¹ The discussions are many and they are in general agreement. For this paper the primary secondary sources have been:Townend, , ‘The Poems‘, in Dorey, T. A. (ed.), Cicero (New York, 1965), 124–5;Google ScholarEwbank, W. W., The Poems of Cicero (New York and London, 1978)Google Scholar;Skutsch, O., The Annals of Q. Ennius (Oxford, 1985)Google Scholar;Bailey, C., Titi Lucreti Cari De Rerum Natura Libri Sex (Oxford, 1947)Google Scholar;Duckworth, G. E., Vergil and Classical Hexameter Poetry(Ann Arbor, 1969)CrossRefGoogle Scholar;Courtney, E.The Fragmentary Latin Poets (Oxford, 1993)Google Scholar;Buchner, K., RE VII. A. 1 12656, s.v. ‘Tullius’Google Scholar;Gratwick, A. S., ‘Ennius′ Annales’, in Cambridge History of Classical Literature (Cambridge, 1982), vol.2, pt. 1, 66–75Google Scholar; and Goldberg, S. M., Epic in Republican Rome (Oxford, 1995). These works are cited by author only in the notes that follow. The specific features of prosody and diction relevant to Furius are discussed below.Google Scholar

² Büchner gives a detailed analysis of Cicero′s advances in ten categories and concludes, ‘Nimmt man alles zusammen so wird C. zu einer wichtigen Brücke zu den Augusteern’(1266).Traglia, , La Lingua di Cicerone Poeta (Bari, 1950), 159–233Google Scholarmakes even more vigorous claims for Cicero's importance. Townend, 124 says,‘Cicer’s importance as a poet, then, is to be sought, if anywhere, in the part he played in the development of Latin poetical technique. He goes on to specify Cicero′s advances over Ennius as the increased proportion of dactyls, the regularization of the third foot caesura, avoidance of complete coincidence of ictus and accent in the first four feet of the hexameter with regular coincidence in the cadence, and refinement in elision. An appreciative view of Cicero's technical importance may be found in Peck, T., ‘Cicero's Hexameters’, TAPA 28 (1897), 60–74;Google Scholar and a less enthusiastic verdict on him as a ‘versifier’ in Merrill, W. W., ‘The Metrical Technique of Lucretius and Cicero’, Cat. Publ. in Cl. Phil. 7.10 (1924), 30–56.Google Scholar Duckworth cites with approval Buchner and Townend, and considers Cicero ‘a kind of milestone’, 43. Courtney, 149–52, avoids any general qualitative judgement of Cicero as a poet and tries to balance an appreciation of Cicero′s place in the development of the hexameter with attention to his technical defects in suppleness and variety. He credits Cicero with having ‘moved forward from the archaic mould still retained by Lucretius towards the Vergjlian norm of regularizing caesura and line end’ (150), and with introducing two adjective, two noun lines which created the possibility of neoteric wordpatterning. Already in 1921, however, Merrill, W. A. in ‘Lucretius and Cicero′s Verse’, University of California Publications in Classical Philology 5 (1921), 144Google Scholar had cautioned,‘It is not probable that the youthful Cicero all alone modified the heavy hexameter of Ennius and Lucilius; it is a more rational view to suppose that the teaching grammarians in the schools brought about a gradual change, and that a slow improvement in technique and vocabulary would be apparent to us if the lost work of the period after Ennius had survive’. For Cicero′s contributions to epic diction, see Allen, W., ‘O fortunatam natam…’, TAPA 87 (1956), 141–143;Google ScholarWigodsky, M., Vergil and Early Latin Poetry, Hermes Einzelschriften 24 (Wiesbaden, 1972), 109–114;Google Scholar and Goldberg, 135–157Google Scholar

³ The date of Cicero′s Aratea is uncertain. It appears to be an early work and Cicero himself says he was admodum adulescentulus when he produced his translation (de nat. de. 2.41.104). In part because of some errors in understanding the Greek, it is generally placed before Cicero′s trip to Greece (797 B.C.), around 86 B.C. See the discussion by Ewbank 224. Cicero′s other poems appear to be much later:de Cons., 60B.C.;Marius prob. after the de Cons,; de temp, meis c. 55 B.C.; and perhaps a revised edition of the Prognostica in 60 B.C.

⁴ While I do believe that the Vergilian hexameter in its full range of contexts and effects displays a flexibility and power not found before or after, I am not arguing that Furius is creating or progressing to a fixed entity called the‘ Vergilian hexamete’. The point is rather that the effects he does achieve are part of the Vergilian repertoire, which is to say that he helped create the range of norms and variations which allowed Vergil to create the effects that suited him

⁵ This is not to say that Furius was a great poet in theme, scope or imagination, only that his echnical accomplishments did not go unnoticed by and were not unuseful to Vergil and Ovid.

⁶ See Holford-Strevens, L., Aulus Gellius (London, 1988), 160.Google Scholar

⁷ See, for example, fr. 8 (Courtney): ′ille gravi subito devictus volnere habenas | misit equi I lapsusque in humum defluxit et armis | reddidit aeratis sonatu‘ (Macrob. Sat. 6.4.10). Of the I, fourteen lines and partial lines cited by Macrobius eight are enjambed. Skutsch, RE VII 322, I seems to be the first to have distinguished Furius Antias and Macrobiu’ Furius on grounds that I Furius Antias shows no Ennian imitation and no enjambment; see also HolfordStrevens j (n. 6), 160 n. 89.Bardon, H., La Littérature latine inconnue (Paris, 1952), 179–81, goes furthest ’ in identifying the features that will concern this essay: ‘art maniéré, mais rafflné, des neoteroil'allitération virescit vulnere virtus combine avec ces recherches quelque raideur héritée des anciens epiques (180).Google Scholar

⁸ I follow the reasoning of Courtney 193 and 198–199.

⁹ Based on a search of the PHI CD ROM 5.3 by Musaios 1.0c, we can say that Gellius uses the term vetus for his Roman ‘ ancestors’ and for Pindar and Ennius, but for no poet later than those in the list of erotic poets at Att. Noct. 19.10: versus cecinit Valeri Aeditui, veteris poetae, item Porcii Licini et Q. Catuli. Unfortunately, we know nothing about Valerius Aedituus beyond this reference and the two epigrams totalling ten lines which Gellius cites. It may be significant that of the three poets listed he alone get the designation ‘vetus poeta’. Gellius′ usage with respect to poets would seem, then, to place Furius Antias in the generation of Catulus, or before, which makes him likely to be the A. Furius to whom Catulus dedicated the book on his consulship. However, in referring to prose writers Gellius places both Cicero (12.13.17 and 13.17.2) and Varro (2.20 pr.; 7.5.10; and 13.17.2) among the veteres (cp. 3.16.19 and 18.7.8). i This means that Furius Antias, if he is not A. Furius (fl. c. 100), could be a poet roughly of Cicero′s generation. In any event, nothing suggests that poets of the generation of Catullus and Lucretius, the generation to which M. Furius Bibaculus belonged, would have been called vetus f by Gellius. On Gellius′ typical faults, see Holford-Strevens, L., Aulus Gellius (London, 1988), 47–58 and 227–235: they include a certain licence in his mise en scène, which may be more verisimilar than historically accurate, and some peculiar errors in names, usually of praenomina.No grounds have ever been given to dispute his identification of Furius Antias.Google Scholar

¹⁰ Neither the identification of Macrobius′ Furius with Furius Bibaculus nor the distinction between Furius Bibaculus and Furius Antias is, of course, universally accepted, and the discussion is cluttered with doubts about the other Furii.Perret, J., Horace (Paris, 1959), 57–58Google Scholar, identified Macrobius′ Furius with Furius Antias as part of a strategy to claim that Horace admired the neoterics: the epic Furius is not the neoteric Furius and, therefore, the Furius Horace disdains is not the neoteric Furius Bibaculus. His reviewer, Bruere, R. T., CPh 56 (1961),Google Scholar122 agreed on the grounds that there was ’nothing neoteric about’ the lines in Macrobius (cp. Bardon (n. 7), above).Lucas, H., on the other hand, in ‘Die Annalen des Furius Antias’, Philologus 92 (1937–1938), 344–8, joins Macrobius′ Furius with Furius Antias and Furius Bibaculus, claiming but not demonstrating ‘die natürliche Auffassung’ between the verses, but still distinguishing them from the object of Horace′s disdain: his reason is that Horace would not condemn a poet Vergil imitated. The essay offers a handy review of the debate, taking it back to Emil Baehrens. See most recently the citations and argument in Courtney, 195200. The evidence and further bibliography are reviewed with some skepticism that Ps.Acro′s identification is correct by Wigodsky (n. 2), 148–150.CrossRefGoogle Scholar

¹¹ The epigrams of Catulus are discussed by Ross, D. O., Style and Tradition in Catullus (Cambridge, MA, 1969), 144–7, who claims that they show little evidence of‘neoteric’ attention to metrical detail. They receive a more appreciative review by Conte (n. 40), 138–9, who emphasizes Catulus′ prominence in the cultural landscape. Conte, 110, accepts the identification of A. Furius and Furius Antias, as do all other scholars. The reference to A. Furius in Cicero is:‘… ex eo libro, quern de consulatu et de rebus gestis suis conscriptum molli et Xenophonteo genere sermonis misit ad A. Furium poetam familiarem suum…’, Brutus 132.CrossRefGoogle Scholar

¹² On the placement of this caesura, see the discussion of the fragment below.

¹³ Probabilities for binomial distributions can be calculated according to the formula discussed and given in Mosteller, F. and Rourke, R. E. K., Sturdy Statistics: Nonparametrics and Order Statistics (Reading, MA, 1973)24–5, or it may be roughly determined by reference to the i chart on p. 319.1 have used the rough determination because there is no need in this argument for fastidious accuracy.Google Scholar

¹⁴ Denned as endings which are neither disyllablic word preceded by a dactylic word or word-ending nor trisyllablic word preceded by a trochaic word or wordending. Percentage from Skutsch, 49. My sources, cited by author only, are listed in note 1.I have not checked the figures myself nor have I sought to reconcile the small differences that sometimes appear between the statistics of different researchers. The reason is simply that agreement is close enough for the general picture I am trying to draw.

¹⁵ Skutsch, 48.

¹⁶ Ewbank, 42. These three criteria are selected because it is a necessary characteristic of a binomial experiment that the outcomes of each trial be independent of one another. One could not, for instance, maintain that observing or breaching Hermann′s bridge was entirely independent of classical line ending.

¹⁷ Ewbank 60–64, 45, and 42.

¹⁸ The odds of finding a dactyl exactly twice out of six is.138, but this is not a useful figure because I am not trying to measure the probability of finding these characteristics in exactly Furius′ proportion, but the odds of finding Ennian lines which display a similar attention to what became the classical norm; in this instance, a lighter opening foot

¹⁹ It is the difference, not the probability figure, which is significant. Theoretically, by specifying enough variables one could arrive at the statistical probability that only the lines under consideration have all the features specified. In that event, however, it would be equally unlikely to find the lines in Ennius or in Vergil

²⁰ Since Ennius allows a three-weak only 10.6% of the time, we would expect that random samples of six lines from Ennius would lack 3-weak almost half the time. The statistics do not speak to particulars, but to general trends and to the unlikelihood of finding such a collection of characteristics before the experiments of the last century of the Republic.

²¹ See Ross, op. cit. 132–137 for the neoteric emphasis on Sperrung (that is, the adjective at major caesura and the noun at line end (or in penultimate position)) and the typical neoteric word-positions. The word-order patterns ‘most indicative of neoteric innovation’ (132), also phrased ‘significant word order’ (133), are:…A/…S, …/A…S, …A/…S…, and A…/…S. See also the remarks of T. E. V. Pearce, ‘The enclosing word order in the Latin Hexameter I’, CQ 16 (1966), 140–171, at 165–166 and ‘The enclosing word order in the Latin Hexameter II’, CQ 16 (1966), 298–320, at 299.

²² In both the number of nounadjective groups and their division Furius conforms to the Vergilian model: in Ennius 331 groups in 623 lines, separated eightyeight times; Vergil has 407 groups in 500 random lines separated 207 times (Skutsch, 67); Furius has 5 groups in 6 lines separated three times. By my own quick count I find 246 separations in the 572 complete lines of Aratea (including fragments) printed in Ewbank. This feature is thought to be an identifiably Ciceronian feature (see Pearce, CQ 16 [1966], 298 and Courtney, 151).Patzer, H.,‘ZumSprachstil des neoterischen Hexameters’, MH 12 (1955), 77–95 shows that one significant difference between Ennius and Catullus is the larger number of adjectives in Catullus.Google Scholar

²³ It would be a mistake to try to work these percentages into the probability statistics above. One would run into conflicts with the second requirement of binomial distributions: the outcomes of the trials must be independent of one another.

²⁴ It would be a mistake to try to work these percentages into the probability statistics above. One would run into conflicts with the second requirement of binomial distributions: the outcomes of the trials must be independent of one another

²⁵ It is not relevant that they are cited for their neologisms; cp. Ennius′ altivolantum (Ann. 76, Skutsch), a neologism of an identifiably archaic stripe (Lucretius adopted it: D.R.N. 5.433).

²⁶ Ennius suppresses final -s once in every five lines: Skutsch, 56

²⁷ See Gratwick, , 68–69, for a brief summary of Ennian characteristics. Skutsch, of course, goes into much greater detailGoogle Scholar

²⁸ Skutsch, 46 and 52.

²⁹ For Matius, Cn., see Courtney, 99–106. Examples of nonFurian hexameters include fr. 1: ‘corpora Graiorum maerebat mandier igni’ (homodyne fourth, archaic infinitive, heavy alliteration), fr. 2: ‘obsceni interpres funestique ominis auctor’ (spondees, no caesura after 3 strong, elision); and fr. 3:‘ dum dat vincendi praepes Victoria palmam’ (spondees; etymologizing and heavy alliteration).Google Scholar

³⁰ ‘After the death of Lucilius the satirist in 103 B.C. there seems to have been a virtual vacuum in Roman poetry; although Accius, born in 170 B.C., survived long enough to converse with the young Cicero in the early eighties. We have a few odd lines of the Homeric translations of Matius and Ninnius, and of the short poems of Laevius—none, in any case, capable of impressing following generations with their importance’: Townend, 110. See the brief discussion of Bardon, H., La litterature latine inconnue (Paris, 1952), 178–81Google Scholar and Wigodsky, M., Vergil and Early Latin Poetry, Hermes Einzelschriften 24 (Wiesbaden, 1972), 98–101Google Scholar. Townend, 124, is wisely cautious of precise evaluation of Cicero's place: ‘An estimate of his achievement in this field is made the more difficult by our relative ignorance of what was written by other poets before Cicero’. More extreme is the judgment of Conte, G. B., Latin Literature: A History (Baltimore and London, 1994), J. B. Solodow (tr.), 110:‘ In the period that extends from the age of the Scipios to the age of Caesar, epic writing appears, on the basis of the very few fragments left, to be completely dominated by the example of Ennius’. He further says of the new schools of poetry that they ‘regard this genre [epic] as a static survival, dusty and empty’. This essay challenges both generalities.Google Scholar

³¹ I use Vergil as my main example of this standard diction because he offers a sufficient number of epic verses and sufficient variety of subject matter to test Furius′ diction. As it turns out, further citations from other authors would only be redundant: Furius′ diction is not just classical but Vergilian.

³² Further checks of Catullus, Tibullus, Propertius and Lucan failed to produce any convincing parallels.

³³ We may note that in comparison with Lucretius, Cicero shows an increase in the number of lines with only 3-strong and 4-strong caesura; Furius in these fragments has the same high percentage as Cicero (fragments 1, 2, 5, 6). The figures come from Merrill, W. A., UCPCPh 7 (1924), 293–306. He notes that Lucretius and Cicero both have 2-strong, 3-strong and 4-strong about onethird of the time, but that Cicero has 3-strong and 4-strong over 50 % of the time while Lucretius has those caesurae only about 12%. He has excluded as a separate category the sequence 2-weak, 3-strong, 4-strong, which is found in Furius fragment 4Google Scholar

³⁴ This occurs in 67 % of Furius′ lines. According to Bailey, 110, the self-contained dactylic first foot increases from 14.5% in Ennius to over 30% in Lucretius and Vergil

³⁵ Other instances from Vergil are too numerous to need citation

³⁶ Courtney, 150. See below on fr. 3 for an example of Cicero's first conformation. There is no representative of the third conformation.

³⁷ In fact, the word is rare in all the Augustan poets, being found only at Tibullus 1.1.40; n 1.8.52; and Ovid, Fasti 3.760.

³⁸ Similar postponements of ater to emphasize the dark result of the action can be found in Vergil; see Aen. 6.272, ‘nox abstulit atra colorem’, see also 10.664.

³⁹ Cp. ‘cedant arma togae, concedant laurea laudi’, Cicero fr. 12 Courtney = 16 M = 11 B, and ‘praevius Aurorae. solis noctisque satelles’, Cicero fr. 1.1.

⁴⁰ Cp. fr. 6 where the alliteration of‘p’ is fairly unobtrusive and fr. 1 above for richer effects. For Cicero's use of alliteration, see Bailey, 146–50.

⁴¹ Ewbank, 45, for Cicero and Catullus; Bailey, 110, for Ennius, Cicero, Lucretius and Vergil; Skutsch, 48, offers similar figures.

⁴² If this is correct, then it is inaccurate to consider this fragment an example of the kind of first foot spondee avoided by later poets. It is being used for a precise effect: to set off the otherwise completely dactylic hexameter

⁴³ See the record of ancient etymology in Maltby, Robert, A Lexicon of Ancient Etymologies (Leeds, 1991), 36, s.v.Google Scholar

⁴⁴ The principle here is the corollary to that by which atrae was postponed above.

⁴⁵ The figures are cited above, n. 21; the percentages are as follows: separation in Ennius occurs in 14% of his lines and accounts for 36% of the noun-adjective groups; in Vergil 41 % of the lines and 51 % of the groups; for Furius 50% of the lines and 60% of the groups.

⁴⁶ Fragment 4 is instructive. If Vergil had Furius in mind, he dropped the comparison (sicut fulca—perhaps as an unhappy or colloquial syncope) and added two adjectives with the result that he had a line of the pattern ab V BA in which the position of the verb was taken by levis volat: ‘caeruleo (a) per summa (b) levis volat (V) aequora (B) curru (A)’. Equally instructive is a line from Cicero: ‘caprigeni pecoris custos de gurgite vasto’ (Prognostka, fr. 9, Ewbank). Had Cicero sought Sperrung, he could easily have written ‘caprigeni custos pecoris de gurgite vasto’ or even ‘caprigeni vasto pecoris de gurgite custos’. This seems to confirm Courtney's observation, 151, that Cicero's ‘plethora of adjectives’ created the opportunity for neoteric Sperrung, not the technique. The basic study of word order remains Conrad, C., ‘Traditional patterns of word order in Latin epic from Ennius to Vergil’, HSCP 69 (1965), 195–258.Google Scholar

⁴⁷ See above, note 14. See Conrad 208–9 for the frequency of …A/…S in Ennius (n. 56), Cicero, Lucretius III, Catullus 64 and Vergil, Aeneid 8. Ennius has the lowest frequency; there is an increase in Lucretius, and a significant increase over Lucretius in Cicero, Catullus, and Vergil. Conrad offers no figures for …/A… S

⁴⁸ On Cicero's general significance, see above n. 2.

⁴⁹ See Phil. 2.20: ‘Nee vero tibi de versibus plura respondebo: tantum dicam breviter, te neque illos neque ullas omnino litteras nosse; me nee rei publicae nee amicis umquam defuisse, et tamen omni genere monumentorum meorum perfecisse operis subsicivis ut meae vigiliae meaeque litterae et iuventuti utilitatis et nomini Romano laudis aliquid adferrent’. See also the the defensive responses to his critics at In Pis. 74 and De Off. 1.77. It is arguable that in his ‘nostra 2 quaedam Aratea’ he intended to win new poetic territory for Latin; see Div. 2.14 and Townend, 117. See also the discussion of Courtney, 149.

⁵⁰ Especially the Invectiva in Ciceronem, but also Seneca, Controv. 3. praef.; Martial 2.89.3; Tacitus, Dialogus 21.6;Quintilian 9.4.41 and 11.1.24; and Juvenal 10.122.

⁵¹ See Att. 23A and De Div. 1.17

⁵² For the social, rather than aesthetic, context of criticism against Cicero's verse, see the argument of Goldberg, 148–154.

⁵³ The evidence is cited and discussed in Courtney, 150–152, and Townend, 124–128.

⁵⁴ I refer to the statistical norm and historical tendency, not to an absolute prohibition or avoidance. The spondaic fifth is, in fact, used by Vergil more often than by Cicero, because the norm against which Vergil writes, having been established by poets like Cicero, allowed Vergil to use the fifth foot spondee for expressive purposes. For Vergilian exceptions to the classical norms, see Wilkinson, L. P., ‘The Augustan rules for dactylic verse’, CQ 34 (1940), 30–43, at 33 n. 4.CrossRefGoogle Scholar

⁵⁵ Percentages of trochaic fourth foot caesura for Ennius, Cicero, Lucretius, and Vergil can be found in Bailey, 112. For avoidance of this caesura, see Courtney, 151.

⁵⁶ As always, figures are a problem. While Townend, 127, notes the absence of post-Aratea spondaic fifths and Courtney, 151, notes the absence of both the spondaic fifth and the trochaic fourth, one must note that there is, in fact, only one spondaic hexameter in the Aratea and only two trochaic fourth caesuras.

⁵⁷ In this practice he corresponds to the practice of Lucretius and Catullus but not to the more refined practice of Vergil. The fourth foot weak caesura, while virtually guaranteeing fourth foot coincidence, meant that the word following must have either the shape or in order to guarantee fifth foot coincidence. Apparently, given his desire for a vigorous cadence, it was too high a price to pay for an easy fourth foot coincidence

⁵⁸ Cicero's allowance of the first foot spondee develops from one line in two in the Aratea to one line in four in the Marius. The figures are from Büchner, 1260–1261. See Townend, 120–122, on the difficulty of dating the Marius. Büchner reports (following Guendel) that the range of spondees drops from 2.55 per verse in Aratea to 2.06 in De Cons.

⁵⁹ See the statistics of Ewbank on the favoured patterns for dactyl and spondee in the first four feet, 46–48; Townend, 126, refers to ‘the advances already achieved by Cicero, if not before’.

⁶⁰ I would like to thank my colleague Dr J. Reed for his careful reading of an early version of my argument, the editors of CQ for their thoughtful advice, and the anonymous readers of CQ for invaluable suggestions. One of them is responsible for pointing out the Ovidian allusions.