis or equestris copias, seque et proximos,’ and ‘seque et Gallias; sibique et posteris, sibique et proelio,’ and ‘sibique et legibus’; add two variations, namely ipsique et coniugi like seque et coniugem above, and regnumque et domum (Ann. XIV, 31, 3), and compare seque et domum above, and seque regnumque in note 95. Cf. Hofmann 663. To the passages cited by Gerber and Greef, one more case of seque et proximos ought to be added if the manuscript's reading is restored at Ann. 1, 34, 2, as is done in all recent editions except in the Oxford Text—one of the few cases where Fisher's judgement seems to be at fault.

⁹⁷ Lex. Tac. 1279, A: seque ac liberos, seque ac maiores. Cf. Hofmann 663. Uterque opibusque atque (Ann. IV, 34, 21) is, to say the least, doubtful.

⁹⁸ Lex. Tac. 395, B: Hist. V, 5, 13, and Ann. 13, 7, 3; cf. Hofmann l.c. and, for Cicero, Madvig on De Fin. V, 64, and Excursus 1, 792.

⁹⁹ Lex. Tac. 1279, A, citing Ann. III, 12, 23.

¹⁰⁰ It matters little whether -que is here taken with et posteris as is done by most editors, or to connect the third member of ‘foveret attolleret sibique et posteris conformaret’, as is suggested by Eriksson, Stud. zu d. Ann. 72. But the fact that the tag existed would give some point to -que et.

¹⁰¹ For examples, see Hofmann, Thes. l.c.

¹⁰² Here I neglect the conjectures which were either withdrawn or made for the sake of discussion only: see above, p. 39; also the two cruces mentioned on the same page, and some instances of corrected spelling. I further leave aside all emendations of his textus vulgatus, whether concealed or not.

¹⁰³ See above pp. 37 ff.

^103a See above, pp. 40 ff., except for 10, 24 (above, p. 43) and 30, 15 (above, p. 45) where Lipsius eventually returned to emendations found in the vulgate.

¹⁰⁴ Six of these were later found in M: see above, p. 37, n. 30, and p. 38.

¹⁰⁵ See above, p. 43.

¹⁰⁶ G. H. Walther, not himself partial to conjectures too different from M's text, said resignedly, ‘difficillimum sane erit meliora proferre.’

¹⁰⁷ He was able to make do with such stumblingblocks as (41, 4) quod tam triste, (ib. 6) et externae fidei, and other difficult readings mentioned below, n. 114, and if he did not explain them he did not expel them from the text.

¹⁰⁸ One may say that he did try to live up to the rule about emendations laid down in the SC de correctoribus in his amusing Satyra Menippea: ‘Siquis e libris bonis fidisque correxerit, laudi semperesse. Siquis e coniecturis, noxae. Nisi eae clarae, liquidae, certae sunt.’

¹⁰⁹ I have mentioned seventeen such cases from the first book, about half of them in deference to the opinions of others though often unacknowledged: see above, p. 39.

¹¹⁰ This is but a rough estimate of the number of Beroaldus's emendations. Lenchantin, who does not profess to give a full apparatus, lists some fifty of them. In obtaining the figure given in the text I have considered Beroaldus's marginalia in the Mediceus as well as his edition; many of the marginalia reappear in the edition. I have excluded, however, the interlinear corrections in the codex, some of which may well be due to Beroaldus. The various hands are, in fact, insufficiently known. I feel doubtful even of some of the marginalia ascribed to M ²: for example, I cannot find any difference between the hand, variously described as M ² or ‘in marg.’, which added sed (for et) at 77, 2, fol. 30R, and that which added dum (for tum) in the line before, and is correctly ascribed to Beroaldus.

¹¹¹ This figure is obtained by selecting only the items not in the codex from the twenty-six mentioned above at p. 48, and n. 104.

¹¹² Rhenanus's emendations have been accepted at 6, 23, reddatur; 8, 28; 28, 10; 30, 15; 35, 3 and 14; 56, 9 (doubtful); 65, 27; 68, 14, and 79, 17. Four of these went under Lipsius's name: see above, p. 35, n. 20. For 32, 17, where Rhenanus was on the right track, see above, p. 45.

¹¹³ The scholars concerned are Victorius, Paulus and Aldus Manutius, Ferrettus, Vertranius, Muretus, Mercerus, Pichena, Acidalius, Grotius, Freinsheimus, Heinsius, and J. Gronovius. The actual number of Muretus's contributions is not easily assessed, see below, p. 51. Aldus Manutius's pontis for the MS. poti (ponti Beroaldus, but pontis in his marginal note in M) (69, 8), needs to be reconsidered: see Löfstedt's discussion of genitive and dative, Synt. 1 (2nd ed.), 209 ff.

¹¹⁴ Wolf's easy alteration at 3, 7, is generally accepted, and Bezzenberger's conjecture is probably the best that can be done at 8, 12. There is a probable emendation also at 57, 16 (Ruperti and Spengel), while the following passages need reconsideration: 4, 16; 8, 1 (where Sörbom, Variatio 151, is more to the point than Lenchantin's reference to Eriksson); 8, 9; 11, 1; 38, 4 (see above, p. 39, n. 39); 41, 4, and 6 (cf. Löfstedt, , Synt. I, 2nd ed., 190Google Scholar, n. 2); 42, 7; 49, 5; 56, 9, and 75, 13.

¹¹⁵ Syme goes so far as to ascribe to Tacitus certain features of spelling found in the Mediceus. He advances the theory (p. 125) that the Medicean scribe's writing in full, or abbreviating, Roman praenomina may represent Tacitus's own usage— a forgotten source of Tacitean variatio.

¹¹⁶ Cf. above, pp. 40 ff., 43 ff. The other emendations which Syme rejects (l.c. 126 and 129) concern diverse cases. There are two, now generally and rightly rejected (if not by H. Fuchs)—31, 15, impellere, and 74, 23, paenitentiae. There are, further, the various nineteenth-century emendations mentioned above, n. 114, the merit of which has still to be argued. There is, lastly, the reading ponti (69, 8) (above, n. 113) which is of the same kind as the other grammatical problems. Syme might have mentioned that Lenchantin, very properly, but against his doctrine, has queried the order of words in quisque cuius (21, 7)—one of the few points of accord between Lenchantin and H. Fuchs.