4. The Metrical Context of κλέοc ἄφθιτον in Epic and Lyric

There is a serious difficulty with the theory that epic formulas are derived from lyric formulas. The attestations of Greek lyric verses, let alone Pherecratic verses specifically, are scant. Furthermore, what little survives is late in comparison with the Iliad and Odyssey. Whereas the Homeric corpus became a fixed text in a prehistoric period, lyric poets like Ibykos or Sappho are historical figures. Moreover, I acknowledge that the surviving fragments of Greek Lyric reveal a profound influence exerted by those Panhellenic marvels, the Iliad and Odyssey. [1] Nevertheless, I wish to argue that the traditional phraseology of lyric versification was originally independent of epic versification.

It is tempting to assume automatically that phraseological parallelisms in Epic and Lyric involve simple borrowings from the first into the second. For an example, let us consider the following two verses:

The lyric verse here functions as the fourth {103|104} member of the epode within the overall hymnic structure of Ibykos 282P. This fourth member, as attested at lines 8, 21, 34, 47 of the poem, is regularly a pher^2d:

Contrast my reconstruction of epic hexameter, pher^3d: As for the formula Τελαμώνιοc ἄλκιμοc Αἴαc, shared in the same position by the epic and lyric verses, I have already described its shape as pher^d. [2] The question now is, did the lyric verse inherit the formula in this position, or do we see here a mere borrowing from epic verse?

We may ask the same sort of question at line 47 of Ibykos 282P, the same lyric poem. This verse too is a pher^2d: Compare the following verses in Epic: {104|105} In terms of the reconstruction which I propose for epic hexameter, the expressions are formulas shaped ^pher^d and pher^d respectively. Notice that the epithet ἄφθιτον is shaped like a dactylic expansion, occurring between ⏑ ⏑ and – ⏓ of a partial pher segment ⏑ ⏑ – ⏓. For attestations of this segment within the larger framework of (^)pher^d, consider the following: As with other formulaic segments shaped ⏑ ⏑ – ⏓, [4] the type κλέοc + ἐcτίν ́εἴή, etc. may occur not only at verse-final position but also before the trochaic caesura: From the internal standpoint of Epic, it seems that the type κλέοc + ἐcτίν/ἔcται/εἴή/εἶναι, shaped ⏑ ⏑ – ⏓, is more basic than the expanded type κλέοc + ἄφθιτον + ἐcτίν/εἴη, shaped ⏑ ⏑ – ⏑ ⏑ – ⏓. Also, the expanded type is far less common. If, then, Ibykos is merely lifting expressions out of Epic in order to insert them into the ending of his pher^2d meter, it seems puzzling that he should choose the less common pattern κλέοc + ἄφθιτον + ἑξεῖc, when κλέοc + ἑξεῖc could be inserted just as easily into the ending. Of course, since the κλέοc + ἄφθιτον + ἔcται of Iliad Ι 413 refers to Achilles himself, we could still counter that Ibykos was inspired by this one particular Homeric verse to coin his κλέοc + ἄφθιτον + ἑξεῖc. Such a line of argumentation is much less tenable, however, in the instance of Τελαμώνιοc ἄλκιμοc Αἴαc. The Homeric corpus features this expression in only two verses (Μ 349 = 362), both seemingly insignificant from the non-formulaic point of view, while the unexpanded equivalent, Τελαμώνιοc Αἴαc, is attested no less than 21 times in verse-final position. Are we to believe that Ibykos has here eschewed a routine epic expression and opted for a highly isolated but (for Ibykos) {106|107} insignificant variant? Even if it were so, we run into a further difficulty in his use of the word μέγαc: In the Homeric corpus, we have twelve attestations of besides the two of but nothing like which on the surface looks like a conflation of the previous two formulas. If we choose to insist that Ibykos borrowed here from Homer, we are forced to assume that he picked an extremely rare formula only to conflate it with another formula. Given the stock meanings of μέγαc and ἄλκιμοc, it hardly seems worth going to all the trouble—at least, from a post-Homeric standpoint.

I find it easier to think of Τελαμώνιοc ἄλκιμοc Αἴαc as an inherited pher^d formula which survives in the pher^2d meter of Ibykos and in the pher^3d meter which evolved into the epic hexameter. Similarly, κλέοc ἄφθιτον ἔcται/ἑξεῖc/etc. would be the component of a (^)pher^d formula. To repeat, the equivalent of Τελαμώνιοc ἄλκιμοc Αἴαc in a plain pher formula would be Τελαμώνιοc Αἴαc. Likewise, κλέοc ἔcται/ ἑξεῖc/etc. {107|108} would be the component of a plain pher formula. For a formally parallel component, compare κλέοc ἐcθλόν, which behaves in its placement like other formulas shaped ⏑ ⏑ – ⏓. [5] Consider the Homeric attestations in the following positions: In the last instance, there is a variant κλέοc εὐρύ, available for situations where the word following our expression starts with a consonant rather than vowel:

If indeed κλέοc ἄφθιτον ἔcται/εἴη/etc. is a component of a (^)pher^d formula while κλέοc ἔcται/εἴη/etc. is a component of a plain pher formula, what is the origin of ἄφθιτον, the word which reflects the dactylic expansion from pher to (^)pher^d? We need not assume that the combination of κλέοc + ἄφθιτον was necessarily caused by the factor of dactylic expansion. Rather, this factor may have simply promoted the preservation of an epithet already inherited by κλέοc but suppressed in simple pher meter. The point is, the ending of a simple pher could not accommodate κλέοc ἄφθιτον. On the other hand, {108|109} the expansion of κλέοc ἔcται to κλέοc – ⏑ ⏑ ἔcται affords an ideal opportunity for the inclusion of ἄφθιτον. In other words, it may be that the combination κλέοc ἄφθιτον is latent within the framework of a pher but overt within the framework of a (^)pher^d.

Actually, we do have proof that κλέοc ἄφθιτον is an inherited expression of the remotest antiquity. As Kuhn noticed in 1853, [6] Greek κλέοc ἄφθιτον is a direct cognate of Indic śráva(s) ákṣitam, as attested in Rig-Veda 1.9.7. Both expressions can be reconstructed to an identical prototype, *klewos n̥dhg ^w hitom. [7] The cognate collocation of ἄφθιτο-/ákṣita- with κλέοc/śrávas is invaluable for my argument, since it constitutes comparative evidence that ἄφθιτον was an epithet of κλέοc since Indo-European times. In view of its archaism, we should search for the pristine metrical position of this expression κλέοc ἄφθιτον. I propose that the type κλέοc ἄφθιτον ἔcται/εἴη/etc. will not reveal the original position, since it is built on the type κλέοc ἔcται/εἴη/etc. In order to find the basic metrical context of κλέοc ἄφθιτον, we should first examine the metrical context of its cognate, śráva(s) ákṣitam. If śráva(s) ákṣitam is positioned in an Indic meter which has a Greek cognate, then we should look for parallel positioning of κλέοc ἄφθιτον in such a Greek meter. {109|110}

The expression śráva(s) ákṣitam, as attested in Rig-Veda 1.9.7, is contained by the syllables 5 6 and 6 7 8 of two successive octosyllables (Gāyatrī): The full text reads as follows: The aim of Chapter 9 is to show in detail why śrávas here is separated from ákṣitam, which occupies the last three syllables of the verse-closing. If śrávas had directly preceded ákṣitam just as κλέοc precedes ἄφθιτον, the resulting pattern would have included a double short sequence, which is obsolescent in the Rig-Veda. [8] In line with the metrical tendency to avoid ⏑ ⏑, there is a phraseological displacement of śráva(s) ákṣitam by ákṣiti śrávas, likewise meaning ‘imperishable fame’ but containing a different and preferable rhythm: {110|111}

In support of the proposition that the śrávas of śráva(s) ákṣitam was displaced from syllables 4 5 of octosyllables, Chapter 9 shows that the word śrávas survives nowhere in syllables 4 5 of Rig-Vedic octosyllables. In fact, among all the Rig-Vedic occurrences of śrávas, its metrical shape is never ⏑ ⏑ in the closing generally, let alone in syllables 4 5 of octosyllables specifically. The second situation is precisely the pattern which I posit in the reconstruction In the verse-closings of the Rig-Veda, śrávas always scans ⏑ – rather than ⏑ ⏑, because its position there is always preconsonantal (śrá̆vās C-), never prevocalic (śrá̆văs V- ). Thus the overall positional behavior of śrávas agrees with two Rig-Vedic metrical tendencies: (1) elimination of ⏑ ⏑ from the verse-closing, even where half of the ⏑ ⏑ was in the verse-opening; and (2) generalization of an iambic rhythm in the verse-closing, ⏑ – ⏑ ⏓. For illustration, I provide here a schema showing the patterns {111|112} of occurrence and nonoccurrence in the closing of Rig-Vedic octosyllables: (Note that when śrávas is in 4 5, half of it is in the closing and the other half in the opening.) In compliance with these tendencies, Rig-Veda 1.9.7 accommodates the whole expression śráva(s) ákṣitam within two metrically regular octosyllables in preference to one metrically irregular octosyllable:

In contrast to the obliteration of śrávas from syllables 4 5 of octosyllables, consider the retention of śrávas in syllables 5 6 or 6 7 of hendecasyllables: I reserve the details for Part II, but the {112|113} essentials must be pointed out here. In line with the Rig–Vedic metrical tendency to generalize ⏑ ⏑ in syllables 5 6 or 6 7 of hendecasyllables, [10] śrávas in these positions is regularly prevocalic and thus scans ⏑ ⏑.

In Part II, my reconstruction of an original śrá̆văs (V-) in syllables 4 5 of octosyllables is based primarily on the declined forms śrávasas, śrávase, and śrávasi, all attested in syllables 4 5 6 of octosyllables. The rhythm ⏑ ⏑ of the base śrávas- in these forms cannot be displaced by ⏑ – as in simple śrávas, where repositioning into odd-even syllables (5 6 or 7 8) and beginning the next word with a consonant automatically resulted in the scansion ⏑ –. Such trisyllabic declined forms of śrávas- simply cannot comply with both Rig-Vedic metrical tendencies of (1) eliminating ⏑ ⏑ from the closing, even where half of the ⏑ ⏑ was in the opening; (2) generalizing an iambic rhythm in the closing, ⏑ – ⏑ ⏓. At best, these trisyllabic forms avoid interference with the prevalent iambic closing by occurring only in syllables 4 5 6 of octosyllables, not 5 6 7 or 6 7 8: On the other hand, these trisyllabic forms resist the metrical tendency removing ⏑ ⏑ in 4 5, and in this respect their positional behavior is more archaic than that of disyllabic śrávas. For example, the genitive singular of máhi śrávas ‘great fame’ remains śrávasa(s) mahás, with ⏑ ⏑ in syllables 4 5 of octosyllables: vs. We see here the survival of a positional switch within the framework of an actual declension. The whole expression máhi śrávas is being declined, and the position of śrávas- switches back and forth from syllables 7 8 to 4 5 in the process. I argue a parallel switch of śrávas in vs.

Paradoxically, when it comes to studying traditional phraseology embedded in various metrical rhythms, the multiple new rigid meters of Greek Lyric may be more valuable than any single corresponding Rig-Vedic meter, even though the latter be older and more flexible. A flexible meter is by definition less resistent to new {114|115} rhythmical tendencies than rigid meters. [11] For an illustration, consider the Rig-Vedic metrical tendency towards generalizing ⏑ – ⏑ ⏓ in the closing of the Gāyatrī octosyllable. In line with this tendency, traditional phrases with the shape … – ⏑ ⏑ ⏓ are likely to be leveled out by other phrases with the shape … ⏑ – ⏑ ⏓. On the other hand, Greek Lyric has at least the potential of preserving traditional phrases with the shape … – ⏑ ⏑ ⏓ as well as those with the shape … ⏑ – ⏑ ⏓, since both may remain securely embedded in two respective separate meters: For another illustration which happens to be more immediate, consider again the Rig-Vedic tendency toward generalizing long over short in syllable 4 of the Gāyatrī octosyllable. In line with this tendency, I have just argued that traditional phrases like śráva(s) ákṣitam, shaped … ⏑ ⏑ – ⏑ ⏓, are likely to be leveled out by other phrases shaped … – ⏑ – ⏑ ⏓, like ákṣiti śrávas.

Let us now review the contrast between the relatively flexible Gāyatrī octosyllable in Indic and its multiple rigid equivalents in Greek Lyric: Notice that κλέοc ἄφθιτον would fit perfectly at the end of a Glyconic. If indeed the Greek Glyconic is related to the Indic Gāyatrī octosyllable, then we might expect the positioning of κλέοc ἄφθιτον at the end of a Glyconic, in view of the potential positioning of śráva(s) ákṣitam at the end of a Gāyatrī octosyllable. To put it another way, we might expect κλέοc ἄφθιτον and śráva(s) ákṣitam to be cognate not only in form but also in metrical context.

The task now is to find an instance of κλέοc ἄφθιτον at the end of a Glyconic: Failing that, the next best thing would be to find κλέοc ἄφθιτον at the end of an internally expanded Glyconic, The latter type is not only a derivative of the plain Glyconic (gl), but also a functional variant. For example, the type gl^d actually alternates with plain gl in the stanzaic structure of Sappho 94LP, the poem well known for the line which follows the pattern gl gl gl^d, as at lines 6 7 8: {116|117} By good fortune, κλέοc ἄφθιτον is actually attested in the closing of an internally expanded Glyconic The expression occurs in a poem consisting of gl^2d verses, Sappho 44LP (The Wedding of Hektor and Andromache). Line 4 reads:

I fully expect an important objection: that Sappho did not inherit this particular expression in this particular metrical position. The reasoning would be that she simply borrowed it from Homeric diction. Although it is hard to deny that the meter of Sappho is far more archaic than the Homeric hexameter, [12] many will make precisely this assumption, that even so the traditional diction contained in such an archaic meter results from mere borrowing. The chapter must end on this moot point as we proceed to look at Sappho’s poem more closely.