- 39 -

HHXAMTER PATTERNS IN VERGIL by George E. Duckworth, M.A., Ph.D.

I attempt, in this brief article, to summarize my recent investigations into Vergil's metrical procedures: the patterns he favors and their frequencies, his use of repeated, opposite, and reverse patterns in adjacent lines, the surprising manner in which Aeneid X-XII differ metrically from the Georgies and Aeneid I-IX,1 and the extent to which his favorite pattern (dsss) is affected by the style and subject matter of the Aeneid.2

1. The Hexameter Patterns. If we exclude the spondaic verses (thirty-two in Vergil's three authentic works) and consider only the verses with a dactyl and spondee (ds) in the last two feet, we have for the first four feet sixteen possible schemata, or patterns (the term I prefer to use). Since Latin poetry favors the spondee rather than the dactyl, I list the sixteen patterns as follows, beginning with all spondees in the first four feet, and ending with all dactyls. The second eight patterns are the opposite of the first eight, in reverse order.3 The underlined dactyls and spondees, as they shift in each group, should also help the reader to keep in mind the sixteen possible combinations. 4 spondees, 0 dactyls: 1. s s s s 3 spondees, 1 dactyl : 2. s s s d 3. s s d s 4. s d s s 5. d s s s 2 spondees, 2 dactyls: 6. s s d d 7. s d s d 8. d s s d (opposite patterns) 9. s d d s 10. d _s d s 11. d d £ s 1 spondee, 3 dactyls: 12. _s d d d 13. d _s d d 14. d d s d 15. d d d s 0 spondees, 4 dactyls: 16. d d d d

The order of the patterns as listed above bears no relation to the frequency of their appearance. If these sixteen hexameter patterns were each used to the same extent, the frequency of each would be 6.25 per cent, but this happens neither in Greek nor in Latin poetry. The four most frequent patterns appear from 40 to 60 per cent of the time, and the percentage for t’ne first eight ranges from 65 to 85 (or more). To illustrate from Homer and Ertnius, the earliest of the Greek and Roman hexameter poets, we have the - 40- following percentages: Patterns Homer Ennius Most frequent dddd 19.93 ssss 15.35 First four: 57.99 41.34 First eight: 84.34 65.35 The eight most frequent patterns in Homer contain (in the first four feet) twenty-two dactyls and ten spondees (8x4= 32); Ennius has the exact opposite, twenty-two spondees and ten dactyls. Ennius* most frequent pattern (ssss) is the opposite of Homer's favorite (dddd), and his percent­ age for the eight most frequent patterns is almost twenty per cent lower than the corresponding percentage in Homer. At the beginning of Latin hexameter poetry there was thus less concentration on the same metrical patterns than had been the case in Homer.

2. The Patterns in Vergil. The pattern appearing most frequently in the fragments of Lucilius is sdss (16.86 per cent). It seems to have been Cicero who established dsss as the favorite pattern for the later Republican poets and they used it with a steadily increasing frequency: Cicero (Aratea and fragments), 18.28 per cent; Lucretius, 20.20; Catullus LXIV, 27.59. The concentration on the first four and the first eight patterns is likewise much greater than it had been in Ennius; the percentages are, respectively: Lucilius, 47.60, 74.21; Cicero, 57.39, 82.26; Lucretius, 54.34, 78.81; Catullus, 67.90, 90.98. This lack of variety in Catullus is amazing; his percentage for the four most frequent patterns is higher than the percentage for the first eight in Ennius, and eight of the sixteen possible patterns have in Catullus a total of only 9.02 per cent.

These figures for the Republican poets enable us to see more clearly Vergil’s interest in greater metrical variety; his innovations appear most strikingly in the Eclogues, where we have the following percentages: first pattern (ddssi), 13.09 (lower even than Ennius); first four, 41.45 (cf. Ennius, 41.34); first eight, 69.09 (cf. Ennius, 65.35). The fact that ddss is the favorite pattern in the Eclogues reflects the more dactylic nature of the pastorals as a whole (partly the influence of Theocritus, partly the result of the two dactylic refrains, ddsd and dddd, in Eclogue VIII). The distribution of the spondees and dactyls in the first four feet of the eight most frequent patterns is sixteen and sixteen (cf. Cicero and Catullus: twenty spondees and twelve dactyls; Lucretius: eighteen spondees and fourteen dactyls); in both the Georgies and the Aeneid it is twenty spondees and twelve dactyls.

Eclogue IV is unique among the pastorals; its percentages in relation to the other nine poems are as follows: Patterns IV I- III, v-x Moat frequent: 24.19 12.09 First four: 72.58 39.71 First eight: 91.93 68.28 The first pattern is thus twice as frequent in IV as in the other pastorals, and the percentage of the first four in IV exceeds by over four per cent friat of the first eight in the other poems. The late Professor Jackson Knight stresses the influence of Catullus LXIV on Eclogue IV: 4 The Fourth Eclogue, perhaps the most Catullan of Vergil's certain poems, recalls the hexameters of Catullus in phrases, in the unity of single lines, in schema tiza 1. i on of balanced - 41-

word-order, and in the quality of the verse-groups. When we recall the Catullan percentages as given above (first pattern, 27.59; first four, 67.90; first eight, 90.98), it becomes apparent that Vergil in Eclogue IV deliberately imitates Catullus also in his metrical frequencies.

In the Georgies Vergil develops a metrical uniformity lacking in the Eclogues and establishes norms which, with minor changes, determine his procedure in the Aeneid. The favorite pattern of Cicero, Lucretius, and Catullus, dsss, which was second in the Eclogues, moves into first place in the Georgies (and in each of the four books), and is likewise first in the Aeneid (and in each of the books except V and VII where ddss has a slight lead).

The first eight patterns in the three works and the relevant percentages (first pattern, first four, second four, and first eight), are as follows: Eclogues Georgies Aeneid ddss 13.09 dsss 15.81 dsss 14.39 dsss dsds ddss dsds ddss dsds (dssd 41.45 sdss 48.99 sdss 46.95 (ddsd dssd ssss sdss ddds ddds ddds ssss ssds sdds 27.64 ssds 24.43 sdds 25.83 69.09 73.42 72.78 Tie least frequent pattern in all three poems is sddd. In the first-eight pattern we find a steady increase in the use of a fourth-foot spondee (Eclogues, 6; Georgies, 7; Aeneid, 8). The percentages of the eight most frequent patterns are slightly higher in the Georgies and the Aeneid than in the Eclogues, but still considerably lower than the corresponding percentages in Cicero, Lucretius, and Catullus.

3. Variety and Repetition. Another approach to the variety in Vergil’s use of metrical patterns is a division of each poem into consecutive units of sixteen verses each (disregarding spondaic verses, interpolations, and, for tlie Aeneid, all incomplete verses) and a listing of the number of the sixteen possible patterns which occur in each of these units. With half of the sixteen patterns appearing in over two-thirds, almost three-fourths, of the total verses, one might expect this number to be relatively small, but this is not the case with Vergil. In the Aeneid, for instance, only nine such sixteen-line units have less than seven patterns, whereas eleven units contain as many as thirteen patterns, as, for example, I 189-204, with no repetitions until we reach 202, the fourteenth line in the unit. The percentage of units containing eight to eleven patterns remains remarkably constant throughout Vergil's three works: Eclogues, 87.23; Georgies, 87.41; Aeneid, 86.89. The numerical average of the different patterns per sixteen-line unit is likewise amazingly consistent: Eclogues, 9.7; Georgies, 9.3; Aeneid, 9.4. With these we may compare Cicero (Aratea), 8.5; Lucretius, 8.6; and Catullus LXIV, 7.0.

The same pattern may occur in two, three,or more lines in succession; these I term repeats . Only twice does Vergil use the same pattern five times in adjacent verses, and both occurrences are in his earlier poetry (Eel. X 36-40, Georg. I 46-50). Elsewhere he allows himself a maximum of - 42 - four repeated patterns and these very rarely - four times in the Georgies and only seven in the Aeneid. Often the same pattern may occur frequently but separated by one or two verses; I call these repetitions "near repeats" and when they appear, as often happens, in the vicinity of the repeats, they produce repeat clusters", which I define as passages in which six or more instances of the same pattern are found in sixteen or fewer lines, e.g. Aen. V 632-647, where dsss appears nine times. Repeat clusters occur ϊ ί Vergil as follows: Eclogues, one every 275 lines; Georgies, one every 145.5 lines; Aeneid, one every 200.1 lines; this is very different from the practice of the Republican poets: Cicero (Aratea), one every 59.6 lines; Lucretius, one every 49.2 lines; and Catullus LXIV, one every 29.0 lines. Again Vergil's avoidance of repetition and his desire for greater metrical variety are readily apparent.

The frequency of the repeats in Vergil is as follows: Eclogues, one every 13.1 lines; Georgies, one every 12.3 lines; Aeneid, one every 12.4 lines. Of the earlier poets Cicero is closest to Vergil in this respect, with one repeat every 11.4 lines, but cf. Lucretius, one every 8.8 lines, and Catullus, one every 7.0 lines. If we include the near repeats (as defined above), we have the following frequency every x lines: Eclogues, 5.1; Georgies, 4.5; Aeneid, 4.6; we find a greater use of repeats and near repeats in the Republican poets, where the corresponding totals are: Cicero, 3.9; Lucretius, 3.6; Catullus, 3.0.

4. Repeated Patterns and Fourth-foot Texture. One effective means of avoiding monotony in repeated patterns is to vary the fourth-foot texture, t.iat is, to shift either from heterodyne (clash of word accent and metrical ictus) to homodyne (coincidence of accent and ictus), or from hornodyne to heterodyne. The importance of fourth-foot texture has been stressed by Jackson Knight, who points out the different effects achieved by heterodyne (resistance, difficulty, effort, weariness, pathos) and homodyne (rapidity, smoothness, freedom from restraint and tension) and shows that Vergil, unlike most Latin hexameter poets (both earlier and later), reduced the percentage of fourth-foot homodyne from fifty or sixty to less than forty.® In order to discover the extent to which change in fourth-foot texture produces variety in repeats and near repeats I have rechecked Vergil's hexameters for homodyne (m) and heterodyne (t); my percentages are slightly higher than those of Jackson Knight, as follows: Eclogues, 39.73; Georgies, 36.08; Aeneid, 37.78.

For an instance of shift from heterodyne to homodyne, see Aen.IV 68-69 (dsss); from homodyne to heterodyne, Aen.IV 161-162 (ddss); when the same imstrical pattern occurs in three verses in succession, we often find a doable shift; that is, one kind of texture is enclosed by the other, e.g. Aen.IV 75-77 (dsds; t, m, t) and 584-586 (sdss; m, t, m); the second illustration is the more unusual as the ratio of homodyne to heterodyne is two to one, almost the reverse of the normal percentage in the Aeneid (37.78 to 62.22). In the repeat cluster in Aen.Ill 709-718, where dsss occurs seven times in ten lines, the ratio of homodyne to heterodyne is four to three, and the change from one texture to the other takes place five times out of a possible six. In such a passage the shift in fourth-foot texture does much to relieve the monotony which would other­ wise result from the excessive repetition of dsss.

The percentages of the change in fourth-foot texture consistently run - 43 - higher than the percentages of absolute fourth-foot homodyne and this would seem to indicate that Vergil deliberately uses change in fourth-foot texture to provide variety and to counteract the monotonous effect resulting from the repetition of the same metrical patterns. The percentage of change in the repeats and the increase over the homodyne percentage are as follows: Eclogues, 49.21, + 9.48; Georgies, 43.50 + 7.42; Aeneid, 44.49 + 6.71. 6

5. Opposite and Reverse Patterns. Of the sixteen metrical patterns, eight are the opposites of the other eight; sddd is the opposite of dsss, sdds is the opposite of dssd, etc. In addition to the eight opposites, there are four patterns which I term "reverses , those in which the sequence of dactyls and spondees appears in reverse order, i.e. dsss and sssd, sdss and ssds, ddds and sddd, ddsd and dsdd. 7

If we list the patterns in the order of their frequency in the Aeneid, we immediately see another aspect of Vergil's metrical practice: as a general rule, the more frequent a pattern, the less frequent its opposite: dsss (1) sddd (16) ddss (2) ssdd (14) dsds (3) sdsd (U) sdss (4) dsdd (12) ssss (5) dddd (15) ddds (6) sssd (13) ssds (7) ddsd (10) sdds (8) dssd ( 9) Even more interesting is the fact that opposite patterns appear so frequently in adjacent lines (e.g. Aen.IV 8-9, sddd and dsss; 13-14, dddd and ssss) and often a pattern is both preceded and followed by its opposite (e.g. Aen.IV 51-53, dsdd framed by sdss; 697-699, sssd framed by ddds; in III 393-395, ssss the more frequent pattern, is enclosed by dddd.)

Opposites appear in adjacent lines with considerable frequency: Eclogues, one every 19.6 lines; Georgies, one every 20.9 lines; Aeneid, one every 23.1 l i n e s . 8 The opposite occurring most often in each of Vergil's three works is sdsd-dsds; its percentage of the total opposites is as follows: Eclogues, 19.05; Georgies, 23.58; Aeneid, 16.04. In the Aeneid sdsd is preceded or fallowed by dsds 18.28 per cent of its occurrences; the percentage of sddd preceded or followed by dsss is considerably higher, 29.38.

Not only do opposites attract each other to a surprising degree (especially sddd and dsss), but the same is true of the four reverse combin­ ations; these also appear frequently in adjacent lines (e.g. Aen.IV 17-18, sdss and ssds; 89-90, sssd and dsss) and sometimes one pattern is framed by its reverse (e.g. Aen.II 385-397, sssd enclosed by dsss; V 382-384, where the less frequent sssd frames dsss).

Vergil's three works differ in regard to the most frequent reverse and its percentage of the total reverses; Eclogues: dsdd-ddsd, 46.67; Georgies, sssd-dsss, 38.30; Aeneid, ssds-sdss, 40.08. In the Aeneid ssds is preceded or followed by sdss 17.21 per cent of its occurrences; the corresponding percentage for sssd in relation to dsss is 29.56. Contrary to the relatively constant frequency of opposites in the three works, the adjacent reverse patterns reveal a steady increase: Eclogues, one every 55.0 lines; Georgies, ons every 46.4 lines; Aeneid, one every 38.9 lines. ® - 44-

6. Aeneid X-XII - Unrevised? My examination of the metrical patterns in the %eneid has had a surprising and unexpected result. In instance after instance my findings for Books X-XII differ from those for the other nine books; in other words; Vergil in these three books seems to depart from his normal procedure elsewhere in the Aeneid and likewise in the Georgies. I list briefly these discrepancies and abnormalities as follows:

1. The frequency of dsss in X-XII is 15.34 per cent, as against 14.02 per cent in I-IX. The fact that X and XII (with thehighest percent­ ages, 15.90 and 16.30 respectively) contain so many battle scenes may account in part for the high frequency of dsss, since this pattern is low in emotional and dramatic scenes and high in descrip­ tive passages, those written in a more objective style (see below, Section 7).

2. There are only nine sixteen-line units in the Aeneid containing less than seven metrical patterns; five of these (55.56 per cent) are in X-XII.

3. Vergil rarely uses the same metrical pattern four times in succession; there are seven instances in the Aeneid, but three of these triple repetitions (42.86 per cent) occur in X-XII.

4. The Aeneid contains forty-nine repeat clusters, an average of one every 200.1 lines. But twenty-three of these clusters (46.94 per cent) appear in X-XII; the average is thus one every 271.0 lines in I-IX; one every 119.9 in X-XII, more than twice as often.

5. Repeats are likewise more numerous in X-XII, an average of one every 11.4 lines; in I-IX the average is one every 12.9 lines.

6. The percentage of repeats in the first four patterns, in relation to the total repeats, varies: I-IX, 66.42; X-XII, 72.61; the variation is greatest in the percentages for dsss and dsds repeats; dsss: I-IX, 20.99; X-XII, 24.90; dsds: I-IX, 17.15; X-XII, 22.41.

7. The same two patterns show in X-XII a higher percentage of repeats in relation to the total occurrences of the two patterns; dsss: I-IX, 11.64; X-XII, 14.18; dsds: I-IX, 12.40; X-XII, 16.12.

8. Opposite patterns appear in adjacent lines with the same approximate frequency, until we reach Aeneid X-XII: Eclogues, one every 19.6 lines; Georgies, one every 20.9 lines; Aeneid,I-IX, one every 21.8 lines; but X-XII, one every 27.6 lines.

9. There is considerable variation in the percentages of the different opposites; sddd is preceded or followed by dsss 28.47 per cent of its occurrences in I-IX, but 32.0 per cent in X-XII; this is perhaps not surprising because the frequency of dsss is highest in X and XII. In the case of ssdd preceded or followed by ddss, the percentages change strikingly in the opposite direction: I-IX, 22.78; X-XII, 13.89; this is true to a lesser degree of sssd and ddds: I-IX, 16.52; X-XII, 12.50.

10. In the case of reverse patterns in adjacent lines, sssd is preceded - 45 -

or followed by dsss 30.36 per cent of the time in I-IX but only 26.39 per cent in X-XII. This change is just the opposite of what the high frequency of dsss in X and XII would lead us to expect. The percentages for sddd preceded or followed by ddds show a greater variation: 16.67 in I-IX; 10.0 in X-XII.

1.1. Perhaps the most significant statistical abnormality appears in the shift from fourth-foot heterodyne to homodyne (or the reverse) in repeated patterns. The percentage of repeats containing such a change in fourth-foot texture runs consistently higher in Vergil's poetry than the normal fourth-foot homodyne percentages, until we come to Aeneid X-XII: Eclogues, 49.21, + 9.48; Georgies, 43.50, + 7.42; Aeneid I-IX, 48.72, + 10.50, but X-XII, 34.85, - 1.79. The greatest discrepancy appears in Book XI, 33.33, - 5.22. This sudden change in the treatment of fourth-foot texture in repeated passages in X-XII is most striking and can hardly be attributed to coincidence.

12. The two patterns where the percentage of change in fourth-foot texture is so very different are ddss: I-IX, 56.52; X-XII, 27.27; and dsds: I-IX, 39.36; X-XII, 14.81; in each case the X-XII percentage is less than half.

Many of the statistical variations listed above might individually seem inconclusive; but their cumulative effect is, in my opinion, overwhelming and indicates without question that, in his choice and treatment of metrical patterns, in the relative frequency of repeated, opposite, and reverse patterns, and especially in his handling of fourth-foot texture in repeated patterns, Vergil displays in Aeneid X-XII a metrical technique very unlike his normal procedure.

How are these many metrical peculiarities and abnormalities to be explained? Since Vergil composed the Aeneid particulatim..., et nihil in ordinem arripiens (Donatus-Suetonius Life, 23) we cannot assume either that X-XII were written first and that Vergil then changed his metrical technique (the many close similarities between the Georgies and Aeneid I-IX would also argue against this possibility) or that they were written last and the poet, after relatively uniform procedure in the Georgies and Aeneid I-IX, suddenly changed his treatment of metrical patterns in so many respects. The most satisfactory explanation seems to be this: Vergil had made metrical revisions through Book IX in an effort to achieve greater variety; X-XII contain so many curious and unusual metrical features simply because his sudden death in 19 B.C. lsft these three books unrevised.

7. DSSS and Vergil's Subjective Style. I return now to dsss, Vergil's favorite metrical pattern, second in the Eclogues (10.67 per cent), first in the Georgies (15.81 per cent) and in each of the four books (with a percentage range from 14.16 in III to 17.38 in II), and first in the Aeneid (14.39 per cent) and in all books except V and VII (with a percentage range from 12.40 in V to 16.30 in XII). When I examined Vergil's treatment of dsss in the Aeneid, I discovered to my surprise that there are numerous episodes and speeches where he uses dsss very sparingly, from three to five per cent; in other passages the percentages rise to twenty-five or thirty. Why do we have this striking variation?

The explanation is to be found in Vergil's "subjective style", which Otis describes as the manner in which the poet shares the emotions of his characters (empathy) and presents his own personal reaction to their emotions - 46 -

(sympathy). This "empathetic-sympathetic style" is new to epic poetry and differs strikingly from the more objective, narrative style of Homer and Apollonius; Otis says: "Virgil not only reads the minds of his characters; he constantly communicates to us his own reactions to them and to their behaviour." ^ I now find an amazing correlation of subject matter and meter everywhere in the epic; the metrical patterns occurring most frequently in the more emotional, subjective speeches and episodes are not the same as those in the more narrative and objective passages. Perhaps the sequence of three spondees, appearing elsewhere only in ssss and the less frequent sssd (the reverse of dsss) provides a sense of dignity and majesty more suitable for certain themes than for others.

I first noticed this correlation of meter and subjective style in the second half of Georgies IV, where we have a story within a story; the tragic tale of Orpheus and Eurydice framed by the story of Aristaeus. Otis says: "the style of the Orpheus is utterly subjective in contrast to the objectivity of Homeric epic or the Aristaeus."11 The pattern dsss is first in the Aristaeus (14.88 per cent) as in the book as a whole (15.63 per cent), but in the Orpheus and Eurydice story dsss drops sharply to 5.41 per cent and is tied for sixth place with three other patterns (ddds, sdsd, and dddd.

This indicates that dsss is low where we have subjective style, in this case in the emotional and dramatic story of Orpheus and E u r y d i c e ; 12 on the other hand, for the more objective style of the Aristaeus, the percentage of dsss is no higher than normal. Under what circumstances does Vergil increase dsss to twenty-five or thirty per cent? To answer this question wa turn now to the Aeneid.

To illustrate the variation of dsss in the epic, I shall give the percentages of the pattern in passages selected chiefly from Books I and IV.

In I, where the dsss percentage is 13.88, we find the following: la-80 Prologue. Juno and Aeolus 17.86 81-156 The Storm and Neptune 6. 58 157-222 Aeneas and the Tr0jans in Africa 7.58 223-296 Venus and Jupiter 17.57 418-519 Carthage and Dido 19.42 418-452 Aeneas sees Carthage 5.71 453-493 Scenes of the Trojan War 29.27 520-656 The Trojans welcomed 12.78 520-560 Speech of Ilioneus 7.69 At the beginning of the book, with its emphasis on Rome, Carthage, and Juio, dsss is well above average; it is low during the storm and the arrival in Africa (and Aeneas' laments), and again high in the Venus-Jupiter scene, which stresses the future greatness of Rome and Augustus. Aeneas' emotions at the sight of Carthage produce a low dsss percentage, but it rises immediately to 29.27 when the scenes of the Trojan War are described. When Ilioneus makes his appeal to Dido and describes the hardships endured by the Trojans, dsss is again low.

For IV (dsss percentage, 13.47) I list the following fluctuations: - 47-

90τ128 Juno-Venus scene 20.51 296-330 Speech of Dido 11.43)

331-361 Speech of Aeneas 3 . 33 ) 8.33 362-392 Speech of Dido 9.68) Framing narrative (279-295, 393-415) 23.08 584-629 Dido's curses 13.04 BUT 584-606 (personal and emotional) 4.35 607-629 (Aeneas, Rome, Carthage) 21.74 630-705 Suicide and Anna's lament 6.58 In the Juno-Venus scene, the dsss percentage is high, as seems usual in passages concerning divinities. The highly emotional speeches of Dido and Aeneas in the central portion of the book differ to a striking degree in their use of dsss from the narrative framework (8.33 per cent as against 23.08); also, dsss occurs only once in Aeneas' central speech (331-361), a percentage of 3,33; perhaps this indicates the depth of his emotion. Another surprising variation appears within Dido's curse; in the first part, more personal and emotional, the dsss percentage is 4.35, but in the second half, which foreshadows Aeneas' later fate and the undying hostility between Rome and Carthage, the percentage rises sharply to 21.74. In the final scene (Dido's suicide, Anna's lament, and the release of Dido'sspirit), dsss falls to half its normal frequency. These fluctuations in Book IV seem especially significant and reveal Vergil's avoidance of dsss in the more dramatic and emotional passages.

To summarize, the frequency of dsss is low in episodes and speeches which are dramatic, psychological, and emotional, where the style is subjective or empathetic-sympathetic , and especially in scenes of death and laments for the dead; cf. also the following percentages: II 730-804 (departure and loss of Creusa), 9.59; V 827-871 (death of Palinurus) 8.89; X 479-509 (death of Pallas), 3.33. The variation is often sudden and striking; cf. IX 367-421 (attempt of Nisus and Euryalus to escape), 29.09, but zero per cent in 422-449 (deaths of the youths); XI 29-58 (Aeneas' speech over the body of Pallas), 3.45, but 24.39 in 59-99 (preparations for the trip to Pallanteum). The abnormally high percentages appear regul­ arly in certain types of episodes: those in a more objective style, especially scenes of mass fighting and individual combat; those in which divinities (Jupiter, Juno, Venus, etc.) appear and speak; those dealing with Rome and Augustuses see also II 559-633 (Aeneas-Venus episode), 20.55; III 274-293 (Actium), 20.0; V 604-663 (Iris and the Trojan women), 27.12; VI 738-807 (Augustus), 20.0; VIII 225-261 (the Hercules-Cacus fight), 21.62; 337-369 (the walk through Rome), 25.48; 675-728 (Augustus), 24.53; X 256- 361 (landing and battle), 21.15; 510-605 (renewal of fighting after death of Pallas), 22,11; XII 697-790 (first encounter of Aeneas and Turnus), 22.34; 791-842 (Jupiter and Juno, and the greatness of Rome), 19.23.

Vergil's metrical patterns, at least in the case of dsss, are thus fitted to his subject matter and to his style. As in individual lines, where the words do not determine the meter but are subordinated to the metrical patterns which the poet so often adapts to the sense, so in his larger units the patterns are not determined by the words. Subject matter, style,and meter go hand in hand. The words are written to the music, not the music to the words. Jackson Knight said some years ago: 16

Rhythm and metre are just as likely to suggest words as words are likely to enforce a certain metrical or rhythmical - 48-

form. There is no doubt of this. Vergil's imagination to a large extent worked by ear; that is, what he said often depended at least partly on the sounds, and not merely on the thoughts, which he remembered.

NOTES. 1. See G.E. Duckworth, "Variety and Repetition in Vergil's Hexameters", ΤΑΡΑ 95 (1964) 9-65; this article contains necessary definitions, illustrative material, and statistical tables.

2. See G.E. Duckworth, "Vergil's Subjective Style and its Relation to Meter", to appear in Vergilius, No. 12 (1966). 3. For instance, sdds is the opposite of dssd, dsds is the opposite of sdsd, ddss is the opposite of ssdd, etc. 4. W.F. Jackson Knight, Roman Vergil (rev. ed, Harmondsworth 1966) 330; see also L.P. Wilkinson, Golden Latin Artistry (Cambridge 1963) 194; R.E.H. Westendorp Boerma, Vergil's Debt to Catullus", Acta Classica 1 (1958) 55, and bibliography there cited. 5. W.F. Jackson Knight, Accentual Symmetry in Vergil (Oxford 1939) esp. 36r43; his percentages for fourth-foot homodyne are: Eclogues, 37.27; Georgies, 33.45, Aeneid, 35.95. 6. If we include the near repeats with the repeats, the percentage of change and the increase over the homodyne percentage are: Eclogues, 44.10, + 4.37; Georgies, 47.10, + 11.02; Aeneid, 45.83, + 8.05. 7. The patterns ssss, dddd, sdds, and dssd have no reverse; in the case of ssdd, ddss. sdsd, and dsds, the reverse is the same as the opposite. 8. This is a form of variety in which the Republican poets were less interested; cf. Cicero (Aratea), one every 79.8 lines; Lucretius, one every 30.8 lines; Catullus LXIV, one every 37.7 lines (surprisingly frequent, however, when we consider that the percentage for the first eight patterns is 90.98 and opposite patterns are therefore very few.) 9. The corresponding figures for the earlier poets are as follows: Cicero (Aratea), 24.0 (i.e., more than three times as frequent as opposites); Lucretius, 51.3; Catullus LXIV, 53.9. In Horace's hexameter poetry reverses appear on an average once every 29.4 lines, considerably more often than in Vergil. 10. B. Otis, Virgil. A Study in Civilized Poetry (Oxford 1963) 88. 11. Op.cit., 200. 12. The Aristaeus and Orpheus stories have often been compared to the Peleus-Thetis and Ariadne stories in Catullus LXIV. There is no such metrical change in Catullus; the most frequent pattern is dsss, with a high percentage of 27.59; in the Peleus-Thetis section the percentage is 24.74, and in the central Ariadne story it rises to 30.48. This is very different from Vergil's procedure in Georgies IV, where the Orpheus differs from the Aristaeus not only stylistically bu't also metrically, as a result of the suppression of dsss. 13. For an analysis of selected passages in the other ten books, see my - 49-

forthcoming article in Vergilius (above, note 2). My divisions and subdivisions of the books are those which appear in G.E. Duckworth, "Tripartite Structure in the Aeneid" , Vergilius, No. 7 (1961) 2-11; Structural Patterns and Proportions in Vergil's Aeneid (Ann Arbor 1962) 25-33. 14. It is significant also that Aeneas' speech is at the very center of Aeneid IV; see G. E. Duckworth, "Mathematical Symmetry in Vergil's Aeneid" , ΤΑΡΑ 91 (1960) 189; Structural Patterns and Proportions, 24. 15. Cf. in the Georgies (average percentage of dsss, 15.81): I 5b-42 (the gods and Octavian), 27.27; II 136-176 (praise of Italy), 26.83; III 1-48 (a temple of song for Octavian), 20.83. 15. Accentual Symmetry in Vergil, 37.