^{page 350 note 2} Philol. xxv (1867), 92 ff.; xxvi (1867), 92 ff.; xxvii (1868), 113 ff.

^{page 350 note 3} Syntactica: Studien und Beitrage zur historischen Syntax des Lateins, ii (Lund, 1933), 283 ff.Google Scholar

^{page 350 note 4} Studien zu den Annalen des Tacitus (Lund, 1934).Google Scholar

^{page 350 note 5} ‘Development of Language and Style in the Annals of Tacitus’, F.R.S. lviii (1968), 22 ff.Google Scholar

^{page 350 note 6} See, e.g., Yule, G. Udny, The Statistical Study of Literary Vocabulary (Cambridge, 1944), 130 and (for examples from specific writers) 149Google Scholar; Young, D., Greece & Rome, vi² (1959), 96 ff.Google Scholar; Dover, K. J., Lysias and the Corpus Lysiacum, Univ. of California Press, 1968, e.g. pp. 87–90Google Scholar; Tränkle, H., ‘Beobachtungen und Erwägungen zum Wandel der livianischen Sprache’, Wien. Stud. lxxxi (1968), 103 ff.Google Scholar (modifying the earlier view that Livy's style becomes less archaizing after the first decade).

^{page 351 note 1} See Sörbom, G., Variatio sermonic Tacitei aliaeque apud eundem quaestiones selectae (Uppsala, 1935).Google Scholar

^{page 351 note 2} The question whether Seneca has remembered the actual wording of the speakers he purports to quote, or whether he recorded the drift of their arguments in his own style, is irrelevant here.

^{page 351 note 3} Löstedt was aware of the need to contrast Tacitean usage with the educated language of the early Empire. He occasionally refers to Petronius as a representative of the latter : see Philologischer Kommentar zur Peregrinatio Aetheriae (Uppsala, 1911), 168 f., 256 f.Google Scholar

^{page 352 note 1} e.g. (a) adusque 14. 58. 4, alimonium = alimentum 11. 16. 3, audentia = audacia 15. 53. 2, citus = cito 12. 12. 3, dehonestamentunt = dedecus 12. 14. 3, 14. 21. 4, disserto = dissero 12. 11. 1, 13. 38. 3, dido 11. 1. 2 (see Lex. Tac. s.v. ‘fama’ for alternatives), diutinus = diuturnus 14. 18. 2, dignus= divido 11. 10. 2, funus = cadaver 13. 17. 3, 15. 71. I (see Norden, E., P. Vergilius Maro, Aeneis Buch VI², p. 178)Google Scholar, genero 15. 23. 1 (see E. Fraenkel, Horace, 293 n. 3), illo = illuc 15. 60. 4, intorqueo 14. 36. 3, inultus = impunitus 13. 25. 2, moenia = portae, urbs 14. 24. 4, obtego = protego 16. 5. 3, partio = partior 12. 30. 2, permeo = pervenio 14. 58. 4, popularis = populus 12. 16. 1, 14. 24. 4, praevalidus = validissimus 15. 20. I, queritor 16. 34. 2, regno trans. 13. 54. 1, retego = ‘reveal’ 15. 74. 1, rebello = descisco 12. 50. 2, innumerus 12. 29. 3, 12. 56. 3, 14. 53. 5, eo infiticts = infitior 15. 2. 3, mergo = mergo 15. 69. 2, subvecto = subveho 15. 43. 3; (b) eligo = deligo (see Wölfflin, , Philol. xxvi (1867), 108)Google Scholar, firmus = validus 3. 18. 5, 4. 62. 1, apparatus = paratus 2. 69. 2, capillus 4. 57. 2 (see Tränkle, H., Wien. Stud. lxxxi (1968), 123)Google Scholar, cadaver 1. 22. 2, 2. 18. 1 (see Norden 1.c.), consimilis 3. 13. 2, satietas 1. 49. 2, firmitas = firmitudo 4. 63. I, taceo = sileo 4. 60. 2, tacitus = silens 2. 43. 5, adpropinquo = propinquo 6. 50. 2.

Synonyms of or alternatives to all the above occur in the opposite half of the Annals.

^{page 352 note 2} See J. P. Krebs and Schmalz, J. H., Antibarbarus der lateinischen Sprache⁷ (Basel, 1905), S.V.Google Scholar

^{page 352 note 3} Philol. XXV (1867), 101.

^{page 352 note 4} The disproportionate frequency of senecta in the ablative, and its absence in the nominative, can be paralleled in other prose authors. Before Tacitus the word is used in prose by Nepos (Vir. Ill. 15, in Peter, H., Historicorum Romanorum Reliquiae, ii)Google Scholar, Varro (Ling. 5. 5), Livy (6 times), Valerius Maximus (4. 3. ext. 3, 8. 13. ext. ), and Pliny the Elder (43 times, often with a concrete meaning). The nominative occurs only 3 times (Liv. 2. 40. 6, Plin. Nat. 8. 116, 30. 69), but the ablative once in Nepos, once in Varro, 4 times in Livy, and 28 times in Pliny. Pliny has senectus 10 times in the nominative, but only 5 times in the ablative. Senecta had originally been an adjective. In Plautus it is still found 7 times in the fossilized ablative expression senecta aetate, as well as 3 times on its own (always in the phrase in senecta). Apparently when detached from aetate to form an independent substantive it continued to be regarded as more acceptable in the ablative than in other cases.

^{page 352 note 5} Op. cit., pp. 17 f.

^{page 353 note 1} The quality of earlier evidence has been decided according to the last three principles listed above. It will be seen that almost all examples of normalization presented by Eriksson have been considered unconvincing.

^{page 353 note 2} Op. cit.

^{page 353 note 3} ‘Quibus and quis in Tacitus’, C.R. N.S. xviii (1968), 144 ff.

^{page 353 note 4} On the identical treatment by Tacitus and Sallust of one type of hyperbaton see Adams, J. N., P.C.P.S. N.S. xvii (1971), 8 f.Google Scholar

^{page 353 note 5} ‘In componendis synonymis quae ratio adhibita sit in Taciti Germania et Agricola’, Studi in onore di Ugo Enrico Paoli (Firenze, 1955), 431.Google Scholar

^{page 353 note 6} Jax was careful to distinguish between examples found in narrative and those in speeches. Throughout most of the Annals the majority of instances are in narrative. In 16, however, 5 of the 6 are in speeches.

^{page 353 note 7} Über den Gebrauch der Anaphora bei Tacitus (Erlangen, 1918), 26, 33, 43.Google Scholar

^{page 354 note 1} It would be difficult to establish objectively that there is any distinction in this respect between the rhetorical prose of the Empire and that of the Republic. Norden, E., Die antike Kunstprosa⁴ (Leipzig-Berlin, 1923), 289Google Scholar accepted the impressionistic statement of Seneca and included tricola among the distinctive features of the ‘new style’.

^{page 354 note 2} Draeger, A., Über Syntax und Stil des Tacitus³ (Leipzig, 1882), 102.Google Scholar

^{page 354 note 3} Lindholm, E., Stilistische Studien zur Erweiterung der Satzglieder im Lateinischen (Lund, 1931), 189Google Scholar discusses only tricola (of an unspecified kind) with members of ascending length.

^{page 354 note 4} Threefold groups of substantives, qualified or unqualified, whether in asyndeton or linked by copulative particles : e.g. Ann. 1. 2. I ‘ubi militem donis, populum annona, cunctos dulcedine otii pellexit’ (in this case there are two series of three substantives ; I also include simple examples such as 1. 4. 1 ‘segue et domum et pacem sustentavit’).

^{page 354 note 5} Threefold groups of adjectives or of word groups with adjectival function : e.g. 1. 3. 1 ‘M. Agrippam, ignobilem loco, bonum militia et victoriae socium’.

^{page 354 note 6} e.g. 1. 4. 2 ‘pauci bona libertatis in cassum disserere, plures bellum pavescere, alii cupere.’ I have deliberately chosen three objectively-describable types of tricolon as test cases. The principles of colometry established by Fraenkel, E. (‘Kolon und Satz I’, Nachr. Gött. Ges. d. Wiss., 1932, 197 ff.Google Scholar ; ‘Kolon und Satz II’, Nachr. Gött. Ges. d. Wiss., 1933, 319 ff.Google Scholar ; ‘Noch einmal Kolon und Satz’, Sitzungsberichte der Bayerischen Akad. d. Wiss., Phil.-hist. Klasse 1965, Heft 2Google Scholar) admit of a degree of subjective interpretation and hence cannot serve as a basis for statistical analysis.

^{page 355 note 1} The examples of antithesis given at Rhet. Her. 4. 21 are of the first type.

^{page 355 note 2} The speeches in I to 3 contain about 21 examples, and those in 13 to 15 about 10.

^{page 355 note 3} The figures comprise examples of parallel finite clauses (e.g. 2. 19. I ‘pugnam volunt, arma rapiunt’) and of parallel phrases or word-groups within finite clauses (e.g. 1. 2. 1 ‘militem donis, populum annona... pellexit’). On parallel constructions in general see J. B. Hofmann and Szantyr, A., Lateinische Syntax and Stilistik (Munich, 1965), 726 ff.Google Scholar

^{page 355 note 4} See Adams, op. cit. 9.

^{page 355 note 5} For a full description of the various types see Kohl, A., Der Satznachtrag bei Tacitus (Würzburg, 1960).Google Scholar

^{page 356 note 1} The degree of pause after the verb varies from case to case.

^{page 356 note 2} See Adams, op. cit. 8 f.

^{page 356 note 3} The whole of books 3, 12, 13, 14, and 15; books i. 1-65, 2. 1-73, and 4. 1-64.

^{page 356 note 4} op. cit. (see above, p. 350 n. 5), p. 30.

^{page 356 note 5} Philol. xxiv (1866), 18. Quasi is the word preferred by the archaizers Fronto and Apuleius: see Hofmann and Szantyr, 596.

^{page 356 note 6} Philol. xxv (1867), 124.

^{page 356 note 7} Philol. xxvii (1868), 148.

^{page 356 note 8} On quo see Hofmann and Szantyr, 679 f.

^{page 356 note 9} A.F.P. xix (1898), 255.Google Scholar

^{page 357 note 1} Op. cit. 13.

^{page 357 note 2} The assertion of Krebs and Schmalz, s.v. ‘epistula’, that litterae could be used of both private and public correspondence, but epistula only of the former, is not borne out by the evidence, at least from the imperial period. The title of one of the imperial secretaries was ab epistulis.

^{page 357 note 3} In Plautus insons is almost as frequent as innocens (7 : 11). Sons is found only at Capt. 476. The monosyllable may have been displaced first. Once nocens established itself as the everyday word its negative form would have tended to oust insons.

^{page 358 note 1} For a full collection of examples see Gotzes, O., De quibusdam sermonis Gelliani pro prietatibus observations (Halis, 1883), 32 f. It may be an indication of the formality of the word that Gellius uses it 42 times in chapter headings.Google Scholar

^{page 358 note 2} amica: Bacch. 177, 367, 562, 607; anu: Cist. 660; ancilla:Cas. 254; vicina: Mil. 1212; filia: Aul. 683.

^{page 359 note 1} Like certain other archaisms (see Axelson, B., Unpoetische Wörter (Lund, 1945), 25 ff.Google Scholar), particularly those associated with the Legal and official languages, it is found only sporadically in the poets. All 3 instances in Horace (Carm. 3. 8. 17, 4. 2. 42, Carm. Saec. 18) are in passages with an official ring.

^{page 359 note 2} See Kalb, W., Roms Furisten, nach ihrer Sprache dargestellt (Leipzig, 1890), 105.Google Scholar

^{page 359 note 3} See Lex. Tac. 1503.

^{page 359 note 4} Simulque is a common Plinian usage which could have been substituted (Nat. 12.116, 15. 91, 15. 124).

^{page 360 note 1} Glotta, xv (1927), 285.

^{page 360 note 2} See Gries, K., Constancy in Livy's Latinity (New York, 1949), 21 ff.Google Scholar

^{page 360 note 3} See van de Heyde, K., ‘Flumen, fluvius, amnis’, Mnem. lx (1932), 535 ff.Google Scholar

^{page 360 note 4} For further examples and discussion, see my forthcoming article (Sect. IV) ‘Latin Words for “Woman” and ‘Wife” ’ in Glotta.

^{page 360 note 5} For a full collection of verbs of this type see Mignot, X., Les verbes dinominatifi latins (Paris, 1969), 352 ff.Google Scholar

^{page 361 note 1} See the comments by Mignot, loc. cit. on causificor, cerifico, fructifico, largifico, ludifico, mitifico, modifico, notifico, orbifico, pacifico, purlfico, turpificatus, and versifico.

^{page 361 note 2} Hofmann and Szantyr, 607 do not sufficiently emphasize the rarity of the word in this sense in ordinary prose.

^{page 361 note 3} The figures given by Löfstedt, , Roman Literary Portraits (Oxford, 1958), 160 are inaccurate.Google Scholar

^{page 361 note 4} See T.L.L. ii. 1568. 1 ff.

^{page 361 note 5} See Lex. Tac. 923b, 935b. Nec . . . aut is preferred in the Histories (20:5), neque . . . aut in the Annals (26: 10).

^{page 361 note 6} See Löfstedt, , Syntactica, i 332 ff.Google Scholar

^{page 361 note 7} See Wölfflin, , Philol. xxvi (1867), 161.Google Scholar However Mölfflin was wrong to say that the word is always so used. See Ann. 13. 30. 2, 14. 57. 3, 16. 14. 5.

^{page 361 note 8} See Löfstedt, Peregrinatio, 71 f. on a similar adjective (parvus) apparently avoided by Tacitus as hackneyed.

^{page 361 note 9} 2.56. 1, 3. 19. 2, 3.40. 2, 4. 6. 2, 4. 10. 1, 6. 39. 3. At 4. 10. 1 for the Med. plurimis maximeque fideis auctoribus Ritter suggested maximaequefidei. Maximeque fidis (Beroaldus) is less plausible, for Tacitus has fidissitnus 9 times, but maxime fidus nowhere else.

^{page 362 note 1} See Hofmann and Szantyr, 603 ff; Krebs and Schmalz, s.v.

^{page 362 note 2} See Hofmann and Szantyr, 602.

^{page 362 note 3} Similar idiolectal variations can be seen in the poets. In the first century the older word succeeded in establishing itself as the stock poeticism only in certain authors : Axelson, Unpoetische Wōrter, 124 n.

^{page 362 note 4} Philol. xxv (1867), 108.

^{page 362 note 5} In books 1-12 lego and deligo tend to occur in clusters, a phenomenon probably due to unconscious association. Deligo is found twice in successive chapters in two different places in Annals 1 (58, 59, 64, 65); 3 times in 8 chapters in 2 (40, 43, 47); twice in successive chapters (47, 48), 3 times in 4 chapters (72, 73, 75), and twice in 5 chapters (52, 56) in 4 ; 4 times in 2 chapters (2, 3) and 6 times in 9 chapters (41-9) in 6; and twice in the same chapter (66) in 12. Lego is found twice in 5 chapters (43, 47) in i ; twice in 5 chapters (48, 52) in 2; 3 times in 6 chapters (30, 32, 35) in 3 ; and twice in the same chapter (16) in 4. On word clusters in Livy, see Gries, K., C.Ph. xlvi (1951), 36 f.Google Scholar

^{page 362 note 6} It is a mistake to assume that compound verbs are uniformly of lower stylistic level than their corresponding simple verbs (if they are synonyms). The reverse is often true. Each pair must therefore be considered as an individual case.

^{page 363 note 1} There are a few places where exterreo has the meaning ‘frighten away’, but these are not numerous enough to alter the general picture given by the statistics. See T.L.L. V. 2. 2025. 65 ff.

^{page 363 note 2} For archaisms or poetic words found at the beginning of the work, then dropped until the later books, see Trankle, H., Wien. Stud. lxxxi (1968), 118 ff.Google Scholar

^{page 363 note 3} Cicero (3 times), Caesar (5 times), Seneca the Elder (once), Curtius (once), Columella (once), and Asconius (once) have only passive forms. Of the poets, Lucretius and Virgil allow only the passive, but Horace (once), Ovid (3), Val. Flacc. (2), Silius (7), and Statius (once) use the active as well.

^{page 363 note 4} Only literal examples are considered here. Struo has a metaphorical use not shared by exstruo.

^{page 363 note 5} For inscriptional examples of exstruo from the late first century A.D. see Epistula Vespasiani ad Saborenses, A.D. 78 (see Bruns, C. G., Fontes Iuris Romani Antiqui⁷ [Tübingen, 1909], 255)Google Scholar; Lex arae urbanae, A.D. 84-96 (Bruns., op. cit. 288).

^{page 364 note 1} 4. 13. 16 sacrificio perpetrato (cf. Liv. 25. 16. 2, 44. 22. 16, 44. 37. 13); 7. 2. 29 perpetrata caedes (cf. Liv. s. 6. I, 45. 5. 5, 45. 5. II).

^{page 364 note 2} See Fraenkel, E., F.R.S. xli (1951), 193Google Scholar; Tränkle, , Wien. Stud. lxxxi (1968), 126.Google Scholar

^{page 364 note 3} Some of the above compounds may at times have nuances of meaning which distinguish them from their simple verbs, but the general picture is clear.

^{page 365 note 1} Noted by Gerber and Greef, Lex. Tac. 911 b; cf. the more detailed statistics given by Goodyear, op. cit. 31.

^{page 365 note 2} See Goodyear, op. cit. 30.

^{page 365 note 3} See Tingdal, G. C., Ändelsen -is i Ackus. Plur. hos de Efteraugusteiska Författarne (Göteborg, 1916), 70 ff.Google Scholar

^{page 365 note 4} See Nutting, H. C., ‘The Use of forem in Tacitus’, Univ. of Cal. Publ. in Class. Philol. vii. 209 ff.Google Scholar

^{page 365 note 5} See Löfstedt, , Syntactica, ii. 285Google Scholar; Martin, R. H., C.R. N.s. xviii (1968), 144 ff.Google Scholar

^{page 365 note 6} See Eriksson's index for these words.

^{page 365 note 7} See Syme, R., Tacitus (Oxford, 1958), 739 n. 7Google Scholar; cf. Burnet, I. G., ‘Linguistic “Normality” in Tacitus : promitto and polliceor’, B.I.C.S. xvi (1969), 63 ff.Google Scholar

^{page 365 note 8} See Brakmann, C., Mnem. liii (1925), 179f., 183f.Google Scholar

^{page 365 note 9} See Wöfflin, , Philol. xxv (1867), 124.Google Scholar

^{page 365 note 10} See Eriksson's index.

^{page 365 note 11} Cf. Wöfflin, 's remarks on dein and deinde (Philol. xxv [1867], 105 f.).Google Scholar

^{page 365 note 12} Ann. 1. 23. 5, 1. 36. 2, 4. 25. 3, 4. 51. 1, 6. I. 2. For adversus cf. 13. 10. 1.

^{page 366 note 1} See T.L.L. i. 85o. 63 ff.

^{page 366 note 2} See Löfstedt, , Syntactica, ii. 291.Google Scholar

^{page 366 note 3} Op. cit. 89. The usage is found frequently in Pliny the Elder (Thesaurus material) and Tacitus' Annals, but elsewhere rarely. On Tacitus, see Wöfflin, , Philol. xxv (1867), 120.Google Scholar

^{page 366 note 4} Op. cit. 16.

^{page 366 note 5} e.g. omnis (6 times Annals, 7 times Histories), tris (3 times Annals, twice Histories), pluris (4 times Annals), partis (3 times Annals), resistentis (twice Annals), insignis (twice Annals), ingruentis (twice Annals), hostilis (twice Annals), hostis (3 times Histories), pontificis (twice Annals). See Tingdal, loc. cit.

^{page 366 note 6} Euphony is one factor which might determine a writer's choice. At 13. 21. 1 ff. Aulus Gellius quotes and discusses a remark by Valerius Probus that in deciding between -es and -is a writer should be guided by his ear.

^{page 367 note 1} See R. Kühner and Stegmann, C., Ausfiirliche Grammatik der lateinischen Sprache: Satzlehre⁴, ii (revised by A. Thierfelder) [Darmstadt, 1962], 33Google Scholar; on Tacitus see Kienzle, L., Die Kopulativpartikeln ‘et, que, atque’ bei Tac., Plin., Sen. (Tübingen, 1906), 17 ff., 21 ff. (on atque).Google Scholar

^{page 367 note 2} See Kühner and Stegmann, loc. cit.; cf. Merten, G., De particularum copulativarum apud veteres Romanorum scriptores usu (Marburg, 1893), 2.Google Scholar

^{page 367 note 3} See T.L.L. ii. 1050, 10 ff.; Löfstedt, Peregrinatio, 85 ff.

^{page 367 note 4} In early Latin it may have been in use. It is used 3 times by Plautus, against 2 instances of ignarus.

^{page 367 note 5} Rhet. Her. 1. 10, 3. 27, Cic. Sull. 28, Fin. 1. 1, 5. 51, de Orat. 1. 45, Fam. 5. 12. 2, 13. 7. 1, Att. 16. 15. 2. There are occasional variations of number and gender. Note also Deiot. 8 non erant nescii, which is of the same structure but is in the third person.

^{page 367 note 6} Font. 2, Att. 15. 11. 4.

^{page 368 note 1} 27. 7. 5 (haud nescius), 43. 13. 1 (non sum nescius).

^{page 368 note 2} On the analogists see Colson, F. H., ‘The Analogist and Anomalist Controversy’, C.Q. xiii (1919), 24 ff.Google Scholar

^{page 368 note 3} On archaic forms as a characteristic of the style of Sallust rather than Livy, see Tränkle, , Wien. Stud. lxxxi (1968), 133.Google Scholar On Virgil's sparing use of archaic forms see Leumann, M., Mus. Helv. iv (1947), 126 f.Google Scholar Sallust was probably following the annalistic tradition in using unusual forms (see, e.g., the fragments of Cn. Gellius).

^{page 369 note 1} On which see Krebs and Schmalz, s.v. ‘tueor’.

^{page 369 note 2} Counting only forms in which both words are used : tutaba(n)tur, tutabor, tutandus, tutans, tuta(n)tur, tutare(n)tur, tutari, tutatus, and the corresponding forms of tueor.

^{page 369 note 3} By Wolff, J., Fahr. xxvi (1901), 637 ff.Google Scholar

^{page 369 note 4} On Curtius, Seneca the Younger (who almost without exception has the word in the clausula), and Pliny the Younger see Axelson, Unpoetische Wörter, 83 n. On the other writers I have consulted the Thesaurus material.

^{page 369 note 5} See Kroll, W., Glotta, xxii (1934), 7 n. 6Google Scholar; Fraenkel, , Eranos, xlix (1951), 56.Google Scholar

^{page 370 note 1} A full collection of examples is to be found in Kunze, A., Sallustiana (Leipzig, 1892), 4 ff.Google Scholar

^{page 370 note 2} See Lex. Tac. 106a; Wölfflin, , Philol. xxv (1867), 134.Google Scholar

^{page 370 note 3} atque periculum (Sall. Cat. 4. I, 4. 4, 30. 5, 52. 2, Tac. Ann. 2. 40. 2); aequabilius atque constantius (Sall. Cat. 2. 3, Tac. Ann. 15. 21. 4).

^{page 370 note 4} See Novak, R., Wien. Stud. xv (1893), 257 ff.Google Scholar

^{page 370 note 5} See Kühner and Stegmann, i. 700 ff.; Löfstedt, , Syntactica, ii. 262 f.Google Scholar; Hofmann and Szantyr, 362; Fritsch, S., Über den Sprachgebrauch des Velleius Paterculus (Arnstadt, 1876), 15.Google Scholar

^{page 370 note 6} See, e.g., Kühnast, L., Die Hauptpunkte der livianischen Syntax ² (Berlin, 1872), 108Google Scholar; Lindgren, H., Studia Curtiana (Uppsala, 1935), 56.Google Scholar See also the examples quoted from Cicero and Caesar by Lebreton, J., Études sur la langue et la grammaire de Cicéron (Paris, 1901), 377 f.Google Scholar

^{page 371 note 1} See Löfstedt, Syntactica, ii. II ff.; Hofmann and Szantyr, 343.

^{page 371 note 2} The following are the examples included in the table: (a) se omitted with future active infinitive: Hist. 1. 50. 3, 2. 9. 2, 2. 32. 2, 3. 13. 3, 3. 20.2 (twice), 3. 77. 1,4. 14. 4, 4. 56. 3, Ann. 1. 7. 4, 1. 19. 4, 1. 25. 3, 1. 35. 4, 2. 71. 5, 2. 79. 1, 2. 83. 3, 3. 14. 4, 3. 47. 2, 3. 58. 2, 3. 72. 2, 4. 40. 3, 5. 9. 1, 6. 48. 2, 11. 25. 3, 11.30.2, 12. 7. 1, 12. 34, 12. 45. 1, 12. 47. 1, 13. 14. 3, 13. 38. 3, 13. 42. 4, 13. 56. 1, 14. I. 2, 14. 29. 1, 15. 5. 4, 15. 29. 1, 15. 42. 2, 15. 43. 2, 15. 52. t, 16. 24. 1, 16. 28. 2; (b) se omitted with other tenses of the infinitive: Hist. 1. 51. 4, 2. 18. 2, 4. 25. 1, Ann. 1. 8. 4,2. 58. 1, 2. 63. 1, 3. 8. 2, 4. 8. 3, 4. 39. 2, 4. 39. 4, 11. 3. 2, 11. 17. 1, 12. 47. 3, 13. 38. 1, 13. 46. 2, 13. 49. 4, 14. 35. 1, 14. 48. 2, 15. 17. 2, 15. 27. 2, 15. 36. 3, 15. 51. 2, 15.55. 1, 15. 62. 1, 16. 25. 1.

^{page 371 note 3} Laughton, E., The Participle in Cicero (Oxford, 1964), 59 ff.Google Scholar

^{page 371 note 4} 1. 8. 6, 1. 16. 1, 1. 19. 5, 1. 24. 1, 33. 3, 1. 36. 2, 1. 42. 1, 1. 52. 1, 1. 59. t, 63, 2, 1. 72. 4, 2. 2. 3, 2. 47. 2, 2. 57. 1, 2. 82. 3, 2. 84.2, 3.5.1,3.9.3, 3.22.1,3.24.1,3.28.4, 3. 31. 2, 3. 46. 4, 3. 50. 3, 3. 52. 1, 3. 52. 3, 3. 54. 2, 3. 54. 5, 3. 71. 2, 13. 9. 3, 14. 14.3.

^{page 371 note 5} See Hofmann and Szantyr, 345 f. for a brief discussion.

^{page 372 note 1} admoneo, agito, certo, cohortor, do, edico, hortor, impello, impetro, induco, mando, mereo, moneo, oro, pango, perpello, postulo, suadeo, subigo, urgeo. I omit permitto as a special case. The infinitive had effectively ousted ut in all genres. Tacitus uses ut with the word only at Ann. 3. 63. 1, where the infinitive would have been awkward.

^{page 372 note 2} See the Lex. Tac. under the words mentioned above and also under accingo, adnitor, aemulor, ambio, conitor, dehortor, denuntio, deposco, exposco, impero, incumbo, inlicio, insto, nitor, nuntio, praeopto, praescribo, scribe.

^{page 372 note 3} On these two words see Krebs and Schmalz, s.v.

^{page 372 note 4} Philol. xxv (1867), 114. Eriksson (op. cit. 25) disregarded Wölfflin's remarks in his discussion of the frequency of the two constructions. Since he merely tabulated all occurrences without reference to their context his statistics are of little value.

^{page 372 note 5} The instances included in the table are : 1. 2. 1, 1. 5. 1, 1. 7. 7, 1. 23. 1, 1. 41. 2, 1. 64. 2, 1. 71. 1, 1. 71. 2, 1. 76.4 (tw1ce), 6. 3. 2, 6. 46. 5, 6. 48. 2, 6. 50. 4, 11. 9. 1, 12. 8. 1, 12. 10. 1, 13. 8.3, 14 5. 1, 14. 23. 1, 14. 36. 3, 14. 38. 1, 14. 39. 1, 15.4. 3, 15. 53. 3, 15. 56. 2, 15. 66. 2, 16. 14. 1 16.19.3,16.22.1,16.24. I (twice), 16.32. 3 (twice). In the following passages it has been possible to see a reason for the avoidance of the dative+gerundive: 2.6. 3 (variation), 2. 43. 4 (Wölfflin, loc. cit.), 2. 79. 1 (ad dicendam causam; this and similar phrases are always used by Tacitus instead of the dative with the gerundive), 3. 3o. 2 (Wölfflin, loc. cit.), 3. 38. 2 (ad dicendam...causam), 3. 39. 1 (variation), 3. 49. 2 (ad dicendum testimonium), 3. 56. 4 (Wölfflin, loc. cit.), 3. 63. 4 (euphony), 4. 16. 4 (Wölfilin, loc. cit.), 4. 41. 1 (used with impello, which always takes ad rather than the dative in Tacitus), 4.48.2 (Wölfflin, loc. cit.), 4. 73. 1 (clarity), 5. it. 1 (variation), 11. 37. 2 (dicendam ad causam), 12. 14. 1 (exuendam ad fidem; Tacitus avoids fides in the dative with a gerundive, but uses ad fidem with a gerundive a number of times), 13. 5. 1 (ad causam orandam), 13. 31. 2 (ad retinendam . . . fidem), 13. 38. 3 (variation), 13. 39. 1 (clarity), 14. 26. 1 (Wölffiin, loc. cit.), 14. 51. 1 (clarity), 15. 58. 3 (dicendam ad causam).

^{page 373 note 1} The figures are only approximate, for Tacitus' intentions may sometimes escape a modern reader. Nevertheless it is reasonably certain that the classical construction is used more freely in the second half of the Annals (with the change coming at 6).

^{page 373 note 2} On the Histories see Wölfflin, loc. cit.

^{page 373 note 3} Quibus condicionibus Tacitus ellipsim verbi admiserit et qua ration excoluerit (Würzburg, 1900).